Multimodal deep learning enhances genomic risk prediction for cardiometabolic diseases in UK Biobank

Cardiometabolic diseases are multifactorial disorders influenced by numerous genetic variants and their complex interactions. Although recent studies have advanced the understanding of genetic risk prediction, current approaches predominantly rely on linear models that may not fully capture the complex, non-linear relationships between genetic factors. Here, we present DeepGP (Deep learning-based Genome-wide Predictor), a novel multimodal deep learning framework that incorporates bidirectional state space modules to predict cardiometabolic disease risk using genome-wide variants and demographic data. We conducted extensive experiments to evaluate DeepGP's performance. First, in simulation studies incorporating joint genetic and environmental interactions, we demonstrated DeepGP's superior prediction performance across varying levels of heritability. When evaluated on eight cardiometabolic diseases in European ancestry cohorts from the UK Biobank, DeepGP achieved significantly higher accuracy compared with conventional polygenic risk scores and machine learning methods. Model interpretability analysis identified both well-established genes and potential new signals contributing to the risk of the disease. Further validation on populations with African and Caribbean ancestries showed robust transferability of the model. Our results demonstrate the potential of cutting-edge deep learning technologies to enhance risk stratification for complex diseases across diverse ancestries and to improve disease understanding.