Modeling of Postprandial Variability in Type 1 Diabetes Mellitus Individuals Using Artificial Intelligence Techniques

Type 1 diabetes is a long-term condition in which the body can no longer make insulin, a hormone that helps control blood sugar. About 1.6 million people in the United States live with this condition, and many rely on multiple daily insulin injections to stay healthy. Managing type 1 diabetes is challenging because blood sugar levels can change quickly due to meals, exercise, stress, or even natural daily body rhythms.

New technologies called artificial pancreas (AP) systems are being developed to improve day-to-day diabetes management. An artificial pancreas system includes three parts: a continuous glucose monitor (CGM) that tracks blood sugar levels throughout the day, an insulin pump that delivers insulin, and a control algorithm, which is a set of computer rules that tells the pump how much insulin to give. These systems can adjust insulin doses automatically, helping reduce high blood sugar (hyperglycemia) and low blood sugar (hypoglycemia), lowering the burden on patients, and reducing human errors.

Although artificial pancreas systems work well in controlled studies, everyday life creates challenges that current systems still struggle with. For example, after meals—especially meals high in carbohydrates—blood sugar can rise quickly. Physical activity and natural sleep-wake cycles also affect blood sugar in ways that are difficult for current systems to predict.

In this project, we will develop computer models that better reflect these real-life situations, especially the changes in blood sugar that occur after eating. To do this, we will use real-world data collected from people with type 1 diabetes going about their daily lives. Because these data are complex and vary widely from person to person, we will use artificial intelligence (AI) to help create models that can recognize patterns and adapt to different conditions.

These new models will allow us to safely test and improve control algorithms through computer simulations before moving to studies in people. This approach will help us design artificial pancreas systems that respond more effectively to meals, exercise, and other daily activities. By improving these systems, we will aim to enhance the quality of life and long-term health of individuals living with type 1 diabetes.