Dataset used for the Machine-learning-derived predictive score for early estimation of COVID-19 mortality risk in hospitalized patients work, promoted by the SPANISH SOCIETY OF HOSPITAL PHARMACY (SEFH). The repository contains: bbdd_complete: Complete observations data set (n = 1,400 rows; d = 38 predictors); bbdd_calibration: Calibration data set (n = 10,008 rows; d = 38 predictors); bbdd_validation: Validation data set (n = 2,501 rows; d = 38 predictors); glossary variable names: glossary of terms translating the Spanish variable names, to the English ones used in the manuscript. Access to the data sets is restricted to the previous agreement via email with the authors of the work and the SEFH.