Publications by authors named "Tsvetan R Yordanov"
Article Synopsis
- Federated learning (FL) is a technique that allows hospitals to develop predictive models without sharing patient records, but it may affect model performance negatively compared to centralized methods.
- The study evaluated four strategies for predicting 30-day mortality in patients undergoing transcatheter aortic valve implantation (TAVI), including centralized learning and various federated approaches.
- The results showed that federated approaches delivered similar predictive performance in terms of the area under the curve (AUC) and calibration, suggesting that FL can be a practical option for developing clinical prediction models.
Objectives: To illustrate in-depth validation of prediction models developed on multicenter data.
Methods: For each hospital in a multicenter registry, we evaluated predictive performance of a 30-day mortality prediction model for transcatheter aortic valve implantation (TAVI) using the Netherlands heart registration (NHR) dataset. We measured discrimination and calibration per hospital in a leave-center-out analysis (LCOA).