A Deep Learning based Approach for Predicting Hypoglycemia in Type-1 Diabetic Patients Using Recurrent Neural Networks

Hypoglycemia is a dangerous scenario arises in the case of patients with Diabetic Meletus, when the blood glucose level falls below a threshold level (70 mg/dL or 3.9 mmol). This is very common among Type-1 diabetic patients. This may cause a lot of negative implications in the affected patient's body, such as cognitive disabilities, seizures, and in the extreme case, death of the patient. We can make use the huge amount data available with new generation diabetic health care instruments like continuous glucose monitors (CGMs) and insulin pumps to predict future events of hypoglycemia, thereby saving the life of a patient. Our paper recommends a deep learning model to predict the upcoming hypoglycemic events using Recurrent Neural Networks (RNN).