Google Research Releases SensorFM: Foundation Model Pretrained on One Trillion Minutes of Wearable Sensor Data

Google Research has introduced SensorFM, a foundation model for wearable health sensing pretrained on an unprecedented one trillion minutes of sensor data. The scale of pretraining data here is the headline: this is orders of magnitude larger than prior wearable health models, which suggests strong generalization across sensor modalities like accelerometers, heart rate, and SpO2. For developers building health and fitness applications, on-device inference tools, or clinical monitoring pipelines, SensorFM represents a powerful starting point that could dramatically reduce the labeled data required for fine-tuning task-specific models. The 'foundation model' framing signals it is designed for transfer learning, meaning developers should evaluate it as a feature extractor or adapter base rather than an end-to-end solution. This is a significant infrastructure-level contribution from Google Research for the wearables and health-tech development community.