Use of Artificial Intelligence to Predict Outcomes in Critically Ill Patients Receiving Extracorporeal Membrane Oxygenation: A Systematic Review
Authors' conclusion
ML-based models hold the potential to be used as decision-support tools for patients receiving ECMO. However, their clinical applicability is limited by variability in discriminatory performance, consistently high risk of bias, and infrequent application of external validation.
Study context
- Population
- Critically ill patients receiving extracorporeal membrane oxygenation (ECMO) (machine learning (ML)-based models predicting ECMO outcomes)14 articles · 31 models
- Outcome
- Predictive performance and quality of ML-based models
Methodological details5 databases (MEDLINE, Web of Science, Scopus, CINAHL, Embase) searched from inception
- 5 databases (MEDLINE, Web of Science, Scopus, CINAHL, Embase) searched from inception
Pooled result
As reported in the abstract
- Most common predicted outcomes: mortality
- 6
- Neurologic outcomes
- 3
- ECMO complications
- 3
- Successful liberation from ECMO
- 1
- Readmission to hospital
- 1
- Random forest models
- 10
- Extreme gradient boosting models
- 7
- Models validated in an external population
- Three (7.3%)
- AUC range
- 0.33 to 0.942
- Models with good to excellent predictive performance
- Twenty (49%)
Original publication
Use of Artificial Intelligence to Predict Outcomes in Critically Ill Patients Receiving Extracorporeal Membrane Oxygenation: A Systematic Review. J Cardiothorac Vasc Anesth. Published 29 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →