Learning Human Health and Diseases from 24-hour Wrist Movement

The paper introduces Sensori, a self‑supervised foundation model that learns health representations from 24‑hour raw tri‑axial wrist movement data. Trained on 122,640 participants across the UK, China, and the US, Sensori captures diverse movement behaviours, demographics, health axes, and physical function. In independent cohorts, the model improved disease classification for 52 of 102 conditions and incident disease risk prediction for 26 of 87 conditions, especially for neurological and psychiatric disorders.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.