Artificial Intelligence and Diabetes Management

Patients with Type 1 diabetes are unable to produce insulin, an essential hormone needed to convert food into energy, and so they must depend on exogenous insulin to survive. Although there is no cure for diabetes, patients can avoid serious complications of the disease by keeping their blood glucose levels as close to normal as possible. Blood glucose data can be automatically collected and sent to a physician for analysis. This results in a large amount of data, and at present, data analysis is a time-consuming, manual process. The sheer volume of data makes it difficult for physicians to provide therapy recommendations in a timely manner. Frank Schwartz, MD, associate professor of specialty medicine and an endocrinologist who cares for diabetes patients, and Cynthia Marling, associate professor of electrical engineering and computer science, are exploring the use of artificial intelligence to analyze data from patients with Type 1 diabetes and provide recommendations for therapy adjustment comparable to those an endocrinologist would make. They and their group aim to provide more frequent feedback for patients, enabling better glycemic control while reducing the time spent by endocrinologists on routine data analysis. They are conducting a preliminary clinical study, involving patients with Type 1 diabetes on insulin pump therapy, to evaluate the feasibility of providing intelligent decision support.