Decision Support System for Type 1 Diabetes Doing Exercise

Type 1 diabetes is a long-term condition in which the body cannot produce insulin, a hormone needed to control blood sugar levels. People with Type 1 diabetes must carefully balance insulin doses, food intake, and physical activity every day to stay healthy. Worldwide, about 9.2 million people live with Type 1 diabetes, including 1.8 million children and teenagers under the age of 20. If blood sugar is not well controlled, it can lead to serious long-term problems such as heart disease, nerve damage, kidney failure, and vision loss.
One of the most difficult parts of living with Type 1 diabetes is managing blood sugar during exercise and after complex meals, especially meals high in fat or protein. Exercise is very good for overall health, but it can cause blood sugar to drop dangerously low, sometimes hours later. Because of this fear, fewer than one in three people with Type 1 diabetes exercise regularly. Similarly, high-fat or high-protein meals can cause blood sugar to rise slowly and unexpectedly several hours after eating, making diabetes management very challenging. Even with modern insulin pumps and glucose sensors, more than 50% of people with Type 1 diabetes still struggle to keep their blood sugar in a healthy range.
These challenges are even greater in low- and middle-income countries, where many people cannot afford advanced diabetes technology or regular access to specialist doctors. As a result, millions of individuals must rely on guesswork to manage their condition, putting them at constant risk of dangerous blood sugar highs and lows.
This research aims to develop an affordable and explainable artificial intelligence advisor to help people with Type 1 diabetes manage their blood sugar more safely during exercise and after complex meals. The system will be powered by a Video Foundation Model which is a type of advanced computer model that can learn from videos and sensor data to understand human activities, such as different types of exercise and daily routines.
The AI advisor will combine information from continuous glucose monitoring systems, insulin records, meal information, and wearable activity data. Continuous glucose monitoring is a technology that measures blood sugar levels every few minutes using a small sensor worn on the body. By learning patterns from real-world data, the system will provide personalized, real-time recommendations that are easy to understand and explain why a suggestion is being made.
This research is necessary because it could reduce fear around exercise, improve long-term health, and make expert-level diabetes guidance accessible to people who currently do not have it. By focusing on affordability and transparency, this project has the potential to improve daily life and health outcomes for people with Type 1 diabetes worldwide, especially in resource-limited settings.