Prediction of CKD Using Choas Based Particle Swarm Optimization (CBPSO) with Ensemble Method

Chronic kidney disease (CKD) is the high prevalence and incidence health issue which is frequently triggered by late diagnoses. It must be identified early in order to prevent expensive end-stage medications such as dialysis and kidney transplantation. A cost-efficient approach is to predicting complications of CKD and death rates which has been developing the effective risk prediction models. This paper focused in implementing modified Particle Swarm Optimization (PSO) by introducing chaos based concept as a guide line for improving the optimization by logistic map search as Choas Based Particle Swarm Optimization (CBPSO). The CBPSO has assisted to improve the feature extraction for better prediction from the ensemble method. However, the ensemble method is implemented for assist in filtering and wrapping to choose the most significant feature based on weighted average concept in diagnosing CKD. Therefore, this research focused on CBPSO with ensemble method using three classification methods namely, Logistic Regression (LR), Random Forest (RF)and K-Nearest Neighborhood (KNN). In addition, the result outcome evaluates the performance of CBPSO with ensemble SVM method with 10-cross validation method generate better accuracy of 98.50% in predicting CKD from the obtained dataset.