Date of Award

Spring 2021

Degree Type


Degree Name

PhD Health Sciences


Health and Medical Sciences


Deborah DeLuca, J.D.

Committee Member

Glenn Beamer, Ph.D.

Committee Member

Genevieve Pinto-Zipp, Ed.D.


health interventions, clinical studies, attrition, health technology, Bayesian Modeling, Cardiovascular Disease, Artificial Intelligence, Translational Medicine


Background and Purpose: The high prevalence and mortality associated with Cardiovascular Diseases (CVD) is a burden on the United States healthcare system due to the millions of people that have and are at risk for CVD, causing clinical, practical, logistical, geographical, and financial difficulties associated with delivering medical care (Mensah & Brown, 2007). Digital Health Interventions (DHI) reduce CVD outcomes and improve CVD healthcare quality, but DHI clinical studies experience high rates of patient attrition (participant drop out), affecting the reliability of the collected data (Khanji et al., 2019; Santo & Redfern, 2020; Widmer et al., 2017). Achieving improvements in DHI patient attrition requires extensive evidence base on the risk and protective factors that contribute to patient attrition in DHI clinical studies. However, DHI literature presents special challenges. Through the development of a Bayesian Network Model, the purpose of the study was to begin to identify the relationships of the various factors associated with patient attrition in CVD DHI clinical studies.

Methods: A mixed-methods study was conducted and consisted of an examination of DHI literature, retrospective data analysis of a Cardiovascular Disease (CVD) digital health study, and expert elicitation to develop a Bayesian Network model.

Results: This study proposes a translational machine learning (ML) framework for exploring the relationships between various risk and protective factors of patient attrition in Digital Health Interventions (DHI) clinical studies and illustrates its application in medicine at the biological, intrapersonal, interpersonal, organizational and policy levels. These findings provide rationale and guidance for future strategies and interventions to facilitate full participation from sample groups and prevent patients/participants from ceasing use or dropping out of DHI studies.

Conclusion: The description and analysis of the risk and protective factors associated with patient attrition in DHI studies, and the relationships of those factors expands the evidence-base on Digital Health patient attrition, adherence, and engagement. Artificial Intelligence (AI) and Machine Learning are promising areas of development for medical research (Ramesh et al., 2004). This study serves as a framework for investigating complex healthcare phenomena utilizing AI and machine learning through an integrated translational approach.

Available for download on Monday, April 27, 2026