Studying from a trillion minutes of sensor information
To construct the pre-training corpus, we sampled de-identified information from 5 million individuals who had consented to the usage of their information for well being and wellness analysis, captured between September 2024 and September 2025. The dataset spans greater than 100 nations, all 50 U.S. states, and over 20 Fitbit and Pixel Watch system fashions. From every individual we drew a number of weeks of knowledge, yielding over two billion hours — greater than a trillion minutes — of minute-resolution alerts.
SensorFM ingests 34 one-minute combination options derived from 5 sensor modalities: photoplethysmography (PPG), accelerometry, electrodermal exercise (EDA), pores and skin temperature, and altimetry. Collectively these seize coronary heart price and heart-rate variability, blood-oxygen saturation, sleep levels, movement and steps, pores and skin conductance, and temperature over a full 24-hour window.
Moderately than counting on labels, SensorFM learns by way of self-supervised reconstruction, constructing on the LSM-2 strategy and its Adaptive and Inherited Masking (AIM) framework. It is a essential design selection, as a result of lacking and fragmented information (e.g., stretches of time the place information shouldn’t be accessible) is the norm with wearable units, attributable to a wide range of components similar to sensors’ power-cycle, units coming off the wrist, energy saving modes of operation, and sensors switching on and off. Standard self-supervised strategies assume full, uninterrupted inputs and so are pressured to both impute the gaps (which may introduce bias) or discard incomplete home windows (which throws away worthwhile information). AIM takes neither path: it treats real-world missingness as a pure artifact and learns instantly from incomplete recordings, combining the tokens inherited from real gaps with these artificially masked for the reconstruction goal and treating the 2 as equal. The result’s a illustration that’s missingness-aware by development. SensorFM doesn’t simply tolerate fragmented information, it makes use of it productively, because the generative outcomes beneath present.
