Authors: Assistant Professor S.Venkateswara Rao, G.Priyanka vanditha
Abstract: Improving patient survival and decreasing healthcare costs requires early diagnosis and precise prediction of cardiac events, because heart disease remains one of the top causes of death globally. There are a lot of machine learning algorithms that may help doctors spot cardiac problems, but they all have their limits when it comes to dealing with datasets that are different. Concerns with computational efficiency, false classification rates, feature selection, and forecast accuracy are common with current classification methods. In order to tackle these problems, this research introduces a hybrid machine learning classification framework that combines the strengths of SVM, Decision Tree J48, ANN, and Hidden Markov Model (HMM). In order to determine which qualities are most important for classification, the suggested approach uses two feature selection methods: Correlation-Based Feature Selection (CFS) and Gain Ratio in conjunction with the Ranker search method. A layered processing technique based on Naïve Bayes is used to integrate the best classification models for better prediction, depending on the algorithms' comparative performance. To begin, several feature subsets are used to assess the effectiveness of various classification methods. To improve prediction accuracy and overall QoS, the most efficient algorithms are then included into the proposed hybrid architecture. A better prediction performance compared to current methodologies is achieved by the suggested hybrid classification strategy, according to experimental study. This makes it a useful decision-support tool for healthcare applications and detection of heart disease.
