Using wearables to enable cardiovascular-kidney-metabolic (CKM) risk profiling for personalized coaching
The paper presents an investigational machine learning framework that converts passive wearable data into a modifiable “coachable health age” and uses a sensor‑world model (SensorWM) trained on over 2.2 million person‑weeks to simulate personalized 12‑week “what‑if” habit trajectories, which are then turned into clinician‑vetted coaching plans by a language‑model agent. The approach also allows integration of routine blood biomarkers to detect early sub‑clinical strain across cardiovascular, kidney, and metabolic systems. This end‑to‑end pipeline bridges continuous tracking with actionable, personalized coaching, offering engineers a concrete blueprint for health‑age modeling and counterfactual simulation.
⚡ Key Takeaways
- Trained on over 2.2 million person‑weeks of wearable data for the SensorWM world model.
- SensorWM world model generates personalized 12‑week habit trajectory simulations.
- A language‑model agent translates simulated trajectories into realistic, clinician‑vetted coaching plans.
- Framework can be expanded to incorporate routine blood biomarkers for multi‑organ CKM health profiling.
Engineers can now build systems that not only predict health risk from wearables but also generate actionable, personalized habit recommendations, accelerating the move from passive monitoring to proactive health coaching in production.
✅ Practical Steps
- Apply the concepts from this article to design a health‑age model that relies solely on modifiable wearable signals.
Want the full story? Read the original article.
Read on Google Research ↗