A systematic exposure‐wide framework leveraging machine learning to identify multidomain exposure factors and their joint influence on cognitive function: Evidence from a neurological cohort

• DOI: 10.1002/alz.14624
• Publication Date: 2025-03-12T01:21:35Z
• Authors: Jingtao Wu; Bowen Yin; Rui Wen; Huanting Pei; Siqi Zhu; Jiaxin Zhao; Yanbing Li; Ming Yang; Yaoyu Hu; Qun Xu; Ang Li; Yuxia Ma

Abstract INTRODUCTION Cognitive decline has become a growing public concern, yet large‐scale exposure data identifying the contributing factors remain limited. METHODS We conducted an exposure‐wide association study involving 1142 participants and 207 exposures, using machine learning to assess relative contribution joint effects of key factors. Cluster analysis intervention simulation trials helped identify high‐risk subpopulations potential benefits targeted interventions. RESULTS In adjusted mixed models, socioeconomic status domain emerged as strongest predictor longitudinal global cognitive score ( β = 2.91, p < 0.0001, q 0.0001), while dietary also played important role in memory function. The cluster found that “unfavorable lifestyle” dominated phenotype was associated with poorest outcomes. Simulation indicated scores could improve by shifting individuals from unfavorable favorable phenotypes. DISCUSSION health requires multidomain interventions, particularly fields, necessitates collaboration between government individuals. Highlights design, which assesses broad range is used novel variables understand their contributions findings indicate most significant contributor function, diet plays largest Increasing proportion phenotypes through interventions can significantly enhance health.