Heart Disease Prediction using Machine Learning and Optimization Techniques

Abstract

The effective management of healthcare facilities, including clinics and hospitals, newlinerelies heavily on the healthcare management system (HMS). In modern society, car- newlinediovascular disease is a pervasive contributor to morbidity and mortality, representing the primary cause of fatality. This condition significantly impacts human life, and its prediction represents a challenging task. In order to tackle this problem, it is necessary to develop a prediction system that can help medical professionals assess a patient s newlinecardiac state by analysing their clinical data or records. newlineMachine learning techniques have been utilized in the healthcare field to tackle the challenge of disease forecasting, specifically emphasizing cardiac disease prediction. newlineMany current predictive modeling approaches employ supervised learning techniques, newlinesuch as Artificial Neural Networks, Decision Trees, Inductive Rule Learning, Logis- newlinetic Regression, K-Nearest Neighbour, Naive Bayesian classifiers, Random Forest, and newlineSupport Vector Machines. However, only a limited number of studies have explored newlinethe use of unsupervised learning techniques in this area. Previous research on heart newlinedisease prediction has demonstrated that a single model-based classification approach newlineis insufficient to yield satisfactory results and early prediction with the best features; newlineand they do not provide the risk level of disease, suggesting that a hybrid model is a newlineviable alternative for improved classification performance.The objective of this work is to develop an effective disease prediction framework that can be effectively utilised in the medical field, which utilizes hybrid models in conjunction with effective feature selection techniques. The primary contribution of this work is to improve the fore cast accuracy based on hybrid Modified RoughK-means++ clustering with Restricted newlineBoltzmann Machine (MRK-means++ - RBM) model, Optimal mayfly based Feature Selection with Adaboost Ensemble based Classification model (OMA-Adaboost) and risk level based Deep bi-LSTM and