Automated insulin pump design based on probabilistic prediction algorithm

Insulin is a hormone secreted by the pancreas that regulates blood sugar in the body. Diabetes mellitus, also known as diabetes, occurs when the pancreas can not produce sufficient insulin for the body. Diabetes is a chronic disease and patients at advanced stages need to inject insulin externally at certain doses during the day. Artificial pancreas designs are aimed to measure these glucose levels and to control these levels by applying insulin to the required amount. Nowadays the systems consist of continuous glucose monitoring systems and automated insulin pumps. In this thesis, an automated insulin pump system is designed. In the system implemented, insulin injection is made to the patient at certain intervals during the day and records of injections made are kept. Thus, the patient's insulin intake is automatically monitored and patient follow-up is facilitated. The intelligent insulin pump realized is able to make injections by predicting and calculating insulin doses solely with the meal information given by the patient using a probabilistic prediction algorithm. In this system there is no need for patient to make blood glucose measurements for the bolus doses. Depending on the predicted and calculated insulin doses, insulin injection with the pump will be performed. In addition to this bolus dose, an injection of the basal insulin dose will also be performed during the day periodically according to the weight information of the patient.