ML driven approach for Post-Transplant Surveillance and Risk Modelling in Heart Patients

Heart transplantation is the most effective treatment for patients with end-stage heart failure, yet several ongoing challenges limit its success. These include improving donor-recipient compatibility, evaluating surgical risk accurately, and forecasting long-term outcomes reliably. Traditional clinical methods often focus on surgical logistics and lack data-driven tools to guide decision-making throughout the transplant journey. To overcome these limitations, this study presents the Hybrid-XR Transplant System—a modular, machine learning–based framework combining XGBoost and Random Survival Forest (RSF) algorithms. The system addresses four key areas: donor-recipient matching, pre-operative risk assessment, post-transplant monitoring, and survival prediction. XGBoost is used for classification tasks such as compatibility and risk level, while RSF is applied to survival analysis, offering improved accuracy over traditional statistical models. Each component was rigorously validated using diverse evaluation metrics. Classification performance was measured using Accuracy, Precision, Recall, F1-Score, AUC-ROC, and Matthews Correlation Coefficient (MCC). For survival analysis, the Concordance Index (C-Index) and Log-Rank Test were used. Hybrid-XR consistently outperformed Support Vector Machines, Logistic Regression, Decision Trees, and the Cox model. By integrating XGBoost-based classification and Random Survival Forest survival analysis, it enables robust interpretation of complex clinical data and supports personalized, data-driven decision-making in heart transplantation.