Clinical decision support system for heart disease prediction using machine learning techniques

Abstract

Cardio-Vascular Disease (CVD) is a severe public health concern newlineglobally. Early and accurate CVD diagnosis is a difficult task but a necessary newlineendeavour required to prevent further damage and protect patients lives. newlineDeveloping an intelligent Clinical Decision Support System (CDSS) is a newlinesignificant focus for researchers in the healthcare domain, as diagnosing heart newlinedisease can be a difficult undertaking. The increasing volume of intricate newlinepatient medical data provides an opportunity to uncover valuable insights that newlinecan enhance the accuracy of decision-making and thereby reducing errors. newlineMachine Learning (ML) based CDSS has the potential to assist healthcare newlineproviders in making accurate CVD diagnoses and treatments. Clinical data newlineusually contain Missing Values (MVs); hence, the incorporated imputation newlinetechniques for ML have become a critical consideration when working with newlinereal-world medical datasets. Furthermore, removing instances with MVs will newlinelead to essential data loss and produce incorrect results. Imputation of missing newlinedata is a common challenge in classification problems, particularly when the newlineclass label is missing. To address these challenges, the thesis proposes a data newlinepreprocessing model and an Ensemble Two-Fold Classification (ETC) newlineframework for effectively handling MVs with approximated values based on newlinethe values in the dataset in order to create a reliable and efficient CDSS for newlineheart disease prediction with improved classification accuracy. newline