Heart Rate and CGM Feature Representation Diabetes Detection From Heart Rate: Learning Joint Features of Heart Rate and Continuous Glucose Monitors Yields Better Representations

Continuous Glucose Monitors (CGMs) provide tremendous value for diabetes detection and management. However, their high cost regulatory complications have prevented the widespread use of CGMs. On other hand, Heart Rate (HR) monitors are in wide growing popularity. In this work, we investigate connection between HR monitor CGM devices to find a cheaper alternative measuring glucose dysregulation. We recruited 550 volunteers that included healthy, type 2 diabetic, pre-diabetic, gestational diabetic cohorts wear 10 days. Although physiological mechanisms underlying regulation heart rate share many commonalities, commonly used features time series analysis yield poor correlations signals. by learning joint representation using Canonical Correlation Analysis (CCA), can learn CCA space respectively statistically significant correlation. Finding maximize objective with enables us about subject's system an alone not require cumbersome Here consider as particular application. representations serve proxy CGMs classification.