AI-driven prediction of severe respiratory sequelae in COVID-19 patients

Our AI-driven predictive model demonstrates strong potential for the early identification of respiratory sequelae in COVID-19 patients. By applying the K-means algorithm to cluster patients based on clinical characteristics, in combination with feature extraction from chest X-rays using the ResNet-50 deep learning model, we accurately stratified patients by their risk of severe respiratory outcomes. However, to evaluate its performance in clinical settings, further validation using larger, independent datasets is essential to confirm the model's reliability and generalizability across diverse populations.