How AI is Helping People with Type 1 Diabetes Stay in Range
Original Article: https://www.medscape.com/viewarticle/how-can-machine-learning-boost-t1d-patients-time-range-2025a10004jv
Making Carb Counting Easier
Dr. Yao Qin, a researcher at UC Santa Barbara in California, who is also a person with Type 1 Diabetes (T1D), is developing a tool called NutriBench to help estimate carbohydrates in meals more accurately. Right now, people with T1D have to manually calculate carbs before eating, which is time-consuming and often inaccurate. Large language models (LLMs), a type of AI, can now estimate carb content instantly based on meal descriptions. In a study, GPT-4o mini provided more accurate estimates than human experts, showing AI’s potential in improving diabetes management.
Smarter Insulin Adjustments for Exercise
Managing blood sugar during exercise is also difficult for people with T1D because different activities affect glucose levels in various ways. This researcher is leading a project to use AI-powered automated insulin delivery (AID) systems to help people stay in range during workouts. By analyzing data on exercise and glucose levels, her team is developing smart insulin presets that adjust insulin delivery based on activity type, reducing the risk of dangerous highs and lows.
The Future of AI in Diabetes Care
Qin’s goal is to create an AI system that learns from each person’s unique lifestyle to provide better, more personalized insulin recommendations over time. If successful, these advancements could help people with T1D manage their blood sugar more easily and live with fewer interruptions from their condition.
