Semester of Graduation
Summer 2026
Degree Type
Thesis
Degree Name
Master of Science in Information Technology
Department
Information Technology - College of Computing and Software Engineering
Committee Chair/First Advisor
Nazmus Sakib
Second Advisor
Sumit Chakravarty
Third Advisor
Maria Valero
Abstract
Alzheimer's Disease and Alzheimer's Disease-Related Dementias (AD/ADRD) present significant difficulties for patients, caregivers, family members, and healthcare organizations. People diagnosed with AD/ADRD could face issues like cognitive decline, memory problems, and loss of independence. The occurrence of such incidents may necessitate quick action by caregivers, as any delay in assisting may expose the person to injuries or even hospitalization. This process may cause stress, fatigue, anxiety, depression, and burnout. Hence, dementia care necessitates technologies that ensure the caregiver's safety and sustainability. The proposed thesis proposes an AI-driven healthcare information system that integrates caregiver burnout risk prediction and SafeCircle, an iOS-based prototype for remote monitoring of patients with AD/ADRD. In the first component, a predictive framework for caregiver burnout risk boundaries is presented using the NHATS database. Various feature selection techniques, such as LightGBM, XGBoost, and Recursive Feature Elimination with Random Forest, have been employed to identify important caregiver-related features. Linear Support Vector Machine (SVM) and Principal Component Analysis (PCA) are employed to separate low- and high-burnout instances geometrically and to identify caregivers at the decision boundary as early-risk individuals. The second component in this thesis introduces SafeCircle, an iOS-based, AI-enabled, and privacy-preserving remote monitoring prototype for AD/ADRD patients. SafeCircle includes radar-supported fall and wandering detection, real-time alerts, GPS location tracking, SOS emergency services, patient and caregiver profile management, and short video clips of event triggers. A preliminary user acceptance test produced an estimated System Usability Scale score of 80.5 out of 100, indicating good perceived usability.
Included in
Computer and Systems Architecture Commons, Health Services Research Commons, Patient Safety Commons
Comments
Dr. Liang Zhao also served as a member of the thesis committee.