Predicting Optimum Insulin Doses for Patients with Type I Diabetes Mellitus

Type I diabetes mellitus (type I DM) is a serious disease, also known as insulin-dependent diabetes. There are two main approaches to treating type I DM patients: treatment with insulin pens and two types of insulin (ultrashort/short and prolonged/basal/long insulin), and treatment with insulin pump therapy and ultrashort insulin. It is still a challenge to develop a system for calculating insulin doses for treating type I DM using insulin pens.

This work addresses the problem of determining the correctness of insulin doses based on historical data. Various means of artificial intelligence (decision trees, gradient boosting, support vectors, and different neural network architectures) are proposed to solve the problem, and a comparative analysis is performed.