Sign in Subscribe
Machine learning

Machine Learning Model Predicts Discrepancies in Oxygen Saturation

Strategic Insights

— 1 min read

Researchers at the University of Tokyo have developed a machine learning model to predict discrepancies between percutaneous oxygen saturation (SpO2) and arterial blood oxygen saturation (SaO2) in critically ill patients. The model was trained on data from the eICU Collaborative Research Database and validated on data from the MIMIC-IV dataset. The study aimed to explore the factors related to the discrepancy and to develop a prediction model that uses readily available bedside information.

Key Takeaways:

  • The study used a dataset of 4,781 admission records from the eICU Collaborative Research Database and 19,804 paired SpO2 and SaO2 measurements.
  • The XGBoost model demonstrated the best predictive performance with an AUROC of 0.73 and a calibration slope of 0.90 in the development cohort.
  • The model was deployed as a web-based application, making it accessible for bedside use.
  • Vital signs, such as low blood pressure and low temperature, were associated with the discrepancy between SpO2 and SaO2.
  • The model was validated on the MIMIC-IV dataset, where it performed with an AUROC of 0.56, but demonstrated improved performance after an exploratory model-updating step, reaching an AUROC of 0.70.
  • The study highlights the importance of considering vital signs in predicting SpO2-SaO2 discrepancies in critically ill patients.

Statistics:

  • 4,781 admission records were used in the eICU Collaborative Research Database.
  • 19,804 paired SpO2 and SaO2 measurements were used in the study.
  • The XGBoost model had an AUROC of 0.73 and a calibration slope of 0.90 in the development cohort.
  • The model had an AUROC of 0.56 in the MIMIC-IV validation cohort, but 0.70 after an exploratory model-updating step.
  • The model was trained on data from 2014 to 2015 and validated on data from 2008 to 2019.

Sources:

  • PLOS One (2025;20(10))
  • Public Library Science (1160 Battery Street, Ste 100, San Francisco, CA 94111, USA)
  • University of Tokyo (Faculty of Medicine, Bunkyo, Tokyo, Japan)
  • eICU Collaborative Research Database
  • MIMIC-IV dataset

Read more