Clinical Data-Driven Explainable AI for COVID-19 Treatment Outcome Analysis

This study aims to predict treatment outcomes for COVID-19 patients hospitalized at An Giang General Hospital using clinical data and machine learning techniques. Clinical data were collected from 2,291 patients to develop predictive models. The methodology integrates CatBoost for both feature selection and outcome classification. This method allows for the identification of key clinical features influencing treatment outcomes while ensuring high predictive accuracy. We also evaluated and compared the performance of CatBoost against other machine learning models, such as Random Forest, XGBoost, and others, after feature selection. Results demonstrate that CatBoost and Random Forest outperform other methods in terms of prediction accuracy, with CatBoost proving especially effective in selecting features that enhance classification efficiency. Furthermore, data visualization techniques were employed to provide insights into clinical characteristics and their impact on patient outcomes.