Insulin Resistance Prediction from Wearables and Routine Blood Biomarkers

Summary

Demonstrates that insulin resistance — a key metabolic health marker and precursor to type 2 diabetes — can be predicted from everyday wearable sensor data combined with routine blood lab values. Using activity (accelerometry), heart rate, SpO2, sleep stage data, and skin temperature from Fitbit/Pixel Watch devices, the model achieves clinically meaningful prediction of insulin resistance in a large population study. Published in Nature (2026, vol.