Socially Assistive Robots in Mental Healthcare: Principles and Conceptual Framework for User-Centered Design (original) (raw)
1 Introduction
Digital mental healthcare is a rapidly evolving field, although there are many different outlooks and possibilities for its future development. Key delivery channels of digital mental healthcare include chatbots, virtual/augmented reality, health monitoring with wearable devices, and comprehensive data management systems for patient health information [[1](/article/10.1007/s12369-025-01323-5#ref-CR1 "Kasoju N, Remya NS, Sasi R, Sujesh S, Soman B, Kesavadas C et al. (2023) Digital health: trends, opportunities and challenges in medical devices, pharma and bio-technology. CSI Trans ICT 11(1):11–30. https://doi.org/10.1007/s40012-023-00380-3
"), [2](/article/10.1007/s12369-025-01323-5#ref-CR2 "Torous J, Bucci S, Bell IH, Kessing LV, Faurholt-Jepsen M, Whelan P et al. (2021) The growing field of digital psychiatry: current evidence and the future of apps, social media, chatbots, and virtual reality. World J. Psychiatry 20(3):318–335.
https://doi.org/10.1002/wps.20883
")\]. Among these solutions, socially assistive robots (SARs) warrant further exploration in digital mental healthcare. SARs refer to robots that provide healthcare assistance through social interaction with users. Although definitions vary slightly across studies, SARs need to fulfill the following two core characteristics: (i) they have to provide assistance to users, and (ii) they have to engage in social interactions to provide that assistance \[[3](/article/10.1007/s12369-025-01323-5#ref-CR3 "Fardeau E, Senghor AS, Racine E (2023) The impact of socially assistive robots on human flourishing in the context of dementia: a scoping review. Int J Soc robot 15(6):1025–1075.
https://doi.org/10.1007/s12369-023-00980-8
"), [4](/article/10.1007/s12369-025-01323-5#ref-CR4 "Feil-Seifer D, Mataric MJ (2005) Defining socially assistive robotics. 9th Int Conf Rehabil robot 465–468.
https://doi.org/10.1109/ICORR.2005.1501143
")\].SARs for mental healthcare vary in appearance and functions depending on their objectives and needs. Currently, the most common design of SARs is humanoid or anthropomorphic robots equipped with human-like eyes/hands and computer screens to support additional tasks (e.g., Aldebaran robotics’ Pepper/Nao [[5](/article/10.1007/s12369-025-01323-5#ref-CR5 "Aldebaran (2025) Pepper. https://aldebaran.com/en/pepper/
. Accessed 31 Mar 2025"), [6](/article/10.1007/s12369-025-01323-5#ref-CR6 "Aldebaran (2025) Na06.
https://aldebaran.com/en/nao6/
. Accessed 31 Mar 2025")\], Misty Robotics’ _Misty_ \[[7](/article/10.1007/s12369-025-01323-5#ref-CR7 "Misty Robotics (2025) Misty II.
https://www.mistyrobotics.com/misty-ii
. Accessed 31 Mar 2025")\]), which are deployed in hospitals or elderly care homes for health monitoring/screening, activity facilitation, or educational support \[[8](/article/10.1007/s12369-025-01323-5#ref-CR8 "Getson C, Nejat G (2021) Socially assistive robots helping older adults through the pandemic and life after COVID-19. Robotics 10(3):106.
https://doi.org/10.3390/robotics10030106
")\]. In contrast, some robots take the form of plush toys (e.g., AISTs _PARO_ \[[9](/article/10.1007/s12369-025-01323-5#ref-CR9 "PARO Robots (2025) PARO Therapeutic robot.
https://www.parorobots.com/
. Accessed 31 Mar 2025")\], Sony’s _aibo_ \[[10](/article/10.1007/s12369-025-01323-5#ref-CR10 "SONY (2025) Aibo.
https://us.aibo.com/
. Accessed 31 Mar 2025")\]), if they aim to foster emotional affinity, especially with children \[[11](/article/10.1007/s12369-025-01323-5#ref-CR11 "Kabacińska K, Prescott TJ, Robillard JM (2021) Socially assistive robots as mental health interventions for children: a scoping review. Int J Soc robot 13(5):919–935.
https://doi.org/10.1007/s12369-020-00679-0
")\]. The potential of SARs is as diverse as their forms and functions, with applications ranging from assisting elderly people and those with cognitive disabilities to aiding in recovery and rehabilitation \[[12](/article/10.1007/s12369-025-01323-5#ref-CR12 "Aymerich-Franch L, Ferrer I (2023) Socially assistive robots’ deployment in healthcare settings: a global perspective. Int J Humanoid RoBot 20(1):2350002.
https://doi.org/10.1142/S0219843623500020
"), [13](/article/10.1007/s12369-025-01323-5#ref-CR13 "Feil-Seifer D, Matarić MJ (2011) Socially assistive robotics. IEEE Robot Autom Mag 18(1):24–31.
https://doi.org/10.1109/MRA.2010.940150
")\]. SARs can offer patient diagnosis and assessment/monitoring, management of medication adherence, education, companionship, stress reduction, and healthcare promotion \[[12](/article/10.1007/s12369-025-01323-5#ref-CR12 "Aymerich-Franch L, Ferrer I (2023) Socially assistive robots’ deployment in healthcare settings: a global perspective. Int J Humanoid RoBot 20(1):2350002.
https://doi.org/10.1142/S0219843623500020
"), [14](/article/10.1007/s12369-025-01323-5#ref-CR14 "Hsieh CJ, Li PS, Wang CH, Lin SL, Hsu TC, Tsai CMT (2023) Socially assistive robots for people living with dementia in long-term facilities: a systematic review and meta-analysis of randomized controlled trials. Gerontology 69(8):1027–1042.
https://doi.org/10.1159/000529849
")\].Despite their strong potential, the effectiveness of SARs in real-world mental healthcare applications are still not unequivocal. Guemghar et al.’s (2022) scoping review inspected intervention studies that employed SARs for mental healthcare and indicated promising outcomes from the included studies [[15](/article/10.1007/s12369-025-01323-5#ref-CR15 "Guemghar I, Pires de Oliveira Padilha P, Abdel-Baki A, Jutras-Aswad D, Paquette J, Pomey MP (2022) Social robot interventions in mental health care and their outcomes, barriers, and facilitators: scoping review. JMIR Ment Health 9(4):e36094. https://doi.org/10.2196/36094
")\]. Also, Hsieh et al.’s (2023) systematic reviews of randomized controlled trials concluded that SARs can improve mental health symptoms of dementia patients \[[14](/article/10.1007/s12369-025-01323-5#ref-CR14 "Hsieh CJ, Li PS, Wang CH, Lin SL, Hsu TC, Tsai CMT (2023) Socially assistive robots for people living with dementia in long-term facilities: a systematic review and meta-analysis of randomized controlled trials. Gerontology 69(8):1027–1042.
https://doi.org/10.1159/000529849
")\]. However, some reviews have raised concerns regarding the study designs and the generalizability of the findings \[[15](#ref-CR15 "Guemghar I, Pires de Oliveira Padilha P, Abdel-Baki A, Jutras-Aswad D, Paquette J, Pomey MP (2022) Social robot interventions in mental health care and their outcomes, barriers, and facilitators: scoping review. JMIR Ment Health 9(4):e36094.
https://doi.org/10.2196/36094
"),[16](#ref-CR16 "Koh WQ, Felding SA, Budak KB, Toomey E, Casey D (2021) Barriers and facilitators to the implementation of social robots for older adults and people with dementia: a scoping review. BMC GeriAtr 21(1):351.
https://doi.org/10.1186/s12877-021-02277-9
"),[17](/article/10.1007/s12369-025-01323-5#ref-CR17 "Papadopoulos I, Koulouglioti C, Lazzarino R, Ali S (2020) Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: a systematic review. BMJ Open 10(1):e033096.
https://doi.org/10.1136/bmjopen-2019-033096
")\]. The reviews suggested the barriers to the implementation of SARs, including the technical issues of SARs, their limited performance (e.g., restricted interaction capabilities or skills of personalization/adaptability), and the negative preconceptions about SARs \[[15](/article/10.1007/s12369-025-01323-5#ref-CR15 "Guemghar I, Pires de Oliveira Padilha P, Abdel-Baki A, Jutras-Aswad D, Paquette J, Pomey MP (2022) Social robot interventions in mental health care and their outcomes, barriers, and facilitators: scoping review. JMIR Ment Health 9(4):e36094.
https://doi.org/10.2196/36094
"), [17](/article/10.1007/s12369-025-01323-5#ref-CR17 "Papadopoulos I, Koulouglioti C, Lazzarino R, Ali S (2020) Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: a systematic review. BMJ Open 10(1):e033096.
https://doi.org/10.1136/bmjopen-2019-033096
")\].To overcome these limitations, it is important to align the technical features and therapeutic concepts of SARs with user needs, as these elements are closely intertwined with overall user experiences [[16](/article/10.1007/s12369-025-01323-5#ref-CR16 "Koh WQ, Felding SA, Budak KB, Toomey E, Casey D (2021) Barriers and facilitators to the implementation of social robots for older adults and people with dementia: a scoping review. BMC GeriAtr 21(1):351. https://doi.org/10.1186/s12877-021-02277-9
"), [18](/article/10.1007/s12369-025-01323-5#ref-CR18 "Wannheden C, Stenfors T, Stenling A, von Thiele Schwarz U (2021) Satisfied or frustrated? A qualitative analysis of need satisfying and need frustrating experiences of engaging with digital health technology in chronic care. Front Public Health 8:623773.
https://doi.org/10.3389/fpubh.2020.623773
")\]. Good SARs should consider user attraction and engagement as well as technological sophistication and evidence-based treatment strategies. An exclusive focus on either the technological capabilities or the therapeutic functionalities risks overlooking user experience-one of the core principles of digital healthcare. Addressing the user-centered approach may be especially crucial for SARs, as their primary role is to interact with users by serving as companions, coaches, or playmates \[[19](/article/10.1007/s12369-025-01323-5#ref-CR19 "Kachouie R, Sedighadeli S, Khosla R, Chu MT (2014) Socially assistive robots in elderly care: a mixed-method systematic literature review. Int J Hum Comput Interact 30(5):369–393.
https://doi.org/10.1080/10447318.2013.873278
"), [20](/article/10.1007/s12369-025-01323-5#ref-CR20 "Rabbitt SM, Kazdin AE, Scassellati B (2015) Integrating socially assistive robotics into mental healthcare interventions: applications and recommendations for expanded use. Clin Psychol Rev 35:35–46.
https://doi.org/10.1016/j.cpr.2014.07.001
")\]. SARs should be designed to mediate social interactions based on the users’ perspectives so that they can change their perceptions, enhance and modify social behaviors, develop structures for interactions, and change how they feel \[[21](/article/10.1007/s12369-025-01323-5#ref-CR21 "Chita-Tegmark M, Scheutz M (2021) Assistive robots for the social management of health: a framework for robot design and human-robot interaction research. Int J Soc robot 13(2):197–217.
https://doi.org/10.1007/s12369-020-00634-z
")\].Therefore, to develop a good SAR, developers must comprehensively understand its core objectives from the users’ perspectives and integrate user-centered design throughout the entire development process. Based on both the general objectives of SARs and the specific goals of the individual product, developers should establish clear plans to guide subsequent development stages to make their product align with their common and therapeutic objectives. The appearance, functional capabilities, and interaction strategies of SARs should be designed in accordance with these guidelines and evidence-based practices, aiming to enhance user autonomy, competence, and treatment outcomes. This review provides a comprehensive overview of user-centered design principles for SARs in mental healthcare. Key topics, including general and mental healthcare-specific objectives of SARs, design strategies of SARs, methods and principles to measure their clinical and real-world effectiveness, and the recent emergence of new technologies such as large language models (LLMs), have been selected from representative literature across relevant fields. We included relevant literature for each topic that provides comprehensive and balanced perspectives, alongside illustrative examples of SAR applications that have implemented these approaches. Finally, the review outlines developmental roadmaps of SARs that highlight critical elements and strategies to be integrated during the design and evaluation processes, emphasizing the future development of SARs in the era of rapidly advancing artificial intelligence (AI) technologies.
Specifically, this review consists of the following sections. Sect. 2, “SARs in Mental Healthcare,” outlines how SARs are currently being implemented in mental healthcare. Sect. 3, “Common Objectives of SARs,” describes the fundamental principles that all user-centered SARs should follow, also elaborating on specific objectives tailored to various types of SAR applications and user needs. Sect. 4, “Design Strategies for User-Centered SARs,” introduces practical strategies to foster adaptive, rich, and engaging interactions between SARs and users, focusing on robot and user autonomy, personalization, gamification and reward systems, contextual and participatory design approaches. This section also discusses methods to sustain effective long-term interactions through appropriate role settings and trust-building mechanisms. Sect. 5, “Evaluation of SARs,” reviews evaluation methods for SARs, including self-report measures, ethnographic and qualitative analyses, and behavioral, experiential, and physiological responses, depending on the specific needs and goals of each SAR. This section also focuses on how these independent measures can be combined to achieve multimodal and holistic evaluation of SARs, including iterative evaluation and refinement of SAR systems to enhance user-centeredness of the product. Lastly, Sect. 6, “SARs and Large Language Models,” introduces recent AI technologies and their impact on SARs, discussing potential ethical issues for their application to mental healthcare and proposing future directions.
2 SARs in Mental Healthcare
SARs represent one of the most promising platforms for mental healthcare, as they effectively provide emotional support through meaningful interactions with users. SARs promote users’ mental health by monitoring and diagnosing user data and offering interventions through supportive interactions. SARs designed for monitoring can diagnose mental health status through conversation history and user data [[22](/article/10.1007/s12369-025-01323-5#ref-CR22 "Zhang J, Chen T (2025) Artificial intelligence based social robots in the process of student mental health diagnosis. Entertain Comput 52:100799. https://doi.org/10.1016/j.entcom.2024.100799
")\], analyze daily activities and their changes to provide user feedback \[[23](/article/10.1007/s12369-025-01323-5#ref-CR23 "Bennett CC, Stanojević C, Šabanović SA, Piatt J, Kim S (2021) When no one is watching: ecological momentary assessment to understand situated social robot use in healthcare. In Proceedings of the 9th International Conference on Human-Agent Interaction, pp 245–251.
https://doi.org/10.1145/3472307.3484670
"), [24](/article/10.1007/s12369-025-01323-5#ref-CR24 "Calatrava-Nicolás FM, Gutiérrez-Maestro E, Bautista-Salinas D, Ortiz FJ, González JR, Vera-Repullo JA et al. (2021) Robotic-based well-being monitoring and coaching system for the elderly in their daily activities. Sens 21(20):6865.
https://doi.org/10.3390/s21206865
")\], and prepare for emergency situations by promptly alerting therapists or relevant institutions when needed \[[25](/article/10.1007/s12369-025-01323-5#ref-CR25 "Laban G, Ben-Zion Z, Cross ES (2022) Social robots for supporting post-traumatic stress disorder diagnosis and treatment. Front Psychiatry 12:752874.
https://doi.org/10.3389/fpsyt.2021.752874
")\]. Some SARs offer more comprehensive monitoring capabilities, extending beyond mere robot-based sensing to include diverse spatiotemporal sensing through the Internet of Things (IoT) and ambient-assisted living environments \[[26](/article/10.1007/s12369-025-01323-5#ref-CR26 "Simoens P, Mahieu C, Ongenae F, De Backere F, De Pestel S, Nelis J et al. (2016) Internet of robotic things: context-aware and personalized interventions of assistive social robots (short paper). In 2016 5th IEEE International Conference on Cloud Networking (Cloudnet), pp 204–207.
https://doi.org/10.1109/CloudNet.2016.27
"), [27](/article/10.1007/s12369-025-01323-5#ref-CR27 "Calderita LV, Vega A, Barroso-Ramírez S, Bustos P, Núñez P (2020) Designing a cyber-physical system for ambient assisted living: a use-case analysis for social robot navigation in caregiving centers. Sens 20(14):4005.
https://doi.org/10.3390/s20144005
")\].Current SARs for mental healthcare are primarily employed to support psychological well-being and alleviate mental health symptoms. SARs designed for interventions are mainly targeted toward elderly individuals, children, or people with disabilities, and they support well-being during hospital stays, deliver companionship, or provide overall life care to enhance the mental well-being of older adults [[28](/article/10.1007/s12369-025-01323-5#ref-CR28 "Cifuentes CA, Pinto MJ, Céspedes N, Múnera M (2020) Social robots in therapy and care. Curr Robot Rep 1:59–74. https://doi.org/10.1007/s43154-020-00009-2
")\]. Interventions delivered through SARs include structured guidance and free-form interactions, while the former adapts traditional psychotherapeutic programs for robotic platforms and the latter aim to enhance well-being through open and natural conversations. For structured psychotherapeutic guidance, SARs can offer psychoeducation, therapeutic play, guided training, or cognitive behavioral therapy for the effective treatment of mental health symptoms \[[11](/article/10.1007/s12369-025-01323-5#ref-CR11 "Kabacińska K, Prescott TJ, Robillard JM (2021) Socially assistive robots as mental health interventions for children: a scoping review. Int J Soc robot 13(5):919–935.
https://doi.org/10.1007/s12369-020-00679-0
"), [29](/article/10.1007/s12369-025-01323-5#ref-CR29 "Rasouli S, Gupta G, Nilsen E, Dautenhahn K (2022) Potential applications of social robots in robot-assisted interventions for social anxiety. Int J Soc robot 14:1–32.
https://doi.org/10.1007/s12369-021-00851-0
")\]. In contrast, free interactions with robots offer users immediate, continual responses and consistent emotional support \[[11](/article/10.1007/s12369-025-01323-5#ref-CR11 "Kabacińska K, Prescott TJ, Robillard JM (2021) Socially assistive robots as mental health interventions for children: a scoping review. Int J Soc robot 13(5):919–935.
https://doi.org/10.1007/s12369-020-00679-0
")\]. Thanks to the recent advancement of AI, robots can now play diverse roles in mental healthcare such as counselors, coaches, companions, or play partners, providing highly accessible interactions \[[20](/article/10.1007/s12369-025-01323-5#ref-CR20 "Rabbitt SM, Kazdin AE, Scassellati B (2015) Integrating socially assistive robotics into mental healthcare interventions: applications and recommendations for expanded use. Clin Psychol Rev 35:35–46.
https://doi.org/10.1016/j.cpr.2014.07.001
")\]. Through interactions with robots, users receive reflective feedback and personalized insights, which may foster perceived intimacy due to continuous engagement. Moreover, robots can encourage better self-disclosure among users, while ongoing interactions can alleviate feelings of loneliness \[[30](/article/10.1007/s12369-025-01323-5#ref-CR30 "Croes EA, Antheunis ML (2021 2020) 36 questions to loving a chatbot: are people willing to self-disclose to a chatbot? In Chatbot Research and Design: 4th Int Workshop, CONVERSATIONS, pp 81–95.
https://doi.org/10.1007/978-3-030-68288-2_6
"), [31](/article/10.1007/s12369-025-01323-5#ref-CR31 "Yen HY, Huang CW, Chiu HL, Jin G (2024) The effect of social robots on depression and loneliness for older residents in long-term care facilities: a meta-analysis of randomized controlled trials. J Am Med Dir assoc 25(6):104979.
https://doi.org/10.1016/j.jamda.2024.02.017
")\].For both monitoring and diagnostic purposes, enabling rich interactions through multimodal approaches—combining verbal and non-verbal methods—is beneficial for optimizing the therapy using SARs [[29](/article/10.1007/s12369-025-01323-5#ref-CR29 "Rasouli S, Gupta G, Nilsen E, Dautenhahn K (2022) Potential applications of social robots in robot-assisted interventions for social anxiety. Int J Soc robot 14:1–32. https://doi.org/10.1007/s12369-021-00851-0
")\]. Such approaches can enhance the depth and quality of user engagement, finally maximizing the effectiveness of robotic mental healthcare. For effective treatment, developers should carefully refine interaction designs and incorporate comprehensive design strategies that consider not only dialogue patterns but also the roles of the robots and the context surrounding robots and users. However, for SARs in mental healthcare to effectively achieve their therapeutic goals, it is first necessary to clearly understand the common objectives of SARs and establish core design principles to guide the entire developmental process of the robot.3 Common Objectives of SARs
An essential first step in developing SARs is to establish clear, user-centered objectives. It is known that the user-centered approach can address challenges in digital healthcare such as low adherence and engagement in digital interventions [[32](/article/10.1007/s12369-025-01323-5#ref-CR32 "Dahlhausen F, Zinner M, Bieske L, Ehlers JP, Boehme P, Fehring L (2022) There’s an app for that, but nobody’s using it: insights on improving patient access and adherence to digital therapeutics in Germany. Digit Health 8:20552076221104672. https://doi.org/10.1177/20552076221104672
"), [33](/article/10.1007/s12369-025-01323-5#ref-CR33 "Yardley L, Morrison L, Bradbury K, Muller I (2015) The person-based approach to intervention development: application to digital health-related behavior change interventions. J Med Internet Res 17(1):e4055.
https://doi.org/10.2196/jmir.4055
")\]. However, SARs must go beyond mere accessibility or engagement and prioritize enhancing the overall user experience, as making users enjoy the usage of SARs are crucial for their effectiveness \[[16](/article/10.1007/s12369-025-01323-5#ref-CR16 "Koh WQ, Felding SA, Budak KB, Toomey E, Casey D (2021) Barriers and facilitators to the implementation of social robots for older adults and people with dementia: a scoping review. BMC GeriAtr 21(1):351.
https://doi.org/10.1186/s12877-021-02277-9
"), [17](/article/10.1007/s12369-025-01323-5#ref-CR17 "Papadopoulos I, Koulouglioti C, Lazzarino R, Ali S (2020) Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: a systematic review. BMJ Open 10(1):e033096.
https://doi.org/10.1136/bmjopen-2019-033096
")\]. The user-centered SARs should understand users’ knowledge, skills, behavior, motivations, and cultural contexts, fostering the “effective engagement” that fulfills the intended outcomes of SARs \[[34](/article/10.1007/s12369-025-01323-5#ref-CR34 "van Gemert-Pijnen Je, Nijland N, van Limburg M, Ossebaard HC, Kelders SM, Eysenbach G, Seydel ER (2011) A holistic framework to improve the uptake and impact of eHealth technologies. J Med Internet Res 13(4):e1672.
https://doi.org/10.2196/jmir.1672
"), [35](/article/10.1007/s12369-025-01323-5#ref-CR35 "Yardley L, Spring BJ, Riper H, Morrison LG, Crane DH, Curtis K et al. (2016) Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med 51(5):833–842.
https://doi.org/10.1016/j.amepre.2016.06.015
")\].Based on the user-centered approach, there are three common objectives of SARs: (i) to promote user autonomy, (ii) to promote user competence, and (ii) to encourage users’ positive and meaningful emotional experiences. First, SARs should enhance user autonomy, so that they can freely select options that are relevant and clinically meaningful to their individual needs. SARs should provide users with as many choices as possible, allowing them to have control over their interactions such as selecting preferred content or deciding when to engage. SARs should also enhance users’ competence. They should facilitate noticeable improvements for users without causing inconvenience or discomfort. Finally, SARs should foster positive emotional experiences and a sense of relatedness. For better user experiences, SARs offer enjoyable interactions to users and establish trust through meaningful engagement. These principles underscore user experience as a central goal throughout SAR development and validation.
Defining the purpose of a SAR involves two core considerations: the general rationale (“_Why SAR?_”) and the specific rationale (“_Why our SAR?_”). The general rationale focuses on common goals, such as enhancing user functioning and quality of life through social interaction. In contrast, the specific rationale involves customizing the features of SARs to meet targeted user needs and therapeutic objectives. Such customization influences elements such as appearance (e.g., humanoid, zoomorphic, or non-human-like), technical functionalities (e.g., chatbots, physiological sensors, IoT integration), and interaction design.
SAR characteristics can be categorized along two key dimensions: morphology and integration level (Fig. 1). Morphologically, SARs can mostly be either anthropomorphic, which employs human-like features to enhance engagement, or non-human-like, which adopts forms tailored to specific interaction goals such as pet-like companions or telepresence robots. The degree of human-likeness is a fundamental determinant of human-robot interaction, especially during initial interactions, as it shapes users’ expectations about how the robot will communicate and perform [[36](/article/10.1007/s12369-025-01323-5#ref-CR36 "Onnasch L, Roesler E (2021) A taxonomy to structure and analyze human-robot interaction. Int J Soc robot 13(4):833–849. https://doi.org/10.1007/s12369-020-00666-5
")\]. Therefore, human-like and non-human-like social robots may adopt significantly different approaches in how they interact with users. Robot anthropomorphism is known to have beneficial effects on user interactions and engagement \[[37](#ref-CR37 "Roesler E, Manzey D, Onnasch L (2021) A meta-analysis on the effectiveness of anthropomorphism in human-robot interaction. Sci robot 6(58):eabj5425.
https://doi.org/10.1126/scirobotics.abj5425
"),[38](#ref-CR38 "Park G, Lee S, Chung J (2023) Do anthropomorphic chatbots increase counseling satisfaction and reuse intention? The moderated mediation of social rapport and social anxiety. Cyberpsychol Behav Soc Netw 26(5):357–365.
https://doi.org/10.1089/cyber.2022.0157
"),[39](/article/10.1007/s12369-025-01323-5#ref-CR39 "Giger JC, Piçarra N, Alves-Oliveira P, Oliveira R, Arriaga P (2019) Humanization of robots: is it really such a good idea? Hum Behav Emerg Technol 1(2):111–123.
https://doi.org/10.1002/hbe2.147
")\]. However, the effectiveness of anthropomorphism can be influenced by several contextual factors \[[37](/article/10.1007/s12369-025-01323-5#ref-CR37 "Roesler E, Manzey D, Onnasch L (2021) A meta-analysis on the effectiveness of anthropomorphism in human-robot interaction. Sci robot 6(58):eabj5425.
https://doi.org/10.1126/scirobotics.abj5425
")\], and challenges may arise in humanizing robots, such as the uncanny valley phenomenon \[[40](/article/10.1007/s12369-025-01323-5#ref-CR40 "Mori M, MacDorman KF, Kageki N (2012) The uncanny valley [from the field]. IEEE Robot Autom Mag 19(2):98–100.
https://doi.org/10.1109/MRA.2012.2192811
")\] or potential distortions in interaction quality, perceived trust, or emotional affinity toward robots \[[39](/article/10.1007/s12369-025-01323-5#ref-CR39 "Giger JC, Piçarra N, Alves-Oliveira P, Oliveira R, Arriaga P (2019) Humanization of robots: is it really such a good idea? Hum Behav Emerg Technol 1(2):111–123.
https://doi.org/10.1002/hbe2.147
"), [41](/article/10.1007/s12369-025-01323-5#ref-CR41 "Szondy M, Fazekas P (2024) Attachment to robots and therapeutic efficiency in mental health. Front psychol 15:1347177.
https://doi.org/10.3389/fpsyg.2024.1347177
")\].Fig. 1
Common objectives and specifications of SAR features categorized by morphology and level of integration [[42](#ref-CR42 "Gouaillier D, Hugel V, Blazevic P, Kilner C, Monceaux J, Lafourcade P (2009) Mechatronic design of NAO humanoid. In 2009 IEEE International Conference on Robotics and Automation, pp 769–774. https://doi.org/10.1109/ROBOT.2009.5152516
"),[43](#ref-CR43 "Pandey AK, Gelin R (2018) A mass-produced sociable humanoid robot: Pepper: the first machine of its kind. IEEE Robot Autom Mag 25(3):40–48.
https://doi.org/10.1109/MRA.2018.2833157
"),[44](#ref-CR44 "Hung L, Liu C, Woldum E, Au-Yeung A, Berndt A, Wallsworth C et al. (2019) The benefits of and barriers to using a social robot PARO in care settings: a scoping review. BMC GeriAtr 19:232.
https://doi.org/10.1186/s12877-019-1244-6
"),[45](#ref-CR45 "Fujita M (2004) On activating human communications with pet-type robot AIBO. Proc IEEE 92(11):1804–1813.
https://doi.org/10.1109/JPROC.2004.835364
"),[46](#ref-CR46 "Desai M, Tsui KM, Yanco HA, Uhlik C (2011) Essential features of telepresence robots. In 2011 IEEE Conference on Technologies for Practical Robot Applications, pp 15–20.
https://doi.org/10.1109/TEPRA.2011.5753474
"),[47](#ref-CR47 "Luperto M, Monroy J, Renoux J, Lunardini F, Basilico N, Bulgheroni M et al. (2022) Integrating social assistive robots, IoT, virtual communities and smart objects to assist at-home independently living elders: the MoveCare project. Int J Soc robot 15(3):517–545.
https://doi.org/10.1007/s12369-021-00843-0
"),[48](#ref-CR48 "Mayer P, Beck C, Panek P (2012) Examples of multimodal user interfaces for socially assistive robots in ambient assisted living environments. In 2012 IEEE 3rd International Conference on Cognitive Infocommunications (CogInfoCom), pp 401–406.
https://doi.org/10.1109/CogInfoCom.2012.6422014
"),[49](/article/10.1007/s12369-025-01323-5#ref-CR49 "Lee OE, Lee H, Park A, Choi NG (2024) My precious friend: human-robot interactions in home care for socially isolated older adults. Clin gerontol 47(1):161–170.
https://doi.org/10.1080/07317115.2022.2156829
")\]. _Note_: for more information about each robot, see the official websites: _Nao_ \[[6](/article/10.1007/s12369-025-01323-5#ref-CR6 "Aldebaran (2025) Na06.
https://aldebaran.com/en/nao6/
. Accessed 31 Mar 2025")\], _pepper_ \[[5](/article/10.1007/s12369-025-01323-5#ref-CR5 "Aldebaran (2025) Pepper.
https://aldebaran.com/en/pepper/
. Accessed 31 Mar 2025")\], _PARO_ \[[9](/article/10.1007/s12369-025-01323-5#ref-CR9 "PARO Robots (2025) PARO Therapeutic robot.
https://www.parorobots.com/
. Accessed 31 Mar 2025")\], _aibo_ \[[10](/article/10.1007/s12369-025-01323-5#ref-CR10 "SONY (2025) Aibo.
https://us.aibo.com/
. Accessed 31 Mar 2025")\], _ohmnicare_ \[[50](/article/10.1007/s12369-025-01323-5#ref-CR50 "O (2025) A new standard in virtual care: OhmniCare.
https://ohmnilabs.com/products/ohmnicare-mobile-telehealth-robot/
. Accessed 31 Mar 2025")\], _hyodol_ \[[51](/article/10.1007/s12369-025-01323-5#ref-CR51 "Hyodol (2025) HYODOL: AI care platform.
https://en.hyodol.com/
. Accessed 31 Mar 2025")\]Regarding integration, some SARs function independently, achieving therapeutic goals solely through direct interaction with users, while others are embedded within broader therapeutic systems, such as ambient assisted living environments, to provide more comprehensive and versatile support [[46](/article/10.1007/s12369-025-01323-5#ref-CR46 "Desai M, Tsui KM, Yanco HA, Uhlik C (2011) Essential features of telepresence robots. In 2011 IEEE Conference on Technologies for Practical Robot Applications, pp 15–20. https://doi.org/10.1109/TEPRA.2011.5753474
"), [52](#ref-CR52 "De Carolis B, Ferilli S, Macchiarulo N (2022) Ambient assisted living and social robots: towards learning relations between user’s daily routines and mood. In Adjunct proceedings of the 30th ACM conference on user modeling, adaptation and personalization, pp 123–129.
https://doi.org/10.1145/3511047.3537691
"),[53](#ref-CR53 "Phillips E, Zhao X, Ullman D, Malle BF (2018) What is human-like? Decomposing robots’ human-like appearance using the anthropomorphic robot (abot) database. In Proc 2018 ACM/IEEE Int Conf Hum-Robot Interact, pp 105–113.
https://doi.org/10.1145/3171221.3171268
"),[54](/article/10.1007/s12369-025-01323-5#ref-CR54 "Calvaresi D, Cesarini D, Sernani P, Marinoni M, Dragoni AF, Sturm A (2017) Exploring the ambient assisted living domain: a systematic review. J Ambient Intell Humaniz Comput 8:239–257.
https://doi.org/10.1007/s12652-016-0374-3
")\]. Whereas independently functioning robots operate as a single robotic device while communicating with users through a limited interaction platform, system-integrated robots function through the cooperation of multiple interdependent devices and agents, aiming to enhance user interactions within robotic ecologies \[[55](/article/10.1007/s12369-025-01323-5#ref-CR55 "Dragone M, Abdel-Naby S, Swords D, O’Hare GM, Broxvall M (2012) A programming framework for multi-agent coordination of robotic ecologies. In Int Workshop on Programming Multi-Agent Systems, pp 72–89.
https://doi.org/10.1007/978-3-642-38700-5_5
")\]. Through these integrative systems, robots and service providers can more effectively monitor health-related factors and achieve rich interactions through multimodal environmental interactions \[[55](#ref-CR55 "Dragone M, Abdel-Naby S, Swords D, O’Hare GM, Broxvall M (2012) A programming framework for multi-agent coordination of robotic ecologies. In Int Workshop on Programming Multi-Agent Systems, pp 72–89.
https://doi.org/10.1007/978-3-642-38700-5_5
"),[56](#ref-CR56 "Marques G, Pires IM, Miranda N, Pitarma R (2019) Air quality monitoring using assistive robots for ambient assisted living and enhanced living environments through internet of things. Electron 8(12):1375.
https://doi.org/10.3390/electronics8121375
"),[57](/article/10.1007/s12369-025-01323-5#ref-CR57 "Bui HD, Chong NY (2018) An integrated approach to human-robot-smart environment interaction interface for ambient assisted living. In 2018 IEEE Workshop on Advanced Robotics and its Social Impacts (ARSO), pp 32–37.
https://doi.org/10.1109/ARSO.2018.8625821
")\]. However, despite the significant advantages of system-integrated robots, they can involve high initial costs and potential barriers concerning user acceptability, as well as greater vulnerability to technical issues, inflexibility, and concerns around privacy and security.4 Design Strategies for User-Centered SARs
The current SARs still leave ample room for improvement for their effective use in mental healthcare. Many existing SARs often resemble interactive robots only in appearance, relying primarily on screens or embedded computers, which cannot fully utilize the robot’s capabilities to access advanced AI-based platforms available these days [[58](#ref-CR58 "Feingold-Polak R, Barzel O, Levy-Tzedek S (2024) Socially assistive robot for stroke rehabilitation: a long-term in-the-wild Pilot randomized controlled trial. IEEE Trans Neural Syst Rehabil Eng 32:1616–1626. https://doi.org/10.1109/TNSRE.2024.3387320
"),[59](#ref-CR59 "Spitale M, Silleresi S, Garzotto F, Matarić MJ (2023) Using socially assistive robots in speech-language therapy for children with language impairments. Int J Soc robot 15(9):1525–1542.
https://doi.org/10.1007/s12369-023-01028-7
"),[60](/article/10.1007/s12369-025-01323-5#ref-CR60 "Elfaki AO, Abduljabbar M, Ali L, Alnajjar F, Mehiar DA, Marei AM et al. (2023) Revolutionizing social robotics: a cloud-based framework for enhancing the intelligence and autonomy of social robots. Robot 12(2):48.
https://doi.org/10.3390/robotics12020048
")\]. Future SARs may require breakthroughs that keep pace with the rapid advancements in AI technologies; therefore, we suggest some design strategies for next-generation SARs that incorporate user-centered principles and offer more effective mental healthcare, as summarized in Fig. [2](/article/10.1007/s12369-025-01323-5#Fig2). First, SARs should be capable of independently interacting with users without the need for direct human intervention. This will require SARs to autonomously understand users’ needs, characteristics, conditions, and contexts in detail and adapt their behavior accordingly, incorporating personalization, multimodal sensing, and behavioral adaptation systems. In addition, for long-term interactions, SARs may need to independently communicate with users to foster their sense of attachment and trust toward the robots \[[41](/article/10.1007/s12369-025-01323-5#ref-CR41 "Szondy M, Fazekas P (2024) Attachment to robots and therapeutic efficiency in mental health. Front psychol 15:1347177.
https://doi.org/10.3389/fpsyg.2024.1347177
")\]. Strategically designing roles and identities of SARs to enhance users’ engagement and sustained interaction can be beneficial. Finally, as SARs should be enjoyable, incorporating gamified content to enhance effective user engagement can also help fulfill the objectives of SARs.Fig. 2
Design principles and framework for SARs in mental healthcare
4.1 Robot Autonomy
Currently, SARs in mental healthcare tend to be more effective when their use is guided by human therapists alongside robotic interventions [[15](/article/10.1007/s12369-025-01323-5#ref-CR15 "Guemghar I, Pires de Oliveira Padilha P, Abdel-Baki A, Jutras-Aswad D, Paquette J, Pomey MP (2022) Social robot interventions in mental health care and their outcomes, barriers, and facilitators: scoping review. JMIR Ment Health 9(4):e36094. https://doi.org/10.2196/36094
")\]. Although collaboration with humans can enhance the effectiveness of SARs, however, it remains ideal for robots to autonomously plan, act, and display affective cues, considering the nature of robots \[[61](/article/10.1007/s12369-025-01323-5#ref-CR61 "Malfaz M, Castro-González Á, Barber R, Salichs MA (2011) A biologically inspired architecture for an autonomous and social robot. IEEE T Auton Ment De 3(3):232–246.
https://doi.org/10.1109/TAMD.2011.2112766
")\]. Autonomous robots capable of independently supporting humans with minimal human intervention can reduce manpower costs for therapists and healthcare providers \[[4](/article/10.1007/s12369-025-01323-5#ref-CR4 "Feil-Seifer D, Mataric MJ (2005) Defining socially assistive robotics. 9th Int Conf Rehabil robot 465–468.
https://doi.org/10.1109/ICORR.2005.1501143
"), [62](/article/10.1007/s12369-025-01323-5#ref-CR62 "Shukla J, Barreda-Ángeles M, Oliver J, Puig D (2017) Effectiveness of socially assistive robotics during cognitive stimulation interventions: impact on caregivers. In 2017 26th IEEE Int Symp Robot Human Interact Commun (RO-MAN), pp 62–67.
https://doi.org/10.1109/ROMAN.2017.8172281
")\]. The autonomy of robots is becoming increasingly important, as advancements in AI allow SARs to integrate more sophisticated AI assistants for behavioral healthcare \[[63](/article/10.1007/s12369-025-01323-5#ref-CR63 "Stade EC, Stirman SW, Ungar LH, Boland CL, Schwartz HA, Yaden DB et al. (2024) Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment Health Res 3:12.
https://doi.org/10.1038/s44184-024-00056-z
")\]. SARs equipped with advanced technologies will have the potential to independently engage in complex interactions with users, complementing existing healthcare, substituting certain tasks, or even expanding healthcare into new areas previously beyond human reach.One of the prerequisites for independent interaction is ensuring that users perceive SARs as an entity capable of human-like communication. Users engage in interactions with SARs when they feel that they can seamlessly interact with robots, and such interactions can be perpetuated when they feel comfortable with SARs. To achieve independent interactions, SARs should understand individual user characteristics with real-time behavioral adaptation systems to autonomously and immediately adjust interactions and behaviors according to users’ reactions. For this, SARs collect information about users through multiple devices and sensors, continuously adapting their behavior based on the processed data. Such multimodal processing includes not only users’ voluntary input but also various physiological, gait/speech, and facial recognition data [[64](/article/10.1007/s12369-025-01323-5#ref-CR64 "Abdollahi H, Mahoor MH, Zandie R, Siewierski J, Qualls SH (2022) Artificial emotional intelligence in socially assistive robots for older adults: a pilot study. IEEE Trans Affective Comput 14(3):2020–2032. https://doi.org/10.1109/TAFFC.2022.3143803
"), [65](/article/10.1007/s12369-025-01323-5#ref-CR65 "Céspedes N, Irfan B, Senft E, Cifuentes CA, Gutierrez LF, Rincon-Roncancio M et al. (2021) A socially assistive robot for long-term cardiac rehabilitation in the real world. Front neurorobot 15:633248.
https://doi.org/10.3389/fnbot.2021.633248
")\]. Abdollahi et al. (2022) introduced an emotion recognition system based on a SAR that integrates multimodal user data, including conversational responses and facial expressions, enabling adaptive robot behaviors \[[64](/article/10.1007/s12369-025-01323-5#ref-CR64 "Abdollahi H, Mahoor MH, Zandie R, Siewierski J, Qualls SH (2022) Artificial emotional intelligence in socially assistive robots for older adults: a pilot study. IEEE Trans Affective Comput 14(3):2020–2032.
https://doi.org/10.1109/TAFFC.2022.3143803
")\]. Heredia et al. (2022) proposed a multimodal emotion recognition system employing a fusion approach combining facial, audio, and text modalities for emotion analysis \[[66](/article/10.1007/s12369-025-01323-5#ref-CR66 "Heredia J, Lopes-Silva E, Cardinale Y, Diaz-Amado J, Dongo I, Graterol W, Aguilera A (2022) Adaptive multimodal emotion detection architecture for social robots. IEEE Access. 10:20727–20744.
https://doi.org/10.1109/ACCESS.2022.3149214
")\].For autonomous communication aimed at promoting mental health, SARs should be capable of adapting to diverse situations and contexts [[67](/article/10.1007/s12369-025-01323-5#ref-CR67 "Di Napoli C, Rossi S (2019) A layered architecture for socially assistive robotics as a service. In 2019 IEEE Int Conf Syst Man Cybern, pp 352–357. https://doi.org/10.1109/SMC.2019.8914532
"), [68](/article/10.1007/s12369-025-01323-5#ref-CR68 "Umbrico A, Cesta A, Cortellessa G, Orlandini A (2020) A holistic approach to behavior adaptation for socially assistive robots. Int J Soc robot 12(3):617–637.
https://doi.org/10.1007/s12369-019-00617-9
")\]. Adaptive decision-making models have been explored to reinforce the creativity and versatile reasoning skills of SARs, which can enhance adaptability and the effectiveness of user interactions \[[68](/article/10.1007/s12369-025-01323-5#ref-CR68 "Umbrico A, Cesta A, Cortellessa G, Orlandini A (2020) A holistic approach to behavior adaptation for socially assistive robots. Int J Soc robot 12(3):617–637.
https://doi.org/10.1007/s12369-019-00617-9
"), [69](/article/10.1007/s12369-025-01323-5#ref-CR69 "Dell’anna D, Jamshidnejad A (2022) Evolving fuzzy logic systems for creative personalized socially assistive robots. Eng Appl Artif Intell 114:105064.
https://doi.org/10.1016/j.engappai.2022.105064
")\]. Developing social skills and emotional expressions is also important for autonomous interactions. To achieve this, biologically inspired models simulating homeostatic features, motivation, affection, and cognition have recently become popular \[[70](/article/10.1007/s12369-025-01323-5#ref-CR70 "Maroto-Gómez M, Alonso-Martín F, Malfaz M, Castro-González Á, Castillo JC, Salichs MÁ (2023) A systematic literature review of decision-making and control systems for autonomous and social robots. Int J Soc robot 15(5):745–789.
https://doi.org/10.1007/s12369-023-00977-3
")\]. These models enable robots to express various internal states, such as emotions, thoughts, or needs. Castro-González et al. (2014) developed a motivation and learning system for a social robot by implementing a decision-making algorithm that considers the robot’s internal drives, external stimuli, and feedback from its previous actions \[[71](/article/10.1007/s12369-025-01323-5#ref-CR71 "Castro-González Á, Malfaz M, Gorostiza JF, Salichs MA (2014) Learning behaviors by an autonomous social robot with motivations. Cybern Syst 45(7):568–598.
https://doi.org/10.1080/01969722.2014.945321
")\]. Similarly, Malfaz et al. (2011) proposed an architecture that incorporates automatic- and deliberate-level processes for implementing drives, motivations, emotions, and self-learning, integrated with long-term and short-term memories, sensors and actuators, and a biologically inspired decision-making system \[[61](/article/10.1007/s12369-025-01323-5#ref-CR61 "Malfaz M, Castro-González Á, Barber R, Salichs MA (2011) A biologically inspired architecture for an autonomous and social robot. IEEE T Auton Ment De 3(3):232–246.
https://doi.org/10.1109/TAMD.2011.2112766
")\].4.2 User Autonomy
One common mistake of current digital interventions is that they often aim to unilaterally deliver treatment or involve rigid, predefined interactions designed to manipulate user behaviors in a single intended direction. While digital interventions should indeed be strategically constructed to guide users toward certain beneficial behaviors, desirable SARs must offer users as many meaningful choices as possible to ensure enjoyable interactions. Since social robots can either promote or inhibit user autonomy, focusing exclusively on therapeutic goals from the developers’ perspectives while neglecting user autonomy may lead to detrimental effects on both users and therapeutic outcomes [[72](/article/10.1007/s12369-025-01323-5#ref-CR72 "Formosa P (2021) Robot autonomy vs. human autonomy: social robots, artificial intelligence (AI), and the nature of autonomy. Minds mach 31(4):595–616. https://doi.org/10.1007/s11023-021-09579-2
")\]. Therefore, developers should design SARs to offer versatile, adaptive interaction content that provides diverse options, enabling users to freely choose content, timing, duration, detailed settings, and modes of interaction. SARs should empower users toward meaningful goals and support their competencies, rather than restrict their choices or offer superficial options without genuine consideration of their autonomy \[[72](/article/10.1007/s12369-025-01323-5#ref-CR72 "Formosa P (2021) Robot autonomy vs. human autonomy: social robots, artificial intelligence (AI), and the nature of autonomy. Minds mach 31(4):595–616.
https://doi.org/10.1007/s11023-021-09579-2
")\]. Van Minkelen et al.’s (2020) study provides an illustrative example of incorporating the principle of autonomy into social robots. In their research, a robot tutor was designed to respect children’s autonomy by offering multiple options during vocabulary tasks, resulting in stronger and longer-lasting engagement \[[73](/article/10.1007/s12369-025-01323-5#ref-CR73 "Van Minkelen P, Gruson C, Van Hees P, Willems M, De Wit J, Aarts R et al. (2020) Using self-determination theory in social robots to increase motivation in L2 word learning. In Proc 2020 ACM/IEEE Int Conf Hum-Robot Interact, pp 369–377.
https://doi.org/10.1145/3319502.3374828
")\].Since user autonomy represents one of the most critical social and ethical considerations, developers should prioritize it as a primary objective in their products. Moreover, it is essential that developers provide users with authentic choices rather than simply expanding superficial or meaningless options. Authentic choices refer specifically to those reflecting users’ genuine motives, desires, preferences, and reasons, enabling users to endorse or acknowledge these decisions upon reflection [[72](/article/10.1007/s12369-025-01323-5#ref-CR72 "Formosa P (2021) Robot autonomy vs. human autonomy: social robots, artificial intelligence (AI), and the nature of autonomy. Minds mach 31(4):595–616. https://doi.org/10.1007/s11023-021-09579-2
")\]. Importantly, these choices also include the option of not using the SAR. Although it is indeed a disadvantageous situation for healthcare providers, under the principle of autonomy, SARs are controlled by users when being used, and users have the right to terminate the treatment if it causes harm \[[74](/article/10.1007/s12369-025-01323-5#ref-CR74 "Feil-Seifer D, Skinner K, Matarić MJ (2007) Benchmarks for evaluating socially assistive robotics. Interact Stud 8(3):423–439.
https://doi.org/10.1075/is.8.3.07fei
")\]. Therefore, we argue that it may be risky to urge its usage by repetitive notifications or even forcing it to technically increase adherence, as it leads to adverse effects if users do not enjoy or feel comfortable using the SAR. A similar mistake is found when a product depends solely on extrinsic rewards for user engagement. Although extrinsic rewards may create positive incentives for user engagement, they can also lead to distraction or undermine intrinsic motivation, especially in long-term interactions, as in the case of SARs \[[75](/article/10.1007/s12369-025-01323-5#ref-CR75 "Gary K, Stoll R, Rallabhandi P, Patwardhan M, Hamel D, Amresh A et al. (2017) Mhealth games as rewards: incentive or distraction? Proc 2017 Int Conf Digit Health 209–210.
https://doi.org/10.1145/3079452.3079459
"), [76](/article/10.1007/s12369-025-01323-5#ref-CR76 "Lewis ZH, Swartz MC, Lyons EJ (2016) What’s the point?: a review of reward systems implemented in gamification interventions. Games Health J 5(2):93–99.
https://doi.org/10.1089/g4h.2015.0078
")\]. If rewards are to be used, it could be better to provide emotional rewards such as positive feedback or compliments, as they inspire positive emotions and finally fortify users’ enjoyment. However, a more effective approach may be to create an environment that naturally fosters interest, encourages challenge, and inspires growth, thereby reinforcing intrinsic motivation. Van Minkelen et al. (2020) also emphasized intrinsic motivation within its intervention principles, providing children with a greater sense of control over their own learning process \[[73](/article/10.1007/s12369-025-01323-5#ref-CR73 "Van Minkelen P, Gruson C, Van Hees P, Willems M, De Wit J, Aarts R et al. (2020) Using self-determination theory in social robots to increase motivation in L2 word learning. In Proc 2020 ACM/IEEE Int Conf Hum-Robot Interact, pp 369–377.
https://doi.org/10.1145/3319502.3374828
")\].4.3 Personalization
As discussed above, automatic feedback and adaptive decision-making based on the multidimensional acquisition and analysis of user data can be beneficial for the future development of SARs. However, fully adaptive behavioral models will require SARs to understand user traits and contextual or cultural factors to deliver services tailored to individual needs and situations. In other words, SARs should fit users’ needs and preferences or contexts/cultural backgrounds, have various interaction channels through multimodal interaction and data collection systems, and automatically adapt their behavior according to the users’ reactions to provide tailored services [[19](/article/10.1007/s12369-025-01323-5#ref-CR19 "Kachouie R, Sedighadeli S, Khosla R, Chu MT (2014) Socially assistive robots in elderly care: a mixed-method systematic literature review. Int J Hum Comput Interact 30(5):369–393. https://doi.org/10.1080/10447318.2013.873278
"), [77](/article/10.1007/s12369-025-01323-5#ref-CR77 "Benedictis RD, Umbrico A, Fracasso F, Cortellessa G, Orlandini A, Cesta A (2023) A dichotomic approach to adaptive interaction for socially assistive robots. User Model User-Adapt Interact 33(2):293–331.
https://doi.org/10.1007/s11257-022-09347-6
"), [78](/article/10.1007/s12369-025-01323-5#ref-CR78 "Shi Z, Groechel TR, Jain S, Chima K, Rudovic O, Matarić MJ (2022) Toward personalized affect-aware socially assistive robot tutors for long-term interventions with children with autism. ACM Trans Hum-Robot Interact 11(4):1–28.
https://doi.org/10.1145/3526111
")\]. Personalization can optimize therapy by tailoring digital interventions based on individual needs, which is now widely discussed in the fields of digital therapeutics and SARs \[[17](/article/10.1007/s12369-025-01323-5#ref-CR17 "Papadopoulos I, Koulouglioti C, Lazzarino R, Ali S (2020) Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: a systematic review. BMJ Open 10(1):e033096.
https://doi.org/10.1136/bmjopen-2019-033096
"), [79](/article/10.1007/s12369-025-01323-5#ref-CR79 "Hornstein S, Zantvoort K, Lueken U, Funk B, Hilbert K (2023) Personalization strategies in digital mental health interventions: a systematic review and conceptual framework for depressive symptoms. Front Digit Health 5:1170002.
https://doi.org/10.3389/fdgth.2023.1170002
")\]. Effective personalization involves three steps: (i) gathering information on a wide range of potential users, (ii) identifying which types of interventions are the most suitable for different user characteristics (or which degrees of intervention levels are the most suitable for specific user profiles), and (iii) variating and individualizing the interventions according to the identified interventions for desirable outcomes.There have been attempts to apply personalization to the behaviors or emotion recognition features of SARs. Moro et al. (2018) designed a personalized SAR by enabling expert caregivers to demonstrate speech and gestures to the robot, which is complemented by a reinforcement learning algorithm to personalize the robot’s behavior based on user functioning, user activity, and robot activity states [[80](/article/10.1007/s12369-025-01323-5#ref-CR80 "Moro C, Nejat G, Mihailidis A (2018) Learning and personalizing socially assistive robot behaviors to aid with activities of daily living. ACM Trans Hum-Robot Interact 7(2):1–25. https://doi.org/10.1145/3277903
")\]. Van Wingerden et al. (2021) also identified and adapted to the specific needs of participants with disabilities by personalizing the robot’s tone of voice, speech speed, intonation, and the color of LED eyes according to the emotional states \[[81](/article/10.1007/s12369-025-01323-5#ref-CR81 "Van Wingerden E, Barakova E, Lourens T, Sterkenburg PS (2021) Robot-mediated therapy to reduce worrying in persons with visual and intellectual disabilities. J Appl Res Intellect Disabil 34(1):229–238.
https://doi.org/10.1111/jar.12801
")\]. These studies collectively illustrate the potential of SARs for data-driven personalization through adaptive adjustments \[[82](/article/10.1007/s12369-025-01323-5#ref-CR82 "Chawla NV, Davis DA (2013) Bringing big data to personalized healthcare: a patient-centered framework. J Gen Intern Med 28:660–665.
https://doi.org/10.1007/s11606-013-2455-8
")\].However, the current level of personalization in robot behaviors also has substantial limitations. In our view, while these efforts depend entirely on big data analysis employing machine learning, their verification processes typically focus on a narrow subset of features among the many potential functionalities of a SAR, probably because of the immense data requirements associated with validating high-dimensional models. This issue might be addressed by technical advances or the establishment of healthcare big data management platforms (e.g., the AI Lab of the National Health Service England [[83](/article/10.1007/s12369-025-01323-5#ref-CR83 "NHS England (2025) The NHS AI Lab - NHS transformation directorate. https://transform.england.nhs.uk/ai-lab/
. Accessed 4 Apr 2025")\]); however, results generated by large datasets may provide little insight into individual users, especially if they are from minority groups or vulnerable populations \[[84](/article/10.1007/s12369-025-01323-5#ref-CR84 "Jackson L, Kuhlman C, Jackson F, Fox PK (2019) Including vulnerable populations in the assessment of data from vulnerable populations. Front Big Data 2:19.
https://doi.org/10.3389/fdata.2019.00019
"), [85](/article/10.1007/s12369-025-01323-5#ref-CR85 "Nimri R, Phillip M, Clements MA, Kovatchev B (2024) Closed-loop control, artificial intelligence-based decision-support Systems, and data science. Diabetes Technol Ther 26:S1.
https://doi.org/10.1089/dia.2024.2505
")\]. Such an issue stems from the fundamental limitation of the nomothetic approach, which only highlights interindividual variability based on the homogeneity assumption within individuals \[[86](/article/10.1007/s12369-025-01323-5#ref-CR86 "Beltz AM, Wright AG, Sprague BN, Molenaar PC (2016) Bridging the nomothetic and idiographic approaches to the analysis of clinical data. Assessment 23(4):447–458.
https://doi.org/10.1177/1073191116648209
"), [87](/article/10.1007/s12369-025-01323-5#ref-CR87 "Haynes SN, Mumma GH, Pinson C (2009) Idiographic assessment: conceptual and psychometric foundations of individualized behavioral assessment. Clin Psychol Rev 29(2):179–191.
https://doi.org/10.1016/j.cpr.2008.12.003
")\]. For personalized SARs, it is promising to adopt an idiographic approach that focuses on individuals’ unique characteristics and differences \[[87](/article/10.1007/s12369-025-01323-5#ref-CR87 "Haynes SN, Mumma GH, Pinson C (2009) Idiographic assessment: conceptual and psychometric foundations of individualized behavioral assessment. Clin Psychol Rev 29(2):179–191.
https://doi.org/10.1016/j.cpr.2008.12.003
"), [88](/article/10.1007/s12369-025-01323-5#ref-CR88 "Piccirillo ML, Rodebaugh TL (2019) Foundations of idiographic methods in psychology and applications for psychotherapy. Clin Psychol Rev 71:90–100.
https://doi.org/10.1016/j.cpr.2019.01.002
")\]. Such an approach can be supported by ethnographic or qualitative research methods, which are discussed in more detail in later sections.4.4 Motivation, Reward, and Gamification
As discussed above, SARs should aim to maximize users’ competence while minimizing the effort required for use. This means that SARs should be designed as easily as possible and every content they provide should enhance users’ sense of improvement, which is relevant to the goals of SARs. To enhance user familiarity and reduce the learning curve, developers may consider adopting intuitive designs for interfaces or interaction systems so that users can draw on their prior knowledge to use a new product [89]. If users need to learn something new, it is important to let them feel that the learning process is enjoyable rather than a burden. Gamifying the tutorials can be an effective strategy to help users learn naturally, which provides dynamics that boost user engagement and make the experience more enjoyable. Moreover, it is beneficial to ensure that the knowledge that users gain is transferable to real-world situations, which is relevant to the objectives of SARs. This approach fosters “meaningful learning” that can be applied to new contexts and situations [90].
Importantly, all interactions between users and the SAR should align with its therapeutic goals. Developers must design interactions beneficial to users and establish graded goals that reinforce desired behaviors. Creating “milestones” for users to gradually achieve toward final goals, coupled with immediate feedback upon their accomplishment, can further motivate users [[33](/article/10.1007/s12369-025-01323-5#ref-CR33 "Yardley L, Morrison L, Bradbury K, Muller I (2015) The person-based approach to intervention development: application to digital health-related behavior change interventions. J Med Internet Res 17(1):e4055. https://doi.org/10.2196/jmir.4055
"), [91](/article/10.1007/s12369-025-01323-5#ref-CR91 "Jaramillo-Mediavilla L, Basantes-Andrade A, Cabezas-González M, Casillas-Martín S (2024) Impact of gamification on motivation and academic performance: a systematic review. Educ Sci 14(6):639.
https://doi.org/10.3390/educsci14060639
")\]. However, interactions and feedback should remain contextually relevant, not distracting users from the main objectives of SARs \[[75](/article/10.1007/s12369-025-01323-5#ref-CR75 "Gary K, Stoll R, Rallabhandi P, Patwardhan M, Hamel D, Amresh A et al. (2017) Mhealth games as rewards: incentive or distraction? Proc 2017 Int Conf Digit Health 209–210.
https://doi.org/10.1145/3079452.3079459
"), [92](/article/10.1007/s12369-025-01323-5#ref-CR92 "Rusz D, Le Pelley ME, Kompier MAJ, Mait L, Bijleveld E (2020) Reward-driven distraction: a meta-analysis. Psychol Bull 146(10):872–899.
https://doi.org/10.1037/bul0000296
"), [93](/article/10.1007/s12369-025-01323-5#ref-CR93 "Smith TW, Pittman TS (1978) Reward, distraction, and the overjustification effect. J Pers Soc psychol 36(5):565–572.
https://doi.org/10.1037/0022-3514.36.5.565
")\], and reward systems must avoid focusing solely on extrinsic rewards or undermining intrinsic motivation, thereby supporting users’ autonomy and competence \[[73](/article/10.1007/s12369-025-01323-5#ref-CR73 "Van Minkelen P, Gruson C, Van Hees P, Willems M, De Wit J, Aarts R et al. (2020) Using self-determination theory in social robots to increase motivation in L2 word learning. In Proc 2020 ACM/IEEE Int Conf Hum-Robot Interact, pp 369–377.
https://doi.org/10.1145/3319502.3374828
")\]. Additionally, successful SARs should ensure content enjoyment \[[16](/article/10.1007/s12369-025-01323-5#ref-CR16 "Koh WQ, Felding SA, Budak KB, Toomey E, Casey D (2021) Barriers and facilitators to the implementation of social robots for older adults and people with dementia: a scoping review. BMC GeriAtr 21(1):351.
https://doi.org/10.1186/s12877-021-02277-9
"), [17](/article/10.1007/s12369-025-01323-5#ref-CR17 "Papadopoulos I, Koulouglioti C, Lazzarino R, Ali S (2020) Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: a systematic review. BMJ Open 10(1):e033096.
https://doi.org/10.1136/bmjopen-2019-033096
")\]; users prefer interactions to be fun and amusing \[[94](/article/10.1007/s12369-025-01323-5#ref-CR94 "Vandemeulebroucke T, de Casterlé BD, Gastmans C (2018) How do older adults experience and perceive socially assistive robots in aged care: a systematic review of qualitative evidence. Aging Ment Health 22(2):149–167.
https://doi.org/10.1080/13607863.2017.1286455
")\], which has led to integrating gamified content with SARs to enhance engagement \[[95](/article/10.1007/s12369-025-01323-5#ref-CR95 "Feingold-Polak R, Barzel O, Levy-Tzedek S (2021) A robot goes to rehab: a novel gamified system for long-term stroke rehabilitation using a socially assistive robot-methodology and usability testing. J Neuroeng Rehabil 18:122.
https://doi.org/10.1186/s12984-021-00915-2
"), [96](/article/10.1007/s12369-025-01323-5#ref-CR96 "Markelius A, Sjöberg S, Bergström M, Ravandi BS, Vivas AB, Khan I, Lowe R (2024) Differential outcomes training of visuospatial memory: a gamified approach using a socially assistive robot. Int J Soc robot 16(2):363–384.
https://doi.org/10.1007/s12369-023-01083-0
")\]. Gamification, the incorporation of game-like elements into non-entertainment contexts to boost user engagement and motivation \[[97](/article/10.1007/s12369-025-01323-5#ref-CR97 "Robson K, Plangger K, Kietzmann JH, McCarthy I, Pitt L (2015) Is it all a game? Understanding the principles of gamification. Bus. Horiz 58(4):411–420.
https://doi.org/10.1016/j.bushor.2015.03.006
"), [98](/article/10.1007/s12369-025-01323-5#ref-CR98 "Seaborn K, Fels DI (2015) Gamification in theory and action: a survey. Int J Hum Comput Stud 74:14–31.
https://doi.org/10.1016/j.ijhcs.2014.09.006
")\], enriches rigid content and creates positive emotions. Effective gamification emphasizes dynamics that offer strategic actions that evoke enjoyable emotions, alongside mechanics specifying setup, rules, and progression \[[97](/article/10.1007/s12369-025-01323-5#ref-CR97 "Robson K, Plangger K, Kietzmann JH, McCarthy I, Pitt L (2015) Is it all a game? Understanding the principles of gamification. Bus. Horiz 58(4):411–420.
https://doi.org/10.1016/j.bushor.2015.03.006
")\].Donnermann et al. (2021) integrated a gamification and reward system into a social robot for educational purposes, awarding points and badges for each completed lesson to address users’ need for competence and enhance engagement [[99](/article/10.1007/s12369-025-01323-5#ref-CR99 "Donnermann M, Lein M, Messingschlager T, Riedmann A, Schaper P, Steinhaeusser S, Lugrin B (2021) Social robots and gamification for technology supported learning: an empirical study on engagement and motivation. Comput Hum Behav 121:106792. https://doi.org/10.1016/j.chb.2021.106792
")\]. However, they found no significant effect of gamification on engagement or motivation and, interestingly, observed even lower engagement when gamification was combined with social robots. Although gamification may have the potential to make products enjoyable and promote healthy behaviors, challenges to the gamification of medical health products come from the lack of adaptation regarding the products’ purposes and target groups \[[100](/article/10.1007/s12369-025-01323-5#ref-CR100 "Sardi L, Idri A, Fernández-Alemán JL (2017) A systematic review of gamification in e-health. J Biomed Inf 71:31–48.
https://doi.org/10.1016/j.jbi.2017.05.011
")\]. Gamification may not by itself enhance user engagement and motivation, especially if overly gamified content causes distraction, makes users divert from relevant content, or overloads their mental resources \[[99](/article/10.1007/s12369-025-01323-5#ref-CR99 "Donnermann M, Lein M, Messingschlager T, Riedmann A, Schaper P, Steinhaeusser S, Lugrin B (2021) Social robots and gamification for technology supported learning: an empirical study on engagement and motivation. Comput Hum Behav 121:106792.
https://doi.org/10.1016/j.chb.2021.106792
")\]. The current challenges in gamified digital interventions and SARs also highlight a lack of well-crafted design principles, which do not consider user autonomy and their alignment with the overarching goals of SARs. To address this, developers should consider user perspectives when designing gamified elements, ensuring that they promote autonomy and intrinsic motivation. Ideally, users should be able to choose whether or not they will play the game or which gamified content they will play among the multiple available options. In addition, gamified content should be relevant to the objectives of the SAR. Applying gamification out of context could let users deviate from the main goals of SARs, thwarting the overall interventions they provide \[[101](/article/10.1007/s12369-025-01323-5#ref-CR101 "Kim B (2015) Designing gamification in the right way. Libr Technol Rep 51(2):29–35")\].4.5 Contextual Design
Another important feature of user-centered SARs is their ability to incorporate contextual factors rather than relying solely on verbal exchanges. Contexts play an important role in user engagement, often becoming a significant determinant of the therapeutic outcomes of robotic interventions [[11](/article/10.1007/s12369-025-01323-5#ref-CR11 "Kabacińska K, Prescott TJ, Robillard JM (2021) Socially assistive robots as mental health interventions for children: a scoping review. Int J Soc robot 13(5):919–935. https://doi.org/10.1007/s12369-020-00679-0
"), [102](/article/10.1007/s12369-025-01323-5#ref-CR102 "Borghouts J, Eikey E, Mark G, De Leon C, Schueller SM, Schneider M et al. (2021) Barriers to and facilitators of user engagement with digital mental health interventions: systematic review. J Med Internet Res 23(3):e24387.
https://doi.org/10.2196/24387
")\]. Understanding contextual factors is crucial for personalizing interventions, as they provide vital clues for effectively profiling users and adapting interventions according to individual profiles \[[103](/article/10.1007/s12369-025-01323-5#ref-CR103 "Miralles I, Granell C (2019) Considerations for designing context-aware mobile apps for mental health interventions. Int J Environ Res Public Health 16(7):1197.
https://doi.org/10.3390/ijerph16071197
"), [104](/article/10.1007/s12369-025-01323-5#ref-CR104 "Aranda Jan CB, Jagtap S, Moultrie J (2016) Towards a framework for holistic contextual design for low-resource settings. Int J Des 10(3.
https://doi.org/10.17863/CAM.7254
")\]. Contextual design enables richer interactions by integrating multiple sensory channels, multimodal stimuli, and context-based feedback, which collectively enhance user engagement \[[105](/article/10.1007/s12369-025-01323-5#ref-CR105 "Feng Y, Perugia G, Yu S, Barakova EI, Hu J, Rauterberg GM (2022) Context-enhanced human-robot interaction: exploring the role of system interactivity and multimodal stimuli on the engagement of people with dementia. Int J Soc robot 14:807–826.
https://doi.org/10.1007/s12369-021-00823-4
")\]. Moreover, it facilitates dynamic and realistic interactions across various modalities, allowing SARs to integrate multiple non-verbal cues, including visual and auditory signals, gestures, facial expressions, and haptic or tactile feedback \[[106](/article/10.1007/s12369-025-01323-5#ref-CR106 "Feng Y, Barakova EI, Yu S, Hu J, Rauterberg GM (2020) Effects of the level of interactivity of a social robot and the response of the augmented reality display in contextual interactions of people with dementia. Sens 20(13):3771.
https://doi.org/10.3390/s20133771
"), [107](/article/10.1007/s12369-025-01323-5#ref-CR107 "Lima MR, Wairagkar M, Natarajan N, Vaitheswaran S, Vaidyanathan R (2021) Robotic telemedicine for mental health: a multimodal approach to improve human-robot engagement. Front Robot AI 8:618866.
https://doi.org/10.3389/frobt.2021.618866
")\].Contexts can encompass both the user’s personal contexts (culture, preferences, emotional state) and the environmental contexts that surround interactions [[104](/article/10.1007/s12369-025-01323-5#ref-CR104 "Aranda Jan CB, Jagtap S, Moultrie J (2016) Towards a framework for holistic contextual design for low-resource settings. Int J Des 10(3. https://doi.org/10.17863/CAM.7254
")\]. In addition, developers may not only identify and adjust to the contexts but also create their own contexts for the objectives of the SAR. A notable example is the context-enhanced robotic design presented by Feng et al. (2022). Their system featured a sheep-like social companion robot that positively responds—by moving and bleating happily—when touched or petted, along with an augmented reality display that presents calming and natural environments. This multimodal, immersive experience provided users with calming contexts, sustaining their attention and interest and increasing their engagement \[[105](/article/10.1007/s12369-025-01323-5#ref-CR105 "Feng Y, Perugia G, Yu S, Barakova EI, Hu J, Rauterberg GM (2022) Context-enhanced human-robot interaction: exploring the role of system interactivity and multimodal stimuli on the engagement of people with dementia. Int J Soc robot 14:807–826.
https://doi.org/10.1007/s12369-021-00823-4
")\]. By providing immersive and dynamic environments with context-based feedback, they could sustain users’ attention and interest and increase user engagement. Similarly, van Wingerden et al. (2021) integrated symbolic objects such as dolls, colored blocks, and beach chairs into SAR interactions, each representing distinct coping strategies for managing worry \[[81](/article/10.1007/s12369-025-01323-5#ref-CR81 "Van Wingerden E, Barakova E, Lourens T, Sterkenburg PS (2021) Robot-mediated therapy to reduce worrying in persons with visual and intellectual disabilities. J Appl Res Intellect Disabil 34(1):229–238.
https://doi.org/10.1111/jar.12801
")\]. The robot in this study also employed dynamic eye color changes to express emotions more effectively, which is known to enhance social perception among users \[[108](/article/10.1007/s12369-025-01323-5#ref-CR108 "Dou X, Yan L, Wu K, Niu J (2022) Effects of voice and lighting color on the social perception of home healthcare robots. Appl Sci 12(23):12191.
https://doi.org/10.3390/app122312191
")\]. These designs intentionally embed meaningful narratives or emotional triggers, facilitating more humanized interactions.One core principle of contextual design is to deliver the right support at the right time and in the right way. The just-in-time adaptive intervention is a design strategy that focuses on providing adaptive support tailored to the user’s continuously changing states and contexts [[109](/article/10.1007/s12369-025-01323-5#ref-CR109 "Nahum-Shani I, Smith SN, Spring BJ, Collins LM, Witkiewitz K, Tewari A, Murphy SA (2018) Just-in-time adaptive interventions (JITAIs) in mobile health: key components and design principles for ongoing health behavior support. ann. behav. Med. https://doi.org/10.1007/s12160-016-9830-8
")\]. Following this principle, SARs continuously monitor user activity and provide real-time contextual feedback through non-verbal channels such as biofeedback, haptic responses, and emotional signals. Developers must thoughtfully design these adaptive multimodal cues, carefully determining the timing, method, and circumstances under which they are delivered. This process also requires iterative refinement of the SAR design based on consistent feedback from users, which is discussed below in detail. Moreover, effective contextual design should consistently address authentic user and stakeholder needs and preferences, underscoring the value of participatory design throughout the entire design process.4.6 Participatory Design
Integrating users’ perspectives in designing SARs may encompass gathering users’ opinions through interviews, surveys, or observing how they interact with the robots. However, the most desirable way for user-centered SAR development is to let its users (or stakeholders including therapists, nurses, social workers, or any others connected to the product usage) participate in the development process so that they can actively contribute to the product design from the beginning. Participatory design emphasizes co-work between developers and users/stakeholders throughout the entire design process. The designers and stakeholders continuously interact and collaborate to plan the project, identify users’ goals and values through structured interaction tools, and design prototypes that technically address their identified needs and goals [110, 111]. Participatory design can play a crucial role in applying the user-centered approach as it not only enhances user empowerment and accessibility but also accelerates the democratization of product development by fostering user-led products and bottom-up development processes [[112](/article/10.1007/s12369-025-01323-5#ref-CR112 "Benz C, Scott-Jeffs W, McKercher KA, Welsh M, Norman R, Hendrie D et al. (2024) Community-based participatory-research through co-design: supporting collaboration from all sides of disability. Res Involv Engagem 10(1):47. https://doi.org/10.1186/s40900-024-00573-3
"), [113](/article/10.1007/s12369-025-01323-5#ref-CR113 "Björgvinsson E, Ehn P, Hillgren PA (2010) Participatory design and “democratizing innovation”. In Proc 11th Biennial Participatory Design Conference, pp 41–50.
https://doi.org/10.1145/1900441.1900448
")\].Several studies have incorporated participatory design for SARs. Šabanović et al. (2015) attempted participatory design workshops to design a user-centered SAR for older adults with depression [[114](/article/10.1007/s12369-025-01323-5#ref-CR114 "Šabanović S, Chang WL, Bennett CC, Piatt JA, Hakken D (2015) A robot of my own: participatory design of socially assistive robots for independently living older adults diagnosed with depression. In: Human aspects of IT for the aged population. Springer, pp 104–114. https://doi.org/10.1007/978-3-319-20892-3_11
")\]. Winkle et al. (2020) integrated participatory design into the development of SAR, proposing a mutual shaping model that involves various stakeholders as a focus group in the development process \[[115](/article/10.1007/s12369-025-01323-5#ref-CR115 "Winkle K, Caleb-Solly P, Turton A, Bremner P (2020) Mutual shaping in the design of socially assistive robots: a case study on social robots for therapy. Int J Soc robot 12:847–866.
https://doi.org/10.1007/s12369-019-00536-9
")\]. In this model, the mutual shaping process is divided into three levels: first, the focus group accepts and reflects the stakeholders’ perspectives so that they can have an impact on improving the SARs’ acceptance and engagement; second, the focus group participates in the design and use of the robot to improve its effectiveness and reduce risks of negative consequences in the deployment process; and finally, the process of participatory design and mutual shaping is iterated to ensure its long-term effectiveness in real-world practices.An important principle of participatory design is to integrate user information based on the ethnographic approach, which gathers the users’ perspectives in the real field of usage. Ethnographic research observes users in the field for a deeper understanding of the product’s real-world usage, thereby deducing qualitative conclusions from the research questions [[116](/article/10.1007/s12369-025-01323-5#ref-CR116 "Leslie M, Paradis E, Gropper MA, Kitto S, Reeves S, Pronovost P (2017) An ethnographic study of health information technology use in three intensive care units. Health Serv Res 52(4):1330–1348. https://doi.org/10.1111/1475-6773.12466
")\]. Developers have the opportunity to closely observe users in real-world settings, allowing them to gather rich, qualitative insights into user experiences. This direct observation process offers a deeper understanding of how the product is used in practice. From our perspective, it provides an advantage over traditional quantitative methods, such as surveys or large-sample data analysis. Large-sample quantitative research based on the nomothetic approach often relies on group-level analysis, which tends to neglect individual-level dimensions that are essential for incorporating personalization and user-centered principles. In contrast, the ethnographic approach aligns well with the idiographic approach, which values individual uniqueness rather than mere group-level differences. Mast et al. (2012) adopted an ethnographic approach to investigate elderly people’s homes and teleassistance centers, and their in-depth fieldwork contributed to a better understanding of the potential challenges and attitudes related to the deployment of robots in these settings \[[117](/article/10.1007/s12369-025-01323-5#ref-CR117 "Mast M, Burmester M, Krüger K, Fatikow S, Arbeiter G, Graf B et al. (2012) User-centered design of a dynamic-autonomy remote interaction concept for manipulation-capable robots to assist elderly people in the home. J Hum Rob Interact 1(1):96–118.
https://doi.org/10.5898/JHRI.1.1.Mast
")\].4.7 Long-Term Interactions
Design strategies for SARs vary according to their intended objectives. While short-term or non-intensive interactions may be sufficient in many cases, we suggest that developers also foster long-term interactions with users if they aim to employ SARs for extended engagements, such as daily care or companionship [[118](/article/10.1007/s12369-025-01323-5#ref-CR118 "Laban G, Kappas A, Morrison V, Cross ES (2024) Building long-term human-robot relationships: examining disclosure, perception and well-being across time. Int J Soc robot 16(5):1–27. https://doi.org/10.1007/s12369-023-01076-z
")\]. Perceived intimacy with robots can lead to emotional comfort for users, thereby supporting treatment objectives \[[119](/article/10.1007/s12369-025-01323-5#ref-CR119 "Prescott TJ, Robillard JM (2021) Are friends electric? The benefits and risks of human-robot relationships. iScience 24(1):101993.
https://doi.org/10.1016/j.isci.2020.101993
")\]. This is particularly beneficial in contexts that require 24/7 support or continuous, persistent care, such as companion robots, robots for intensive psychotherapy, or at-home care robots \[[41](/article/10.1007/s12369-025-01323-5#ref-CR41 "Szondy M, Fazekas P (2024) Attachment to robots and therapeutic efficiency in mental health. Front psychol 15:1347177.
https://doi.org/10.3389/fpsyg.2024.1347177
")\]. Generally, the concept of intimacy or attachment to robots draws from standard relationship science ideas, where users experience responsiveness, support, and care—perceiving the robot as a “safe haven” \[[120](/article/10.1007/s12369-025-01323-5#ref-CR120 "Birnbaum GE, Mizrahi M, Hoffman G, Reis HT, Finkel EJ, Sass O (2016) What robots can teach us about intimacy: the reassuring effects of robot responsiveness to human disclosure. Comput Hum Behav 63:416–423.
https://doi.org/10.1016/j.chb.2016.05.064
")\]. Additionally, promoting “rapport,” understood here as the user’s sense of comfort, familiarity, and trust when interacting with the robot counselor, may enhance the effectiveness of socially assistive robots, especially when they function in counseling roles \[[121](/article/10.1007/s12369-025-01323-5#ref-CR121 "Lin TH, Dinner H, Leung TL, Mutlu B, Trafton JG, Sebo S (2025) Connection-coordination rapport (CCR) scale: a dual-factor scale to measure human-robot rapport. arXiv preprint arXiv: 2501.11887.
https://doi.org/10.48550/arXiv.2501.11887
")\].However, developers must carefully consider potential risks and ethical concerns associated with prolonged human-robot interactions [[119](/article/10.1007/s12369-025-01323-5#ref-CR119 "Prescott TJ, Robillard JM (2021) Are friends electric? The benefits and risks of human-robot relationships. iScience 24(1):101993. https://doi.org/10.1016/j.isci.2020.101993
"), [122](/article/10.1007/s12369-025-01323-5#ref-CR122 "De Graaf MM (2016) An ethical evaluation of human-robot relationships. Int J Soc robot 8:589–598.
https://doi.org/10.1007/s12369-016-0368-5
")\]. Intense interactions, as opposed to superficial ones, inherently increase ethical risks, such as overdependence on robots \[[119](/article/10.1007/s12369-025-01323-5#ref-CR119 "Prescott TJ, Robillard JM (2021) Are friends electric? The benefits and risks of human-robot relationships. iScience 24(1):101993.
https://doi.org/10.1016/j.isci.2020.101993
")\]. Explicitly designing robots for intimate interactions may introduce ethical issues related to deception, leading to unsafe or unintended consequences \[[123](/article/10.1007/s12369-025-01323-5#ref-CR123 "Borenstein J, Arkin R (2019) Robots, ethics, and intimacy: the need for scientific research. In: Berkich D, d’Alfonso M (eds) On the cognitive, ethical, and scientific dimensions of artificial intelligence. Philosophical studies series, vol 134. Springer, Cham.
https://doi.org/10.1007/978-3-030-01800-9_16
")\]. Consequently, long-term interactions with robots require a significantly more careful and ethically informed design compared to short-term engagements. The topic of long-term human-robot interaction remains debatable and necessitates further rigorous scientific investigation \[[123](/article/10.1007/s12369-025-01323-5#ref-CR123 "Borenstein J, Arkin R (2019) Robots, ethics, and intimacy: the need for scientific research. In: Berkich D, d’Alfonso M (eds) On the cognitive, ethical, and scientific dimensions of artificial intelligence. Philosophical studies series, vol 134. Springer, Cham.
https://doi.org/10.1007/978-3-030-01800-9_16
")\]. Here, two central aspects of long-term human-robot interaction are explored: trust and role-based interactions.4.7.1 Trust-Building
Fostering trust in SARs can help sustain user interactions and maintain ongoing relationships with users. Establishing trust is a complex process, however, considering its subjective (user-perceived) nature and impacts of both controllable factors (such as robot design and performance) and uncontrollable factors (such as user characteristics and contexts). Trust is a multifaceted construct, shaped by both internal robot features and situational factors, including technical safety, functional reliability, sensitivity to users’ emotions, and the capacity for appropriate communication. Users’ perceived trust is impacted by the robot’s trustworthiness and contextual or human-related factors, which can cause mismatches between actual robot capabilities and perceived trust [[124](/article/10.1007/s12369-025-01323-5#ref-CR124 "Kok BC, Soh H (2020) Trust in robots: challenges and opportunities. Curr Robot Rep 1(4):297–309. https://doi.org/10.1007/s43154-020-00029-y
"), [125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239.
https://doi.org/10.1016/j.neubiorev.2019.07.014
")\]. When the robot ensures security, privacy, and safety, performs reliably as expected, demonstrates awareness of users’ needs and social contexts and provides clear and transparent feedback regarding their capabilities and intentions, these elements collectively shape the user’s overall sense of trust, which in turn can enhance sustained and meaningful long-term interactions with the robot \[[124](/article/10.1007/s12369-025-01323-5#ref-CR124 "Kok BC, Soh H (2020) Trust in robots: challenges and opportunities. Curr Robot Rep 1(4):297–309.
https://doi.org/10.1007/s43154-020-00029-y
")\]. However, due to the multidimensional nature of trust, optimizing trust levels can be challenging, and excessively high trust may increase user vulnerability and raise ethical concerns \[[124](/article/10.1007/s12369-025-01323-5#ref-CR124 "Kok BC, Soh H (2020) Trust in robots: challenges and opportunities. Curr Robot Rep 1(4):297–309.
https://doi.org/10.1007/s43154-020-00029-y
"), [125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239.
https://doi.org/10.1016/j.neubiorev.2019.07.014
")\].Nevertheless, it remains essential to design SARs and environments carefully to foster appropriate trust. Reliable and trustworthy SARs should be safe, consistent, responsive, and comforting, which enhances user competence and quality of life. Initial reliance on robots develops from users’ dispositions, contexts, and initial experiences; trust builds over time as users observe robot performance and behavior [[124](/article/10.1007/s12369-025-01323-5#ref-CR124 "Kok BC, Soh H (2020) Trust in robots: challenges and opportunities. Curr Robot Rep 1(4):297–309. https://doi.org/10.1007/s43154-020-00029-y
"), [126](/article/10.1007/s12369-025-01323-5#ref-CR126 "Hoff KA, Bashir M (2015) Trust in automation: integrating empirical evidence on factors that influence trust. Hum Factors 57(3):407–434.
https://doi.org/10.1177/0018720814547570
")\]. SARs require technical completeness and personalization capabilities, as well as accurately understanding users’ emotions and intentions to foster trust effectively \[[125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239.
https://doi.org/10.1016/j.neubiorev.2019.07.014
")\]. Maintaining reliable and predictable performance without system errors is crucial, as even minor malfunctions can lead to dissatisfaction \[[94](/article/10.1007/s12369-025-01323-5#ref-CR94 "Vandemeulebroucke T, de Casterlé BD, Gastmans C (2018) How do older adults experience and perceive socially assistive robots in aged care: a systematic review of qualitative evidence. Aging Ment Health 22(2):149–167.
https://doi.org/10.1080/13607863.2017.1286455
"), [127](/article/10.1007/s12369-025-01323-5#ref-CR127 "Hancock PA, Kessler TT, Kaplan AD, Brill JC, Szalma JL (2021) Evolving trust in robots: specification through sequential and comparative meta-analyses. Hum Factors 63(7):1196–1229.
https://doi.org/10.1177/0018720820922080
"), [128](/article/10.1007/s12369-025-01323-5#ref-CR128 "Wang X, Zhou R, Zhang R (2020) The impact of expectation and disconfirmation on user experience and behavior intention. In: Design, user experience, and usability. Interaction design: 9th int conf, DUXU 2020. Springer, Cham, pp 464–475.
https://doi.org/10.1007/978-3-030-49713-2_32
")\].Moreover, it is important to make AI or adaptive decision-making systems in SARs explainable to users. To improve transparency, they should enable users to clearly understand how decisions are made, thereby fostering trust and enhancing user performance [[129](/article/10.1007/s12369-025-01323-5#ref-CR129 "Leichtmann B, Humer C, Hinterreiter A, Streit M, Mara M (2023) Effects of explainable artificial intelligence on trust and human behavior in a high-risk decision task. Comput Hum Behav 139:107539. https://doi.org/10.1016/j.chb.2022.107539
"), [130](/article/10.1007/s12369-025-01323-5#ref-CR130 "Mahbooba B, Timilsina M, Sahal R, Serrano M (2021) Explainable artificial intelligence (XAI) to enhance trust management in intrusion detection systems using decision tree model. Complexity 6634811.
https://doi.org/10.1155/2021/6634811
")\]. The robot’s appearance, perceived personality, and interaction style should also be carefully designed, as human-like, familiar appearances and natural interactions can promote greater trust \[[131](/article/10.1007/s12369-025-01323-5#ref-CR131 "Lockey S, Gillespie N, Holm D, Someh IA (2021) A review of trust in artificial intelligence: challenges, vulnerabilities and future directions. Proc 54th Hawaii Int Conf Syst Sci 5463–5472")\]. When users trust SARs, they can provide transparent feedback, evoke positive emotions, and motivate users toward mental health and quality-of-life improvements \[[125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239.
https://doi.org/10.1016/j.neubiorev.2019.07.014
"), [126](/article/10.1007/s12369-025-01323-5#ref-CR126 "Hoff KA, Bashir M (2015) Trust in automation: integrating empirical evidence on factors that influence trust. Hum Factors 57(3):407–434.
https://doi.org/10.1177/0018720814547570
")\]. However, developers should also be mindful of the delicate balance between trust, user vulnerability, and autonomy. It is crucial to foster user attachment and trust without letting their trust compromise user autonomy. Nevertheless, finding the “sweet spot” of trust becomes increasingly challenging, and the rapid development of LLM chatbots has further amplified concerns about overtrust and its associated ethical implications, which will be discussed in more detail below.4.7.2 Role Setting
For effective long-term interactions, it is also important to clearly define the roles, identities, or relationship settings of the robots to enhance perceived intimacy and sustained interaction [[41](/article/10.1007/s12369-025-01323-5#ref-CR41 "Szondy M, Fazekas P (2024) Attachment to robots and therapeutic efficiency in mental health. Front psychol 15:1347177. https://doi.org/10.3389/fpsyg.2024.1347177
")\]. A common approach for the role setting of SARs involves mimicking real-world human relationships. This process may include designing robots with human-like (or pet-like) appearances and incorporating interactive behaviors such as handshaking or stroking \[[132](/article/10.1007/s12369-025-01323-5#ref-CR132 "Ferrario A, Sedlakova J, Trachsel M (2024) The role of humanization and robustness of large language models in conversational artificial intelligence for individuals with depression: a critical analysis. JMIR Ment Health 11:e56569.
https://doi.org/10.2196/56569
"), [133](/article/10.1007/s12369-025-01323-5#ref-CR133 "Spatola N, Cherif E (2023) Spontaneous humanization of robots in passive observation of human-robot interaction: a path toward ethical consideration and human-robot cooperation. Comput Hum Behav Artif Humans 1(2):100012.
https://doi.org/10.1016/j.chbah.2023.100012
")\]. However, robot morphology alone may not ensure sustained user engagement or long-term usage \[[134](/article/10.1007/s12369-025-01323-5#ref-CR134 "Bradwell HL, Winnington R, Thill S, Jones RB (2021) Morphology of socially assistive robots for health and social care: a reflection on 24 months of research with anthropomorphic, zoomorphic and mechanomorphic devices. In 2021 30th IEEE Int Conf Robot Human Interact Commun (RO-MAN), pp 376–383.
https://doi.org/10.1109/RO-MAN50785.2021.9515446
"), [135](/article/10.1007/s12369-025-01323-5#ref-CR135 "Randall N, Bennett CC, Šabanović S, Nagata S, Eldridge L, Collins S, Piatt JA (2019) More than just friends: in-home use and design recommendations for sensing socially assistive robots (SARs) by older adults with depression. Paladyn J Behav robot 10(1):237–255.
https://doi.org/10.1515/pjbr-2019-0020
")\]. For effective role setting, SARs should not only mimic human-like (or animal-like) appearances but also carefully develop verbal, non-verbal, visual, and relational cues to authentically replicate human relationships during interactions \[[132](/article/10.1007/s12369-025-01323-5#ref-CR132 "Ferrario A, Sedlakova J, Trachsel M (2024) The role of humanization and robustness of large language models in conversational artificial intelligence for individuals with depression: a critical analysis. JMIR Ment Health 11:e56569.
https://doi.org/10.2196/56569
"), [136](/article/10.1007/s12369-025-01323-5#ref-CR136 "Van Straten CL, Peter J, Kühne (2020) Child-robot relationship formation: a narrative review of empirical research. Int J Soc robot 12(2):325–344.
https://doi.org/10.1007/s12369-019-00569-0
")\]. Through careful role setting, SARs can imitate close relationships, such as companions, those between family members, or therapists and patients, to deepen user engagement \[[137](/article/10.1007/s12369-025-01323-5#ref-CR137 "Grodniewicz JP, Hohol M (2023) Waiting for a digital therapist: three challenges on the path to psychotherapy delivered by artificial intelligence. Front Psychiatry 14:1190084.
https://doi.org/10.3389/fpsyt.2023.1190084
")\].An illustrative strategy for this approach is demonstrated in Hyodol, a SAR developed in South Korea for the mental healthcare of elderly people through behavioral activation. Hyodol is designed to resemble a child and addresses its users as “grandma” or “grandpa,” allowing users to touch its hands, ears, or body and responding to their physical affection. Research on Hyodol shows that users perceive it as their real grandchild, expressing intimacy through actions like cuddling it or covering it with blankets [[138](/article/10.1007/s12369-025-01323-5#ref-CR138 "Kim SK, Jang JW, Hwang YS, Lee OE, Jo HS (2023) Investigating the effectiveness of socially assistive robot on depression and cognitive functions of community dwelling older adults with cognitive impairments. Assist Technol 1–9. https://doi.org/10.1080/10400435.2023.2237554
")\]. Likewise, Wang et al.’s (2024) pillow robot is designed to simulate a parental role to effectively aid children’s sleep \[[139](/article/10.1007/s12369-025-01323-5#ref-CR139 "Wang Y, Wang Q, Sun X, Cao C, Sun R (2024) Sleep elf: pillow robot that pats and sings child to sleep mimicking a parent. In Companion of the 2024 ACM/IEEE Int Conf Hum-Robot Interact, pp 1100–1104.")\]. Although this robot does not appear like humans, it communicates with children by playing parent-like lullabies and gently patting them with a robotic arm until they fall asleep.However, researchers also argue that making robots look and act like humans is not always an ideal strategy. Beyond well-known issues such as the uncanny valley effect [[40](/article/10.1007/s12369-025-01323-5#ref-CR40 "Mori M, MacDorman KF, Kageki N (2012) The uncanny valley [from the field]. IEEE Robot Autom Mag 19(2):98–100. https://doi.org/10.1109/MRA.2012.2192811
")\], even when human-like robots are effectively accepted by users, their presence may not always yield desirable outcomes. Concerns have been raised about the potential for humanized robots to distort user interactions by fostering specious relationships, whereby users may develop unrealistic expectations or become over-dependent on robots \[[39](/article/10.1007/s12369-025-01323-5#ref-CR39 "Giger JC, Piçarra N, Alves-Oliveira P, Oliveira R, Arriaga P (2019) Humanization of robots: is it really such a good idea? Hum Behav Emerg Technol 1(2):111–123.
https://doi.org/10.1002/hbe2.147
"), [140](/article/10.1007/s12369-025-01323-5#ref-CR140 "Robert LP (2017) The growing problem of humanizing robots. Int Robot Autom J 3(1):00043.
https://doi.org/10.15406/iratj.2017.03.00043
")\]. This issue is especially prominent for recent chatbots based on LLMs that perfectly mimic human conversations, as discussed in the previous section. When designing SARs to simulate human-like relationships, we suggest avoiding relationship settings that may lead to overdependence, such as those resembling counselors or healthcare providers. Alternatively, developers can mimic the intimacy found in relationships with animals by designing robots to resemble plush toys, as mentioned earlier. No matter which types of relationships they mimic, it is important that SARs mimicking intimate relationships should lead to a perceived therapeutic alliance that lets users feel that the robots take care of them and are predictable \[[137](/article/10.1007/s12369-025-01323-5#ref-CR137 "Grodniewicz JP, Hohol M (2023) Waiting for a digital therapist: three challenges on the path to psychotherapy delivered by artificial intelligence. Front Psychiatry 14:1190084.
https://doi.org/10.3389/fpsyt.2023.1190084
")\], which indicates the necessity of building trust in SARs’ independent user interactions.5 Evaluation of SARs
Once the initial SAR design process is completed, the prototyped SAR should be tested to determine whether the product is operating according to its objectives and effectively contributing to users’ mental health. The evaluation of SARs requires diverse and multidimensional criteria, drawing from multidisciplinary fields including robotics, human-computer interaction, and digital healthcare. Feil-Seifer et al. (2007) proposed a set of diverse benchmarks for evaluating SARs, focusing on safety and privacy, the promotion of autonomy, adaptive social interactions, the impact on users and caregivers, and whether laboratory-designed features can be generalized to real-world settings [[74](/article/10.1007/s12369-025-01323-5#ref-CR74 "Feil-Seifer D, Skinner K, Matarić MJ (2007) Benchmarks for evaluating socially assistive robotics. Interact Stud 8(3):423–439. https://doi.org/10.1075/is.8.3.07fei
")\]. Sim and Loo (2015) introduced various evaluation models of SARs through a comprehensive literature review, based on their aim of measurement (e.g., empathy, user acceptance, friendship, behavioral adaptation, human-like features) and methodology (e.g., self-report, behavioral, physiological, task performances) \[[141](/article/10.1007/s12369-025-01323-5#ref-CR141 "Sim DYY, Loo CK (2015) Extensive assessment and evaluation methodologies on assistive social robots for modelling human-robot interaction-A review. Inf Sci 301:305–344.
https://doi.org/10.1016/j.ins.2014.12.017
")\]. These criteria can be broadly categorized into three dimensions: the technical completeness of robotics, the reliability and effectiveness of their social interaction features, and their psychological impact on users.To recapitulate, the evaluation of SARs encompasses assessing whether these platforms function effectively (i) to enhance user experiences, such as autonomy, competence, engagement, satisfaction, enjoyment, and trust; (ii) to ensure user safety or perceived safety, encompassing ethical considerations, preservation of user autonomy, and fostering trust; and (iii) to improve users’ mental health and psychological well-being, especially within real-world settings. A variety of methodologies can be employed for the holistic evaluation model of SARs, including subjective or self-reported measures, behavioral data and usage metrics, physiological indicators, and qualitative approaches such as interviews (Fig. 3).
Fig. 3
The multimodal and holistic evaluation framework of SARs
5.1 Primary Criteria
The primary evaluation criteria for SAR effectiveness in mental healthcare typically focus on endpoint outcomes, assessing whether the product successfully achieved its objectives of improving users’ mental health, psychological well-being, or quality of life. The main purpose of this criterion is to demonstrate the effectiveness of SARs in treatment or mental healthcare, particularly when seeking approval for a digital therapeutic or healthcare product, rather than conducting independent or investigator-initiated studies [[142](/article/10.1007/s12369-025-01323-5#ref-CR142 "Guo C, Ashrafian H, Ghafur S, Fontana G, Gardner C, Prime M (2020) Challenges for the evaluation of digital health solutions-a call for innovative evidence generation approaches. NPJ Digit Med 3(1):110. https://doi.org/10.1038/s41746-020-00314-2
")\]. Such evaluations are often conducted to obtain clearance from the US Food and Drug Administration (FDA) or its international counterparts as medical devices, including _PARO_, which has successfully received FDA approval as a biofeedback device \[[143](/article/10.1007/s12369-025-01323-5#ref-CR143 "Shibata T, Hung L, Petersen S, Darling K, Inoue K, Martyn K et al. (2021) PARO as a biofeedback medical device for mental health in the COVID-19 Era. Sustainability 13(20):11502.
https://doi.org/10.3390/su132011502
")\]. However, most current SARs do not seek formal medical device approval, positioning themselves instead primarily as companions, educational resources, or service-oriented platforms.To measure the primary efficacy, validated clinical assessment tools for mental health symptoms or quality of life can be utilized, which are self-reported by users or assessed by clinicians. Research designs for clinical trials can be used to assess the efficacy of robots (e.g., their relative superiority/non-inferiority compared to the control group/alternative treatment) or cost-benefit analyses (e.g., reducing caregiving costs or alleviating the burden on human caregivers compared to their costs). However, evidence from controlled settings may provide limited information about the effectiveness in real-world settings, unless the product’s contexts, implementation methods, and potential benefits are relatively clear [[144](/article/10.1007/s12369-025-01323-5#ref-CR144 "Murray E, Hekler EB, Andersson G, Collins LM, Doherty A, Hollis C et al. (2016) Evaluating digital health interventions: key questions and approaches. Am J Prev Med 51(5):843–851. https://doi.org/10.1016/j.amepre.2016.06.008
")\]. Therefore, for developers, adopting a user-centered approach for evaluations is crucial, as adequately incorporating users’ perspectives into evaluations can ensure that measured effectiveness can be more reliably generalized to real-world settings. We also recommend that researchers consider applying a qualitative or mixed-methods design when evaluating the effectiveness of SARs, as this approach combines objective evidence with users’ perspectives, offering thorough understandings and valuable insights for improving future iterations of the product \[[145](/article/10.1007/s12369-025-01323-5#ref-CR145 "Asl AM, Ulate MM, Franco Martin M, van der Roest H (2022) Methodologies used to study the feasibility, usability, efficacy, and effectiveness of social robots for elderly adults: scoping review. J Med Internet Res 24(8):e37434.
https://doi.org/10.2196/37434
")\]. Chen et al. (2020) integrated qualitative interviews with quantitative, self-reported scales to evaluate the effectiveness of social robot interventions. In addition to the improvements in quantitative measures, their findings on interviews demonstrated that participants perceived the robot as a genuine companion, leading to enhanced social interactions and improved mental well-being \[[146](/article/10.1007/s12369-025-01323-5#ref-CR146 "Chen SC, Moyle W, Jones C, Petsky H (2020) A social robot intervention on depression, loneliness, and quality of life for Taiwanese older adults in long-term care. Int PsychogeriAtr 32(8):981–991.
https://doi.org/10.1017/S1041610220000459
")\].For a comprehensive evaluation, the effectiveness of SARs should be assessed using multidimensional and holistic criteria, including but not limited to overall user experiences, user autonomy and competence, and observable changes in mental health and psychological well-being in real-life contexts. Therefore, especially for the purposes of internal validation or product improvement, developers should prioritize a diverse set of criteria such as safety, user experiences, and trust, as elaborated in the subsequent subsections, rather than relying solely on primary criteria. Moreover, incorporating follow-up evaluations or longitudinal studies is essential to accurately gauge the long-term effectiveness of SARs, which can provide deeper insights into users’ sustained experiences and changes throughout extended periods of interaction.
5.2 User Experiences
User experiences are germane to the main objectives of SARs and should be considered the most important aspect of SAR evaluation, perhaps more than its “primary” measures. The evaluation of user experiences begins with assessing the usability of the product and extends to user engagement, finally focusing on the trust between users and SARs. In traditional human-computer interaction, usability has often been evaluated from the perspective of the robot, focusing on its effectiveness, efficiency in interactions, and user satisfaction (e.g., preferences, ease of use, user attitudes/perception) to identify the robot’s functional capabilities [[147](/article/10.1007/s12369-025-01323-5#ref-CR147 "Hornbæk K (2006) Current practice in measuring usability: challenges to usability studies and research. Int J Hum Comput Stud 64(2):79–102. https://doi.org/10.1016/j.ijhcs.2005.06.002
")\]. In contrast, user engagement focuses on the interaction process between users and robots, such as how users feel during interactions and how much the users are involved in the interactions between them. User engagement focuses on not only user engagement with SARs but also engagement with behavior change and the socio-contextual factors around them \[[35](/article/10.1007/s12369-025-01323-5#ref-CR35 "Yardley L, Spring BJ, Riper H, Morrison LG, Crane DH, Curtis K et al. (2016) Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med 51(5):833–842.
https://doi.org/10.1016/j.amepre.2016.06.015
")\]. Trust should also be evaluated with regard to the long-term interactions between users and SARs, as it plays a crucial role in ensuring sustained use of robots and ultimately contributes to improvements in users’ mental health and quality of life through continuous interactions \[[125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239.
https://doi.org/10.1016/j.neubiorev.2019.07.014
")\]. Trust is affected by even more complex aspects, encompassing user-related factors and diverse contextual factors, which cannot be explained merely by robot-related factors of the “trustworthiness” of SARs \[[127](/article/10.1007/s12369-025-01323-5#ref-CR127 "Hancock PA, Kessler TT, Kaplan AD, Brill JC, Szalma JL (2021) Evolving trust in robots: specification through sequential and comparative meta-analyses. Hum Factors 63(7):1196–1229.
https://doi.org/10.1177/0018720820922080
")\].Typically, usability and user engagement are initially inspected by developers through cognitive walkthrough or heuristic evaluation and then objectively tested through user testing or task completion/performance analysis. In many cases, behavioral data (e.g., time/duration, frequency/patterns of use, user inputs) are used to evaluate a digital health product’s usability and user engagement during interactions. For SARs, a comprehensive and dynamic evaluation model can be used, including attention, eye-tracking/body movements, and rhythm of interaction [[148](/article/10.1007/s12369-025-01323-5#ref-CR148 "Anzalone SM, Boucenna S, Ivaldi S, Chetouani M (2015) Evaluating the engagement with social robots. Int J Soc robot 7:465–478. https://doi.org/10.1007/s12369-015-0298-7
")\]. In addition to relatively simple indicators of user engagement such as the frequency of user inputs and interactions \[[149](/article/10.1007/s12369-025-01323-5#ref-CR149 "Daley K, Hungerbuehler I, Cavanagh K, Claro HG, Swinton PA, Kapps M (2020) Preliminary evaluation of the engagement and effectiveness of a mental health chatbot. Front Digit Health 2:576361.
https://doi.org/10.3389/fdgth.2020.576361
"), [150](/article/10.1007/s12369-025-01323-5#ref-CR150 "Perugia G, Rodríguez-Martín D, Boladeras MD, Mallofré AC, Barakova E, Rauterberg M (2018) Quantity of movement as a measure of engagement for dementia: the influence of motivational disorders. Am J Alzheimers Dis Other demen 33(2):112–121.
https://doi.org/10.1177/1533317517739700
")\], measures like gaze patterns, backchannel responses (e.g., nodding, verbal acknowledgments such as “uh-huh”), facial expressions, and body postures can effectively serve as attentional or affective markers of engagement \[[151](/article/10.1007/s12369-025-01323-5#ref-CR151 "Perugia G, Díaz-Boladeras M, Catala-Mallofré A, Barakova EI, Rauterberg M (2020) ENGAGE-DEM: a model of engagement of people with dementia. IEEE Trans Affect Comput 13(2):926–943.
https://doi.org/10.1109/TAFFC.2020.2980275
")\]. It is crucial to select behavioral markers that are specifically relevant to the targeted symptoms addressed by SARs. For children with autism or attention-deficit/hyperactivity disorder, behavioral markers related to social skills can be particularly relevant for engagement, such as joint attention, social communication, or eye-contact \[[152](/article/10.1007/s12369-025-01323-5#ref-CR152 "Puglisi A, Caprì T, Pignolo L, Gismondo S, Chilà P, Minutoli R et al. (2022) Social humanoid robots for children with autism spectrum disorders: a review of modalities, indications, and pitfalls. Children 9(7):953.
https://doi.org/10.3390/children9070953
"), [153](/article/10.1007/s12369-025-01323-5#ref-CR153 "Rakhymbayeva N, Amirova A, Sandygulova A (2021) A long-term engagement with a social robot for autism therapy. Front Robot AI 8:669972.
https://doi.org/10.3389/frobt.2021.669972
")\]. In the context of individuals with schizophrenia, evaluating synchronization with the robot’s behaviors can provide meaningful insights \[[154](/article/10.1007/s12369-025-01323-5#ref-CR154 "Cohen L, Khoramshahi M, Salesse RN, Bortolon C, Słowiński P, Zhai C et al. (2017) Influence of facial feedback during a cooperative human-robot task in schizophrenia. Sci Rep 7(1):15023.
https://doi.org/10.1038/s41598-017-14773-3
")\].Self-reports are a commonly used method to identify perceived usability and user engagement. Self-assessment of user experiences is usually based on the United Theory of Acceptance and Use of Technology [[155](/article/10.1007/s12369-025-01323-5#ref-CR155 "Venkatesh V, Morris MG, Davis GB, Davis FD (2003) User acceptance of information technology: toward a unified view. Mis Q 27(3):425–478. https://doi.org/10.2307/30036540
")\], and there are also established measures for usability \[[156](/article/10.1007/s12369-025-01323-5#ref-CR156 "Bangor A, Kortum PT, Miller JT (2008) An empirical evaluation of the system usability scale. Int J Hum Comput Interact 24(6):574–594.
https://doi.org/10.1080/10447310802205776
")\], user engagement \[[35](/article/10.1007/s12369-025-01323-5#ref-CR35 "Yardley L, Spring BJ, Riper H, Morrison LG, Crane DH, Curtis K et al. (2016) Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med 51(5):833–842.
https://doi.org/10.1016/j.amepre.2016.06.015
")\], and trust \[[125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239.
https://doi.org/10.1016/j.neubiorev.2019.07.014
")\], which allows developers to use self-reported questionnaires to measure perceived user engagement. Self-report measures offer convenience due to their ease of implementation through straightforward surveys, which can also be administered online. When established measures are employed, self-reports are generally considered reliable and valid. However, they may lack the depth necessary to capture the full complexity of latent psychological constructs, such as user autonomy or competence, in which case they are better assessed through indirect assessment rather than explicit questioning. Furthermore, traditional self-report surveys are often not conducted in real-time, which may introduce recall bias and compromise the accuracy of subjective user responses \[[35](/article/10.1007/s12369-025-01323-5#ref-CR35 "Yardley L, Spring BJ, Riper H, Morrison LG, Crane DH, Curtis K et al. (2016) Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med 51(5):833–842.
https://doi.org/10.1016/j.amepre.2016.06.015
")\]. To address this limitation, the ecological momentary assessment (EMA) can be employed, which captures user responses repeatedly and in real-time during actual use, thereby enhancing ecological validity and more accurately reflecting users’ immediate experiences \[[23](/article/10.1007/s12369-025-01323-5#ref-CR23 "Bennett CC, Stanojević C, Šabanović SA, Piatt J, Kim S (2021) When no one is watching: ecological momentary assessment to understand situated social robot use in healthcare. In Proceedings of the 9th International Conference on Human-Agent Interaction, pp 245–251.
https://doi.org/10.1145/3472307.3484670
"), [157](/article/10.1007/s12369-025-01323-5#ref-CR157 "Shiffman S, Stone AA, Hufford MR (2008) Ecological momentary assessment. Annu Rev Clin psychol 4(1):1–32.
https://doi.org/10.1146/annurev.clinpsy.3.022806.091415
")\].Physiological measures can serve as effective circumplex indicators of affect associated with user experiences. The physiological properties capture changes in the body’s central or peripheral nervous system considered to correlate with mental and emotional states. They reveal how the body’s internal responses fluctuate in tandem with psychological processes. Common physiological indicators include heart rate [measured through electrocardiography (ECG) or photoplethysmography (PPG)], brain activity measured via electroencephalography (EEG), and electrodermal activity assessed through galvanic skin response (GSR). These signals can be measured by laboratory sensing devices or commercial products such as smartphones or wearable devices. The physiological signals are processed and analyzed to interpret stress levels, emotional valence, arousal, calmness, and excitement [[141](/article/10.1007/s12369-025-01323-5#ref-CR141 "Sim DYY, Loo CK (2015) Extensive assessment and evaluation methodologies on assistive social robots for modelling human-robot interaction-A review. Inf Sci 301:305–344. https://doi.org/10.1016/j.ins.2014.12.017
"), [151](/article/10.1007/s12369-025-01323-5#ref-CR151 "Perugia G, Díaz-Boladeras M, Catala-Mallofré A, Barakova EI, Rauterberg M (2020) ENGAGE-DEM: a model of engagement of people with dementia. IEEE Trans Affect Comput 13(2):926–943.
https://doi.org/10.1109/TAFFC.2020.2980275
")\]. Klęczek et al. (2024), for instance, combined EEG and GSR measurements with self-reported data on expectancies, enjoyment, satisfaction, and trust to evaluate user arousal and emotional valence when interacting with a mental health coaching robot \[[158](/article/10.1007/s12369-025-01323-5#ref-CR158 "Klęczek K, Rice A, Alimardani M (2024) Robots as mental health coaches: a study of emotional responses to technology-assisted stress management tasks using physiological signals. Sens 24(13):4032.
https://doi.org/10.3390/s24134032
")\].Finally, qualitative methods can be employed to gain rich and nuanced insights into deeper aspects of user experiences. Mixed-method approaches that integrate both quantitative and qualitative research methods have become increasingly common in contemporary user experience studies, with interviews being frequently employed [[159](/article/10.1007/s12369-025-01323-5#ref-CR159 "Lanius C, Weber R, Robinson J (2021) User experience methods in research and practice. J Tech Writ Comm 51(4):350–379. https://doi.org/10.1177/00472816211044499
"), [160](/article/10.1007/s12369-025-01323-5#ref-CR160 "Shourmasti ES, Colomo-Palacios R, Holone H, Demi S (2021) User experience in social robots. Sens 21(15):5052.
https://doi.org/10.3390/s21155052
")\]. Interviews with users or stakeholders are beneficial for early-stage SAR design as they help shape design principles, strategies, and therapeutic objectives, as well as identify potential development challenges. However, it is critical to clearly define the objectives of the interview and thoughtfully select the types of interview and the guiding questions to be presented \[[160](/article/10.1007/s12369-025-01323-5#ref-CR160 "Shourmasti ES, Colomo-Palacios R, Holone H, Demi S (2021) User experience in social robots. Sens 21(15):5052.
https://doi.org/10.3390/s21155052
")\]. As an illustrative example of qualitative research for evaluating social robots, de Graaf et al. (2014) employed a mixed-method design combining questionnaires and interviews to evaluate the long-term acceptance of robots. Their findings highlighted the significant roles of perceived usefulness, social presence, enjoyment, and attractiveness in fostering sustained user acceptance \[[161](/article/10.1007/s12369-025-01323-5#ref-CR161 "De Graaf MM, Ben Allouch S, Van Dijk JA (2016) Long-term evaluation of a social robot in real homes. Interact Stud 17(3):462–491.
https://doi.org/10.1075/is.17.3.08deg
")\].To evaluate a SAR, it is beneficial to adopt diverse research designs that incorporate multiple evaluation methods, rather than relying on a narrow range of measurements that may limit the accurate identification of user engagement or trust in real-world contexts [[124](/article/10.1007/s12369-025-01323-5#ref-CR124 "Kok BC, Soh H (2020) Trust in robots: challenges and opportunities. Curr Robot Rep 1(4):297–309. https://doi.org/10.1007/s43154-020-00029-y
")\]. Given adequate evaluation resources, researchers are encouraged to utilize multimodal and holistic evaluation approaches, which may include behavioral assessments, momentary evaluations, usage history analysis, physiological measurements, qualitative interviews, and ethnographic observations or studies. Ideally, employing comprehensive models that integrate various multimodal evaluation criteria can provide a holistic, context-rich understanding of user experiences, facilitating more robust, flexible, and personalized insights into user interactions. For instance, Perugia et al. (2022) proposed an engagement model for individuals with dementia. This model aimed at a comprehensive assessment of engagement levels by combining self-reported measures, behavioral and expressive data, and physiological signals, which were gathered while participants were undertaking activities alongside an animal-like social robot, such as puzzle-solving tasks or spontaneous interactions with the robot \[[151](/article/10.1007/s12369-025-01323-5#ref-CR151 "Perugia G, Díaz-Boladeras M, Catala-Mallofré A, Barakova EI, Rauterberg M (2020) ENGAGE-DEM: a model of engagement of people with dementia. IEEE Trans Affect Comput 13(2):926–943.
https://doi.org/10.1109/TAFFC.2020.2980275
")\].5.3 Safety
Safety is also one of the key criteria for robots and medical interventions. SARs should ensure that the design and interactions prioritize user well-being and avoid any potential harm. Developers should evaluate the potential risks in every aspect of the robot’s interactions and functionalities to guarantee a secure and supportive environment for users. For SARs, safety can be divided into three aspects: physical safety, psychological safety, and privacy and data security. To ensure physical safety, robots should be designed to avoid injury or disease during use. In particular, developers should carefully craft their physical interaction system to control interactions and manage collisions [[162](/article/10.1007/s12369-025-01323-5#ref-CR162 "Haddadin S (2015) Physical safety in robotics. In: Koopman P, Schröder L, Bardenhagen F, Böhm S, Lüttgen G (eds) Formal modeling and verification of cyber-physical systems. Springer, Cham, pp 249–271. https://doi.org/10.1007/978-3-658-09994-7_9
")\]. Psychological safety involves creating an environment in which users feel comfortable and free from distress during interactions. Users’ perceived psychological safety is affected by various factors, but a recent study highlighted that technical functionality, design and handling, and a sense of control primarily impact users’ perceived safety, indicating that enhancing user-centered design can also improve psychological safety \[[163](/article/10.1007/s12369-025-01323-5#ref-CR163 "Vöcking M, Karrenbrock A, Beckmann A, Vondeberg C, Obert L, Hemming B et al. (2024) Emotional and psychological safety in healthcare digitalization: a design ethnographic study. Int J Public Health 69:1607575.
https://doi.org/10.3389/ijph.2024.1607575
")\]. Finally, developers should evaluate and improve how user data are collected, stored, processed, and accessed to protect their personal information, as privacy and data security is a growing issue in digital healthcare \[[164](/article/10.1007/s12369-025-01323-5#ref-CR164 "Filkins BL, Kim JY, Roberts B, Armstrong W, Miller MA, Hultner ML et al. (2016) Privacy and security in the era of digital health: what should translational researchers know and do about it? Am J Transl Res 8(3):1560")\]. The collected data should be anonymized or pseudonymized, managed transparently, used solely for specified purposes, and not shared with third parties without the user’s explicit consent \[[165](/article/10.1007/s12369-025-01323-5#ref-CR165 "Grande D, Marti XL, Feuerstein-Simon R, Merchant RM, Asch DA, Lewson A, Cannuscio CC (2020) Health policy and privacy challenges associated with digital technology. JAMA Netw Open 3(7):e208285.
https://doi.org/10.1001/jamanetworkopen.2020.8285
")\].The physical safety of SARs can be evaluated using established risk assessment frameworks for machines and robots such as ISO 12,100 [166]. For privacy and data security, developers can adhere to regulatory guidelines such as the General Data Protection Regulation (GDPR) provided by the European Union [167], although there are currently no specific standard guidelines for safety or ethical issues of SARs. For social robots for mental healthcare, the most important consideration is psychological safety, which is usually assessed as users’ perceived safety through surveys or structured interviews [[163](/article/10.1007/s12369-025-01323-5#ref-CR163 "Vöcking M, Karrenbrock A, Beckmann A, Vondeberg C, Obert L, Hemming B et al. (2024) Emotional and psychological safety in healthcare digitalization: a design ethnographic study. Int J Public Health 69:1607575. https://doi.org/10.3389/ijph.2024.1607575
"), [168](/article/10.1007/s12369-025-01323-5#ref-CR168 "Hattori T, Nakamura M, Kawamura K, Nihei M (2021) Determining possible risks of introducing socially assistive robots with mobility functions to aged care facilities. Int Conf Hum-Comput Interact 145–155.
https://doi.org/10.1007/978-3-030-78108-8_11
")\]. Users who perceive a product as safe will experience comfort, familiarity, predictability, and a sense of control \[[169](/article/10.1007/s12369-025-01323-5#ref-CR169 "Akalin N, Kiselev A, Kristoffersson A, Loutfi A (2023) A taxonomy of factors influencing perceived safety in human-robot interaction. Int J Soc robot 15(12):1993–2004.
https://doi.org/10.1007/s12369-023-01027-8
")\]. Trust can also significantly contribute to perceived safety, which makes enhancing trust crucial not only for user engagement and improved experiences but also for overall user safety \[[170](/article/10.1007/s12369-025-01323-5#ref-CR170 "Akalin N, Kristoffersson A, Loutfi A (2022) Do you feel safe with your robot? Factors influencing perceived safety in human-robot interaction based on subjective and objective measures. Int J Hum Comput Stud 158:102744.
https://doi.org/10.1016/j.ijhcs.2021.102744
")\]. In addition, ensuring accessibility for individuals with disabilities or vulnerable populations is essential for inclusivity. To support the needs of marginalized or underserved users, developers should incorporate accommodations and embed adaptable, personalized access within interaction methods.As mentioned above, trust can be measured using self-reported scales [[125](/article/10.1007/s12369-025-01323-5#ref-CR125 "Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239. https://doi.org/10.1016/j.neubiorev.2019.07.014
")\], but incorporating diverse methodologies can lead to a more accurate assessment of safety. As with effectiveness and user experiences, implementing qualitative methods can help identify aspects of safety that self-reports alone may fail to capture. For instance, Nyholm et al.’s (2021) qualitative research highlighted users’ ambivalent attitudes toward robot safety, noting that users perceive humanoid robots as both safe and unsafe, which is difficult to identify with quantitative surveys. The study suggested that prioritizing user-centered design and enhancing transparency in information sharing and interaction are important for effective person-centered robotic care \[[171](/article/10.1007/s12369-025-01323-5#ref-CR171 "Nyholm L, Santamäki-Fischer R, Fagerström L (2021) Users’ ambivalent sense of security with humanoid robots in healthcare. Inf Health Soc Care 46(2):218–226.
https://doi.org/10.1080/17538157.2021.1883027
")\]. Behavioral and physiological measures can also help assess safety. Wang et al. (2024) employed eye-tracking data to evaluate safety during the interactions with robots, analyzing participants’ gaze patterns, fixation durations on specific robot-hand areas, and gaze shifts throughout each motion. They identified that frequent or prolonged fixation can indicate concerns regarding the particular movements of the robot \[[172](/article/10.1007/s12369-025-01323-5#ref-CR172 "Wang Y, Wang G, Ge W, Duan J, Chen Z, Wen L (2024) Perceived safety assessment of interactive motions in human-soft robot interaction. Biomimetics 9(1):58.
https://doi.org/10.3390/biomimetics9010058
")\].5.4 Iterative Evaluation and Refinement
User-centered SARs should be continuously improved by iterating the development and evaluation process. Iterative design is another important principle for user-centered and participatory design that emphasizes repetitive user testing and refinement through users’ feedback [[173](/article/10.1007/s12369-025-01323-5#ref-CR173 "Larman C, Basili VR (2003) Iterative and incremental developments. A brief history. Computer 36(6):47–56. https://doi.org/10.1109/MC.2003.1204375
")\]. By conducting frequent informal evaluations—gathering feedback from users or stakeholders and observing how the prototype is used in practice—developers and researchers gain valuable insights into how to refine and improve the product based on users’ perspectives. Iterative design processes aim to continuously refine the product by enhancing usability, minimizing errors and workloads, which foster effective communication and patient involvement, increase user adoption, engagement, and sustainability, and ultimately lead to improved clinical outcomes \[[174](/article/10.1007/s12369-025-01323-5#ref-CR174 "Henderson K, Reihm J, Koshal K, Wijangco J, Sara N, Miller N et al. (2024) A closed-loop digital health tool to improve depression care in multiple sclerosis: iterative design and Cross-sectional Pilot randomized controlled trial and its impact on depression care. JMIR Form Res 8:e52809.
https://doi.org/10.2196/52809
"), [175](/article/10.1007/s12369-025-01323-5#ref-CR175 "Katsulis Z, Ergai A, Leung WY, Schenkel L, Rai A, Adelman J et al. (2016) Iterative user centered design for development of a patient-centered fall prevention toolkit. Appl Ergon 56:117–126.
https://doi.org/10.1016/j.apergo.2016.03.011
")\]. Mast et al. (2012) adopted an iterative design process for at-home elderly care robots. After the initial preparatory stage involving focus group studies, surveys, ethnographic research, expert technical reviews, and usability tests, the researchers undertook several iterations of evaluation and refinement based on user feedback regarding the robots’ acceptability and usability, which led to more refined, user-validated robot prototypes \[[117](/article/10.1007/s12369-025-01323-5#ref-CR117 "Mast M, Burmester M, Krüger K, Fatikow S, Arbeiter G, Graf B et al. (2012) User-centered design of a dynamic-autonomy remote interaction concept for manipulation-capable robots to assist elderly people in the home. J Hum Rob Interact 1(1):96–118.
https://doi.org/10.5898/JHRI.1.1.Mast
")\].For the effective iterative refinement of SARs, we recommend that developers tailor the iteration process to align with their specific objectives and requirements. Current objectives may include generating new ideas, enhancing design quality, identifying and addressing errors or obstacles, or managing and improving coordination and information flows. It is also crucial to recognize whether the current development stage is exploratory, characterized by high uncertainty, or focused on the final refinement of already viable solutions. Based on these factors, developers can define the scope of iterative evaluations and employ feedback methods that best align with their specific goals and development contexts [[176](/article/10.1007/s12369-025-01323-5#ref-CR176 "Wynn DC, Eckert CM (2017) Perspectives on iteration in design and development. Res Eng Des 28:153–184. https://doi.org/10.1007/s00163-016-0226-3
")\]. Employing a variety of methods—such as questionnaires, interviews, usability testing, and behavioral or physiological measures—can also enrich the understanding of user needs. To enhance efficiency, we also suggest that developers seek feedback from early prototypes. By establishing multiple feedback loops, developers can refine the product more effectively and maintain more agile iteration cycles.We also suggest that iterative evaluation can help complement the rigidness of formal trials and provide valuable insights that formal evaluations may not capture. Although formal and standardized assessments are essential for ensuring the validity of the product, relying exclusively on long-term formal trials creates temporal rigidity that may quickly render the evidence outdated, particularly in rapidly evolving fields like SARs and AI chatbots where continuous updates and improvements are required [[142](/article/10.1007/s12369-025-01323-5#ref-CR142 "Guo C, Ashrafian H, Ghafur S, Fontana G, Gardner C, Prime M (2020) Challenges for the evaluation of digital health solutions-a call for innovative evidence generation approaches. NPJ Digit Med 3(1):110. https://doi.org/10.1038/s41746-020-00314-2
")\]. Although recent discussion on real-world evidence \[[177](/article/10.1007/s12369-025-01323-5#ref-CR177 "Sherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL et al. (2016) Real-world evidence-what is it and what can it tell us. N Engl J Med 375(23):2293–2297.
https://doi.org/10.1056/nejmsb1609216
")\] and pragmatic trials \[[178](/article/10.1007/s12369-025-01323-5#ref-CR178 "Ford I, Norrie J (2016) Pragmatic trials. N Engl J Med 375(5):454–463.
https://doi.org/10.1056/NEJMra1510059
")\] that reflect real-world settings may help mitigate this issue, we suggest that developers and researchers also conduct frequent informal assessments or closed testing for each task and function of the product, incorporating feedback from users and stakeholders. This process supports further optimization of the product based on user-centered perspectives. The overall process of the iterative development model of SARs is outlined in Fig. [4](/article/10.1007/s12369-025-01323-5#Fig4).Fig. 4
The iterative design process of SARs based on the user-centered approach
6 SARs and Large Language Models
The emergence of LLMs, such as OpenAI’s ChatGPT, is triggering significant changes in almost every field in current society, including digital healthcare and SARs. Their extensive and versatile conversations and the automatic interaction system that seamlessly mimics human communications provide an opportunity for both robotics and mental healthcare [[179](/article/10.1007/s12369-025-01323-5#ref-CR179 "Blease C, Torous J (2023) ChatGPT and mental healthcare: balancing benefits with risks of harms. BMJ Ment Health 26(1. https://doi.org/10.1136/bmjment-2023-300884
"), [180](/article/10.1007/s12369-025-01323-5#ref-CR180 "Vemprala SH, Bonatti R, Bucker A, Kapoor A (2024) ChatGPT for robotics: design principles and model abilities. IEEE Access. 12:55682–55696.
https://doi.org/10.1109/ACCESS.2024.3387941
")\]. They have the potential to substantially resolve the current challenges of SARs, such as personalization, multimodal and adaptive interactions, dynamic interaction content, and the establishment of long-term relationships with users. LLMs hold significant potential for future mental healthcare in areas such as psychological evaluation, psychoeducation, counseling and coaching, and therapeutic interventions such as cognitive reframing \[[63](/article/10.1007/s12369-025-01323-5#ref-CR63 "Stade EC, Stirman SW, Ungar LH, Boland CL, Schwartz HA, Yaden DB et al. (2024) Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment Health Res 3:12.
https://doi.org/10.1038/s44184-024-00056-z
")\]. Moreover, recent advancements in speech recognition/text-to-speech technologies and AI agents have significantly expanded the scope of automation achievable in the operation and interaction capabilities of AI platforms, including SARs. Breakthroughs in LLMs have enhanced their potential to perform tasks using natural language-based interactions, particularly in the domain of mental healthcare.Several preliminary studies have attempted to integrate LLMs with social robots for mental healthcare. Bertacchini et al. (2023) incorporated ChatGPT into a SAR to facilitate real-time, open-ended dialogue and multimodal communication for cognitive care in autism [[181](/article/10.1007/s12369-025-01323-5#ref-CR181 "Bertacchini F, Demarco F, Scuro C, Pantano P, Bilotta E (2023) A social robot connected with chatGPT to improve cognitive functioning in ASD subjects. Front psychol 14:1232177. https://doi.org/10.3389/fpsyg.2023.1232177
")\]. Similarly, Lee et al. (2023) developed a social robot equipped with a language model to enable empathetic and context-aware interactions \[[182](/article/10.1007/s12369-025-01323-5#ref-CR182 "Lee YK, Jung Y, Kang G, Hahn S (2023) Developing social robots with empathetic non-verbal cues using large language models. arXiv preprint arXiv: 2308.16529.
https://doi.org/10.48550/arXiv.2308.16529
")\]. These studies demonstrated promising results, but they also faced technical limitations and constraints in scalability. While there are significant potential advantages of LLMs such as improved linguistic capabilities, adaptive interactions, and personalized care, challenges remain in fully integrating LLMs into SARs, which include complexities in system integration, reliability issues, limited depth of support, and risks related to hallucinations, biases, safety, and ethical concerns \[[183](/article/10.1007/s12369-025-01323-5#ref-CR183 "Voultsiou E, Vrochidou E, Moussiades L, Papakostas GA (2025) The potential of large language models for social robots in special education. Prog Artif Intell.
https://doi.org/10.1007/s13748-025-00363-2
"), [184](/article/10.1007/s12369-025-01323-5#ref-CR184 "Shi Z, Landrum E, O’Connell A, Kian M, Pinto-Alva L, Shrestha K et al. (2024) How can large language models enable better socially assistive human-robot interaction: a brief survey. AAAI Symp Ser 3(1):401–404.
https://doi.org/10.1609/aaaiss.v3i1.31245
")\].We propose, nevertheless, that these limitations can be addressed through recent technological advancements. There have been efforts to strengthen the accessibility and reliability of AI-generated information, including the integration of advanced search functionalities into LLMs and developing spatial AI systems that enable seamless connectivity between robotic systems and environmental technologies such as IoT. These efforts also include explicitly designing AI models to consistently access relevant and up-to-date information through advanced retrieval techniques, such as retrieval-augmented generation [185]. Especially, the recently introduced Model Context Protocol (MCP) enables AI assistants to securely connect with diverse data sources and tools without complex pre-configurations [[186](/article/10.1007/s12369-025-01323-5#ref-CR186 "Anthropic (2024) Introducing the Model context protocol. https://www.anthropic.com/news/model-context-protocol
. Accessed 11 Apr 2025")\]. By facilitating dynamic interactions and seamless access to real-world information, MCP helps AI agents generate more relevant, rich, and context-aware responses, enhancing their adaptability, robustness, responsibility in unexpected situations, and personalization \[[187](/article/10.1007/s12369-025-01323-5#ref-CR187 "Hou X, Zhao Y, Wang S, Wang H (2025) Model context protocol (MCP): landscape, security threats, and future research directions. arXiv preprint arXiv: 2503.23278.
https://doi.org/10.48550/arXiv.2503.23278
"), [188](/article/10.1007/s12369-025-01323-5#ref-CR188 "Singh A, Ehtesham A, Kumar S, Khoei TT (2025) A survey of the Model context protocol (MCP): standardizing context to enhance large language models (LLMs.
https://doi.org/10.20944/preprints202504.0245.v1
. Accessed 11 Apr 2025")\]. These developments significantly enhance the integration and information accessibility of AI-enabled SARs, thereby facilitating critical mental healthcare functions such as clinical administration, diagnostics and treatment evaluations, education, feedback provision, and targeted interventions.Among these benefits, one of the most critical is the capability of LLMs to enhance SARs by enabling automated care and support through free-flowing conversations. Unlike previous chatbot-based psychotherapeutic content, which largely relies on structured and scripted interactions to deliver therapeutic principles, recent chatbots and assistive AIs powered by LLMs can offer guided yet flexible conversations that resemble natural communications and are individually tailored, representing a significant paradigm shift in mental healthcare delivery [[189](/article/10.1007/s12369-025-01323-5#ref-CR189 "Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48. https://doi.org/10.1038/s44184-024-00097-4
")\]. These advancements allow users to interact with SARs in a human-like manner, improving user familiarity and acceptance by helping users feel more comfortable with SARs. Moreover, LLMs by themselves work as a closed-loop reasoning system that deploys and improves dialogues based on users’ reactions, which is also one of the critical attributes for autonomous and personalized digital therapeutic systems \[[85](/article/10.1007/s12369-025-01323-5#ref-CR85 "Nimri R, Phillip M, Clements MA, Kovatchev B (2024) Closed-loop control, artificial intelligence-based decision-support Systems, and data science. Diabetes Technol Ther 26:S1.
https://doi.org/10.1089/dia.2024.2505
"), [180](/article/10.1007/s12369-025-01323-5#ref-CR180 "Vemprala SH, Bonatti R, Bucker A, Kapoor A (2024) ChatGPT for robotics: design principles and model abilities. IEEE Access. 12:55682–55696.
https://doi.org/10.1109/ACCESS.2024.3387941
")\].Ultimately, these advancements may pave the way for fully automated, long-term clinical care in which SARs directly intervene, support, and empower users toward meaningful behavioral change. Current AI chatbots have long been considered inadequate to substitute for human counselors, with the ability to form therapeutic relationships considered available only after the emergence of artificial general intelligence (AGI) [[63](/article/10.1007/s12369-025-01323-5#ref-CR63 "Stade EC, Stirman SW, Ungar LH, Boland CL, Schwartz HA, Yaden DB et al. (2024) Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment Health Res 3:12. https://doi.org/10.1038/s44184-024-00056-z
"), [190](/article/10.1007/s12369-025-01323-5#ref-CR190 "Fulmer R (2019) Artificial intelligence and counseling: four levels of implementation. Theory psychol 29(6):807–819.
https://doi.org/10.1177/0959354319853045
")\]. However, OpenAI has recently indicated that their technologies are nearing Level 2 of a five-stage progression toward AGI, projecting the emergence of AGI within just a few years \[[191](/article/10.1007/s12369-025-01323-5#ref-CR191 "PYMNTS (2024) OpenAI Director says artificial General intelligence May Be 5 years out.
https://www.pymnts.com/artificial-intelligence-2/2024/openai-board-member-agi-is-five-to-15-years-away/
. Accessed 14 Oct 2024")\]. Fully automated robot therapists can offer improved patient interactions, consistent and highly reliable counseling, and therapeutic support—benefits that are particularly valuable to individuals with limited access to human therapists \[[192](/article/10.1007/s12369-025-01323-5#ref-CR192 "Fiske A, Henningsen P, Buyx A (2019) Your robot therapist will see you now: ethical implications of embodied artificial intelligence in psychiatry, psychology, and psychotherapy. J Med Internet Res 21(5):e13216.
https://doi.org/10.2196/13216
")\]. Furthermore, in certain contexts, the anonymity and non-judgmental nature of AI chatbots can promote increased acceptance and self-disclosure from users \[[30](/article/10.1007/s12369-025-01323-5#ref-CR30 "Croes EA, Antheunis ML (2021 2020) 36 questions to loving a chatbot: are people willing to self-disclose to a chatbot? In Chatbot Research and Design: 4th Int Workshop, CONVERSATIONS, pp 81–95.
https://doi.org/10.1007/978-3-030-68288-2_6
"), [189](/article/10.1007/s12369-025-01323-5#ref-CR189 "Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48.
https://doi.org/10.1038/s44184-024-00097-4
")\].However, the rapid advancement of technology also raises significant concerns about its associated ethical issues. The fundamental problem may arise from the deceptive nature of AI chatbots and anthropomorphic robots. Although they can mimic actual relationships between users and human therapists, they indeed do not have the intentionality to treat users, which raises risks in SARs as well [[193](/article/10.1007/s12369-025-01323-5#ref-CR193 "Sedlakova J, Trachsel M (2023) Conversational artificial intelligence in psychotherapy: a new therapeutic tool or agent? Am J bioeth 23(5):4–13. https://doi.org/10.1080/15265161.2022.2048739
"), [194](/article/10.1007/s12369-025-01323-5#ref-CR194 "Winkle K, Caleb-Solly P, Leonards U, Turton A, Bremner P (2021) Assessing and addressing ethical risk from anthropomorphism and deception in socially assistive robots. In Proc 2021 ACM/IEEE Int Conf Hum-Robot Interact, pp 101–109.
https://doi.org/10.1145/3434073.3444666
")\]. As AI chatbots only simulate therapeutic relationships without genuinely understanding the principles of psychotherapy \[[132](/article/10.1007/s12369-025-01323-5#ref-CR132 "Ferrario A, Sedlakova J, Trachsel M (2024) The role of humanization and robustness of large language models in conversational artificial intelligence for individuals with depression: a critical analysis. JMIR Ment Health 11:e56569.
https://doi.org/10.2196/56569
")\], the ongoing interactions between users and chatbots may leave users vulnerable to becoming overly dependent on robots that do not actually care about them. This leads to the issue of “overtrust” where users develop unrealistic expectations or excessive dependence on the chatbot’s abilities or intentions compared to what the robots actually provide \[[195](/article/10.1007/s12369-025-01323-5#ref-CR195 "Boada JP, Maestre BR, Genís CT (2021) The ethical issues of social assistive robotics: a critical literature review. Technol Soc 67:101726.
https://doi.org/10.1016/j.techsoc.2021.101726
")\]. Considering that trust is basically grounded in vulnerability \[[59](/article/10.1007/s12369-025-01323-5#ref-CR59 "Spitale M, Silleresi S, Garzotto F, Matarić MJ (2023) Using socially assistive robots in speech-language therapy for children with language impairments. Int J Soc robot 15(9):1525–1542.
https://doi.org/10.1007/s12369-023-01028-7
"), [60](/article/10.1007/s12369-025-01323-5#ref-CR60 "Elfaki AO, Abduljabbar M, Ali L, Alnajjar F, Mehiar DA, Marei AM et al. (2023) Revolutionizing social robotics: a cloud-based framework for enhancing the intelligence and autonomy of social robots. Robot 12(2):48.
https://doi.org/10.3390/robotics12020048
")\], the users’ wrong or unfulfillable expectations of SARs without real compassion or empathy may rather exacerbate the users’ state \[[193](#ref-CR193 "Sedlakova J, Trachsel M (2023) Conversational artificial intelligence in psychotherapy: a new therapeutic tool or agent? Am J bioeth 23(5):4–13.
https://doi.org/10.1080/15265161.2022.2048739
"),[194](#ref-CR194 "Winkle K, Caleb-Solly P, Leonards U, Turton A, Bremner P (2021) Assessing and addressing ethical risk from anthropomorphism and deception in socially assistive robots. In Proc 2021 ACM/IEEE Int Conf Hum-Robot Interact, pp 101–109.
https://doi.org/10.1145/3434073.3444666
"),[195](/article/10.1007/s12369-025-01323-5#ref-CR195 "Boada JP, Maestre BR, Genís CT (2021) The ethical issues of social assistive robotics: a critical literature review. Technol Soc 67:101726.
https://doi.org/10.1016/j.techsoc.2021.101726
")\].Researchers warn that overestimating or overanticipating the role of chatbots could undermine users’ autonomy, particularly concerning vulnerable populations [[193](/article/10.1007/s12369-025-01323-5#ref-CR193 "Sedlakova J, Trachsel M (2023) Conversational artificial intelligence in psychotherapy: a new therapeutic tool or agent? Am J bioeth 23(5):4–13. https://doi.org/10.1080/15265161.2022.2048739
"), [196](/article/10.1007/s12369-025-01323-5#ref-CR196 "Khawaja Z, Bélisle-Pipon JC (2023) Your robot therapist is not your therapist: understanding the role of AI-powered mental health chatbots. Front Digit Health 5:1278186.
https://doi.org/10.3389/fdgth.2023.1278186
")\]. Indeed, this issue is already becoming a reality. Studies that analyzed Reddit posts indicate that chatbots’ humanized functions make people overdependent on them, although they often provide re-prompted answers based on the same queries instead of offering truly reliable responses \[[197](/article/10.1007/s12369-025-01323-5#ref-CR197 "Hou H, Leach K, Huang Y (2024) ChatGPT giving relationship advice-How reliable is it? Proc Int AAAI Conf Web Soc Media 18:610–623.
https://doi.org/10.1609/icwsm.v18i1.31338
"), [198](/article/10.1007/s12369-025-01323-5#ref-CR198 "Laestadius L, Bishop A, Gonzalez M, Illenčík D, Campos-Castillo C (2024) Too human and not human enough: a grounded theory analysis of mental health harms from emotional dependence on the social chatbot replika. New Media Soc 26(10):5923–5941.
https://doi.org/10.1177/14614448221142007
")\]. This issue demands prior verification of their potential long-term impact on users and an assessment of the associated risks \[[63](/article/10.1007/s12369-025-01323-5#ref-CR63 "Stade EC, Stirman SW, Ungar LH, Boland CL, Schwartz HA, Yaden DB et al. (2024) Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment Health Res 3:12.
https://doi.org/10.1038/s44184-024-00056-z
")\]. In addition, guidelines or criteria to monitor the risk posed by chatbots should be established to evaluate their ethical compliance and ensure the protection of users’ autonomy \[[192](/article/10.1007/s12369-025-01323-5#ref-CR192 "Fiske A, Henningsen P, Buyx A (2019) Your robot therapist will see you now: ethical implications of embodied artificial intelligence in psychiatry, psychology, and psychotherapy. J Med Internet Res 21(5):e13216.
https://doi.org/10.2196/13216
"), [196](/article/10.1007/s12369-025-01323-5#ref-CR196 "Khawaja Z, Bélisle-Pipon JC (2023) Your robot therapist is not your therapist: understanding the role of AI-powered mental health chatbots. Front Digit Health 5:1278186.
https://doi.org/10.3389/fdgth.2023.1278186
")\].One of the most fundamental solutions to this issue is to ensure that chatbot responses consistently align with established therapeutic principles and evidence-based practices [[63](/article/10.1007/s12369-025-01323-5#ref-CR63 "Stade EC, Stirman SW, Ungar LH, Boland CL, Schwartz HA, Yaden DB et al. (2024) Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment Health Res 3:12. https://doi.org/10.1038/s44184-024-00056-z
"), [189](/article/10.1007/s12369-025-01323-5#ref-CR189 "Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48.
https://doi.org/10.1038/s44184-024-00097-4
")\]. Although verifying these therapeutic principles and effectiveness in LLM-based chatbots is challenging due to their limited explainability (the “black box” problem), efforts have been undertaken to embed these principles within chatbot interactions \[[189](/article/10.1007/s12369-025-01323-5#ref-CR189 "Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48.
https://doi.org/10.1038/s44184-024-00097-4
")\]. Among them, we propose a promising approach that segments psychotherapy into distinct components and validates each separately. Ciarrochi et al. (2024) introduced an AI tool integrated with a chatbot based on third-wave cognitive behavioral therapy, capable of independently assessing users’ psychological states according to an established psychotherapeutic model \[[199](/article/10.1007/s12369-025-01323-5#ref-CR199 "Ciarrochi J, Hernández C, Hill D, Ong C, Gloster AT, Levin ME et al. (2024) Process-based therapy: a common ground for understanding and utilizing therapeutic practices. J Psychother Integr 34(3):265–290.
https://doi.org/10.1037/int0000348
")\]. This method is particularly promising for multi-agent chatbot systems, in which each agent specializes in a specific psychotherapeutic function such as evaluation, information delivery, or intervention. These agents can be individually trained and validated, ensuring that they maintain strong theoretical foundations and effectiveness.Currently, this technology remains in its early stages, and much uncertainty persists regarding how LLMs will function in SAR interactions and their potential impact on the broader field of digital healthcare. We recommend that developers conduct thorough examinations of user-robot interaction characteristics and their potential psychological effects for mitigating risks associated with psychotherapy using LLMs. Moreover, rigorous yet timely ethical scrutiny and regulatory oversight will be critical to addressing the emerging challenges posed by these rapidly advancing robotic and generative AI technologies. Further research and responsible development are required to ensure that generative AI tools provide meaningful and safe mental health support without overpromising benefits or inadvertently causing harm [[189](/article/10.1007/s12369-025-01323-5#ref-CR189 "Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48. https://doi.org/10.1038/s44184-024-00097-4
")\].6.1 Future Directions
We propose future directions for SARs in mental healthcare based on the user-centered principles for SARs and leveraging recent breakthroughs in LLM technology. The integration of enhanced multimodal interactions and personalized adaptive interventions in SARs will facilitate richer, more engaging user interactions. As discussed above, recent breakthroughs in LLMs will enable SARs to autonomously handle comprehensive mental healthcare tasks through natural interactions, possibly allowing them to deliver free-flowing, personalized, and theoretically validated psychotherapeutic care. Voice-based communication will serve as the primary medium for these interactions, offering particular advantages for SARs. This progress will substantially enhance the accessibility and usability of SARs, ultimately enriching user experiences and promoting user autonomy [[200](/article/10.1007/s12369-025-01323-5#ref-CR200 "Smit K, Leewis S, Almoustafa H, Yildrim K, Uymaz T (2024) Enhancing educational dynamics integrating large language models with a social robot. Proc 2024 8th Int Conf Softw E-Bus 87–94. https://doi.org/10.1145/3715885.3715889
")\].These innovative advancements entail several significant implications. First, they will accelerate the integration of pragmatic therapeutic principles into real-world applications. Future SARs will be capable of providing comprehensive user care, broadly supporting user autonomy, competence, and psychological well-being rather than merely managing and improving target symptoms. This aligns with not only user-centered design principles but also the transdiagnostic approach in mental healthcare, which emphasizes focusing treatment on common underlying mechanisms rather than symptom-specific interventions. Consequently, the transdiagnostic approach enables more effective, tailored, and flexible treatments in real-world settings [[201](/article/10.1007/s12369-025-01323-5#ref-CR201 "Dalgleish T, Black M, Johnston D, Bevan A (2020) Transdiagnostic approaches to mental health problems: current status and future directions. J Consult Clin Psycho 88(3):179–195. https://doi.org/10.1037/ccp0000482
")\].The second implication is interoperability, which refers to seamless communication and collaboration among robots, devices, databases, networks, and information systems. Integrated multimodal adaptive interactions and agentic AI systems will enable SARs to effectively incorporate with existing healthcare infrastructures, thereby optimizing therapeutic processes and enhancing the quality of the psychotherapeutic support they provide. This progress not only improves their efficiency, adaptability, and collaborative knowledge but also addresses ethical concerns related to patient safety, trust, and accountability [[202](/article/10.1007/s12369-025-01323-5#ref-CR202 "Naik N, Hameed BM, Shetty DK, Swain D, Shah M, Paul R et al. (2022) Legal and ethical consideration in artificial intelligence in healthcare: who takes responsibility? Front Surg 9:862322. https://doi.org/10.3389/fsurg.2022.862322
"), [203](/article/10.1007/s12369-025-01323-5#ref-CR203 "Pavithra N, Afza N (2024) Harnessing the power of artificial intelligence and robotics impact on attaining competitive advantage for sustainable development in hospitals with conclusions for future research approaches. GMS Hyg Infect Control 19:15.
https://doi.org/10.3205/dgkh000470
")\]. This facilitates rapid clinical decision-making, efficient utilization of clinical data, stronger collaboration with human therapists, and greater flexibility in healthcare delivery. For users, the enhanced reliability and scalability of SARs will ensure timely, personalized support across diverse situations.Finally, these advancements will significantly increase accessibility for users who have traditionally been underserved by technological interfaces. Voice-based natural communication simplifies interactions with SARs, lowering technical barriers and allowing more users to benefit from the diverse support provided by SARs. Such enhancement promotes inclusivity and strengthens person-centered principles in technological and robotic designs. However, developers will still have to consider alternative interaction methods for users with speech or hearing impairments who may not be able to utilize voice interactions. Furthermore, developers must carefully address potential improvements and ethical considerations to fully realize the potential of SARs outlined here, including their responsiveness, the refinement of “guardrails” for empathic and nuanced support, and the establishment of the robust regulatory framework for long-term sustainability [[189](/article/10.1007/s12369-025-01323-5#ref-CR189 "Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48. https://doi.org/10.1038/s44184-024-00097-4
")\].7 Conclusion
This review presents the rationale of user-centered SARs in mental healthcare, including core objectives, design principles, and conceptual frameworks for their development and implementation. This review also proposes practical guidelines for the user-centered design of SARs in development and evaluation phases, which can fortify user autonomy and competence, enhance mental health outcomes, and improve psychological well-being and quality of life through continuous meaningful interactions. This review provides novel insights into interdisciplinary research across the fields of robotics, digital healthcare, and mental health by suggesting therapeutic principles, personalized interventions, adaptive multimodal interactions, and primary ethical issues for the deployment of SARs in mental healthcare. Integrating user-centered principles into SARs in mental healthcare can also ensure that the robots function in accordance with therapeutic objectives, as they enable SARs to provide empathetic and reliable psychotherapeutic support. It underscores developers and healthcare providers’ responsibility to deliver person-centered care through innovative digital solutions, which will eventually foster the accessibility and sustainability of mental healthcare.
This review also has several limitations. First, the proposed design principles and guidelines are largely derived from general user experience or digital health studies rather than SAR-specific research, potentially limiting their applicability to unique contexts like mental healthcare SARs. Additionally, the literature searches for each topic were not systematically conducted, although we have incorporated contrasting perspectives where applicable to introduce perspectives in an impartial way and reduce bias in the conclusions. Ideally, user-centered design guidelines for SARs should be developed through consensus among researchers within the field, while also encouraging interdisciplinary consensus, especially in light of the growing influence of LLMs on SARs.
Accordingly, the development of SARs is an interdisciplinary effort requiring collaboration among experts from diverse fields. Successful integration of their knowledge is achievable when all contributors align with the product’s objectives and purpose, while also understanding and applying user-centered principles throughout the development process. Here, the user-centered principles should not be treated as merely instrumental for enhancing usability and product effectiveness, nor should they reflect “usability from the developers’ perspective,” where usability is asserted through development or updates without genuine consideration of the users’ needs and experiences. The user-centered principles should be viewed as an end in itself, as they align with the core purpose of SARs. This is not only a social responsibility for healthcare providers to promote person-centered healthcare, but it might also be the only way to create digital products that are genuinely effective for users—despite being more arduous and endurance-testing than traditional development processes. Just as we see technological innovation in robotics and AI, we should also expect innovation in the principles and philosophies guiding their development.
References
- Kasoju N, Remya NS, Sasi R, Sujesh S, Soman B, Kesavadas C et al. (2023) Digital health: trends, opportunities and challenges in medical devices, pharma and bio-technology. CSI Trans ICT 11(1):11–30. https://doi.org/10.1007/s40012-023-00380-3
Article Google Scholar - Torous J, Bucci S, Bell IH, Kessing LV, Faurholt-Jepsen M, Whelan P et al. (2021) The growing field of digital psychiatry: current evidence and the future of apps, social media, chatbots, and virtual reality. World J. Psychiatry 20(3):318–335. https://doi.org/10.1002/wps.20883
Article Google Scholar - Fardeau E, Senghor AS, Racine E (2023) The impact of socially assistive robots on human flourishing in the context of dementia: a scoping review. Int J Soc robot 15(6):1025–1075. https://doi.org/10.1007/s12369-023-00980-8
Article Google Scholar - Feil-Seifer D, Mataric MJ (2005) Defining socially assistive robotics. 9th Int Conf Rehabil robot 465–468. https://doi.org/10.1109/ICORR.2005.1501143
- Aldebaran (2025) Pepper. https://aldebaran.com/en/pepper/. Accessed 31 Mar 2025
- Aldebaran (2025) Na06. https://aldebaran.com/en/nao6/. Accessed 31 Mar 2025
- Misty Robotics (2025) Misty II. https://www.mistyrobotics.com/misty-ii. Accessed 31 Mar 2025
- Getson C, Nejat G (2021) Socially assistive robots helping older adults through the pandemic and life after COVID-19. Robotics 10(3):106. https://doi.org/10.3390/robotics10030106
Article Google Scholar - PARO Robots (2025) PARO Therapeutic robot. https://www.parorobots.com/. Accessed 31 Mar 2025
- SONY (2025) Aibo. https://us.aibo.com/. Accessed 31 Mar 2025
- Kabacińska K, Prescott TJ, Robillard JM (2021) Socially assistive robots as mental health interventions for children: a scoping review. Int J Soc robot 13(5):919–935. https://doi.org/10.1007/s12369-020-00679-0
Article Google Scholar - Aymerich-Franch L, Ferrer I (2023) Socially assistive robots’ deployment in healthcare settings: a global perspective. Int J Humanoid RoBot 20(1):2350002. https://doi.org/10.1142/S0219843623500020
Article Google Scholar - Feil-Seifer D, Matarić MJ (2011) Socially assistive robotics. IEEE Robot Autom Mag 18(1):24–31. https://doi.org/10.1109/MRA.2010.940150
Article Google Scholar - Hsieh CJ, Li PS, Wang CH, Lin SL, Hsu TC, Tsai CMT (2023) Socially assistive robots for people living with dementia in long-term facilities: a systematic review and meta-analysis of randomized controlled trials. Gerontology 69(8):1027–1042. https://doi.org/10.1159/000529849
Article Google Scholar - Guemghar I, Pires de Oliveira Padilha P, Abdel-Baki A, Jutras-Aswad D, Paquette J, Pomey MP (2022) Social robot interventions in mental health care and their outcomes, barriers, and facilitators: scoping review. JMIR Ment Health 9(4):e36094. https://doi.org/10.2196/36094
Article Google Scholar - Koh WQ, Felding SA, Budak KB, Toomey E, Casey D (2021) Barriers and facilitators to the implementation of social robots for older adults and people with dementia: a scoping review. BMC GeriAtr 21(1):351. https://doi.org/10.1186/s12877-021-02277-9
Article Google Scholar - Papadopoulos I, Koulouglioti C, Lazzarino R, Ali S (2020) Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: a systematic review. BMJ Open 10(1):e033096. https://doi.org/10.1136/bmjopen-2019-033096
Article Google Scholar - Wannheden C, Stenfors T, Stenling A, von Thiele Schwarz U (2021) Satisfied or frustrated? A qualitative analysis of need satisfying and need frustrating experiences of engaging with digital health technology in chronic care. Front Public Health 8:623773. https://doi.org/10.3389/fpubh.2020.623773
Article Google Scholar - Kachouie R, Sedighadeli S, Khosla R, Chu MT (2014) Socially assistive robots in elderly care: a mixed-method systematic literature review. Int J Hum Comput Interact 30(5):369–393. https://doi.org/10.1080/10447318.2013.873278
Article Google Scholar - Rabbitt SM, Kazdin AE, Scassellati B (2015) Integrating socially assistive robotics into mental healthcare interventions: applications and recommendations for expanded use. Clin Psychol Rev 35:35–46. https://doi.org/10.1016/j.cpr.2014.07.001
Article Google Scholar - Chita-Tegmark M, Scheutz M (2021) Assistive robots for the social management of health: a framework for robot design and human-robot interaction research. Int J Soc robot 13(2):197–217. https://doi.org/10.1007/s12369-020-00634-z
Article Google Scholar - Zhang J, Chen T (2025) Artificial intelligence based social robots in the process of student mental health diagnosis. Entertain Comput 52:100799. https://doi.org/10.1016/j.entcom.2024.100799
Article Google Scholar - Bennett CC, Stanojević C, Šabanović SA, Piatt J, Kim S (2021) When no one is watching: ecological momentary assessment to understand situated social robot use in healthcare. In Proceedings of the 9th International Conference on Human-Agent Interaction, pp 245–251. https://doi.org/10.1145/3472307.3484670
- Calatrava-Nicolás FM, Gutiérrez-Maestro E, Bautista-Salinas D, Ortiz FJ, González JR, Vera-Repullo JA et al. (2021) Robotic-based well-being monitoring and coaching system for the elderly in their daily activities. Sens 21(20):6865. https://doi.org/10.3390/s21206865
Article Google Scholar - Laban G, Ben-Zion Z, Cross ES (2022) Social robots for supporting post-traumatic stress disorder diagnosis and treatment. Front Psychiatry 12:752874. https://doi.org/10.3389/fpsyt.2021.752874
Article Google Scholar - Simoens P, Mahieu C, Ongenae F, De Backere F, De Pestel S, Nelis J et al. (2016) Internet of robotic things: context-aware and personalized interventions of assistive social robots (short paper). In 2016 5th IEEE International Conference on Cloud Networking (Cloudnet), pp 204–207. https://doi.org/10.1109/CloudNet.2016.27
- Calderita LV, Vega A, Barroso-Ramírez S, Bustos P, Núñez P (2020) Designing a cyber-physical system for ambient assisted living: a use-case analysis for social robot navigation in caregiving centers. Sens 20(14):4005. https://doi.org/10.3390/s20144005
Article Google Scholar - Cifuentes CA, Pinto MJ, Céspedes N, Múnera M (2020) Social robots in therapy and care. Curr Robot Rep 1:59–74. https://doi.org/10.1007/s43154-020-00009-2
Article Google Scholar - Rasouli S, Gupta G, Nilsen E, Dautenhahn K (2022) Potential applications of social robots in robot-assisted interventions for social anxiety. Int J Soc robot 14:1–32. https://doi.org/10.1007/s12369-021-00851-0
Article Google Scholar - Croes EA, Antheunis ML (2021 2020) 36 questions to loving a chatbot: are people willing to self-disclose to a chatbot? In Chatbot Research and Design: 4th Int Workshop, CONVERSATIONS, pp 81–95. https://doi.org/10.1007/978-3-030-68288-2_6
- Yen HY, Huang CW, Chiu HL, Jin G (2024) The effect of social robots on depression and loneliness for older residents in long-term care facilities: a meta-analysis of randomized controlled trials. J Am Med Dir assoc 25(6):104979. https://doi.org/10.1016/j.jamda.2024.02.017
Article Google Scholar - Dahlhausen F, Zinner M, Bieske L, Ehlers JP, Boehme P, Fehring L (2022) There’s an app for that, but nobody’s using it: insights on improving patient access and adherence to digital therapeutics in Germany. Digit Health 8:20552076221104672. https://doi.org/10.1177/20552076221104672
Article Google Scholar - Yardley L, Morrison L, Bradbury K, Muller I (2015) The person-based approach to intervention development: application to digital health-related behavior change interventions. J Med Internet Res 17(1):e4055. https://doi.org/10.2196/jmir.4055
Article Google Scholar - van Gemert-Pijnen Je, Nijland N, van Limburg M, Ossebaard HC, Kelders SM, Eysenbach G, Seydel ER (2011) A holistic framework to improve the uptake and impact of eHealth technologies. J Med Internet Res 13(4):e1672. https://doi.org/10.2196/jmir.1672
Article Google Scholar - Yardley L, Spring BJ, Riper H, Morrison LG, Crane DH, Curtis K et al. (2016) Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med 51(5):833–842. https://doi.org/10.1016/j.amepre.2016.06.015
Article Google Scholar - Onnasch L, Roesler E (2021) A taxonomy to structure and analyze human-robot interaction. Int J Soc robot 13(4):833–849. https://doi.org/10.1007/s12369-020-00666-5
Article Google Scholar - Roesler E, Manzey D, Onnasch L (2021) A meta-analysis on the effectiveness of anthropomorphism in human-robot interaction. Sci robot 6(58):eabj5425. https://doi.org/10.1126/scirobotics.abj5425
Article Google Scholar - Park G, Lee S, Chung J (2023) Do anthropomorphic chatbots increase counseling satisfaction and reuse intention? The moderated mediation of social rapport and social anxiety. Cyberpsychol Behav Soc Netw 26(5):357–365. https://doi.org/10.1089/cyber.2022.0157
Article Google Scholar - Giger JC, Piçarra N, Alves-Oliveira P, Oliveira R, Arriaga P (2019) Humanization of robots: is it really such a good idea? Hum Behav Emerg Technol 1(2):111–123. https://doi.org/10.1002/hbe2.147
Article Google Scholar - Mori M, MacDorman KF, Kageki N (2012) The uncanny valley [from the field]. IEEE Robot Autom Mag 19(2):98–100. https://doi.org/10.1109/MRA.2012.2192811
Article Google Scholar - Szondy M, Fazekas P (2024) Attachment to robots and therapeutic efficiency in mental health. Front psychol 15:1347177. https://doi.org/10.3389/fpsyg.2024.1347177
Article Google Scholar - Gouaillier D, Hugel V, Blazevic P, Kilner C, Monceaux J, Lafourcade P (2009) Mechatronic design of NAO humanoid. In 2009 IEEE International Conference on Robotics and Automation, pp 769–774. https://doi.org/10.1109/ROBOT.2009.5152516
- Pandey AK, Gelin R (2018) A mass-produced sociable humanoid robot: Pepper: the first machine of its kind. IEEE Robot Autom Mag 25(3):40–48. https://doi.org/10.1109/MRA.2018.2833157
Article Google Scholar - Hung L, Liu C, Woldum E, Au-Yeung A, Berndt A, Wallsworth C et al. (2019) The benefits of and barriers to using a social robot PARO in care settings: a scoping review. BMC GeriAtr 19:232. https://doi.org/10.1186/s12877-019-1244-6
Article Google Scholar - Fujita M (2004) On activating human communications with pet-type robot AIBO. Proc IEEE 92(11):1804–1813. https://doi.org/10.1109/JPROC.2004.835364
Article Google Scholar - Desai M, Tsui KM, Yanco HA, Uhlik C (2011) Essential features of telepresence robots. In 2011 IEEE Conference on Technologies for Practical Robot Applications, pp 15–20. https://doi.org/10.1109/TEPRA.2011.5753474
- Luperto M, Monroy J, Renoux J, Lunardini F, Basilico N, Bulgheroni M et al. (2022) Integrating social assistive robots, IoT, virtual communities and smart objects to assist at-home independently living elders: the MoveCare project. Int J Soc robot 15(3):517–545. https://doi.org/10.1007/s12369-021-00843-0
Article Google Scholar - Mayer P, Beck C, Panek P (2012) Examples of multimodal user interfaces for socially assistive robots in ambient assisted living environments. In 2012 IEEE 3rd International Conference on Cognitive Infocommunications (CogInfoCom), pp 401–406. https://doi.org/10.1109/CogInfoCom.2012.6422014
- Lee OE, Lee H, Park A, Choi NG (2024) My precious friend: human-robot interactions in home care for socially isolated older adults. Clin gerontol 47(1):161–170. https://doi.org/10.1080/07317115.2022.2156829
Article Google Scholar - O (2025) A new standard in virtual care: OhmniCare. https://ohmnilabs.com/products/ohmnicare-mobile-telehealth-robot/. Accessed 31 Mar 2025
- Hyodol (2025) HYODOL: AI care platform. https://en.hyodol.com/. Accessed 31 Mar 2025
- De Carolis B, Ferilli S, Macchiarulo N (2022) Ambient assisted living and social robots: towards learning relations between user’s daily routines and mood. In Adjunct proceedings of the 30th ACM conference on user modeling, adaptation and personalization, pp 123–129. https://doi.org/10.1145/3511047.3537691
- Phillips E, Zhao X, Ullman D, Malle BF (2018) What is human-like? Decomposing robots’ human-like appearance using the anthropomorphic robot (abot) database. In Proc 2018 ACM/IEEE Int Conf Hum-Robot Interact, pp 105–113. https://doi.org/10.1145/3171221.3171268
- Calvaresi D, Cesarini D, Sernani P, Marinoni M, Dragoni AF, Sturm A (2017) Exploring the ambient assisted living domain: a systematic review. J Ambient Intell Humaniz Comput 8:239–257. https://doi.org/10.1007/s12652-016-0374-3
Article Google Scholar - Dragone M, Abdel-Naby S, Swords D, O’Hare GM, Broxvall M (2012) A programming framework for multi-agent coordination of robotic ecologies. In Int Workshop on Programming Multi-Agent Systems, pp 72–89. https://doi.org/10.1007/978-3-642-38700-5_5
- Marques G, Pires IM, Miranda N, Pitarma R (2019) Air quality monitoring using assistive robots for ambient assisted living and enhanced living environments through internet of things. Electron 8(12):1375. https://doi.org/10.3390/electronics8121375
Article Google Scholar - Bui HD, Chong NY (2018) An integrated approach to human-robot-smart environment interaction interface for ambient assisted living. In 2018 IEEE Workshop on Advanced Robotics and its Social Impacts (ARSO), pp 32–37. https://doi.org/10.1109/ARSO.2018.8625821
- Feingold-Polak R, Barzel O, Levy-Tzedek S (2024) Socially assistive robot for stroke rehabilitation: a long-term in-the-wild Pilot randomized controlled trial. IEEE Trans Neural Syst Rehabil Eng 32:1616–1626. https://doi.org/10.1109/TNSRE.2024.3387320
Article Google Scholar - Spitale M, Silleresi S, Garzotto F, Matarić MJ (2023) Using socially assistive robots in speech-language therapy for children with language impairments. Int J Soc robot 15(9):1525–1542. https://doi.org/10.1007/s12369-023-01028-7
Article Google Scholar - Elfaki AO, Abduljabbar M, Ali L, Alnajjar F, Mehiar DA, Marei AM et al. (2023) Revolutionizing social robotics: a cloud-based framework for enhancing the intelligence and autonomy of social robots. Robot 12(2):48. https://doi.org/10.3390/robotics12020048
Article Google Scholar - Malfaz M, Castro-González Á, Barber R, Salichs MA (2011) A biologically inspired architecture for an autonomous and social robot. IEEE T Auton Ment De 3(3):232–246. https://doi.org/10.1109/TAMD.2011.2112766
Article Google Scholar - Shukla J, Barreda-Ángeles M, Oliver J, Puig D (2017) Effectiveness of socially assistive robotics during cognitive stimulation interventions: impact on caregivers. In 2017 26th IEEE Int Symp Robot Human Interact Commun (RO-MAN), pp 62–67. https://doi.org/10.1109/ROMAN.2017.8172281
- Stade EC, Stirman SW, Ungar LH, Boland CL, Schwartz HA, Yaden DB et al. (2024) Large language models could change the future of behavioral healthcare: a proposal for responsible development and evaluation. Npj Ment Health Res 3:12. https://doi.org/10.1038/s44184-024-00056-z
Article Google Scholar - Abdollahi H, Mahoor MH, Zandie R, Siewierski J, Qualls SH (2022) Artificial emotional intelligence in socially assistive robots for older adults: a pilot study. IEEE Trans Affective Comput 14(3):2020–2032. https://doi.org/10.1109/TAFFC.2022.3143803
Article Google Scholar - Céspedes N, Irfan B, Senft E, Cifuentes CA, Gutierrez LF, Rincon-Roncancio M et al. (2021) A socially assistive robot for long-term cardiac rehabilitation in the real world. Front neurorobot 15:633248. https://doi.org/10.3389/fnbot.2021.633248
Article Google Scholar - Heredia J, Lopes-Silva E, Cardinale Y, Diaz-Amado J, Dongo I, Graterol W, Aguilera A (2022) Adaptive multimodal emotion detection architecture for social robots. IEEE Access. 10:20727–20744. https://doi.org/10.1109/ACCESS.2022.3149214
Article Google Scholar - Di Napoli C, Rossi S (2019) A layered architecture for socially assistive robotics as a service. In 2019 IEEE Int Conf Syst Man Cybern, pp 352–357. https://doi.org/10.1109/SMC.2019.8914532
- Umbrico A, Cesta A, Cortellessa G, Orlandini A (2020) A holistic approach to behavior adaptation for socially assistive robots. Int J Soc robot 12(3):617–637. https://doi.org/10.1007/s12369-019-00617-9
Article Google Scholar - Dell’anna D, Jamshidnejad A (2022) Evolving fuzzy logic systems for creative personalized socially assistive robots. Eng Appl Artif Intell 114:105064. https://doi.org/10.1016/j.engappai.2022.105064
Article Google Scholar - Maroto-Gómez M, Alonso-Martín F, Malfaz M, Castro-González Á, Castillo JC, Salichs MÁ (2023) A systematic literature review of decision-making and control systems for autonomous and social robots. Int J Soc robot 15(5):745–789. https://doi.org/10.1007/s12369-023-00977-3
Article Google Scholar - Castro-González Á, Malfaz M, Gorostiza JF, Salichs MA (2014) Learning behaviors by an autonomous social robot with motivations. Cybern Syst 45(7):568–598. https://doi.org/10.1080/01969722.2014.945321
Article Google Scholar - Formosa P (2021) Robot autonomy vs. human autonomy: social robots, artificial intelligence (AI), and the nature of autonomy. Minds mach 31(4):595–616. https://doi.org/10.1007/s11023-021-09579-2
Article Google Scholar - Van Minkelen P, Gruson C, Van Hees P, Willems M, De Wit J, Aarts R et al. (2020) Using self-determination theory in social robots to increase motivation in L2 word learning. In Proc 2020 ACM/IEEE Int Conf Hum-Robot Interact, pp 369–377. https://doi.org/10.1145/3319502.3374828
- Feil-Seifer D, Skinner K, Matarić MJ (2007) Benchmarks for evaluating socially assistive robotics. Interact Stud 8(3):423–439. https://doi.org/10.1075/is.8.3.07fei
Article Google Scholar - Gary K, Stoll R, Rallabhandi P, Patwardhan M, Hamel D, Amresh A et al. (2017) Mhealth games as rewards: incentive or distraction? Proc 2017 Int Conf Digit Health 209–210. https://doi.org/10.1145/3079452.3079459
- Lewis ZH, Swartz MC, Lyons EJ (2016) What’s the point?: a review of reward systems implemented in gamification interventions. Games Health J 5(2):93–99. https://doi.org/10.1089/g4h.2015.0078
Article Google Scholar - Benedictis RD, Umbrico A, Fracasso F, Cortellessa G, Orlandini A, Cesta A (2023) A dichotomic approach to adaptive interaction for socially assistive robots. User Model User-Adapt Interact 33(2):293–331. https://doi.org/10.1007/s11257-022-09347-6
Article Google Scholar - Shi Z, Groechel TR, Jain S, Chima K, Rudovic O, Matarić MJ (2022) Toward personalized affect-aware socially assistive robot tutors for long-term interventions with children with autism. ACM Trans Hum-Robot Interact 11(4):1–28. https://doi.org/10.1145/3526111
Article Google Scholar - Hornstein S, Zantvoort K, Lueken U, Funk B, Hilbert K (2023) Personalization strategies in digital mental health interventions: a systematic review and conceptual framework for depressive symptoms. Front Digit Health 5:1170002. https://doi.org/10.3389/fdgth.2023.1170002
Article Google Scholar - Moro C, Nejat G, Mihailidis A (2018) Learning and personalizing socially assistive robot behaviors to aid with activities of daily living. ACM Trans Hum-Robot Interact 7(2):1–25. https://doi.org/10.1145/3277903
Article Google Scholar - Van Wingerden E, Barakova E, Lourens T, Sterkenburg PS (2021) Robot-mediated therapy to reduce worrying in persons with visual and intellectual disabilities. J Appl Res Intellect Disabil 34(1):229–238. https://doi.org/10.1111/jar.12801
Article Google Scholar - Chawla NV, Davis DA (2013) Bringing big data to personalized healthcare: a patient-centered framework. J Gen Intern Med 28:660–665. https://doi.org/10.1007/s11606-013-2455-8
Article Google Scholar - NHS England (2025) The NHS AI Lab - NHS transformation directorate. https://transform.england.nhs.uk/ai-lab/. Accessed 4 Apr 2025
- Jackson L, Kuhlman C, Jackson F, Fox PK (2019) Including vulnerable populations in the assessment of data from vulnerable populations. Front Big Data 2:19. https://doi.org/10.3389/fdata.2019.00019
Article Google Scholar - Nimri R, Phillip M, Clements MA, Kovatchev B (2024) Closed-loop control, artificial intelligence-based decision-support Systems, and data science. Diabetes Technol Ther 26:S1. https://doi.org/10.1089/dia.2024.2505
Article Google Scholar - Beltz AM, Wright AG, Sprague BN, Molenaar PC (2016) Bridging the nomothetic and idiographic approaches to the analysis of clinical data. Assessment 23(4):447–458. https://doi.org/10.1177/1073191116648209
Article Google Scholar - Haynes SN, Mumma GH, Pinson C (2009) Idiographic assessment: conceptual and psychometric foundations of individualized behavioral assessment. Clin Psychol Rev 29(2):179–191. https://doi.org/10.1016/j.cpr.2008.12.003
Article Google Scholar - Piccirillo ML, Rodebaugh TL (2019) Foundations of idiographic methods in psychology and applications for psychotherapy. Clin Psychol Rev 71:90–100. https://doi.org/10.1016/j.cpr.2019.01.002
Article Google Scholar - Hurtienne J, Blessing L (2007) Design for intuitive use-testing image schema theory for user interface design. In: DS 42: proc ICED 2007, 16th int conf eng Des. Paris, pp 829–830
- Fisk AD, Czaja SJ, Rogers WA, Charness N, Sharit J (2009) Designing for older adults: principles and creative human factors approaches, 2nd edn edn. CRC Press, Boca Raton
- Jaramillo-Mediavilla L, Basantes-Andrade A, Cabezas-González M, Casillas-Martín S (2024) Impact of gamification on motivation and academic performance: a systematic review. Educ Sci 14(6):639. https://doi.org/10.3390/educsci14060639
Article Google Scholar - Rusz D, Le Pelley ME, Kompier MAJ, Mait L, Bijleveld E (2020) Reward-driven distraction: a meta-analysis. Psychol Bull 146(10):872–899. https://doi.org/10.1037/bul0000296
Article Google Scholar - Smith TW, Pittman TS (1978) Reward, distraction, and the overjustification effect. J Pers Soc psychol 36(5):565–572. https://doi.org/10.1037/0022-3514.36.5.565
Article Google Scholar - Vandemeulebroucke T, de Casterlé BD, Gastmans C (2018) How do older adults experience and perceive socially assistive robots in aged care: a systematic review of qualitative evidence. Aging Ment Health 22(2):149–167. https://doi.org/10.1080/13607863.2017.1286455
Article Google Scholar - Feingold-Polak R, Barzel O, Levy-Tzedek S (2021) A robot goes to rehab: a novel gamified system for long-term stroke rehabilitation using a socially assistive robot-methodology and usability testing. J Neuroeng Rehabil 18:122. https://doi.org/10.1186/s12984-021-00915-2
Article Google Scholar - Markelius A, Sjöberg S, Bergström M, Ravandi BS, Vivas AB, Khan I, Lowe R (2024) Differential outcomes training of visuospatial memory: a gamified approach using a socially assistive robot. Int J Soc robot 16(2):363–384. https://doi.org/10.1007/s12369-023-01083-0
Article Google Scholar - Robson K, Plangger K, Kietzmann JH, McCarthy I, Pitt L (2015) Is it all a game? Understanding the principles of gamification. Bus. Horiz 58(4):411–420. https://doi.org/10.1016/j.bushor.2015.03.006
Article Google Scholar - Seaborn K, Fels DI (2015) Gamification in theory and action: a survey. Int J Hum Comput Stud 74:14–31. https://doi.org/10.1016/j.ijhcs.2014.09.006
Article Google Scholar - Donnermann M, Lein M, Messingschlager T, Riedmann A, Schaper P, Steinhaeusser S, Lugrin B (2021) Social robots and gamification for technology supported learning: an empirical study on engagement and motivation. Comput Hum Behav 121:106792. https://doi.org/10.1016/j.chb.2021.106792
Article Google Scholar - Sardi L, Idri A, Fernández-Alemán JL (2017) A systematic review of gamification in e-health. J Biomed Inf 71:31–48. https://doi.org/10.1016/j.jbi.2017.05.011
Article Google Scholar - Kim B (2015) Designing gamification in the right way. Libr Technol Rep 51(2):29–35
Google Scholar - Borghouts J, Eikey E, Mark G, De Leon C, Schueller SM, Schneider M et al. (2021) Barriers to and facilitators of user engagement with digital mental health interventions: systematic review. J Med Internet Res 23(3):e24387. https://doi.org/10.2196/24387
Article Google Scholar - Miralles I, Granell C (2019) Considerations for designing context-aware mobile apps for mental health interventions. Int J Environ Res Public Health 16(7):1197. https://doi.org/10.3390/ijerph16071197
Article Google Scholar - Aranda Jan CB, Jagtap S, Moultrie J (2016) Towards a framework for holistic contextual design for low-resource settings. Int J Des 10(3. https://doi.org/10.17863/CAM.7254
- Feng Y, Perugia G, Yu S, Barakova EI, Hu J, Rauterberg GM (2022) Context-enhanced human-robot interaction: exploring the role of system interactivity and multimodal stimuli on the engagement of people with dementia. Int J Soc robot 14:807–826. https://doi.org/10.1007/s12369-021-00823-4
Article Google Scholar - Feng Y, Barakova EI, Yu S, Hu J, Rauterberg GM (2020) Effects of the level of interactivity of a social robot and the response of the augmented reality display in contextual interactions of people with dementia. Sens 20(13):3771. https://doi.org/10.3390/s20133771
Article Google Scholar - Lima MR, Wairagkar M, Natarajan N, Vaitheswaran S, Vaidyanathan R (2021) Robotic telemedicine for mental health: a multimodal approach to improve human-robot engagement. Front Robot AI 8:618866. https://doi.org/10.3389/frobt.2021.618866
Article Google Scholar - Dou X, Yan L, Wu K, Niu J (2022) Effects of voice and lighting color on the social perception of home healthcare robots. Appl Sci 12(23):12191. https://doi.org/10.3390/app122312191
Article Google Scholar - Nahum-Shani I, Smith SN, Spring BJ, Collins LM, Witkiewitz K, Tewari A, Murphy SA (2018) Just-in-time adaptive interventions (JITAIs) in mobile health: key components and design principles for ongoing health behavior support. ann. behav. Med. https://doi.org/10.1007/s12160-016-9830-8
Article Google Scholar - Muller MJ, Kuhn S (1993) Participatory design. Commun. ACM 36(6):24–28
Article Google Scholar - Spinuzzi C (2005) The methodology of participatory design. Tech Commun 52(2):163–174
Google Scholar - Benz C, Scott-Jeffs W, McKercher KA, Welsh M, Norman R, Hendrie D et al. (2024) Community-based participatory-research through co-design: supporting collaboration from all sides of disability. Res Involv Engagem 10(1):47. https://doi.org/10.1186/s40900-024-00573-3
Article Google Scholar - Björgvinsson E, Ehn P, Hillgren PA (2010) Participatory design and “democratizing innovation”. In Proc 11th Biennial Participatory Design Conference, pp 41–50. https://doi.org/10.1145/1900441.1900448
- Šabanović S, Chang WL, Bennett CC, Piatt JA, Hakken D (2015) A robot of my own: participatory design of socially assistive robots for independently living older adults diagnosed with depression. In: Human aspects of IT for the aged population. Springer, pp 104–114. https://doi.org/10.1007/978-3-319-20892-3_11
Chapter Google Scholar - Winkle K, Caleb-Solly P, Turton A, Bremner P (2020) Mutual shaping in the design of socially assistive robots: a case study on social robots for therapy. Int J Soc robot 12:847–866. https://doi.org/10.1007/s12369-019-00536-9
Article Google Scholar - Leslie M, Paradis E, Gropper MA, Kitto S, Reeves S, Pronovost P (2017) An ethnographic study of health information technology use in three intensive care units. Health Serv Res 52(4):1330–1348. https://doi.org/10.1111/1475-6773.12466
Article Google Scholar - Mast M, Burmester M, Krüger K, Fatikow S, Arbeiter G, Graf B et al. (2012) User-centered design of a dynamic-autonomy remote interaction concept for manipulation-capable robots to assist elderly people in the home. J Hum Rob Interact 1(1):96–118. https://doi.org/10.5898/JHRI.1.1.Mast
Article Google Scholar - Laban G, Kappas A, Morrison V, Cross ES (2024) Building long-term human-robot relationships: examining disclosure, perception and well-being across time. Int J Soc robot 16(5):1–27. https://doi.org/10.1007/s12369-023-01076-z
Article Google Scholar - Prescott TJ, Robillard JM (2021) Are friends electric? The benefits and risks of human-robot relationships. iScience 24(1):101993. https://doi.org/10.1016/j.isci.2020.101993
Article Google Scholar - Birnbaum GE, Mizrahi M, Hoffman G, Reis HT, Finkel EJ, Sass O (2016) What robots can teach us about intimacy: the reassuring effects of robot responsiveness to human disclosure. Comput Hum Behav 63:416–423. https://doi.org/10.1016/j.chb.2016.05.064
Article Google Scholar - Lin TH, Dinner H, Leung TL, Mutlu B, Trafton JG, Sebo S (2025) Connection-coordination rapport (CCR) scale: a dual-factor scale to measure human-robot rapport. arXiv preprint arXiv: 2501.11887. https://doi.org/10.48550/arXiv.2501.11887
Book Google Scholar - De Graaf MM (2016) An ethical evaluation of human-robot relationships. Int J Soc robot 8:589–598. https://doi.org/10.1007/s12369-016-0368-5
Article Google Scholar - Borenstein J, Arkin R (2019) Robots, ethics, and intimacy: the need for scientific research. In: Berkich D, d’Alfonso M (eds) On the cognitive, ethical, and scientific dimensions of artificial intelligence. Philosophical studies series, vol 134. Springer, Cham. https://doi.org/10.1007/978-3-030-01800-9_16
Chapter Google Scholar - Kok BC, Soh H (2020) Trust in robots: challenges and opportunities. Curr Robot Rep 1(4):297–309. https://doi.org/10.1007/s43154-020-00029-y
Article Google Scholar - Langer A, Feingold-Polak R, Mueller O, Kellmeyer P, Levy-Tzedek S (2019) Trust in socially assistive robots: considerations for use in rehabilitation. Neurosci Biobehav Rev 104:231–239. https://doi.org/10.1016/j.neubiorev.2019.07.014
Article Google Scholar - Hoff KA, Bashir M (2015) Trust in automation: integrating empirical evidence on factors that influence trust. Hum Factors 57(3):407–434. https://doi.org/10.1177/0018720814547570
Article Google Scholar - Hancock PA, Kessler TT, Kaplan AD, Brill JC, Szalma JL (2021) Evolving trust in robots: specification through sequential and comparative meta-analyses. Hum Factors 63(7):1196–1229. https://doi.org/10.1177/0018720820922080
Article Google Scholar - Wang X, Zhou R, Zhang R (2020) The impact of expectation and disconfirmation on user experience and behavior intention. In: Design, user experience, and usability. Interaction design: 9th int conf, DUXU 2020. Springer, Cham, pp 464–475. https://doi.org/10.1007/978-3-030-49713-2_32
Chapter Google Scholar - Leichtmann B, Humer C, Hinterreiter A, Streit M, Mara M (2023) Effects of explainable artificial intelligence on trust and human behavior in a high-risk decision task. Comput Hum Behav 139:107539. https://doi.org/10.1016/j.chb.2022.107539
Article Google Scholar - Mahbooba B, Timilsina M, Sahal R, Serrano M (2021) Explainable artificial intelligence (XAI) to enhance trust management in intrusion detection systems using decision tree model. Complexity 6634811. https://doi.org/10.1155/2021/6634811
- Lockey S, Gillespie N, Holm D, Someh IA (2021) A review of trust in artificial intelligence: challenges, vulnerabilities and future directions. Proc 54th Hawaii Int Conf Syst Sci 5463–5472
- Ferrario A, Sedlakova J, Trachsel M (2024) The role of humanization and robustness of large language models in conversational artificial intelligence for individuals with depression: a critical analysis. JMIR Ment Health 11:e56569. https://doi.org/10.2196/56569
Article Google Scholar - Spatola N, Cherif E (2023) Spontaneous humanization of robots in passive observation of human-robot interaction: a path toward ethical consideration and human-robot cooperation. Comput Hum Behav Artif Humans 1(2):100012. https://doi.org/10.1016/j.chbah.2023.100012
Article Google Scholar - Bradwell HL, Winnington R, Thill S, Jones RB (2021) Morphology of socially assistive robots for health and social care: a reflection on 24 months of research with anthropomorphic, zoomorphic and mechanomorphic devices. In 2021 30th IEEE Int Conf Robot Human Interact Commun (RO-MAN), pp 376–383. https://doi.org/10.1109/RO-MAN50785.2021.9515446
- Randall N, Bennett CC, Šabanović S, Nagata S, Eldridge L, Collins S, Piatt JA (2019) More than just friends: in-home use and design recommendations for sensing socially assistive robots (SARs) by older adults with depression. Paladyn J Behav robot 10(1):237–255. https://doi.org/10.1515/pjbr-2019-0020
Article Google Scholar - Van Straten CL, Peter J, Kühne (2020) Child-robot relationship formation: a narrative review of empirical research. Int J Soc robot 12(2):325–344. https://doi.org/10.1007/s12369-019-00569-0
Article Google Scholar - Grodniewicz JP, Hohol M (2023) Waiting for a digital therapist: three challenges on the path to psychotherapy delivered by artificial intelligence. Front Psychiatry 14:1190084. https://doi.org/10.3389/fpsyt.2023.1190084
Article Google Scholar - Kim SK, Jang JW, Hwang YS, Lee OE, Jo HS (2023) Investigating the effectiveness of socially assistive robot on depression and cognitive functions of community dwelling older adults with cognitive impairments. Assist Technol 1–9. https://doi.org/10.1080/10400435.2023.2237554
- Wang Y, Wang Q, Sun X, Cao C, Sun R (2024) Sleep elf: pillow robot that pats and sings child to sleep mimicking a parent. In Companion of the 2024 ACM/IEEE Int Conf Hum-Robot Interact, pp 1100–1104.
- Robert LP (2017) The growing problem of humanizing robots. Int Robot Autom J 3(1):00043. https://doi.org/10.15406/iratj.2017.03.00043
Article MathSciNet Google Scholar - Sim DYY, Loo CK (2015) Extensive assessment and evaluation methodologies on assistive social robots for modelling human-robot interaction-A review. Inf Sci 301:305–344. https://doi.org/10.1016/j.ins.2014.12.017
Article Google Scholar - Guo C, Ashrafian H, Ghafur S, Fontana G, Gardner C, Prime M (2020) Challenges for the evaluation of digital health solutions-a call for innovative evidence generation approaches. NPJ Digit Med 3(1):110. https://doi.org/10.1038/s41746-020-00314-2
Article Google Scholar - Shibata T, Hung L, Petersen S, Darling K, Inoue K, Martyn K et al. (2021) PARO as a biofeedback medical device for mental health in the COVID-19 Era. Sustainability 13(20):11502. https://doi.org/10.3390/su132011502
Article Google Scholar - Murray E, Hekler EB, Andersson G, Collins LM, Doherty A, Hollis C et al. (2016) Evaluating digital health interventions: key questions and approaches. Am J Prev Med 51(5):843–851. https://doi.org/10.1016/j.amepre.2016.06.008
Article Google Scholar - Asl AM, Ulate MM, Franco Martin M, van der Roest H (2022) Methodologies used to study the feasibility, usability, efficacy, and effectiveness of social robots for elderly adults: scoping review. J Med Internet Res 24(8):e37434. https://doi.org/10.2196/37434
Article Google Scholar - Chen SC, Moyle W, Jones C, Petsky H (2020) A social robot intervention on depression, loneliness, and quality of life for Taiwanese older adults in long-term care. Int PsychogeriAtr 32(8):981–991. https://doi.org/10.1017/S1041610220000459
Article Google Scholar - Hornbæk K (2006) Current practice in measuring usability: challenges to usability studies and research. Int J Hum Comput Stud 64(2):79–102. https://doi.org/10.1016/j.ijhcs.2005.06.002
Article Google Scholar - Anzalone SM, Boucenna S, Ivaldi S, Chetouani M (2015) Evaluating the engagement with social robots. Int J Soc robot 7:465–478. https://doi.org/10.1007/s12369-015-0298-7
Article Google Scholar - Daley K, Hungerbuehler I, Cavanagh K, Claro HG, Swinton PA, Kapps M (2020) Preliminary evaluation of the engagement and effectiveness of a mental health chatbot. Front Digit Health 2:576361. https://doi.org/10.3389/fdgth.2020.576361
Article Google Scholar - Perugia G, Rodríguez-Martín D, Boladeras MD, Mallofré AC, Barakova E, Rauterberg M (2018) Quantity of movement as a measure of engagement for dementia: the influence of motivational disorders. Am J Alzheimers Dis Other demen 33(2):112–121. https://doi.org/10.1177/1533317517739700
Article Google Scholar - Perugia G, Díaz-Boladeras M, Catala-Mallofré A, Barakova EI, Rauterberg M (2020) ENGAGE-DEM: a model of engagement of people with dementia. IEEE Trans Affect Comput 13(2):926–943. https://doi.org/10.1109/TAFFC.2020.2980275
Article Google Scholar - Puglisi A, Caprì T, Pignolo L, Gismondo S, Chilà P, Minutoli R et al. (2022) Social humanoid robots for children with autism spectrum disorders: a review of modalities, indications, and pitfalls. Children 9(7):953. https://doi.org/10.3390/children9070953
Article Google Scholar - Rakhymbayeva N, Amirova A, Sandygulova A (2021) A long-term engagement with a social robot for autism therapy. Front Robot AI 8:669972. https://doi.org/10.3389/frobt.2021.669972
Article Google Scholar - Cohen L, Khoramshahi M, Salesse RN, Bortolon C, Słowiński P, Zhai C et al. (2017) Influence of facial feedback during a cooperative human-robot task in schizophrenia. Sci Rep 7(1):15023. https://doi.org/10.1038/s41598-017-14773-3
Article Google Scholar - Venkatesh V, Morris MG, Davis GB, Davis FD (2003) User acceptance of information technology: toward a unified view. Mis Q 27(3):425–478. https://doi.org/10.2307/30036540
Article Google Scholar - Bangor A, Kortum PT, Miller JT (2008) An empirical evaluation of the system usability scale. Int J Hum Comput Interact 24(6):574–594. https://doi.org/10.1080/10447310802205776
Article Google Scholar - Shiffman S, Stone AA, Hufford MR (2008) Ecological momentary assessment. Annu Rev Clin psychol 4(1):1–32. https://doi.org/10.1146/annurev.clinpsy.3.022806.091415
Article Google Scholar - Klęczek K, Rice A, Alimardani M (2024) Robots as mental health coaches: a study of emotional responses to technology-assisted stress management tasks using physiological signals. Sens 24(13):4032. https://doi.org/10.3390/s24134032
Article Google Scholar - Lanius C, Weber R, Robinson J (2021) User experience methods in research and practice. J Tech Writ Comm 51(4):350–379. https://doi.org/10.1177/00472816211044499
Article Google Scholar - Shourmasti ES, Colomo-Palacios R, Holone H, Demi S (2021) User experience in social robots. Sens 21(15):5052. https://doi.org/10.3390/s21155052
Article Google Scholar - De Graaf MM, Ben Allouch S, Van Dijk JA (2016) Long-term evaluation of a social robot in real homes. Interact Stud 17(3):462–491. https://doi.org/10.1075/is.17.3.08deg
Article Google Scholar - Haddadin S (2015) Physical safety in robotics. In: Koopman P, Schröder L, Bardenhagen F, Böhm S, Lüttgen G (eds) Formal modeling and verification of cyber-physical systems. Springer, Cham, pp 249–271. https://doi.org/10.1007/978-3-658-09994-7_9
Chapter Google Scholar - Vöcking M, Karrenbrock A, Beckmann A, Vondeberg C, Obert L, Hemming B et al. (2024) Emotional and psychological safety in healthcare digitalization: a design ethnographic study. Int J Public Health 69:1607575. https://doi.org/10.3389/ijph.2024.1607575
Article Google Scholar - Filkins BL, Kim JY, Roberts B, Armstrong W, Miller MA, Hultner ML et al. (2016) Privacy and security in the era of digital health: what should translational researchers know and do about it? Am J Transl Res 8(3):1560
Google Scholar - Grande D, Marti XL, Feuerstein-Simon R, Merchant RM, Asch DA, Lewson A, Cannuscio CC (2020) Health policy and privacy challenges associated with digital technology. JAMA Netw Open 3(7):e208285. https://doi.org/10.1001/jamanetworkopen.2020.8285
Article Google Scholar - Chinniah Y (2016) Robot safety: overview of risk assessment and reduction. Adv Robot Autom 5(1):1–5
Google Scholar - Voigt P, Von Dem Bussche A (2017) The EU general Data Protection regulation (GDPR): a practical Guide. Springer, Cham
Book Google Scholar - Hattori T, Nakamura M, Kawamura K, Nihei M (2021) Determining possible risks of introducing socially assistive robots with mobility functions to aged care facilities. Int Conf Hum-Comput Interact 145–155. https://doi.org/10.1007/978-3-030-78108-8_11
- Akalin N, Kiselev A, Kristoffersson A, Loutfi A (2023) A taxonomy of factors influencing perceived safety in human-robot interaction. Int J Soc robot 15(12):1993–2004. https://doi.org/10.1007/s12369-023-01027-8
Article Google Scholar - Akalin N, Kristoffersson A, Loutfi A (2022) Do you feel safe with your robot? Factors influencing perceived safety in human-robot interaction based on subjective and objective measures. Int J Hum Comput Stud 158:102744. https://doi.org/10.1016/j.ijhcs.2021.102744
Article Google Scholar - Nyholm L, Santamäki-Fischer R, Fagerström L (2021) Users’ ambivalent sense of security with humanoid robots in healthcare. Inf Health Soc Care 46(2):218–226. https://doi.org/10.1080/17538157.2021.1883027
Article Google Scholar - Wang Y, Wang G, Ge W, Duan J, Chen Z, Wen L (2024) Perceived safety assessment of interactive motions in human-soft robot interaction. Biomimetics 9(1):58. https://doi.org/10.3390/biomimetics9010058
Article Google Scholar - Larman C, Basili VR (2003) Iterative and incremental developments. A brief history. Computer 36(6):47–56. https://doi.org/10.1109/MC.2003.1204375
Article Google Scholar - Henderson K, Reihm J, Koshal K, Wijangco J, Sara N, Miller N et al. (2024) A closed-loop digital health tool to improve depression care in multiple sclerosis: iterative design and Cross-sectional Pilot randomized controlled trial and its impact on depression care. JMIR Form Res 8:e52809. https://doi.org/10.2196/52809
Article Google Scholar - Katsulis Z, Ergai A, Leung WY, Schenkel L, Rai A, Adelman J et al. (2016) Iterative user centered design for development of a patient-centered fall prevention toolkit. Appl Ergon 56:117–126. https://doi.org/10.1016/j.apergo.2016.03.011
Article Google Scholar - Wynn DC, Eckert CM (2017) Perspectives on iteration in design and development. Res Eng Des 28:153–184. https://doi.org/10.1007/s00163-016-0226-3
Article Google Scholar - Sherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL et al. (2016) Real-world evidence-what is it and what can it tell us. N Engl J Med 375(23):2293–2297. https://doi.org/10.1056/nejmsb1609216
Article Google Scholar - Ford I, Norrie J (2016) Pragmatic trials. N Engl J Med 375(5):454–463. https://doi.org/10.1056/NEJMra1510059
Article Google Scholar - Blease C, Torous J (2023) ChatGPT and mental healthcare: balancing benefits with risks of harms. BMJ Ment Health 26(1. https://doi.org/10.1136/bmjment-2023-300884
- Vemprala SH, Bonatti R, Bucker A, Kapoor A (2024) ChatGPT for robotics: design principles and model abilities. IEEE Access. 12:55682–55696. https://doi.org/10.1109/ACCESS.2024.3387941
Article Google Scholar - Bertacchini F, Demarco F, Scuro C, Pantano P, Bilotta E (2023) A social robot connected with chatGPT to improve cognitive functioning in ASD subjects. Front psychol 14:1232177. https://doi.org/10.3389/fpsyg.2023.1232177
Article Google Scholar - Lee YK, Jung Y, Kang G, Hahn S (2023) Developing social robots with empathetic non-verbal cues using large language models. arXiv preprint arXiv: 2308.16529. https://doi.org/10.48550/arXiv.2308.16529
- Voultsiou E, Vrochidou E, Moussiades L, Papakostas GA (2025) The potential of large language models for social robots in special education. Prog Artif Intell. https://doi.org/10.1007/s13748-025-00363-2
Article Google Scholar - Shi Z, Landrum E, O’Connell A, Kian M, Pinto-Alva L, Shrestha K et al. (2024) How can large language models enable better socially assistive human-robot interaction: a brief survey. AAAI Symp Ser 3(1):401–404. https://doi.org/10.1609/aaaiss.v3i1.31245
Article Google Scholar - Lewis P, Perez E, Piktus A, Petroni F, Karpukhin V, Goyal N et al. (2020) Retrieval-augmented generation for knowledge-intensive nlp tasks. NeurIPS 33:9459–9474
Google Scholar - Anthropic (2024) Introducing the Model context protocol. https://www.anthropic.com/news/model-context-protocol. Accessed 11 Apr 2025
- Hou X, Zhao Y, Wang S, Wang H (2025) Model context protocol (MCP): landscape, security threats, and future research directions. arXiv preprint arXiv: 2503.23278. https://doi.org/10.48550/arXiv.2503.23278
- Singh A, Ehtesham A, Kumar S, Khoei TT (2025) A survey of the Model context protocol (MCP): standardizing context to enhance large language models (LLMs. https://doi.org/10.20944/preprints202504.0245.v1. Accessed 11 Apr 2025
- Siddals S, Torous J, Coxon A (2024) “It happened to be the perfect thing”: experiences of generative AI chatbots for mental health. Npj Ment Health Res 3:48. https://doi.org/10.1038/s44184-024-00097-4
Article Google Scholar - Fulmer R (2019) Artificial intelligence and counseling: four levels of implementation. Theory psychol 29(6):807–819. https://doi.org/10.1177/0959354319853045
Article Google Scholar - PYMNTS (2024) OpenAI Director says artificial General intelligence May Be 5 years out. https://www.pymnts.com/artificial-intelligence-2/2024/openai-board-member-agi-is-five-to-15-years-away/. Accessed 14 Oct 2024
- Fiske A, Henningsen P, Buyx A (2019) Your robot therapist will see you now: ethical implications of embodied artificial intelligence in psychiatry, psychology, and psychotherapy. J Med Internet Res 21(5):e13216. https://doi.org/10.2196/13216
Article Google Scholar - Sedlakova J, Trachsel M (2023) Conversational artificial intelligence in psychotherapy: a new therapeutic tool or agent? Am J bioeth 23(5):4–13. https://doi.org/10.1080/15265161.2022.2048739
Article Google Scholar - Winkle K, Caleb-Solly P, Leonards U, Turton A, Bremner P (2021) Assessing and addressing ethical risk from anthropomorphism and deception in socially assistive robots. In Proc 2021 ACM/IEEE Int Conf Hum-Robot Interact, pp 101–109. https://doi.org/10.1145/3434073.3444666
- Boada JP, Maestre BR, Genís CT (2021) The ethical issues of social assistive robotics: a critical literature review. Technol Soc 67:101726. https://doi.org/10.1016/j.techsoc.2021.101726
Article Google Scholar - Khawaja Z, Bélisle-Pipon JC (2023) Your robot therapist is not your therapist: understanding the role of AI-powered mental health chatbots. Front Digit Health 5:1278186. https://doi.org/10.3389/fdgth.2023.1278186
Article Google Scholar - Hou H, Leach K, Huang Y (2024) ChatGPT giving relationship advice-How reliable is it? Proc Int AAAI Conf Web Soc Media 18:610–623. https://doi.org/10.1609/icwsm.v18i1.31338
Article Google Scholar - Laestadius L, Bishop A, Gonzalez M, Illenčík D, Campos-Castillo C (2024) Too human and not human enough: a grounded theory analysis of mental health harms from emotional dependence on the social chatbot replika. New Media Soc 26(10):5923–5941. https://doi.org/10.1177/14614448221142007
Article Google Scholar - Ciarrochi J, Hernández C, Hill D, Ong C, Gloster AT, Levin ME et al. (2024) Process-based therapy: a common ground for understanding and utilizing therapeutic practices. J Psychother Integr 34(3):265–290. https://doi.org/10.1037/int0000348
Article Google Scholar - Smit K, Leewis S, Almoustafa H, Yildrim K, Uymaz T (2024) Enhancing educational dynamics integrating large language models with a social robot. Proc 2024 8th Int Conf Softw E-Bus 87–94. https://doi.org/10.1145/3715885.3715889
- Dalgleish T, Black M, Johnston D, Bevan A (2020) Transdiagnostic approaches to mental health problems: current status and future directions. J Consult Clin Psycho 88(3):179–195. https://doi.org/10.1037/ccp0000482
Article Google Scholar - Naik N, Hameed BM, Shetty DK, Swain D, Shah M, Paul R et al. (2022) Legal and ethical consideration in artificial intelligence in healthcare: who takes responsibility? Front Surg 9:862322. https://doi.org/10.3389/fsurg.2022.862322
Article Google Scholar - Pavithra N, Afza N (2024) Harnessing the power of artificial intelligence and robotics impact on attaining competitive advantage for sustainable development in hospitals with conclusions for future research approaches. GMS Hyg Infect Control 19:15. https://doi.org/10.3205/dgkh000470
Article Google Scholar