A Comprehensive Study on Novel Hybrid Approach for Decision Support System in Disease Diagnosis Sharmila K.1, Shanthi C.1, Devi R.1, Kannan T. Kamala1 1Assistant Professors, Department of Computer Science, VISTAS, Chennai Online published on 20 March, 2019. Abstract Large quantity of Data Mining techniques have been anticipated through many authors but they are not able to provide better classification. Therefore modifications in the machine learning techniques were made many people in the past which received extraordinary interest in the healthcare sector. Therefore, various combinations of classification techniques, called Hybrid algorithms, were made and achieved a high classification accuracy by the fusion of algorithms and also by the removal of inappropriate attributes. Thus many best hybrid classifiers were made which attracted the attention of many scientists. Even though a good number of hybrid algorithms have been proposed, the hybrid approach given by Sharmila and Vedha Manickam(2016) achieved the highest, 100% accuracy using MRK-SVM hybrid in classifying diabetic dataset. Guvenir and Emeksiz (1998) showed the second level of classification accuracy with 99.25%, using voting feature intervals-5, Nearest Neighbor and Naïve Bayes, during the analysis of erythemato-squamous diseases. Dinesh K. Sharma (2016) also obtained 99.25% of classification accuracy by using Artificial Neural Network) with Support Vector Machine. Top Keywords Healthcare analytics, Data mining, Hybrid algorithm, Disease Diagnosis, Classification, Attribute selection. Top |