Mini toolsSensorFM
By Google
SensorFM is a Large Sensor Foundation Model built to learn general-purpose representations of human physiology from wearable sensor data. It is pre-trained via self-supervised, missingness-aware masked reconstruction on 34 one-minute aggregate features across five sensor modalities: photoplethysmography, accelerometry, electrodermal activity, skin temperature, and altimetry. The model transfers to 35 discriminative health tasks covering cardiovascular health, metabolic risk, mental health, sleep, demographics, and lifestyle, outperforming feature-engineered supervised baselines on 34 of 35 tasks using only frozen linear probes. It supports label-efficient adaptation and robust daily-metric estimation despite missing data, can be paired with an agentic system that automatically builds prediction heads, and can ground a Personal Health Agent, producing clinician-rated health summaries comparable to those based on ground-truth measurements.
New
Structured_data
Overview
SensorFM is a foundation model for wearable health data, pre-trained on more than one trillion minutes of multimodal sensor signals from five million people. It learns a general-purpose representation of human physiology from heart rate, blood oxygen, sleep, motion, and skin conductance signals, transferring across 35 health prediction tasks spanning cardiovascular, metabolic, sleep, and mental health domains.
About Google
At Google, we think that AI can meaningfully improve people's lives and that the biggest impact will come when everyone can access it.
Industry:
Technology, Information and Internet
Company Size:
190820
Location:
Mountain View, CA, US
Website:
ai.google
Last updated: July 10, 2026
