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, synaptic dysfunction, and interacting genetic, environmental, and lifestyle factors. Transcriptomic and metabolomic studies have revealed molecular disruptions relevant to AD, supporting integrative approaches for biomarker discovery. Objective To integrate genetically imputed whole-blood transcriptomics and measured plasma metabolomics to predict cognitive performance, assessed using the PACC3 score, and identify influential genes and metabolites associated with cognition. Methods A machine learning model integrated transcriptomic and metabolomic data from 1046 participants in the Wisconsin Registry for Alzheimer's Prevention (WRAP). Performance was evaluated in a WRAP holdout test set and independently validated in 85 participants from the Wisconsin Alzheimer's Disease Research Center (ADRC). Feature importance was used to identify molecular contributors to prediction. Results The model achieved a normalized root mean squared error of 0.707 and an R 2 of 0.338 in the WRAP holdout dataset (p = 5.93 × 10 −30 ), and corresponding values of 0.915 and 0.061 in ADRC (p = 4.71 × 10 −2 ). Higher imputed expression of RIPK1, IL6ST, and BIN1 was associated with poorer cognitive performance, whereas UGP2, NDUFB5, and TMOD2 were associated with better performance. Predictive metabolites included benzoate, 3-phenylpropionate, imidazolelactate, hexanoylcarnitine, and propionate-related metabolites. Conclusions Multi-omics integration identified candidate biomarkers reflecting inflammatory signaling, mitochondrial dysfunction, and lipid metabolism. Together, these findings demonstrate complementary biological information captured across both omics layers. These convergent signals support improved molecular characterization of AD and biomarker prioritization for future mechanistic and translational studies.