Heart Disease Prediction Using AI

As the world keeps grappling with untimely deaths caused by heart diseases, due to lack of advanced diagnostics, AI is seen as a solution to this problem. Traditional methods of diagnosing cardiovascular diseases are no longer seen as adequate, greater improvement is needed to minimize preventable deaths caused by it and this is where AI comes in. Cardiovascular diagnostic models have been constructed in this paper to assess how accurate these models are and what issues they currently face. In this research, the primary focus was to develop an AI-based cardiovascular diagnostics model that cannot only meet but also exceed the sensitivity and specificity levels of the traditional diagnostics tools. Various algorithms were employed, some of them are as follows: Naive Bayes, Gradient Boosting, Decision Tree, Random Forest, Logistic Regression, K-Nearest Neighbors. Each algorithm is trained, retested, and altered through hyperparameter optimization. Cross-validation was also incorporated to validate the models while simultaneously reducing bias. The use of these multiple algorithms in this form assures the end user of the effectiveness of the designed models. Random Forest secured the best performance in the K-fold evaluation and achieved the highest accuracy level of 94 %. The F1 Score was equal to 93.95 further supporting the model's effectiveness in distinguishing patterns from the trained dataset.Gradient Boosting was also able to showcase convincing results with a precision equal to 94.94 and F1 score equal to 92.59.