Google Research Unveils 12-Week CKM Coaching Model Trained on 2.28 Million Person-Weeks
Updated
Updated · Google Research · Nov 11
Google Research Unveils 12-Week CKM Coaching Model Trained on 2.28 Million Person-Weeks
2 articles · Updated · Google Research · Nov 11
Summary
Google Research introduced an investigational framework that turns wearable data into a “coachable health age” and simulates personalized 12-week habit changes to guide cardiovascular-kidney-metabolic health coaching.
2.28 million person-weeks of de-identified data from about 100,000 consenting Fitbit users trained its Sensor World Model, which forecasts 20 physiological markers from two weeks of baseline data under different sleep, steps and cardio routines.
18,423 disease-free participants in the All of Us cohort helped validate the health-age model, with healthier age gaps linked to higher 8-year disease-free survival and measurable improvement over 3-month windows.
49 participant profiles reviewed by five lifestyle coaches and five physicians showed counterfactual-grounded plans were preferred nearly 3-to-1 over generic wearable-based plans, while meeting 99.2% clinical safety standards.
819 WEAR-ME participants were used to extend the system with blood biomarkers for cardiovascular, kidney and metabolic tracking, though Google said the models remain offline research prototypes rather than clinical diagnostic tools.