Prediction of subcutaneous glucose concentration for type-1 diabetic patients using a feed forward neural network

Insulin Dependent Diabetes Mellitus (IDDM) is a chronic disease characterized by the inability of the pancreas to produce sufficient amount of insulin. Daily compensation of the deficiency requires 4–6 insulin injections to be taken every day. The aim of this insulin therapy is to maintain normoglycemia—i.e., a blood glucose level between 4–7 [mmol/L]. To determine the quantity and timing of these injections, several different approaches are used. Prediction of future glucose values can be used for early hypoglycemic/hyperglycemic alarms for adjustment of insulin injections or insulin infusion rates of manual or automated pumps. Recent developments in continuous glucose monitoring (CGM) devices open new opportunities for glycemia management of diabetic patients. CGM technologies provide glucose readings at high frequency and consequently detailed insight into the subject's glucose variations. The objective of this research is to use glucose readings that are obtained from CGM devices, to develop a feed forward neural network model (NNM) to predict future glucose values. This NNM can be used in model predictive control systems to automatically adjust the glucose level in type-1 diabetic patients. The results of our research indicate that the NNM can be used to accurately predict future glucose values for prediction horizons of 30 minutes or less without time delay between the predicted output and the real glucose samples.