A novel generative AI model, Delphi-2M, trained on extensive UK Biobank data from over 400,000 individuals, can simultaneously predict the likelihood and timing of developing more than 1,000 diseases up to 20 years into the future. Applying large language model techniques centered around multimodal, multi-disease forecasting, this tool incorporates patient demographics, genetics, and lifestyle factors to provide comprehensive personal health risk profiles. Detailed in Nature, the model’s broad predictive capacity could transform risk stratification, preventive care, and personalized medicine, although further validation and integration into clinical workflows remain forthcoming.