Prediction and Avoidance of Nocturnal Hypoglycemia in Type 1 Diabetic Patients Using Deep Learning

Nocturnal hypoglycemia (NH) is a significant risk factor for people with diabetes because the body's immune system is too low to function effectively during sleep power. In this study, we present a new method that uses continuous glucose monitoring (CGM) data and patient-specific parameters such as age, weight, insulin dose, duration of medication, physical activity, stress level, dietary habits and sleep patterns in small amounts. Reliable evidence for early warning Recommendations such as insulin adjustment or snacking to reduce the risk of NH In addition to preventing severe NH, this approach also helps control diabetes in general and improves diabetes control in terms of patient outcomes and quality of life. Our model integrates RNN and LSTM to capture temporal glucose trends and subtle fluctuations. The proposed model achieved an accuracy of 83% and the LSTM-based network performed well in most hypoglycemic events and was found to be stable for the sixth time. These findings highlight the potential of deep learning in improving diabetes care by providing predictive and preventive solutions for NH management.