This project aims to detect hypoglycaemic seizures at an early stage to alert those affected in good time and prevent serious consequences. Hypoglycemia is a state of low blood sugar that can lead to unconsciousness, coma, and, in the worst case, death. Patients with type 1 diabetes are particularly at risk as they are dependent on external insulin, the inappropriate dosage of which can easily lead to low blood sugar levels.
AI methods can recognize potential hypoglycaemic seizures by analyzing historical data patterns. Continuously measured glucose levels and other physiological data, such as heart rate, can be used as input values for the machine learning methods. Our professorship is initially researching the precise classification of the risk up to two hours before hypoglycemia. In the next step, longer prediction periods of up to 24 hours will be considered to recognize short-term risks, postprandial hypoglycemia, nocturnal hypoglycemia, and hypoglycemia induced by increased activity. The project aims to develop personalized approaches that adapt to individual data, age, and specific ratio of individuals.
Letzte Änderung: 7. February 2025