Hybrid Approach for Heart Disease Detection using Classification Algorithms

Cardiovascular disease stands as one of the most formidable health conditions worldwide, impacting a vast number of people globally. In this paper, we introduce a hybrid system that utilizes machine learning methods for identifying cardiac problems. Machine learning (ML) is extensively implemented in the medical industry worldwide, aiding in the prevention of heart conditions and locomotor disorders by analyzing extensive volumes of complex healthcare data. Our focus in this study revolves around identifying individuals with a higher likelihood of developing cardiovascular disease based on specific medical criteria. For this study, we utilized a 303-row, 14-property online UCI dataset. To validate approximately 7 of these 14 qualities, it's essential to test the effectiveness of several methods. Our analysis is based on supervised learning techniques—K-nearest neighbor, support vector machines, random forests, and decision trees. Experimental findings demonstrate the effectiveness, accuracy, sensitivity, and precision of K-nearest neighbor (8 neighbors) compared to SVM, Naive Bayes, Decision Tree Classifier with four or eighteen features, and Random Forest classifiers. We employed feature selection methods to enhance classification accuracy and reduce execution time. Additionally, we fine-tuned hyperparameters and applied best practices for model evaluation using the LOOCV method. Classifier performances were evaluated using performance measurement metrics. Implementing this approach significantly improves the model's ability to predict cardiac arrest accurately for any individual. In contrast to previously used classifiers such as Naive Bayes, the current model's accuracy, when utilizing logistic regression and KNN, proved highly promising in predicting indications of a cardiac ailment in a person. Evaluating the classification algorithm's ability to accurately detect cardiovascular disease is pivotal, reducing stress and enhancing treatment while cutting costs. This program has provided insights into predicting the prevalence of cardiovascular disease and was executed using Python.