Objective:

To evaluate three predictive models—Random Forest, Support Vector Machine, and a multilayer perceptron deep neural network—for the prediction of dataset-defined myocardial infarction risk using a merged public heart disease dataset.

Approach:

• Study Design: Retrospective computational modeling based on secondary, publicly available heart disease data.
• Dataset: Merged public heart disease dataset consisting of 1,888 records and 14 variables derived from five publicly available datasets.
• Outcome Definition: Binary target label indicating higher (1) or lower (0) likelihood of heart disease/heart attack risk.

Key Findings:

• Machine learning methods can model complex nonlinear relationships among clinical variables.
• Random Forest, Support Vector Machine, and deep learning models show strong performance in cardiovascular prediction tasks.
• The dataset used reflects a structured diagnostic-risk dataset rather than a purely baseline screening scenario.

Interpretation:

The models evaluated should be considered as risk-classification decision-support tools rather than definitive diagnostic tools for acute myocardial infarction.

Limitations:

• Models evaluated primarily using accuracy, which can be misleading with low disease prevalence.
• Lack of external validation and comparison with established clinical risk scores.
• Dataset does not reflect the lower prevalence of acute MI typically observed in real-world settings.

Conclusion:

The study positions its results as an internally validated computational model rather than a clinically deployable MI diagnostic tool.