Identifying Clinical Concepts in Unstructured Medical Records based on Disease Prediction using SVM in Comparison with Logistic Regression

The main objective of this work is to find the clinical concepts in unstructured medical records related to disease prediction using Support Vector Machine (SVM) and Logistics Regression (LR). To achieve higher accuracy, a novel SVM classifier is used. Dataset used in this research is taken from Kaggle library. The total sample size is 20, and the two groups are SVM algorithm (N=10) and LR algorithm (N=10). The accuracy of S VM is 86.59% and loss is 13.4%, which appears to be better than the accuracy of LR which is 79.13% and loss which is 20.8% respectively. The two algorithms LR and SVM are statistically satisfied with the independent sample T-Testvalue (p = 1.26,2 tailedin SPSS analysis) with a confidence level of 95%. Therefore it is concluded that detecting medical speciality significantly seems to be better in SVM than in LR.