eICU and MIMIC-IV used for predictive modeling of SpO₂-SaO₂ discrepancy

This dataset contains clinical data from two large healthcare databases, eICU (2014-2015) and MIMIC-IV (2008-2019), aimed at developing and validating a machine learning model for predicting discrepancies between oxygen saturation levels measured by pulse oximetry (SpO₂) and arterial blood gas measurements (SaO₂). The dataset includes a variety of patient variables such as demographic information, vital signs, and clinical indicators relevant to ICU settings.