Comparison of machine learning methods for a diabetes prediction information system

Abstract

Diabetes is a disease for which there is no permanent cure; therefore, methods and information systems are required for its early detection. This paper proposes an information system for predicting diabetes based on the use of data mining methods and machine learning algorithms. The paper discusses a number of machine learning methods such as random forest, AdaBoost algorithm, multilayer perceptron, neuro–like structure of Consecutive Geometric Transformations Models (CGTM), linear re-gression based on the stochastic gradient descent, generalized regression neural network and regression based on the support vector machine. The Pima Indian Diabetes dataset collected from the UCI machine learning repository was used in the research. The dataset contains information about 768 patients and their corresponding nine unique attributes: the number of times of pregnancy; plasma glucose concentration for two hours in an oral glucose tolerance test; diastolic blood pressure; the thickness of the folds of the skin of the triceps; the concentration of serum insulin for two hours; body mass index; a function of diabetes heredity; the age of a person; the result of a variable class (0 – no diabetes, 1 – a sick person). The research has been carried out to improve the prediction index based on the Recursive Feature Elimination method. It was found that the logistic regression model performed well in predicting diabetes. It has been shown that in order to use the created model to predict the likelihood of diabetes mellitus with an accuracy of 78%, it is necessary and sufficient to use such indicators of the patient's health status as the number of times of pregnancy, the concentration of glucose in the blood plasma during the oral glucose tolerance test, the BMI index and the result of the calculation of the heredity functions "Diabetes Pedigree Function".