Quantum SVM-based predictive analytics: transforming classification methods in healthcare and beyond

Abstract

This study explored a Quantum Support Vector Machine (QSVM) and its application in the improvement of predictive modeling for diabetes and critical healthcare applications. Quantum computing can provide QSVMs with capabilities such as estimation of the quantum kernel, mapping a high-dimensional feature space, and robustness in noisy data that cannot be equaled by traditional SVMs. A hybrid quantum–classical pipeline was presented to evaluate QSVM classification performance by incorporating dimensionality reduction techniques (PCA) with feature scaling and quantum feature mapping. Several datasets were chosen to assess the classification performance, including diabetes, wine, prostate cancer, breast cancer, and IRIS datasets. Performance was measured using metrics such as accuracy, F1-score, ROC AUC, and R². As can be seen from the results, QSVM can outperform or match classical SVMs in a few scenarios, especially when dealing with complex medical data, which shows great promise for this quantum machine learning application to advance healthcare and other areas. The proposed system model integrates QSVM with healthcare services through secure quantum-encoded data transfer between the hospital, cloud server, and patient, to enhance the outcomes of predictive modeling and classification. This further opens up the scope for future studies, which must consider combining QSVMs with other quantum algorithms and extending them to cover more healthcare-related datasets to utilize the full capacity of QSVMs in completely transforming medical predictive modeling.