Description:
Summary
This hypoglycemia prediction algorithm calculates the risk of possible hypoglycemic episodes before the patient goes to bed.
Technology Overview
Utilizing a support regression algorithm (SVR) to predict minimum glucose levels and a decision theory/cost-benefit analysis to estimate nocturnal glucose alert levels, this algorithm predicts nighttime hypoglycemia before the patients goes to sleep. By using this algorithm, type 1 diabetes patients can minimize the risk of overnight hypoglycemic episodes as it can predict and notify patients to consume proper amounts of carbohydrates before going to sleep. This can eliminate the risk of missing nighttime alerts, causing greater risk for hypoglycemic episodes.
Currently, there are no nocturnal hypoglycemia algorithms on the market. This algorithm was formulated using Tidepool big data donation datasets of 76,000+ nights of continuous glucose monitoring data from 124 patients. When used correctly, this algorithm could reduce nocturnal hypoglycemia from 6.3% down to 0.9%.
Publication
Mosquera-Lopez et al., "Predicting and Preventing Nocturnal Hypoglycemia in Type 1 Diabetes Using Big Data Analytics and Decision Theroretic Analysis." Diabetes Technol Ther. 22(2020): 801-811. Link
Licensing Opportunity
This technology is available for exclusive and non-exclusive licensing.