Date of Award
Fall 2021
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Computer Science
First Advisor
Ahamed, Sheikh
Second Advisor
Madiraju, Praveen
Third Advisor
Pinto, Daniel
Abstract
Diabetic Retinopathy (DR) is one major complication of diabetes and is the leading cause of blindness worldwide. Progression of DR and complete vision loss can be prevented by keeping diabetes in control and by early diagnosis through annual eye screenings. However, cost, healthcare disparities, cultural limitations, lack of motivation, etc., are the main barriers against regular screening, especially for a few ethnically and racially minority communities. On the other hand, to well-manage and control diabetes, the diabetic population needs to be physically active and keep their weight healthy. From the perspective of Behavioral Science, Some self-management techniques based on motivational interviewing can be utilized to motivate people to take preventive and mandatory measures to control diabetes. However, technical solutions based on `Motivational Interviewing' are still not sufficiently available to healthcare providers who work with the diabetic population. Thus, collaborative teamwork of Computer Science and Behavioral Science is contemporary to improve eye health and the overall health of the diabetic population. In this dissertation, a community telemedicine framework has been proposed and designed which can connect clinicians with community partners to organize retinal screenings in community settings rather than traditional clinical settings. Secondly, automating the initial retinal screenings utilizing Deep Learning models, particularly Convolutional Neural Network (CNN), can reduce ophthalmologists' workload and cost of screening. However, such Machine Learning models lack transparency and cannot explain how these models make particular decisions. Thus, an explainable retinal screening model has been developed to facilitate the recommended annual screening to overcome this limitation. Finally, a Computer-guided Action Planning (CAP) tool has been designed and developed to motivate the diabetic population to adopt healthier behaviors through Brief Action Planning, a self-management support technique. Through several feasibility studies, it is evident that the contribution of this dissertation could be combined to help prevent vision loss from diabetes.