Machine learning (ML) has revolutionized healthcare, including, the classification of heart diseases. Traditional ML techniques often struggle with complex and high-dimensional datasets of heart diseases. Graph-based techniques have emerged as a promising approach to enhance the results by capturing intricate relationships between data points. The aim of this article is to apply and improve ML classification of heart diseases using graph-based techniques. This study utilizes a dataset of 1190 samples with 23 features, including features derived from graphs. Several ML models are employed, and their performance is evaluated using accuracy, precision, and recall. The results demonstrate significant advancements in the classification of heart diseases, with the graph-based network model achieving an accuracy of 95%. The superior performance of the graph-based model can be attributed to its ability to take into account complex indirect relationships between risk factors and disease outcomes. Further improvements can be made by considering advanced properties of complex networks, such as their small-world or scale-free characteristics.
