736-Predicting Insulin Resistance Using Wearables
This research study introduces a scalable method for detecting insulin resistance (IR) by combining data from consumer wearables with standard blood biomarkers. Using a large cohort, the researchers developed deep neural networks and a specialized wearable foundation model to predict metabolic risk more accurately than traditional demographic assessments. The results demonstrate that integrating heart rate and activity patterns with glucose and lipid panels significantly improves the identification of individuals at risk for type 2 diabetes. To make these findings actionable, the study…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.