Tag
This paper introduces EHR-MPC, a framework that decouples learning patient dynamics from treatment optimization by training a generative digital twin of patients using electronic health records, then applying model predictive control at inference time, achieving comparable or improved performance over RL baselines on a multi-hospital sepsis cohort.
Tampa General Hospital partnered with Palantir to build the AI-powered Sepsis Hub, which reduced sepsis mortality by 68% and saved over 700 lives by late 2025.
Professor Clare Bryant at the University of Cambridge is using DeepMind's Co-Scientist to rapidly identify molecular switches that cause severe diseases like sepsis when pathogens jump between species, cutting research time from years to months.