transfer_iacovbr

This repository contains the code used in a study that applied a transfer learning method to predict Intensive Care Unit admission (ICU) in patients with COVID-19 across 12 hospitals included in the multicentric study IACOV-BR. We trained and optimized an XGBoost for each hospital, identified the best-performing hospital and evaluated its generalization capacity through external validation in the other 11 hospitals. Additionally, this best-performing hospital acted as the source domain for transfer learning, allowing us to fine-tune the algorithm by incorporating 50 new decision trees trained with local data from each remaining hospital.