Providing a Diagnosis System based on Hybrid Approach of Artificial Immune System and Velocity Bounded Boolean Particle Swarm Optimization Algorithm

One of the practical usages of data mining is design of intelligent medical diagnosticsystems. The most important challenge in designing such systems is increasing detectionaccuracy; therefore this paper is proposed with the aim of increasing the detectionaccuracy. In this paper an intelligent diagnosis system is presented by using a hybridapproach of artificial immune system algorithm and velocity bounded boolean particleswarm optimization approach. In the proposed system, firstly, most informative featuresare detected by using the velocity bounded boolean particle swarm optimizationalgorithm, and then the classification of disease is done by using the artificial immunesystem algorithm according to the values of detected features. UCI medical datasets areused for evaluating the proposed approach; also, 5-fold cross validation with 10 repeatsis used for more accurate evaluation of the proposed approach. Simulation results showsthat the approaches based on artificial immune system have a good ability in modelingmedical data and can yield acceptable classification accuracy. Furthermore, the proposedapproach significantly improves the classification accuracy due to the use of featureselection algorithm based on velocity bounded boolean particle swarm optimizationalgorithm.

Keywords:

diagnosis , artificial immune system algorithm , boolean particle swarm optimization algorithm , feature selection , k-fold cross validation