Teaching AI to understand a trillion minutes of health data

Girish Narayanswamy, Maxwell A. Xu, A. Ali Heydari, Samy Abdel-Ghaffar, Marius Guerard, Kara Vaillancourt, Zhihan Zhang, Jake Garrison, Levi Albuquerque, Dimitris Spathis, Hong Yu, Hamid Palangi, Xuhai "Orson" Xu, David G. T. Barrett, Joseph Breda, Jed McGiffin, Yubin Kim, Yuwei Zhang, Naghmeh Rezaei, Samuel Solomon, Karan Ahuja, Tim Althoff, Jake Sunshine, Ming-Zher Poh, Benjamin Yetton, Ari Winbush, Nicholas B. Allen, James M. Rehg, Isaac Galatzer-Levy, Yun Liu, John Hernandez, Anupam Pathak, Conor Heneghan, Yuzhe Yang, Ahmed A. Metwally, Pushmeet Kohli, Mark Malhotra, Shwetak Patel, Xin Liu, Daniel McDuff

Wearable devices collect vast amounts of health data, but converting raw sensor signals into useful predictions is hard because everyone's baseline physiology differs. This team built a foundation model pretrained on over one trillion minutes of unlabeled wearable data from 5 million people, then fine-tuned it on 35 different health tasks spanning cardiovascular, metabolic, sleep, and mental health. The scaling worked: more data and bigger models meant steady performance gains, few-shot learning on new tasks, and the ability to deploy an AI agent that autonomously optimizes predictions for each domain. Clinicians rated the system's outputs as more relevant and safer than baselines.