Lloyd and Minkowski Based K-Means Clustering for Effective Diagnosis of Heart Disease and Stroke

Abstract

Computer Aided Data mining based decision support system plays a major role in research to easily diagnose medical disease at an early stage. Automatic annotation approach grouped similar medical semantic terms, however, false detection lead to incorrect merging increasing the computational complexity. In this work, a method called Lloyd and Minkowski based K-Means Clustering (LMK-MC) is designed to organize various features in heart disease, with the aim of increasing the features to be incorporated improving the clustering efficiency and reducing the average computational complexity. The Lloyd and Minkowski based K-Means Clustering includes a function that performs a Minkowski based K-Means Clustering to organize similar type of symptom features for easy prediction. By applying Minkowski based K-Means Clustering, similar type of features causing heart disease is analyzed for any set of k centers, aiming at improving the clustering efficiency. Next, the method LMK-MC, applies pair-wise proximities between all pairs of features reducing the computational complexity for intra-cluster. Finally, Lloyd's algorithm based clustering on medical data is designed that moves every center point to the centroid and performs updates. Lloyd's algorithm based clustering obtains local minima solution that easily merges similar features leading to heart disease and strokes, various disease features for labeling is identified effectively with high recognition accuracy. In order to measure the efficiency of LMK-MC, experiments are conducted using Cleveland Clinic Foundation Heart disease data set. Experimental results demonstrate that the proposed Lloyd and Minkowski based K-Means Clustering achieves efficient amount of identification improving the clustering efficiency, reducing the computational complexity while measuring intra-cluster distances for different clusters and, heart disease and stroke identification rate with minimum processing time.