Historical Data-Based Heart Disease Analysis Using Machine Learning Techniques

Abstract

Healthcare solutions can be provided to every human being with the advancement of machine learning techniques, irrespective of age. Utilizing classification and clustering techniques, diseases can be predicted using a dataset of that specific disease, thereby reducing costs. Due to a lack of knowledge and skills to provide first aid to heart patients, emergency fatalities may occur. This research studies various datasets to identify different features or characteristics causing heart disease. Analysis of these features or the interrelationships between these features can play a vital role in the prediction of heart disease using machine learning algorithms and data mining techniques. The research aims to develop an accurate predictive model that can effectively identify individuals at high risk of developing heart disease. The study utilizes a diverse dataset consisting of various clinical and demographic features, including age, gender, blood pressure, cholesterol levels, diabetes, thalassemia, electrocardiogram readings, etc. The objective of this paper is to propose an integrated framework for pre-processing (as and when required), mining, training, and testing. This research implements three classification algorithms to analyze various historical datasets to make accurate predictions. The classifiers K-Nearest Neighbour, Naïve Bayes, and Decision Tree are employed to train and evaluate predictive models. The dataset is pre-processed, including handling missing values, normalizing features, and addressing class imbalances if present. In order to compare the accuracy of various datasets, a range of evaluation metrics, such as accuracy, precision, recall, and F1-score, are measured, and a performance evaluation confusion matrix is prepared. The results of the study demonstrate that a decision tree classifier with the selected features in the chosen dataset can be used to effectively predict heart disease. The novelty of this research is to select important features causing heart disease with the highest probability.