Sensitivity Analysis of a Predictive Pump Suspension System to Treat People with Type 1 Diabetes

摘要: The primary goal of a low glucose suspend system is to reduce the risk overnight hypoglycemia (low blood glucose) in individuals with type 1 diabetes by reducing/suspending insulin infusion. We have developed Kalman filter-based algorithm, combined number safety rules, implement predictive that shuts off an pump based on prediction 30-70 minutes future. This has been studied over 2,000 nights outpatient-home environment. In this paper, analysis data, we isolate effects individual rules part simulating their removal from existing data. Specifically, decompose basal into small boluses and, using model pharmacodynamic action (the time effect glucose), alter real data corresponding addition or via simulation. Our results show limiting total suspension 180 per night prevents excessive cases where average calibration 58 mg/dl, above mean 18 mg/dl. Further, also simple threshold algorithm suspends below 100 mg/dl if level flat falling, comparable performance. Lastly, filter at heart reduces spent 70 50% expense rise 12 morning levels.