An enhanced ant colony optimization with Gini index for predicting type 2 diabetes.

Diabetes is referred to a group of metabolic disorder that is caused by irregular amount of insulin produced in body. There are two types of diabetes in major clinical cases that is type 1 diabetes (T1D) and type 2 diabetes (T2D). Most of the cases reported is T2D and this is caused by a number of people that are unaware of their possibility in having T2D and lack of knowledge regarding its risks. It is important to perform an early diagnosis of T2D to ensure that people will become more aware and also reduce the chances of them getting high risk or complications such as stroke, blindness and damage to body and nerves. To do a classification task, a tool namely machine learning will be used. This tool is used by researcher to analyze a large amount of data to obtain a data pattern. This pattern then will be interpreted to get a knowledge that can help in decision making process. For the medical diagnosis, there are two things that is important which is accuracy and comprehensibility. Ant Colony Optimization (ACO) is a meta-heuristic method that is branch of swarm intelligence and it has been used to solve many complex problems and generate comprehensible result. Hence, this paper will introduce the new proposed algorithm that will hybrid between ACO with Gini Index. Gini Index is a heuristic measure that is widely used in Classification and Regression Trees (CART). The proposed algorithm was execute using the BlueJ application that is written in Java language. The aim of this paper is to investigate the performance of the new proposed algorithm in terms of accuracy and computation time. Experimental results on Pima Indian Diabetes data set show that the proposed algorithm can improve the predictive accuracy and computation time compared to the original ACO. In addition, it also leads to a smaller number of rules and terms. [ABSTRACT FROM AUTHOR]