Data Sharing Technique for Electronic Health Record (EHR) Classification using Support Vector Machine Algorithm

Abstract

The Electronic Health Record (EHR) integrates information about medical history in patients, complications, and history of drug use efficiently, which demands optimality and speed of service for efficiency and effectiveness of services, especially in determining outpatient and inpatient services on accurate patient history data. In efforts to improve data accuracy, this study combined the c, γ, and degree kernels in the Linear, Polynomial, and Radial Basis Function (RBF) kernels as well as data sharing techniques 10-fold cross-validation, k-medoids, and Hold- out (70 % 30%) resulted in superior K-Medoids data sharing techniques for each Polynomial kernel with an accuracy of 75.76% and a Radial Basis Function (RBF) kernel with an accuracy of 75.56% so that it can be said that the combination of K-Medoids and Polynominal kernel in the algorithm Support Vector Machine (SVM) can be used in this research case