Hybrid quantum–classical ensemble model for enhanced cardiovascular disease risk prediction using clinical data

Abstract In terms of global health, cardiovascular diseases remain an emerging challenge, for which early estimation for timely medical treatment proves crucial. In this paper, an estimate of this challenge has been presented using a machine learning model that combines clinically engineered features such as body mass index, pulse pressure, blood pressure risk stratification, obesity status, and age-group stratification to improve reliability. With these objectives, using a structured data set of 70,000 patient records, three separately optimized classifier models of Random Forest, Light GBM, and XG Boost can be trained and tested, along with an ensemble model using weighted soft voting to minimize model-dependent variability. From the experimentally tested data, it has been observed that the accuracy of XG Boost was found to be maximum at 0.7380 with an F1 score of 0.7220, owing to its exceptional ability of handling non-linear relationships. For Light GBM, maximum AUC value of ROC was achieved at 0.8021, thereby displaying its outstanding discriminative power for distinguishing between positively and negatively classified cardiovascular diseases. For the ensemble model, balanced accuracy of 0.7376, F1 score of 0.7221, and AUC value of ROC of 0.8012 can be achieved, displaying considerable minimized variability. When compared with standard baseline classifiers, including Logistic Regression and Support Vector Machine, the proposed ensemble model achieves noticeable performance improvements, increasing Accuracy by up to 2.52% and ROC–AUC by up to 3.68%. These gains indicate a clear and practically meaningful enhancement in cardiovascular risk prediction performance.