Acute cardiac care & resuscitationDiagnostic or prognostic study
Machine Learning Based Prediction of Hospital Mortality in Post-Cardiotomy Cardiogenic Shock with Mechanical Support
- development and evaluation of machine learning-based models
- registry-based
- 111 centres
- training (60%), validation (20%), and test (20%) cohorts
Authors' conclusion
Machine learning models demonstrated moderate predictive performance for in-hospital mortality in post-cardiotomy cardiogenic shock patients on extracorporeal life support. The random forest model demonstrated moderate discriminative performance, highlighting the relevance of readily available clinical variables. External validation and calibration analysis are required before clinical implementation. The model is best interpreted as a tool for dynamic risk assessment during ECMO support.
Study context
- Population
- Adult patients who received extracorporeal life support for post-cardiotomy cardiogenic shock, Extracorporeal Life Support Organization (ELSO) registry, January 2010 to December 2020adult patients from the ELSO registry
- Participants
- 5,982 patients
- Comparison
- Six machine learning algorithms predicting in-hospital mortality
Performance
As reported in the abstract
boosting AUC (highest)
0.759
random forest AUC
0.688
Further findings
Age
Lactate during support
BMI
ECMO duration
Mean
| Outcome | survivors | non-survivors |
|---|---|---|
| Arterial pH | 7.302 (SD 0.122) | 7.278 (SD 0.139) |
- Age: p < 0.001
- Lactate during support: p < 0.001
- Arterial pH: p < 0.001
- BMI: p < 0.001
- ECMO duration: p = 0.023
- ECMO duration: on-support variable reflecting clinical trajectory rather than a pre-initiation predictor
Sensitivity analysis excluding ECMO duration
- random forest AUC
- 0.697
Transplant-related outcomes
- Prediction limited by class imbalance
Original publication
Summarised from the published abstract; every number is checked against it