Multi-Agent System Based on Machine Learning for Early Diagnosis of Diabetes

Abstract

Diabetes is increasing all over the world. In Morocco, more than 2 million people aged 18 and over are diabetic. In order to diagnose and treat diabetes, early detection is needed. Thus, the aim of this qualitative study was to overcome this difficulty by giving more autonomy and initiative to the different software modules specialized in the medical diagnosis. To enable robust, reliable medical diagnostic support, Multi-Agent System can be the tool for distributed diagnostic. This article will attempt to create a new multi-agent system that evaluates the performance of three well-known machine learning algorithms: the artificial neural network (ANN), support vector machines (SVM), and logistic regression or logit model (LR), based on the diabetes Database. Then the system aggregates the classifications of these algorithms with a controller agent to increase the accuracy of the classification using a majority vote. In addition, this article discusses the current gap and the challenges of adopting machine learning algorithms in multi-agent systems.