Open AccessA Hybrid System to Improve the Performance of Diabetes Disease Prediction using Genetic Algorithm
Author(s) -
Emrana Kabir Hashi,
Muhammad Zaman
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.b7374.129219
Subject(s) - decision tree , naive bayes classifier , support vector machine , machine learning , computer science , artificial intelligence , bayes' theorem , construct (python library) , genetic algorithm , diabetes mellitus , decision support system , data mining , clinical decision support system , disease , statistical classification , medicine , bayesian probability , programming language , endocrinology , pathology
Currently, data mining is playing a significant role in the healthcare system. It helps to extract the hidden pattern from the clinical dataset for further analysis. Also, it can be used to build a tool to manage the medical management system. Among the life-threatening diseases, diabetes mellitus is treated as a serious disease worldwide. Due to its mortality rate, early prediction and diagnosis are very important. Several research works are going on the mentioned issues to reduce the complications caused by diabetes as well as the mortality rate. The medical science needs to analyze an enormous quantity of clinical data for diagnosis purposes using machine learning techniques. In recent approaches, the disease datasets may contain insignificant and digressive features causing less accurate results. The aim of this paper is to analyze the existing prediction systems and hence develop a hybrid disease prediction model using the Genetic Algorithm for Naïve Bayes, Decision Tree and Support Vector Machine classifiers for better accuracy. This proposed diabetes prediction model produces the accuracies of 0.8182, 0.8052, and 0.8312 when Naïve Bayes, Decision Tree, and Support Vector Machine classifiers are used respectively. From the experimental results, it can be demonstrated that for all cases Support Vector Machine provides higher accuracy comparing to the other classifiers. In the analysis, the Pima Indian diabetes dataset is used to construct the proposed model.