Neel Shah - Academia.edu (original) (raw)
Papers by Neel Shah
Pediatric Research
The field of pediatric critical care has been hampered in the era of precision medicine by our in... more The field of pediatric critical care has been hampered in the era of precision medicine by our inability to accurately define and subclassify disease phenotypes. This has been caused by heterogeneity across age groups that further challenges the ability to perform randomized controlled trials in pediatrics. One approach to overcome these inherent challenges include the use of machine learning algorithms that can assist in generating more meaningful interpretations from clinical data. This review summarizes machine learning and artificial intelligence techniques that are currently in use for clinical data modeling with relevance to pediatric critical care. Focus has been placed on the differences between techniques and the role of each in the clinical arena. The various forms of clinical decision support that utilize machine learning are also described. We review the applications and limitations of machine learning techniques to empower clinicians to make informed decisions at the bedside.
Open Forum Infectious Diseases
Background Staphylococcus aureus is the most common cause of native septic arthritis. Few studies... more Background Staphylococcus aureus is the most common cause of native septic arthritis. Few studies have characterized this disease during the US opioid epidemic. The role of methicillin-resistant Staphylococcus aureus (MRSA) nasal screening in this disease has not been elucidated. We sought to identify risk factors and outcomes for S. aureus native septic arthritis and to evaluate MRSA screening in this disease. Methods A retrospective cohort study of native septic arthritis patients (2012–2016) was performed. Demographics, risk factors, and outcomes were compared between Staphylococcus aureus and other native septic arthritis infections. Sensitivity, specificity, and predictive values of MRSA screening were assessed. Results Two hundred fifteen cases of native septic arthritis were included. S. aureus was cultured in 64% (138/215). MRSA was cultured in 23% (50/215). S. aureus was associated with injection drug use (odds ratio [OR], 4.33; 95% CI, 1.74–10.81; P = .002) and switching a...
Advances in Therapy
Substance use disorders (SUDs) place a heavy cost burden on healthcare systems and society at lar... more Substance use disorders (SUDs) place a heavy cost burden on healthcare systems and society at large. Many barriers to effective treatment of SUDs may be overcome with prescription digital therapeutics (PDTs) delivering evidence-based, FDA-approved treatments to patients via mobile devices. This study evaluated the real-world 6-month impact on healthcare resource utilization (HCRU) in 101 patients with SUDs treated with the reSET Ò PDT. What was learned from this study? Comparing the 6 months before treatment to the 6 months after treatment with the PDT revealed significant decreases in overall hospital encounters, inpatient stays, partial hospitalizations, and emergency department visits. Reductions in facility-encounter HCRU drove 6-month per-patient cost reductions of $3591 post-treatment compared to pre-treatment. DIGITAL FEATURES This article is published with digital features, including an infographic to facilitate understanding of the article. To view digital features for this article go to
Advances in Therapy
Background and Aims: reSET-O, an FDA-authorized prescription digital therapeutic (PDT) delivering... more Background and Aims: reSET-O, an FDA-authorized prescription digital therapeutic (PDT) delivering cognitive behavioral therapy and contingency management to patients with opioid u Ò se disorder (OUD), may help improve clinical outcomes. One-year differences in healthcare resource utilization (HCRU) and costs post-PDT initiation were evaluated. Methods: Retrospective analysis of healthcare claims data compared all-cause HCRU (across hospital facility encounters [sum of inpatient stays, treat-and-release emergency department [ED] visits, partial hospitalizations, and hospital outpatient department visits] and clinician services [procedure categories]) after PDT initiation (index) between reSET-O patients and controls. Overall and Medicaid-specific differences in HCRU, costs, and buprenorphine adherence were evaluated. Findings: Cohorts included 901 reSET-O patients (median age 36 years, 62.4% female, 73.9% Medicaid) and 978 controls (median age 38 years, 51.1% female, 65.4% Medicaid). Compared to the control group, the reSET-O group experienced 12% fewer total unique hospital encounters (non-significant), driven by 28% fewer inpatient stays (IRR 0.72; 95% CI 0.55-0.96; P = 0.02), 56% fewer hospital readmissions [IRR 0.44; 95% CI 0.20-0.93; P = 0.033]), and 7% fewer ED visits (IRR 0.93; 95% CI 0.79-1.09; P = 0.386). Total clinician services increased by 1391 events versus controls. Differences were greater among the Medicaid patients. Adjustment for
Obstetrics & Gynecology
Maternal and Child Health Journal
Introduction In the US, there are striking inequities in maternal health outcomes between racial ... more Introduction In the US, there are striking inequities in maternal health outcomes between racial and ethnic groups. Community-based organizations (CBOs) provide social support services that are critical in addressing the needs of clients of color during and after pregnancy. Methods We conducted a descriptive, cross-sectional mixed methods study of CBOs in Greater Boston that provide social support services to pregnant and postpartum clients. In May-August 2020, we administered an online survey about organizational characteristics, client population, and services offered. In July-August 2020, we conducted semi-structured interviews focused on services provided, gaps in services, and the impact of structural racism on clients. We used descriptive statistics to characterize CBOs and services and used thematic analysis to extract themes from the qualitative data. Results A total of 21 unique CBOs participated with 17 CBOs completing the survey and 14 participating in interviews. CBOs served between 10 and 35,000 pregnant and postpartum clients per year (median = 200), and about half (n = 8) focused their programming on pregnant and postpartum clients. The most significant gaps in social support services were housing and childcare. Respondents identified racism and lack of coordination among organizations as the two primary barriers to accessing social support. Discussion CBOs face multiple challenges to providing social support to pregnant and postpartum clients of color, and significant gaps exist in the types of services currently provided. Improved coordination among CBOs and advocacy efforts to develop community-informed solutions are needed to reduce barriers to social support.
Cureus, 2022
Tailgut cysts, or retrorectal cystic hamartomas, are rare congenital abnormalities that develop w... more Tailgut cysts, or retrorectal cystic hamartomas, are rare congenital abnormalities that develop when the embryologic tailgut fails to involute. They are found in the presacral space, which is an area with quite a complex anatomy. Tailgut cysts can be symptomatic due to their mass effect and can even result in complications, including malignancy. Because of their rarity and varied presentations, tailgut cysts are frequently misdiagnosed. CT scans and MRI are useful in the diagnosis of these retrorectal masses, and surgical resection is the definitive treatment. Multiple surgical approaches can be used, with the treatment tailored to suit each individual patient's anatomy and suspected lesion diagnosis.
Cryogenic sprays have many applications in modern engineering. Cooling of electronic equipment su... more Cryogenic sprays have many applications in modern engineering. Cooling of electronic equipment subject to high heat flows, surgical ablation of gastrointestinal mucosae or orbital maneuvering are a few examples of their versatility. However, the atomization of a cryogenic liquid is a complex process. During such an event, aerodynamic effects associated with secondary atomization are further affected by thermodynamic flashing. A better understanding of the characteristics of cryogenic sprays is then necessary to allow for improved design and optimization in applications. The overarching objective of this study is to document such characteristics. The numerical simulation was performed over cryogenic nitrogen spray using an Eulerian-Lagrangian approach. In other words, while the gas phase of the flow is treated as a continuum, the nitrogen droplets are tracked individually in a Lagrangian sense. Models for evaporation, atomization, and breakups capture the physical processes experienc...
Figure S1. Gene specific proliferation values. (TIFF 274 kb)
This is a pre release for the homework exercise.
Annals of Internal Medicine, 2020
Electronic health records (EHRs) are now widely adopted in the United States, but health systems ... more Electronic health records (EHRs) are now widely adopted in the United States, but health systems have barely begun using them to deliver high-value care. More directed and rigorous research is needed to fulfill the promise of EHRs to not only store information but also support the delivery of better care. This article describes 4 potential benefits of EHR-based research: improving clinical decisions, supporting triage decisions, enabling collaboration among the care team (including patients), and increasing productivity via automation of tasks. Six recommendations are made for conducting and reporting research to catalyze value creation: develop interventions systematically by using usercentered design and a building-block approach; assess value in terms of cost, quality, outcomes, and work required of providers and patients; consider the time horizon for the intervention; test best practices for implementation in a range of real-world contexts; assess subtleties of behavior change tools used to improve high-value behaviors; and report the intervention in enough detail that it can be replicated, including context. Just as research played a critical role in developing early EHR prototypes and demonstrating their value to justify dissemination, research will continue to be essential in the next phase: expanding EHRbased interventions and maximizing their role in creating value.
Op-Ed and Correspondence Op-Ed Are Physicians Ready for Accountable Care? 167
Birth, 2022
BACKGROUND Shared decision-making (SDM) may improve communication, teamwork, patient experience, ... more BACKGROUND Shared decision-making (SDM) may improve communication, teamwork, patient experience, respectful maternity care, and safety during childbirth. Despite these benefits, SDM is not widely implemented, and strategies for implementing SDM interventions are not well described. We assessed the acceptability and feasibility of TeamBirth, an SDM solution that centers the birthing person in decision-making through simple tools that structure communication among the care team. We identified and described implementation strategies that bridge the gap between knowledge and practice. METHODS We conducted a qualitative study among four hospitals in the United States to understand the acceptability and feasibility of TeamBirth. We interviewed 103 clinicians and conducted 16 focus group discussions with 52 implementers between June 2018 and October 2019. We drew on the Consolidated Framework for Implementation Research to understand acceptability and feasibility, and to identify and describe the underlying contextual factors that affected implementation. RESULTS We found that clinicians and implementers valued TeamBirth for promoting clarity about care plans among the direct care team and for centering the birthing person in decision-making. Contextual factors that affected implementation included strength of leadership, physician practice models, and quality improvement culture. Effective implementation strategies included regular data feedback and adapting "flexible" components of TeamBirth to the local context. DISCUSSION By identifying and describing TeamBirth's contextual factors and implementation strategies, our findings can help bridge the implementation gap of SDM interventions. Our in-depth analysis offers tangible lessons for other labor and delivery unit leaders as they seek to integrate SDM practices in their own settings.
Proceedings of the 19th International Middleware Conference, 2018
Cloud applications are appealing targets to attackers, yet current cloud infrastructures have few... more Cloud applications are appealing targets to attackers, yet current cloud infrastructures have few ways of helping defend their customers from attacks. However, the use of virtual machines, and the economy of scale found in cloud platforms, provides an opportunity to offer strong security guarantees to tenants at low cost to the cloud provider. We present CRIMES, an evidence based, modular security framework for cloud platforms that uses speculative execution coupled with memory introspection tools to detect malicious behavior in real time. By buffering VM outputs (i.e., outgoing network packets and disk writes) until a scan has been completed, CRIMES gives strong guarantees about the amount of damage an attack can do, while minimizing overheads. When an attack is detected, CRIMES rolls back to a recent checkpoint and performs automated forensic analysis to help pinpoint the source of an attack. Our evaluation demonstrates that CRIMES incurs less overhead compared to memory protectio...
Journal of Evolution of medical and Dental Sciences, 2015
We report a case of 22 years old male presented with pain in left thigh, fever and dyspnea. On Im... more We report a case of 22 years old male presented with pain in left thigh, fever and dyspnea. On Imaging (MRI), an osteomyelitic focus in left thigh was noted. Venous Doppler had revealed a large thrombus in left lower limb. Curettage was done. After curettage patient become clinically asymptomatic. Staphyloccus aureus was identified from purulent material, Piperacilline and tazobactum antibiotics were given. It was a rare presentation of Deep Vein Thrombosis with Acute hematogenous Osteomyelitis in Adult. Cases are reported in Children. 1-6 But not in Adults.
This work would not have been possible without the copious support of my parents, Harish and Rita... more This work would not have been possible without the copious support of my parents, Harish and Rita. Also, I would like to thank my classmates Sonia Marginean and Asim Saleem for encouraging me to pursue my ideas. vi
EngRN: Biomedical Engineering (Topic), 2020
Cancer prediction at an early stage is very crucial as the patient can then prepare for dealing w... more Cancer prediction at an early stage is very crucial as the patient can then prepare for dealing with it. There are several Machine Learning models that help in predicting cancer by identifying samples of independent persons at high risk, facilitating the design and planning of cancer trials. These models use biomarkers like age, menopause, tumor-size, invnodes, breast, breast-quad dimensions to predict breast cancer. However, these models had major drawbacks of late prediction as well as low accuracy. So here presenting the system which uses gene expression profiles (genomic data) to predict breast cancer at an early stage. This model is built using different machine learning algorithms like a highly versatile support vector machine (SVM), Naive Bayes theorem, Decision tree and nearest neighbors approach to predict breast cancer using gene expression profiles.
Pediatric Research
The field of pediatric critical care has been hampered in the era of precision medicine by our in... more The field of pediatric critical care has been hampered in the era of precision medicine by our inability to accurately define and subclassify disease phenotypes. This has been caused by heterogeneity across age groups that further challenges the ability to perform randomized controlled trials in pediatrics. One approach to overcome these inherent challenges include the use of machine learning algorithms that can assist in generating more meaningful interpretations from clinical data. This review summarizes machine learning and artificial intelligence techniques that are currently in use for clinical data modeling with relevance to pediatric critical care. Focus has been placed on the differences between techniques and the role of each in the clinical arena. The various forms of clinical decision support that utilize machine learning are also described. We review the applications and limitations of machine learning techniques to empower clinicians to make informed decisions at the bedside.
Open Forum Infectious Diseases
Background Staphylococcus aureus is the most common cause of native septic arthritis. Few studies... more Background Staphylococcus aureus is the most common cause of native septic arthritis. Few studies have characterized this disease during the US opioid epidemic. The role of methicillin-resistant Staphylococcus aureus (MRSA) nasal screening in this disease has not been elucidated. We sought to identify risk factors and outcomes for S. aureus native septic arthritis and to evaluate MRSA screening in this disease. Methods A retrospective cohort study of native septic arthritis patients (2012–2016) was performed. Demographics, risk factors, and outcomes were compared between Staphylococcus aureus and other native septic arthritis infections. Sensitivity, specificity, and predictive values of MRSA screening were assessed. Results Two hundred fifteen cases of native septic arthritis were included. S. aureus was cultured in 64% (138/215). MRSA was cultured in 23% (50/215). S. aureus was associated with injection drug use (odds ratio [OR], 4.33; 95% CI, 1.74–10.81; P = .002) and switching a...
Advances in Therapy
Substance use disorders (SUDs) place a heavy cost burden on healthcare systems and society at lar... more Substance use disorders (SUDs) place a heavy cost burden on healthcare systems and society at large. Many barriers to effective treatment of SUDs may be overcome with prescription digital therapeutics (PDTs) delivering evidence-based, FDA-approved treatments to patients via mobile devices. This study evaluated the real-world 6-month impact on healthcare resource utilization (HCRU) in 101 patients with SUDs treated with the reSET Ò PDT. What was learned from this study? Comparing the 6 months before treatment to the 6 months after treatment with the PDT revealed significant decreases in overall hospital encounters, inpatient stays, partial hospitalizations, and emergency department visits. Reductions in facility-encounter HCRU drove 6-month per-patient cost reductions of $3591 post-treatment compared to pre-treatment. DIGITAL FEATURES This article is published with digital features, including an infographic to facilitate understanding of the article. To view digital features for this article go to
Advances in Therapy
Background and Aims: reSET-O, an FDA-authorized prescription digital therapeutic (PDT) delivering... more Background and Aims: reSET-O, an FDA-authorized prescription digital therapeutic (PDT) delivering cognitive behavioral therapy and contingency management to patients with opioid u Ò se disorder (OUD), may help improve clinical outcomes. One-year differences in healthcare resource utilization (HCRU) and costs post-PDT initiation were evaluated. Methods: Retrospective analysis of healthcare claims data compared all-cause HCRU (across hospital facility encounters [sum of inpatient stays, treat-and-release emergency department [ED] visits, partial hospitalizations, and hospital outpatient department visits] and clinician services [procedure categories]) after PDT initiation (index) between reSET-O patients and controls. Overall and Medicaid-specific differences in HCRU, costs, and buprenorphine adherence were evaluated. Findings: Cohorts included 901 reSET-O patients (median age 36 years, 62.4% female, 73.9% Medicaid) and 978 controls (median age 38 years, 51.1% female, 65.4% Medicaid). Compared to the control group, the reSET-O group experienced 12% fewer total unique hospital encounters (non-significant), driven by 28% fewer inpatient stays (IRR 0.72; 95% CI 0.55-0.96; P = 0.02), 56% fewer hospital readmissions [IRR 0.44; 95% CI 0.20-0.93; P = 0.033]), and 7% fewer ED visits (IRR 0.93; 95% CI 0.79-1.09; P = 0.386). Total clinician services increased by 1391 events versus controls. Differences were greater among the Medicaid patients. Adjustment for
Obstetrics & Gynecology
Maternal and Child Health Journal
Introduction In the US, there are striking inequities in maternal health outcomes between racial ... more Introduction In the US, there are striking inequities in maternal health outcomes between racial and ethnic groups. Community-based organizations (CBOs) provide social support services that are critical in addressing the needs of clients of color during and after pregnancy. Methods We conducted a descriptive, cross-sectional mixed methods study of CBOs in Greater Boston that provide social support services to pregnant and postpartum clients. In May-August 2020, we administered an online survey about organizational characteristics, client population, and services offered. In July-August 2020, we conducted semi-structured interviews focused on services provided, gaps in services, and the impact of structural racism on clients. We used descriptive statistics to characterize CBOs and services and used thematic analysis to extract themes from the qualitative data. Results A total of 21 unique CBOs participated with 17 CBOs completing the survey and 14 participating in interviews. CBOs served between 10 and 35,000 pregnant and postpartum clients per year (median = 200), and about half (n = 8) focused their programming on pregnant and postpartum clients. The most significant gaps in social support services were housing and childcare. Respondents identified racism and lack of coordination among organizations as the two primary barriers to accessing social support. Discussion CBOs face multiple challenges to providing social support to pregnant and postpartum clients of color, and significant gaps exist in the types of services currently provided. Improved coordination among CBOs and advocacy efforts to develop community-informed solutions are needed to reduce barriers to social support.
Cureus, 2022
Tailgut cysts, or retrorectal cystic hamartomas, are rare congenital abnormalities that develop w... more Tailgut cysts, or retrorectal cystic hamartomas, are rare congenital abnormalities that develop when the embryologic tailgut fails to involute. They are found in the presacral space, which is an area with quite a complex anatomy. Tailgut cysts can be symptomatic due to their mass effect and can even result in complications, including malignancy. Because of their rarity and varied presentations, tailgut cysts are frequently misdiagnosed. CT scans and MRI are useful in the diagnosis of these retrorectal masses, and surgical resection is the definitive treatment. Multiple surgical approaches can be used, with the treatment tailored to suit each individual patient's anatomy and suspected lesion diagnosis.
Cryogenic sprays have many applications in modern engineering. Cooling of electronic equipment su... more Cryogenic sprays have many applications in modern engineering. Cooling of electronic equipment subject to high heat flows, surgical ablation of gastrointestinal mucosae or orbital maneuvering are a few examples of their versatility. However, the atomization of a cryogenic liquid is a complex process. During such an event, aerodynamic effects associated with secondary atomization are further affected by thermodynamic flashing. A better understanding of the characteristics of cryogenic sprays is then necessary to allow for improved design and optimization in applications. The overarching objective of this study is to document such characteristics. The numerical simulation was performed over cryogenic nitrogen spray using an Eulerian-Lagrangian approach. In other words, while the gas phase of the flow is treated as a continuum, the nitrogen droplets are tracked individually in a Lagrangian sense. Models for evaporation, atomization, and breakups capture the physical processes experienc...
Figure S1. Gene specific proliferation values. (TIFF 274 kb)
This is a pre release for the homework exercise.
Annals of Internal Medicine, 2020
Electronic health records (EHRs) are now widely adopted in the United States, but health systems ... more Electronic health records (EHRs) are now widely adopted in the United States, but health systems have barely begun using them to deliver high-value care. More directed and rigorous research is needed to fulfill the promise of EHRs to not only store information but also support the delivery of better care. This article describes 4 potential benefits of EHR-based research: improving clinical decisions, supporting triage decisions, enabling collaboration among the care team (including patients), and increasing productivity via automation of tasks. Six recommendations are made for conducting and reporting research to catalyze value creation: develop interventions systematically by using usercentered design and a building-block approach; assess value in terms of cost, quality, outcomes, and work required of providers and patients; consider the time horizon for the intervention; test best practices for implementation in a range of real-world contexts; assess subtleties of behavior change tools used to improve high-value behaviors; and report the intervention in enough detail that it can be replicated, including context. Just as research played a critical role in developing early EHR prototypes and demonstrating their value to justify dissemination, research will continue to be essential in the next phase: expanding EHRbased interventions and maximizing their role in creating value.
Op-Ed and Correspondence Op-Ed Are Physicians Ready for Accountable Care? 167
Birth, 2022
BACKGROUND Shared decision-making (SDM) may improve communication, teamwork, patient experience, ... more BACKGROUND Shared decision-making (SDM) may improve communication, teamwork, patient experience, respectful maternity care, and safety during childbirth. Despite these benefits, SDM is not widely implemented, and strategies for implementing SDM interventions are not well described. We assessed the acceptability and feasibility of TeamBirth, an SDM solution that centers the birthing person in decision-making through simple tools that structure communication among the care team. We identified and described implementation strategies that bridge the gap between knowledge and practice. METHODS We conducted a qualitative study among four hospitals in the United States to understand the acceptability and feasibility of TeamBirth. We interviewed 103 clinicians and conducted 16 focus group discussions with 52 implementers between June 2018 and October 2019. We drew on the Consolidated Framework for Implementation Research to understand acceptability and feasibility, and to identify and describe the underlying contextual factors that affected implementation. RESULTS We found that clinicians and implementers valued TeamBirth for promoting clarity about care plans among the direct care team and for centering the birthing person in decision-making. Contextual factors that affected implementation included strength of leadership, physician practice models, and quality improvement culture. Effective implementation strategies included regular data feedback and adapting "flexible" components of TeamBirth to the local context. DISCUSSION By identifying and describing TeamBirth's contextual factors and implementation strategies, our findings can help bridge the implementation gap of SDM interventions. Our in-depth analysis offers tangible lessons for other labor and delivery unit leaders as they seek to integrate SDM practices in their own settings.
Proceedings of the 19th International Middleware Conference, 2018
Cloud applications are appealing targets to attackers, yet current cloud infrastructures have few... more Cloud applications are appealing targets to attackers, yet current cloud infrastructures have few ways of helping defend their customers from attacks. However, the use of virtual machines, and the economy of scale found in cloud platforms, provides an opportunity to offer strong security guarantees to tenants at low cost to the cloud provider. We present CRIMES, an evidence based, modular security framework for cloud platforms that uses speculative execution coupled with memory introspection tools to detect malicious behavior in real time. By buffering VM outputs (i.e., outgoing network packets and disk writes) until a scan has been completed, CRIMES gives strong guarantees about the amount of damage an attack can do, while minimizing overheads. When an attack is detected, CRIMES rolls back to a recent checkpoint and performs automated forensic analysis to help pinpoint the source of an attack. Our evaluation demonstrates that CRIMES incurs less overhead compared to memory protectio...
Journal of Evolution of medical and Dental Sciences, 2015
We report a case of 22 years old male presented with pain in left thigh, fever and dyspnea. On Im... more We report a case of 22 years old male presented with pain in left thigh, fever and dyspnea. On Imaging (MRI), an osteomyelitic focus in left thigh was noted. Venous Doppler had revealed a large thrombus in left lower limb. Curettage was done. After curettage patient become clinically asymptomatic. Staphyloccus aureus was identified from purulent material, Piperacilline and tazobactum antibiotics were given. It was a rare presentation of Deep Vein Thrombosis with Acute hematogenous Osteomyelitis in Adult. Cases are reported in Children. 1-6 But not in Adults.
This work would not have been possible without the copious support of my parents, Harish and Rita... more This work would not have been possible without the copious support of my parents, Harish and Rita. Also, I would like to thank my classmates Sonia Marginean and Asim Saleem for encouraging me to pursue my ideas. vi
EngRN: Biomedical Engineering (Topic), 2020
Cancer prediction at an early stage is very crucial as the patient can then prepare for dealing w... more Cancer prediction at an early stage is very crucial as the patient can then prepare for dealing with it. There are several Machine Learning models that help in predicting cancer by identifying samples of independent persons at high risk, facilitating the design and planning of cancer trials. These models use biomarkers like age, menopause, tumor-size, invnodes, breast, breast-quad dimensions to predict breast cancer. However, these models had major drawbacks of late prediction as well as low accuracy. So here presenting the system which uses gene expression profiles (genomic data) to predict breast cancer at an early stage. This model is built using different machine learning algorithms like a highly versatile support vector machine (SVM), Naive Bayes theorem, Decision tree and nearest neighbors approach to predict breast cancer using gene expression profiles.