Theses and Dissertations

Date of Award

8-1-2026

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Applied Statistics and Data Science

First Advisor

Hansapani Rodrigo

Second Advisor

Thoa Thieu

Third Advisor

Tamer Oraby

Abstract

Alzheimer's disease is a progressive neurodegenerative disorder that requires accurate assessment of disease severity and a comprehensive understanding of disease progression to support effective clinical management. This thesis investigates Alzheimer's disease severity using longitudinal multimodal data from the OASIS-3 cohort by integrating statistical disease-progression modelling with machine learning-based disease-stage classification. Disease progression was analyzed using longitudinal transition analysis, continuous-time multi-state models, Bayesian Hidden Markov Model, longitudinal biomarker trajectory analysis, and recurrent neural network forecasting. Current disease-stage classification was evaluated using static and temporal multimodal models, including a controlled evaluation of structural magnetic resonance imaging (MRI) through temporally MRI-matched cohorts and hierarchical MRI pretraining. The findings demonstrate that Alzheimer's disease progresses gradually, with diagnostic uncertainty concentrated in the intermediate stages. Longitudinal information substantially improves forecasting of future disease severity, while multimodal classification benefits from complementary clinical and imaging information. The proposed framework provides a comprehensive approach for understanding disease progression and supporting current disease-stage assessment in Alzheimer's disease.

Comments

Copyright 2026 Christopher Attafuah. All Rights Reserved. https://proquest.com/docview/3409016482

Download

Included in

Mathematics Commons

Share

COinS