Graduation Date

Spring 5-10-2025

Document Type

Thesis

Degree Name

Master of Science (MS)

Programs

Molecular Genetics & Cell Biology

First Advisor

Shibiao Wan

Abstract

Alzheimer’s disease (AD), the most common subtype of dementia, is characterized by a progressive decline in cognitive functions. Early diagnosis enables timely interventions to reduce or slow disease progression, preventing individuals from severe brain function decline. The current AD diagnosis framework depends on A/T/(N) biomarkers detection from cerebrospinal fluid or imaging, which are invasive and expensive. Meanwhile, the pathophysiological changes of AD accumulate in metabolism, neuroinflammation, etc., resulting in heterogeneity in newly registered patients. Recently, next-generation sequencing (NGS) technologies for blood samples have been found to be a non-invasive, cost-effective alternative for AD screening. However, most of the existing studies rely on single omics only. To address this, we develop WIMOAD, a stacking ensemble and weighted integration of multi-omics data for AD diagnosis. It leverages specialized classifiers for paired gene expression and methylation data, followed by meta learning for performance enhancement during classification. The prediction results of two distinct meta models were weighted for final decision-making. Remarkably, WIMOAD outperforms single-omics models and existing integration methods, highlighting its ability to effectively discern intricate patterns in multi-omics data and their correlations with clinical diagnosis results. In addition, WIMOAD also stands out as a biologically interpretable model by leveraging the SHapley Additive exPlanations (SHAP) to elucidate the contributions of genes from each omics to the model output. We believe WIMOAD is a very promising tool for accurate AD diagnosis and effective biomarker discovery across different cognitive stages, which eventually will have consequential impacts on early treatment intervention and personalized therapy design on AD.

Comments

2025 Copyright, the authors

Available for download on Wednesday, April 14, 2027

Share

COinS