A customized donor-matching model for patients requiring heart transplants is described in Scientific Reports this week. The study, which analysed nearly 60,000 heart transplant patients, suggests that the model can improve the ability to select the best donor matches and avoid the worst-case matches in a clinical setting.
Heart transplantation is the main therapy for some patients with end-stage heart failure, and although survival after a heart transplant has improved significantly in recent years, donor scarcity remains problematic and sub-optimal recipient-donor matching can result in rejection and other problems that can lead to mortality. Johan Nilsson and colleagues devised a computational learning approach that can match donors with patients requiring heart transplants by combining simulated annealing and artificial neural networks. Using this method, which the authors call CODUSA (Customize Optimal Donor Using Simulated Annealing), they analysed 59,698 adult heart transplant patients, considering factors including age, gender and blood group.
The findings suggest that donor age matching is the variable mostly strongly associated with long-term donor survival. Optimizing the donor profile using CODUSA could potentially improve patient survival by up to 33 months, the authors suggest.