Google Research uses wearables and AI to forecast personalized health habits

According to Google Research:

Background

Heart disease, kidney conditions, and metabolic disorders like type 2 diabetes form an interconnected web known as cardiovascular-kidney-metabolic (CKM) syndrome. These issues often develop silently years before symptoms appear, and while smartwatches easily track past activity, translating that data into meaningful, actionable guidance remains a major challenge.

Framework Details

Google's new framework introduces a coachable health age model, which avoids simply guessing a user's birth year. Instead, it relies on published epidemiological studies linking behaviors like resting heart rate, steps, and sleep to disease risk. It is paired with Sensor World Model, an action-conditioned model pre-trained on millions of user-weeks of data to forecast how 20 physiological markers will evolve over 12 weeks under different daily routines.

Impact

Evaluated across multiple cohorts and through blind tests with medical professionals, this multimodal approach combines continuous wearable data with routine blood biomarkers to map multi-organ aging. While currently an investigational research prototype, it lays the groundwork for future prospective trials aimed at helping people build lasting, healthy habits.