Multi-Omics Integration of Transcriptomics and Metabolomics with Machine Learning Uncovers Novel Risk Factors for Alzheimer's disease

Background: Alzheimer's disease (AD) is a neurodegenerative disorder characterized by cognitive decline, memory impairment, and functional deterioration. Its complex pathogenesis involves amyloid plaques, tau tangles, neuroinflammation, and synaptic dysfunction, yet the precise molecular mechanisms remain incompletely understood. Genetic, environmental, and lifestyle factors contribute to AD risk, but their interactions are poorly characterized. Advances in transcriptomics and metabolomics have revealed gene expression and metabolic alterations associated with disease progression, highlighting the need for integrative approaches to better define AD-relevant biology and inform biomarker discovery. Objectives: To integrate genetically imputed whole blood transcriptomics and plasma metabolomics to predict cognitive performance (PACC3) and identify genes and metabolites contributing to model prediction. Methods: A machine learning algorithm was applied to integrate genetically imputed transcriptomics and measured plasma metabolomics from the Wisconsin Registry for Alzheimer's Prevention (WRAP; N = 1,046) to predict PACC3. Predictive performance was evaluated using an independent dataset from the Wisconsin Alzheimer's Disease Research Center (ADRC; N = 85). Feature importance was used to prioritize influential genes and metabolites. Results: The model achieved an NRMSE of 0.743 {+/-} 0.037 and R2 of 0.311 {+/-} 0.016 across 5-fold holdout tests in WRAP (p = 5.93 x -10), and an NRMSE of 0.915 and R2 of 0.061 in ADRC. Influential genes included RIPK1, IL6ST, and BIN1 (poorer cognition), while UGP2, NDUFB5, and TMOD2 aligned with cognitive resilience. Predictive metabolites included benzoate and acyl-carnitine species. Conclusion: Multi-omics integration identifies molecular features linked to cognitive performance in AD, supporting its utility for biomarker prioritization.