Chelsea Dobbins | The University of Queensland, Australia (original) (raw)

Papers by Chelsea Dobbins

Sensors, 2019

Mobile and wearable devices are capable of quantifying user behaviors based on their contextual s... more Mobile and wearable devices are capable of quantifying user behaviors based on their contextual sensor data. However, few indexing and annotation mechanisms are available, due to difficulties inherent in raw multivariate data types and the relative sparsity of sensor data. These issues have slowed the development of higher level human-centric searching and querying mechanisms. Here, we propose a pipeline of three algorithms. First, we introduce a spatio-temporal event detection algorithm. Then, we introduce a clustering algorithm based on mobile contextual data. Our spatio-temporal clustering approach can be used as an annotation on raw sensor data. It improves information retrieval by reducing the search space and is based on searching only the related clusters. To further improve behavior quantification, the third algorithm identifies contrasting events withina cluster content. Two large real-world smartphone datasets have been used to evaluate our algorithms and demonstrate the utility and resource efficiency of our approach to search.

2019 IEEE International Conference on Pervasive Computing and Communications (PerCom), 2019

Stress is a negative emotion that occurs in everyday life, such as driving. Recurrent exposure to... more Stress is a negative emotion that occurs in everyday life, such as driving. Recurrent exposure to stress can be detrimental to cardiovascular health in the long term. Nevertheless, the development of adaptive coping strategies can mitigate the influence of everyday stress on cardiovascular health. Understanding context is essential to modelling the occurrence of stress and other negative emotions during everyday life. However, driving is a highly dynamic environment, whereby the context is often described using ambiguous linguistic terms, which can be difficult to quantify. This paper proposes a Fuzzy Logic Mamdani Model to automatically estimate different categories of driving context. The system is comprised of two Membership Functions (MFs), which converts the inputs of speed and traffic density into linguistic variables. Our approach then uses these data to identify six states of driving – Idling, Journey Impedance, High Urban Workload, Low Urban Workload, High Non-Urban Workload and Low Non-Urban Workload. An interactive visualization has then been implemented that links this fuzzy logic model with psychophysiological data to identify the context of stress experienced on the road. The system has been validated using real-world data that has been collected from eight participants during their daily commuter journeys.

— Repeated experiences of negative emotions, such as stress, anger or anxiety, can have long-term... more — Repeated experiences of negative emotions, such as stress, anger or anxiety, can have long-term consequences for health. These episodes of negative emotion can be associated with inflammatory changes in the body, which are clinically relevant for the development of disease in the long-term. However, the development of effective coping strategies can mediate this causal chain. The proliferation of ubiquitous and unobtrusive sensor technology supports an increased awareness of those physiological states associated with negative emotion and supports the development of effective coping strategies. Smartphone and wearable devices utilise multiple on-board sensors that are capable of capturing daily behaviours in a permanent and comprehensive manner, which can be used as the basis for self-reflection and insight. However, there are a number of inherent challenges in this application, including unobtrusive monitoring, data processing, and analysis. This paper posits a mobile lifelogging platform that utilises wearable technology to monitor and classify levels of stress. A pilot study has been undertaken with six participants, who completed up to ten days of data collection. During this time, they wore a wearable device on the wrist during waking hours to collect instances of heart rate (HR) and Galvanic Skin Resistance (GSR). Preliminary data analysis was undertaken using three supervised machine learning algorithms: Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA) and Decision Tree (DT). An accuracy of 70% was achieved using the Decision Tree algorithm.

— Driving is an activity that can induce significant levels of negative emotion, such as stress a... more — Driving is an activity that can induce significant levels of negative emotion, such as stress and anger. These negative emotions occur naturally in everyday life, but frequent episodes can be detrimental to cardiovascular health in the long term. The development of monitoring systems to detect negative emotions often rely on labels derived from subjective self-report. However, this approach is burdensome, intrusive, low fidelity (i.e. scales are administered infrequently) and places huge reliance on the veracity of subjective self-report. This paper explores an alternative approach that provides greater fidelity by using psychophysiological data (e.g. heart rate) to dynamically label data derived from the driving task (e.g. speed, road type). A number of different techniques for generating labels for machine learning were compared: 1) deriving labels from subjective self-report and 2) labelling data via psychophysiological activity (e.g. heart rate (HR), pulse transit time (PTT), etc.) to create dynamic labels of high vs. low anxiety for each participant. The classification accuracy associated with both labelling techniques was evaluated using Linear Discriminant Analysis (LDA) and Support Vector Machines (SVM). Results indicated that classification of driving data using subjective labelled data (1) achieved a maximum AUC of 73%, whilst the labels derived from psychophysiological data (2) achieved equivalent performance of 74%. Whilst classification performance was similar, labelling driving data via psychophysiology offers a number of advantages over self-reports, e.g. implicit, dynamic, objective, high fidelity.

Background Video games are the most effective form of distraction from procedural pain compared t... more Background Video games are the most effective form of distraction from procedural pain compared to other distraction techniques, such as watching television or reading a book (Hussein, 2015). The degree of cognitive engagement with the game is a strong influence on the capacity of game-playing to distract from pain. By increasing game demand to a level that demands maximum levels of attention, it is possible to optimise distraction from pain; however, if the game becomes too difficult, it will fail to act as a distraction.

Mobile and wearable devices now have a greater capability of sensing human activity ubiquitously ... more Mobile and wearable devices now have a greater capability of sensing human activity ubiquitously and unobtrusively through advancements in miniaturization and sensing abilities. However, outstanding issues remain around the energy restrictions of these devices when processing large sets of data. This paper presents our approach that uses feature selection to refine the clustering of accelerometer data to detect physical activity. This also has a positive effect on the computational burden that is associated with processing large sets of data, as energy efficiency and resource use is decreased because less data is processed by the clustering algorithms. Raw accelerometer data, obtained from smartphones and smartwatches, have been preprocessed to extract both time and frequency domain features. Principle component analysis feature selection (PCAFS) and correlation feature selection (CFS) have been used to remove redundant features. The reduced feature sets have then been evaluated against three widely used clustering algorithms, including hierarchical clustering analysis (HCA), k-means, and density-based spatial clustering of applications with noise (DBSCAN). Using the reduced feature sets resulted in improved separability, reduced uncertainty, and improved efficiency compared with the baseline, which utilized all features. Overall, the CFS approach in conjunction with HCA produced higher Dunn Index results of 9.7001 for the phone and 5.1438 for the watch features, which is an improvement over the baseline. The results of this comparative study of feature selection and clustering, with the specific algorithms used, has not been performed previously and provides an optimistic and usable approach to recognize activities using either a smartphone or smartwatch.

Stress is a negative emotion that is part of everyday life. However, frequent episodes or prolong... more Stress is a negative emotion that is part of everyday life. However, frequent episodes or prolonged periods of stress can be detrimental to long-term health. Nevertheless, developing self-awareness is an important aspect of fostering effective ways to self-regulate these experiences. Mobile lifelogging systems provide an ideal platform to support self-regulation of stress by raising awareness of negative emotional states via continuous recording of psychophysiological and behavioural data. However, obtaining meaningful information from large volumes of raw data represents a significant challenge because these data must be accurately quantified and processed before stress can be detected. This work describes a set of algorithms designed to process multiple streams of lifelogging data for stress detection in the context of real world driving. Two data collection exercises have been performed where multimodal data, including raw cardiovascular activity and driving information, were collected from twenty-one people during daily commuter journeys. Our approach enabled us to 1) pre-process raw physiological data to calculate valid measures of heart rate variability, a significant marker of stress, 2) identify/correct artefacts in the raw physiological data and 3) provide a comparison between several classifiers for detecting stress. Results were positive and ensemble classification models provided a maximum accuracy of 86.9% for binary detection of stress in the real-world.

— The experience of negative emotions in everyday life, such as anger and anxiety, can have adver... more — The experience of negative emotions in everyday life, such as anger and anxiety, can have adverse effects on long-term cardiovascular health. However, objective measurements provided by mobile technology can promote insight into this psychobiological process and promote self-awareness and adaptive coping. It is postulated that the creation of a mobile lifelogging platform can support this approach by continuously recording personal data via mobile/wearable devices and processing this information to measure physiological correlates of negative emotions. This paper describes the development of a mobile lifelogging system that measures anxiety and anger during real-life driving. A number of data streams have been incorporated in the platform, including cardiovascular data, speed of the vehicle and first-person photographs of the environment. In addition, thirteen participants completed five days of data collection during daily commuter journeys to test the system. The design of the system hardware and associated data streams are described in the current paper, along with the results of preliminary data analysis.

This paper is a report of an investigation in which mobile technologies were used during large le... more This paper is a report of an investigation in which mobile technologies were used during large lecture-type sessions in tandem with the Textwall™ software. This online program is able to receive and present messages from stu-dents' devices that the lecturer may then share. It also can be used to collect student votes on multiple-choice questions, facilitating dynamic formative assessments during class. A study has been undertaken around the MyWallMate mobile application. This program has been developed by Liverpool John Moores University to expedite the process of sending messages and votes to Textwall™. Results of this study indicated that students had a positive reaction to both Textwall™ and the MyWallMate application. They reported being more comfortable in expressing their opinions via the MyWallMate mobile application and felt that using mobile technology within their lectures enhanced their learning. It has been concluded that the use of mobile technologies and BYOD-type approaches are avenues worth further exploration in global higher education.

Currently, personal assistant systems, run on smartphones and use natural language interfaces. Ho... more Currently, personal assistant systems, run on smartphones and use natural language interfaces. However, these systems rely mostly on the web for finding information. Mobile and wearable devices can collect an enormous amount of contextual personal data such as sleep and physical activities. These information objects and their applications are known as quantified-self, mobile health or personal informatics, and they can be used to provide a deeper insight into our behavior. To our knowledge, existing personal assistant systems do not support all types of quantified-self queries. In response to this, we have undertaken a user study to analyze a set of " textual questions/queries " that users have used to search their quantified-self or mobile health data. Through analyzing these questions, we have constructed a lightweight natural language based query interface-including a text parser algorithm and a user interface-to process the users' queries that have been used for searching quantified-self information. This query interface has been designed to operate on small devices, i.e. smartwatches, as well as augmenting the personal assistant systems by allowing them to process end users' natural language queries about their quantified-self data.

Obesity is a global health issue that affects 2.1 billion people worldwide and has an economic im... more Obesity is a global health issue that affects 2.1 billion people worldwide and has an economic impact of approximately $2 trillion. It is a disease that can make the aging process worse by impairing physical function, which can lead to people becoming more frail and immobile. Nevertheless, it is envisioned that technology can be used to aid in motivating behavioural changes to combat this preventable condition. The ubiquitous presence of wearable and mobile devices has enabled a continual stream of quantifiable data (e.g. physiological signals) to be collected about ourselves. This data can then be used to monitor physical activity to aid in self-reflection and motivation to alter behaviour. However, such information is susceptible to noise interference, which makes processing and extracting knowledge from such data challenging. This paper posits our approach that collects and processes physiological data that has been collected from tri-axial accelerometers and a heart-rate monitor, to detect physical activity. Furthermore, an end-user use case application has also been proposed that integrates these findings into a smartwatch visualisation. This provides a method of visualising the results to the user so that they are able to gain an overview of their activity. The goal of the paper has been to evaluate the performance of supervised machine learning in distinguishing physical activity. This has been achieved by (i) focusing on wearable sensors to collect data and using our methodology to process this raw lifelogging data so that features can be extracted/selected. (ii) Undertaking an evaluation between ten supervised learning classifiers to determine their accuracy in detecting human activity. To demonstrate the effectiveness of our method, this evaluation has been performed across a baseline method and two other methods. (iii) Undertaking an evaluation of the processing time of the approach and the smartwatch battery and network cost analysis between transferring data from the smartwatch to the phone. The results of the classifier evaluations indicate that our approach shows an improvement on existing studies, with accuracies of up to 99% and sensitivities of 100%.

This work introduces a set of scalable algorithms to identify patterns of human daily behaviors. ... more This work introduces a set of scalable algorithms to identify patterns of human daily behaviors. These patterns are extracted from multivariate temporal data that have been collected from smartphones. We have exploited sensors that are available on these devices, and have identified frequent behavioral patterns with a temporal granularity, which has been inspired by the way individuals segment time into events. These patterns are helpful to both end-users and third parties who provide services based on this information. We have demonstrated our approach on two real-world datasets and showed that our pattern identification algorithms are scalable. This scalability makes analysis on resource constrained and small devices such as smartwatches feasible. Traditional data analysis systems are usually operated in a remote system outside the device. This is largely due to the lack of scalability originating from software and hardware restrictions of mobile/wearable devices. By analyzing the data on the device, the user has the control over the data, i.e. privacy, and the network costs will also be removed.

Globally, the rate of preterm births are increasing, thus resulting in significant health, develo... more Globally, the rate of preterm births are increasing, thus resulting in significant health, development and economic problems. Current methods for the early detection of such births are inadequate. Nevertheless, there has been some evidence that the analysis of uterine electrical signals, collected from the abdominal surface, could provide an independent and easier way to diagnose true labour and detect the onset of preterm delivery. Using advanced machine learning algorithms, in conjunction with Electro-hysterography signal processing, numerous studies have focused on detecting true labour several days prior to the event. However, in this paper, the Electrohysterography signals have been used to detect preterm births. This has been achieved using an open dataset, which contains 262 records for women who delivered at term and 38 who delivered prematurely. Several new features from Electromyography studies have been utilised, as well as feature-ranking techniques to determine their discriminative capabilities in detecting term and preterm records. Seven different artificial neural networks were then used to identify these records. The results illustrate that the combination of the Levenberg–Marquardt trained Feed-Forward Neural Network, Radial Basis Function Neural Network and the Random Neural Network classifiers performed the best, with 91% for sensitivity, 84% for specificity, 94% for the area under the curve and 12% for the mean error rate.

The increasing number of devices within the IoT is raising concerns over the efficiency and explo... more The increasing number of devices within the IoT is
raising concerns over the efficiency and exploitability of existing authentication methods. The weaknesses of such methods, in particular passwords, are well documented. Although alternative methods have been proposed, they often rely on users being able to accurately recall complex and often unmemorable information. With the profusion of separate online accounts, this can often be a difficult task. The emerging digital memories concept involves the creation of a repository of memories specific to individuals. We believe this abundance of personal data can be utilised as a form of authentication. In this paper, we propose our digital memories based two-factor authentication mechanism, and also present our promising evaluation results.

The explosion of smaller and more powerful wearable sensing devices has allowed us to continually... more The explosion of smaller and more powerful
wearable sensing devices has allowed us to continually record and quantify our lives. Undertaking such activities is becoming very popular and has grown into a community called the Quantified Self (QS). Utilizing this outlet has the potential to benefit many aspects of our lives and is gaining momentum within the health sector. However, whilst we can easily collect data, interpreting this information is more challenging. Without extensive data analysis, this information is essentially meaningless in its raw form. This paper posits an approach to quantify and cluster levels of physical activity from accelerometer and heartrate data, which has been obtained from four wearable devices (3 accelerometers and 1 hear rate monitor). The approach details our method for pre-processing the data, extracting and selecting the features and a comparison between hierarchical clustering analysis (HCA) and k-means. The results illustrate
that, whilst both methods are capable of
successfully separating the data, the k-means approach out- performed the HCA method at clustering the data.

Preterm birth brings considerable emotional and economic costs to families and society. However, ... more Preterm birth brings considerable emotional and
economic costs to families and society. However, despite extensive research into understanding the risk factors, the prediction of patient mechanisms and improvements to obstetrical practice, the UK National Health Service still annually spends more than £2.95 billion on this issue. Diagnosis of
labour in normal
pregnancies is important for minimizing unnecessary hospitalisations, interventions and expenses. Moreover, accurate identification of spontaneous preterm labour would also allow clinicians to start necessary treatments early in women with true labour and avert unnecessary treatment and hospitalisation for women who are simply having preterm contractions, but who are not in true labour. In this research, the Electrohysterography signals have been used to detect preterm births, because Electrohysterography signals provide a strong basis for objective prediction and diagnosis of
preterm birth. This has been
achieved using an open dataset, which contains 262 records for women who delivered at term and 38 who delivered prematurely. Three different machine learning algorithm were used to identify these records. The results illustrate that the Random Forest performed the best of sensitivity 97%, specificity of 85%, Area under the Receiver Operator curve (AUROC) of 94% and mean square error rate of 14%.

The ubiquity and affordability of mobile and wearable devices has enabled us to continually and d... more The ubiquity and affordability of mobile and wearable devices has enabled us to continually and digitally record our daily life activities. Consequently, we are seeing the growth of data collection experiments in several scientific disciplines. Although these have yielded promising results, mobile and wearable data collection experiments are often restricted to a specific configuration that has been designed for a unique study goal. These approaches do not address all the real-world challenges of “continuous data collection” systems. As a result, there have been few discussions or reports about such issues that are faced when “implementing these platforms” in a practical situation. To address this, we have summarized our technical and user-centric findings from three lifelogging and Quantified Self data collection studies, which we have conducted in real-world settings, for both smartphones and smartwatches. In addition to (i) privacy and (ii) battery related issues; based on our findings we recommend further works to consider (iii) implementing multivariate reflection of the data; (iv) resolving the uncertainty and data loss; and (v) consider to minimize the manual intervention required by users. These findings have provided insights that can be used as a guideline for further Quantified Self or lifelogging studies.

The World Wide Web (WWW) has radically changed the way in which we access, generate and dissemina... more The World Wide Web (WWW) has radically changed the way in which we access, generate and disseminate information. Its presence is felt daily and with more internet-enabled devices being connected the web of knowledge is growing. We are now moving into era where the WWW is capable of ‘understanding’ the actual/intended meaning of our content. This is being achieved by creating links between distributed data sources using the Resource Description Framework (RDF). In order to find information in this web of interconnected sources, complex query languages are often employed, e.g. SPARQL. However, this approach is limited as exact query matches are often required. In order to overcome this challenge, this paper presents a probabilistic approach to searching RDF documents. The developed algorithm converts RDF data into a matrix of features and treats searching as a machine learning problem. Using a number of artificial neural network algorithms, a successfully developed prototype has been developed that demonstrates the applicability of the approach. The results illustrate that the Voted Perceptron classifier (VPC), perceptron linear classifier (PERLC) and random neural network classifier (RNNC) performed particularly well, with accuracies of 100%, 98% and 93% respectively.

Telemedicine and e-Health, 2013

A healthy lifestyle has the ability not only to give you more energy and help you look and feel b... more A healthy lifestyle has the ability not only to give you more energy and help you look and feel better, but it also has the ability to help you live longer and prevent disease, such as obesity and pressure ulcers. This is particularly important for the elderly population, as a healthier lifestyle would enable independent living to occur for a longer period of time. However, providing a direct link between increasing physical activity and positive health outcomes is a problem. The effect of leading an increasing sedentary lifestyle is also not evident straightaway. Effects of this behavior often occur over years and decades, as opposed to days or months. Therefore, there is very little willingness to change, if instant results are not seen. There is a need to provide a mechanism that is able to monitor an individual and provide a visual indication of his or her behavior. It is envisioned that the area of human digital memories is capable of providing such a system. This article explores how sedentary behavior and journey information can be collected, from different environments, so that an illustration of a user&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;s habits can be seen and changes can occur. A successful prototype has also been developed that evaluates the applicability of the approach.

Sensors, 2019

2019 IEEE International Conference on Pervasive Computing and Communications (PerCom), 2019

Telemedicine and e-Health, 2013