Tracking Blood Glucose Concentration of Type 1 Diabetic Patients Using Reinforcement Learning

Blood glucose control in diabetic patients has become one of the main concerns of researchers due to its widespread growth in society in recent years. Additionally, during the COVID-19 pandemic, research indicates that individuals who have diabetes have a higher susceptibility to contracting COVID-19 and experiencing more severe complications. This is attributed to a compromised immune system, resulting in a greater number of fatalities compared to individuals without diabetes. Therefore, the desired controller for blood glucose regulation should be very stable against hypoglycemia and perturbations such as meals, noise, and uncertainty, such as the pancreas of a healthy person. The metabolic disorders that cause diabetes in people are covered first in this article. Then Reinforcement Learning will be introduced as one of the machine learning methods to control the blood glucose concentration, and its robustness in the presence of disturbance will be investigated.