Using wearables to enable cardiovascular-kidney-metabolic (CKM) risk profiling for personalized coaching
Summary
- Google Research introduces an investigational machine learning framework translating passive wearable signals into a modifiable coachable health age.
- The framework utilizes an action-conditioned world model (SensorWM) to simulate personalized 12-week what-if habit trajectories for CKM health management.
- Combined with routine blood biomarkers, the multimodal framework detects early subclinical strain across cardiovascular, kidney, and metabolic systems.
- Blinded evaluations showed lifestyle coaches and clinicians strongly preferred counterfactual-grounded plans over generic wearable recommendations.
Why it matters
This research bridges passive health tracking and actionable behavior change by using predictive world models to simulate personalized health outcomes, potentially improving long-term management of cardiovascular-kidney-metabolic health.