Search papers, labs, and topics across Lattice.
NVIDIANational Center for Childhood DiabetesPheno.AISchneider Children's Medical Center of IsraelTAUThe Jesse Z and Sara Lea ShaferWeizmannZayed University of ArtificialApr 30, 2026arXiv:2604.27899HealthFormer, a decoder-only transformer, is trained on longitudinal multi-modal health data from 15,000 individuals to model human physiological trajectories. This generative model forecasts individual health trajectories across 667 measurements in seven domains and transfers to independent cohorts, improving prediction for incident-disease and mortality endpoints compared to clinical risk scores. The model can also simulate clinical interventions in silico, accurately predicting biomarker changes in a personalized-nutrition trial and aligning with results from randomized controlled trials.
A generative model of human physiology not only beats existing clinical risk scores at predicting disease, but also accurately simulates the effects of clinical interventions, paving the way for personalized medicine.
Understanding how human health changes over time, and why responses to interventions vary between individuals, remains a central challenge in medicine. Here we present HealthFormer, a decoder-only transformer that models the human physiological trajectory generatively, by training on data from the Human Phenotype Project, a multi-visit cohort of over 15,000 deeply phenotyped individuals. We tokenise each participant's health trajectory across 667 measurements spanning seven domains: blood biomarkers, body composition, sleep physiology, continuous glucose monitoring, gut microbiome, wearable-derived physiology, and behaviour and medication exposure. We train HealthFormer to forecast individual physiological trajectories across these domains, and from this single generative objective a range of clinically relevant tasks can be expressed as queries on the model. We show that, without task-specific training, HealthFormer transfers to four independent cohorts and improves prediction for 27 of 30 incident-disease and mortality endpoints, exceeding established clinical risk scores in every comparison. We further show that the model can simulate interventions in silico: in a held-out personalised-nutrition trial, intervention-conditioned predictions recover individual six-month biomarker changes (e.g., Pearson r = 0.78 for diastolic blood pressure). Across 41 randomised intervention-outcome comparisons drawn from published trials, our results show that the predicted direction of effect agrees in every case, and the predicted mean falls within the reported 95% confidence interval in 30 cases. We position HealthFormer as an initial health world model, from which forecasting, risk stratification, and intervention-conditioned simulation arise as queries, providing a basis for clinical digital twins.
