Diagnostics of diabetic polyneuropathy in children and adolescents using data mining methods

Abstract

The article considers the application of data mining methods to develop a diagnostic model for one of the most common and dangerous diabetes mellitus complications – diabetic polyneuropathy characterized by damage to peripheral nerve fibers. The article explores the possibility of diagnosing diabetic polyneuropathy with the use of machine learning methods. The data base of the study includes 3204 anonymized medical records of children and adolescents with type 1 diabetes residing in the territory of Altai krai. 1100 records hold data on diabetic polyneuropathy. Medical records contain different information: patient complaints, medical case history, test results. The attribute space is represented by 40 different indicators. In the course of the study, we considered the differences between the attribute values in two groups, built the attribute space ensuring the best classification quality. The article presents the results of applying various methods to transform the source data, evaluates the quality of the resulting model. The implementation of all the study phases was carried out with the use of the Python high-level programming language.