Machine learning-based mortality prediction models using national liver transplantation registries are feasible but have limited utility across countries

Data from liver transplantation registries in Canada, the United Kingdom, and the United States were used to develop machine-learning models that predict 90-day mortality after primary transplantation. Ridge regression performed best across both individual and harmonized registry datasets. Performance declined when models were trained on harmonized variables and was poor when models were transferred between countries. The findings show that registry-based models have fair discrimination within their source databases but limited external validity, and suggest that standardizing registry variables could improve their value in clinical decision-making.

American Journal of Transplantation, 23(1), 64–71