Ensemble-Based Predictive Modeling Machine Learning Framework for Live-kidney Donation Prediction

Predicting outcomes in live-kidney donation (LKD) is crucial for optimizing donor-recipient matching and improving post-transplantation success rates. Machine learning (ML) techniques, particularly ensemble methods, have shown promise in various biomedical applications due to their ability to combine multiple models for enhanced predictive accuracy. This paper proposes an ensemble-based predictive modeling framework tailored specifically for LKD prediction, incorporating the Synthetic Minority Oversampling Technique (SMOTE). The framework integrates diverse ML algorithms and data preprocessing techniques to effectively handle the complexities inherent in medical data, thus providing robust predictions for decision support in clinical settings. Experimental results using patient data from Avera Transplant Institute’s Live-donor Program demonstrate that our proposed model outperforms existing techniques with a testing accuracy of 98%. Thus, revealing its potential to support clinical decision-making in transplantation medicine and thereby improving patient care and donor selection processes.