SensorFM: A Wearable Health Foundation Model

ymatias · x · 2026-07-11

The author introduces **SensorFM**, a new foundation model for wearable health sensors, pretrained on over **1 trillion minutes** of data from **5 million participants**. The model learns a reusable physiological representation of the human body that can transfer across: - Cardiovascular - Metabolic - Sleep - Mental health - Lifestyle and demographic factors The author believes this work signals a trend: wearable health research is shifting from single-task models toward flexible, highly efficient, and scalable general physiological representations.

Related event: Google Releases SensorFM for Wearable Health Data(2 posts)→

Original post →

More from Research

Research channel →