Scaling Wearable Foundation Models
This work addresses the problem of making sense of continuous wearable data for health tracking and scientific insights, representing a novel method for a known bottleneck in the domain.
The paper tackled the challenge of deriving insights from large-scale wearable sensor data by investigating scaling properties of sensor foundation models, resulting in the creation of LSM, a multimodal foundation model that establishes scaling laws for tasks like imputation and enables sample-efficient downstream learning for activities such as exercise recognition.
Wearable sensors have become ubiquitous thanks to a variety of health tracking features. The resulting continuous and longitudinal measurements from everyday life generate large volumes of data; however, making sense of these observations for scientific and actionable insights is non-trivial. Inspired by the empirical success of generative modeling, where large neural networks learn powerful representations from vast amounts of text, image, video, or audio data, we investigate the scaling properties of sensor foundation models across compute, data, and model size. Using a dataset of up to 40 million hours of in-situ heart rate, heart rate variability, electrodermal activity, accelerometer, skin temperature, and altimeter per-minute data from over 165,000 people, we create LSM, a multimodal foundation model built on the largest wearable-signals dataset with the most extensive range of sensor modalities to date. Our results establish the scaling laws of LSM for tasks such as imputation, interpolation and extrapolation, both across time and sensor modalities. Moreover, we highlight how LSM enables sample-efficient downstream learning for tasks like exercise and activity recognition.