Chukwuma Uduku, United Kingdom
Imperial College London Department of Metabolism, Digestion and ReproductionPresenter of 1 Presentation
INDEPENDENT PREDICTORS OF HYPOGLYCAEMIA AND IMPENDING HYPOGLYCAEMIA USING A WEARABLE PHYSIOLOGICAL DATA ACQUISITION SENSOR
Abstract
Background and Aims
Hypoglycaemia remains a prevalent complication with deleterious consequences among individuals with diabetes. We aimed to identify independent predictors of hypoglycaemia and impending hypoglycaemia using real-time continuous glucose monitoring and a physiological data acquisition wristband.
Methods
Six-week longitudinal analysis of 12 adults with type 1 diabetes using real-time continuous glucose monitoring (Dexcom G6) and a clinically validated physiological data acquisition sensor (Empatica E4). A mixed effects logistic regression model was applied to predict hypoglycaemia using measurements recorded during blood glucose levels below 72 mg/dL and 54mg/dL. Measurements within 1-hour before glucose levels fell below 72mg/dL were used to predict impeding hypoglycaemia.
Results
Participants had a median age (IQR) of 40 (30-39) years and were equally stratified by gender and mode of insulin delivery (multiple daily injections and continuous subcutaneous insulin infusion). Hypoglycaemia was negatively predicted by a higher electrodermal activity standard deviation (SD) (p=0.03), higher heart rate (SD) (<0.01), and higher mean skin temperature, (p<0.05). While greater maximum phasic skin conductance responses and mean heart rate increased the odds of hypoglycaemia (p<0.01). Elevation in mean skin temperature and physical activity (SD) were both significant positive predictive factors for impeding hypoglycaemia but not established biochemical hypoglycaemia.
Conclusions
Measurements obtained from wearable physiological wristband data sensors could be integrated alongside CGM data to improve identification and prediction of hypoglycaemia.