Ali Cinar, a professor of Chemical Engineering at Illinois Tech, has received $1.2 million from the National Institutes of Health to develop his project that aims to help people with Type 1 diabetes. The project will create an AI system integrated into his artificial pancreas to enhance the accuracy of this life-saving device.
A typical person with Type 1 diabetes has to make around 100 to 200 decisions every day just to keep the glucose levels in their body stable.
“Pancreatic cells in their brain communicate with the pancreas through the blood to regulate glucose,” says Cinar, who is also the Hyosung S. R. Cho Endowed Chair in Engineering.
When people with diabetes don’t take the appropriate dose of insulin, they’re at risk of fainting, feeling dizzy, or experiencing severe lows. If their glucose levels are chronically out of range or are too high, they may experience long-term complications like heart disease, kidney failure, and blindness.
Cinar has been on the leading edge of this technology for many years. His research group is the first to incorporate data about physical activity received through a sensor, such as a sports wristband, into the control system of an artificial pancreas, which controls insulin flow.
The machine can analyze a person’s past behavior more extensively to make good decisions about their glucose levels.
Current insulin is not an ‘instant’ solution for low blood sugar. There is a delay between the time the insulin is administered and when it begins to work, meaning that the symptoms of low blood sugar will continue to appear.
“If someone usually eats lunch at noon, and the meal usually has about 20 to 30 grams of carbohydrates, and their blood glucose level is not very low, we could say that 11:45 is a typical weekday for this person. And so, it makes sense to give them a little insulin before the meal instead of giving the full dose because it would produce a significant meeting in their glucose,” said Cinar.
Deep machine learning and artificial intelligence technologies developed in partnership with Associate Professor of Computer Science Mustafa Bilgic will recognize your behavior pattern. The algorithm will match the current day’s pattern to yours, specifically.
The system will assign a probability to the likelihood of the person having their lunch based on their current behavior, and then administer insulin accordingly. It will also monitor the glucose levels to make sure they stay in control with adjustments as needed. If it detects that the glucose level is too high because the person is eating, it will administer more insulin.
Current automated insulin-delivery systems require the user to calculate the carbohydrates in their meals and report it to the system manually. They also expect that the user will make manual adjustments when exercising. This takes time, effort, and makes human error a possible risk.
Some groups such as children and elderly adults may have difficulty entering or remembering their calorie information.
Current monitoring systems often miss complex factors that can change your glucose levels. Beyond food and exercise, stress, sleep, and other variables can either increase or decrease glucose levels.
Cinar attempts to design an artificial pancreas that senses and incorporates factors such as a person’s weight into their automated decision-making.
Stress and exercise have opposite effects on the body. If we wanted to monitor a person’s physical activity levels, but they’re under stress, our system will mistake that person for exercising and lower the amount of insulin in their body.
Multiple factors can come into play at the same time. For runners, for example, their glucose levels could fluctuate based on the combined effects of exercise, stress, and any food that they eat during the run.
Cinar continues, “This is why we’ve started to shift our focus from simply detecting exercise to detecting the state of the person. And it’s becoming more and more interesting each day.”
Cinar’s machine learning system, with enough historical data, could even learn to predict the behavior of a person with irregular habits.
Cinar said, “The advantage of powerful machine learning tools is to be able to tease out secondary relations that exist. No matter how erratic people claim their behavior is, there’s always certain patterns that can be captured.” The system looks at how the day is being developed and then looks at the dictionary to find similar patterns. It will then base its assumptions on that pattern.
Our product, Cinar, wants to give people with diabetes the chance to live their lives without constantly reviewing what they’re doing and how it will affect their insulin delivery system.
Greg and his team are designing a fully automated insulin system. There’s a strong need for this, as people have different schedules and need to adjust their insulin medications when they may be commuting or traveling.
A collaboration between the Cinar Company at Illinois Tech, Bilgic and Rashid are all on our team of world-class designers and developers.
Disclaimer:As is the case with any publicly available material, content published in this report can’t be guaranteed to be accurate or reliable. It doesn’t necessarily represent the official views of NIH.
Ali Cinar’s essay, “Integrating AI and System Engineering for Glucose Regulation in Diabetes,” was awarded by the National Institutes of Health. Award Number 1R01DK135116-01
Disclaimer: Review the official announcement for Exact news.
