<p>Cardiovascular diseases (CVD) have emerged as a leading cause of morbidity and mortality globally due to sedentary lifestyles and other inactive habits. Advancements in deep learning methods have markedly improved medical diagnostics; however, some challenges, such as lack of labeled datasets, high computational requirements, and limited interpretability, still persist and affect their usability, particularly in ECG-based cardiac disease prediction. In this study, we leverage the emerging capabilities of quantum machine learning (QML) to perform both multiclass and binary CVD classification across multiple datasets. Notably, two QML techniques are implemented, namely, the quantum support vector machine (QSVM) and the quantum convolution neural network (QCNN), and their performance is compared against their classical counterparts and other baseline methods under identical experimental conditions. Training and testing these models demand advanced computational resources because of the rapid growth in quantum circuit states, the high cost of constructing kernel matrices, and the large number of parallel circuit runs needed for gradient estimation. All of these factors require parallel and distributed execution on high-performance platforms to maintain practical run time and support latency-sensitive clinical decision-making. Extensive experimentation is carried out using IBM’s Qiskit library, and results show that QCNN outperformed classical CNN by approximately 5% achieving accuracies of 95.3%, 95.6%, and 96.5% across different datasets. Similarly, QSVM surpasses classical SVM by approximately 8% with 91.3%, 93.10%, and 95.20% accuracies, respectively. Beyond precision, cross-dataset validation, ablation study, error analysis, and statistical analysis are performed to validate the outcomes.</p>