COMPARATIVE ORDINAL LONGITUDINAL DATA ANALYSIS TO PREDICT A DIAGNOSIS OF ALZHEIMER'S DISEASE BY USING MULTIMODAL DATA

Date

2024-9-04

Author

Emen, Aycan Çağrı

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

433
views

216
downloads

Cite This

Alzheimer’s Disease (AD) is one of the most common neurological disorders affecting elderly adults. The disease often progresses significantly before symptoms become apparent, making early diagnosis crucial to slowing its progression. Consequently, it is essential to employ appropriate longitudinal analysis methods for predicting the disease. However, previous studies have often neglected the longitudinal nature of the data. This thesis addresses these shortcomings by applying various statistical and machine learning algorithms to the OASIS-3 dataset. The OASIS-3 dataset, which includes demographics, cognitive assessments, MRI and PET scan outputs, was integrated to form a multimodal perspective, allowing for a comprehensive evaluation of the underlying factors influencing AD. The study utilizes marginal models (Generalized Estimating Equations, GEE), random effect models (both frequentist and Bayesian Generalized Linear Mixed Models, GLMMs), and a machine learning algorithm (Ordinal Mixed Effect Random Forest, OMERF) to classify the severity of the disease and identify influential factors. The findings indicate that the Bayesian approach of the Generalized Mixed Effect Model outperforms other methods in identifying severe AD cases among ordinal data. This research highlights the importance of using advanced longitudinal approaches to better understand and predict Alzheimer’s Disease progression, providing valuable insights for early diagnosis and intervention.

Subject Keywords

Generalized Linear Mixed Effect Models, Marginal Models, Ordinal Response, OASIS-3, Alzheimer’s Disease Prediction

URI

https://hdl.handle.net/11511/111491

Collections

Graduate School of Natural and Applied Sciences, Thesis