Google's SensorFM trained on 5M people's body data

By PulseAugur Editorial · [1 sources] · 2026-07-19 17:01

Google developed SensorFM, a foundation model trained on a massive dataset of body data from 5 million individuals. This model processes continuous data streams from wearable devices, moving beyond simple step-counting to generate predictions. The data was collected over trillions of minutes, highlighting the scale of information used for training. AI

IMPACT This development could lead to more sophisticated predictive models for health and behavior based on continuous physiological data.

RANK_REASON The item describes the training of a foundation model on a large dataset, which falls under research. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Google's SensorFM trained on 5M people's body data

COVERAGE [1]

Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] · 2026-07-19 17:01

Google's SensorFM trained on a trillion minutes of body data from 5M people. The device was never the product — the continuous stream off your wrist was. A foun

Google's SensorFM trained on a trillion minutes of body data from 5M people. The device was never the product — the continuous stream off your wrist was. A foundation model turns that exhaust into prediction, not step-counting. # AI # MachineLearning # LLM # Threadverse # Tech

COVERAGE [1]

Google's SensorFM trained on a trillion minutes of body data from 5M people. The device was never the product — the continuous stream off your wrist was. A foun

RELATED ENTITIES

RELATED TOPICS