Artificial intelligence and physiologic models for the management of Type 1 Diabetes Public Deposited
Type 1 diabetes is a medical condition characterized by pancreatic β-cell dysfunction and insulin depletion. Management of type 1 diabetes requires precise injections of exogenous insulin in order to bring glucose within a target range (70-180 mg/dL), however this is difficult to achieve. Factors that affect optimal management of diabetes include incorrect insulin therapy parameters, difficulty to calculate insulin boluses, physiologic variations in insulin sensitivity, adherence with care, and infrequent visits with medical providers. Incorporation of regular exercise has been shown to improve short-term glycemic outcomes and reduce the risk of cardiovascular disease in type 1 diabetes. However, exercise also increases the risk of a medical complication called hypoglycemia (glucose < 70 mg/dL) which is difficult to predict and prevent. In this dissertation, I present artificial intelligence algorithms and physiologic model systems that are designed to provide decision support for the management of type 1 diabetes.