Forecasting consent in organ donation: early assessment of machine-learning techniques

Accurately predicting whether consent for organ donation will be granted is essential for optimizing timing and resource use in donor management. This study develops and evaluates machine learning models to estimate the likelihood of obtaining consent based on donor and contextual factors. The goal is to support early clinical decision-making by identifying cases where consent is more or less likely. Using real-world data from a regional transplant center operating under an opt-in system, we conduct data preprocessing, feature selection, and model training with various algorithms. Model performance is assessed using standard classification metrics, and key predictors of consent outcomes are identified. Results show accuracy levels exceeding 80%, highlighting the importance of including information about the relatives responsible for the decision. We also find that prediction accuracy varies with donor nationality, being higher for non-Italian donors. These findings demonstrate the value of predictive analytics in improving organ procurement efficiency and reducing unnecessary costs.