OpenMHC Dataset and Models Advance Wearable Health AI Research.

The field of wearable health AI has been hampered by the lack of publicly available large-scale datasets and open-source foundation models. To address this, researchers have released OpenMyHeartCounts (OpenMHC), a groundbreaking resource designed to democratize and accelerate scientific progress in this domain. OpenMHC is now the most extensive open-access wearable health dataset available. It contains over 60 million hours of data collected over a decade from 11,894 consenting participants through the My Heart Counts study app. This rich dataset includes information from 19 sensor channels, such as step count, heart rate, and sleep, along with up to 169 linked variables covering health, lifestyle, mood, and behavior. In addition to the data, the release includes open-source implementations of recent wearable foundation models and a standardized benchmark. This benchmark allows for consistent comparison of models across three key tasks: downstream health and behavior prediction, multivariate data imputation, and time-series forecasting. By providing this unprecedented scale of open data, code, and model weights, OpenMHC aims to foster collaborative research and drive innovation in wearable health AI.