Heart Disease Prediction Using Machine Learning

One of the main reasons for death worldwide is heart disease, and early detection of the condition can help lower the risk of having a cardiac arrest. This research paper aims to suggest a machine learning-based method for estimating the risk of developing cardiac disease. First recent advancements in the field have been reviewed and then an ML model has been implemented to work on the limitations of the previous research work. To train and evaluate several ML algorithms, Framingham Heart Study (FHS) dataset has been used. The algorithms trained in this research paper are decision trees, random forest, SVM and ANN. It has been demonstrated that ML techniques can significantly improve the accuracy and timeliness of heart disease diagnosis, with SVM achieving the highest accuracy of 86.53%. This research paper provides valuable insights for clinicians, researchers, and data scientists working in the field of cardiovascular disease diagnosis and treatment.

I. Introduction

Heart disorders are a major global public health concern that account for a large portion of fatalities. According to the WHO, 17.9 million deaths worldwide are caused by cardiovascular illnesses each year, or 31% of all fatalities [1]. Effective disease prevention and management depend on the early detection and precise prediction of CVDs. Many risk factors for heart conditions, including diabetes, high blood pressure, smoking, high cholesterol, and obesity are blamed for the rise in heart conditions in the general population [2]. The Framingham Risk Score and other conventional risk assessment models, in particular when applied to populations with diverse ethnic and genetic origins, have limitations in their ability to predict the risk of CVDs [3]. By combining big and complicated datasets from many sources, machine learning (ML) techniques can be utilised to increase the precision of heart disease risk prediction [4]. In order to forecast heart failure disease, Alotaibi (2019) [5] proposed a machine learning model that makes use of demographic, clinical, and laboratory data. In order to forecast cardiac illness, Ali et al. (2020) [6] created a smart healthcare monitoring system that combines ensemble deep learning with feature fusion. The clinical uses of ML in cardiovascular disease and its connection to cardiac imaging were examined by Al'Aref et al. (2019) [7]. In a research study with 423,604 participants from the UK Biobank, Alaa et al. (2019) [8] employed automated machine learning to predict cardiovascular disease risk. These studies show the potential of ML to increase the precision and effectiveness of heart disease prediction. Then there are various studies that offer principal component analysis (PCA) and other statistical methods as feature selection methods. For instance, Garate-Escamila et al. (2020) [9] developed feature selection and PCA-based classification models for the prediction of cardiac disease. The most recent developments in cardiac disease prediction using ML and image fusion were evaluated by