Nick Phillips

Our new arXiv paper: We built a generative AI model of blood glucose levels from 10 million continuous glucose monitoring measurements of 11,000 people from the Human Phenotype Project We show that our model can predict clinical parameters such as liver-related parameters, blood lipids, and sleep-related indices. The model can also predict onset of future health outcomes even 4 years in advance, as well as outcomes of randomized clinical trials We also integrated dietary data into the models and show that the enhanced model can accurately generate blood glucose levels using only dietary intake data, simulate outcomes of dietary interventions, and predict individual responses to specific foods. The model also performs well on 15 different external datasets from 5 different geographical regions and 6 different continuous glucose monitoring devices Paper: arxiv.org/abs/2408.11876 Data: humanphenotypeproject.org/ Great work led by @GLutsker, Gal Sapir, @nastya_godneva, @smadarshilo, @H_Rossman in collaboration with @SamochaBonetD and @nvidia researchers @GalChechik @MannorShie @eli_meirom