A Hybrid Mining and Optimization Based Framework for Heart Disease Prediction

Abstract

The most recent trend in next-generation computing for data mining and analysis is Artificial newlineIntelligence (AI), especially for the predictive analysis of high-risk diseases such as heart-related newlinediseases. AI applications are made possible by machine and deep learning techniques for better newlinereal-time data management. These techniques have the ability to prevent disease outbreaks, newlineidentify and diagnose illnesses and save operating costs for hospital administration and patients. newlineVarious classification methods have been created by researchers to recognize heart-related newlinediseases based on numeric data. A massive amount of data related to heart diseases are collected newlineevery year from different medical universities and hospitals worldwide, but it has not been newlineadequately utilized to link it with symptoms and disease risk. Patients can currently undergo newlinevarious costly tests such as stress testing, chest X-rays, coronary angiograms, cardiac magnetic newlineresonance imaging and electrocardiograms to determine the extent of their heart diseases. newlineThis thesis aims to survey the incidences of heart diseases in different geographical regions on the newlinebasis of death rates as well as to analyze the risk levels of heart diseases in different age groups. newlineReviewing the advantages and disadvantages of the previously used algorithms and methods for newlineheart diseases classification. Evaluate the impact of various Machine Learning (ML) algorithms newlineof data mining with and without feature optimization and also find the impact of deep learning newline(DL) methods on different testing parameters for heart diseases classification. The experiment has newlinebeen performed using Cleveland (303 instances), Statlog (270 instances), Cardio (65000 newlineinstances), NIH X ray chest image (1.2 lakh images) datasets newline