Google Research stellt mit SensorFM ein Foundation Model vor, das aus mehr als einer Billion Minuten Wearable-Daten von fünf Millionen Fitbit- und Pixel-Watch-Nutzern lernt. Bei 34 von 35 Gesundheits-...

Google updates Pixel Watch 3, 4 and 5 with Health Guardian. Powered by the SensorFM AI model, it now tracks blood pressure and insulin resistance trends from the wrist. Emergency oxygen drop detection

Google's SensorFM AI model can learn reusable patterns from fragmented wearable data across 35 tasks.

Most wearable health models are built one outcome at a time. That approach breaks down at thirty-five endpoints. Labels are expensive and retrospective annotation is infeasible. Google Research introduced SensorFM, a foundation model for wearable health pre-trained on more than 1 trillion minutes of sensor data from 5 million people. What is SensorFM? SensorFM is a Large Sensor foundation Model for wearable time-series representation learning. It ingests 34 one-minute aggregate features drawn from five sensors: PPG, accelerometer, EDA, skin temperature, and altimeter. Those features are organized into seven categories, over a 24-hour context window. The backbone is a ViT-1D encoder trained with a masked-autoencoder objective and a patch size of [20, 1]. Pretraining used 5,000,000 consented participants, sampled between September 2024 and September 2025. That corpus spans 100+ countries, all 50 U.S. states, and 20+ Fitbit and Pixel Watch models. It totals over two billion hours, or more than one trillion minutes. Four variants exist, each paired with a proportional data volume. Evaluation uses separate data. It covers 13,985 subjects across three prospective IRB-approved studies. Those are metabolic, cardiac and respiratory health (N = 1,655), sleep (N = 6,377), and mental health (N = 5,953). The 35 tasks cover cardiovascular...

Google's trillion-minute SensorFM data shows wearables shifting from raw metrics to AI health agents, a trend already reshaping Whoop and Oura.

Google Research SensorFM pretrains on one trillion minutes of wearable sensor data, transferring to 35 health prediction tasks.

Компания <a href="https://research.google/blog/sensorfm-towards-a-general-intelligence-and-interface-for-wearable-health-data/" target="_blank">представила</a> SensorFM — базовую модель искусственного...

Google Research introduced SensorFM, a foundation model for wearable health data, pretrained on over a trillion minutes of sensor signals from about 5 million people. It learns a general representation of human physiology, transfers to 35 health prediction tasks, and is meant to

Google Research has just released two specialized foundation models that directly address these weaknesses: SensorFM for wearable sensor data and TabFM for tabular data.

O Google revelou o SensorFM, um modelo de inteligência artificial treinado com mais de um trilhão de minutos de dados de wearables que promete transformar o monitoramento contínuo da saúde, oferecendo uma compreensão mais ampla da fisiologia humana e abrindo caminho para agentes pessoais de saúde baseados em IA. O post Google apresenta SensorFM, modelo de IA treinado com um trilhão de minutos de dados de wearables apareceu primeiro em Google Discovery.

O Google revelou o SensorFM, um modelo de inteligência artificial treinado com mais de um trilhão de minutos de dados de wearables que promete transformar o monitoramento contínuo da saúde, oferecendo uma compreensão mais ampla da fisiologia humana e abrindo caminho para agentes pessoais de saúde baseados em IA.

SensorFM 소개 스마트워치나 밴드 같은 웨어러블 기기를 며칠만 차고 있어도, 심박수와 걸음 수, 수면, 피부 온도 같은 신호가 분 단위로 쌓입니다. 그런데 이렇게 방대하게 모인 원시 신호를 막상 "이 사람의 건강 상태가 어떤가" 라는 질문의 답으로 바꾸는 일은 여전히 어렵습니다. 사람마다 기저 생리와 생활 습관이 크게 달라서 어떤 사람에게 위험 신호인 패턴이 다른 사람에게는 정상일 수 있고, 무엇보다 모델 학습에 필요한 정답(확진 진단, 검사 결과, 검증된 설문)을 모으는 일 자체가 비싸고 느리며 과거 데이터에 소급해서 붙이는 것은 사실상 불가능하기 때문입니다. 이 논문은 이러한 한계를 정면으로 다루기 위해, 라벨 없는 웨어러블 센서 데이터만으로 인구 규모의 표현을 학습하는 웨어러블 헬스 파운데이션 모델(Foundation Model) SensorFM 을 제안합니다. Google Research와 Google DeepMin...

SensorFM, el nuevo modelo fundacional de Google Research, ha sido entrenado con más de un billón de minutos de datos de wearables de cinco millones de usuarios.

Google Research's SensorFM is a foundation model trained on more than a trillion minutes of wearable data from five million Fitbit and Pixel Watch users. It beats existing benchmarks on 34 of 35 health and behavioral tasks. SensorFM could eventually power Google's AI health coach, but the company hasn't announced any integration plans yet.