UB Research Aims to Improve Diabetes Care By Using AI

Its goal is to better understand the relationship between meals, infused insulin and blood glucose, empowering people with Type 1 diabetes to better manage the condition and improve their quality of life.

The effect that food has on blood glucose levels in people with Type 1 diabetes is well established. Less clear, however, is the role that stress, time of day, activity levels and other factors play in regulating blood glucose.

“We’re developing new tools — combining data collected from diabetes monitoring tools with AI systems, as well as traditional time-series modeling approaches — that could greatly improve how people manage their Type 1 diabetes,” says the project’s leader, Tarunraj Singh, PhD, professor of mechanical and aerospace engineering in the School of Engineering and Applied Sciences.

The project is supported by a $200,000 grant from JDRF, a New York-based nonprofit that funds Type 1 diabetes research.

Lucy D. Mastrandrea, MD, PhD, associate professor of pediatrics and chief of the Division of Endocrinology/Diabetes, is also a principal investigator on the study.

Until recently, people with diabetes had to perform a finger stick several times a day to obtain a blood sample to monitor their blood sugar.

Now, many people rely on continuous glucose monitors, which typically involve inserting a tiny sensor under the skin. The sensor measures glucose levels and sends that data wirelessly to a receiver. Patients can receive hundreds of updates throughout the day.

A nonprofit organization called Tidepool has been collecting such data from volunteers, de-identifying it, and making it available to researchers through the Tidepool Big Data Donation Project.

The UB research team will draw upon this data to validate the AI-driven technology it is working on.

The technology combines a machine learning model — machine learning is a subset of AI that involves getting computers to act intelligently without being explicitly programmed — with a problem-solving technique called first principles thinking.

This hybrid approach, Singh says, will allow the two components to inform each other. Ultimately, it can provide people with diabetes a more nuanced analysis of their blood sugar, especially as it relates to previously unaccounted for factors such as stress, time of day and how active someone is.

It is possible that this technology could be integrated with wearable devices that track heart rate, sleep, steps and other measurements.

Varun Chandola, PhD, assistant professor of computer science and engineering in the School of Engineering and Applied Sciences, is also a principal investigator on the study.