A new machine learning algorithm to predict veno-arterial ECMO implantation after post-cardiotomy low cardiac output syndrome
Authors' conclusion
We developed an explainable ensemble machine learning algorithm that could help clinicians predict the risk of deterioration and the need for VA-ECMO implantation in moderate to severe PC-LCOS patients.
Study context
- Population
- Patients with moderate to severe post-cardiotomy low cardiac output syndrome (vasoactive inotropic score (VIS) > 10 with impaired organ perfusion or need for mechanical circulatory support after cardiac surgery), two university hospitals in Paris, France
- Participants
- 285 vs 190
- Comparison
- Development dataset vs External validation dataset
- Development datasetn = 285
- External validation datasetn = 190
- Outcome
- Need for VA-ECMO implantation
Methodological detailsmachine learning (Deep Super Learner ensemble), two university hospitals, January 2016 to December 2019
- Machine learning (Deep Super Learner ensemble)
- Two university hospitals
- January 2016 to December 2019
Performance
As reported in the abstract
Prediction of VA-ECMO implantation (external validation dataset)
| Test or model | ROC AUC |
|---|---|
| External validation dataset | 0.863 (0.793-0.928) |
External validation dataset
ROC AUC0.863 (0.793-0.928)
Further findings
VA-ECMO implantation incidence
Most important features (Shapley values)
- First postoperative arterial lactate, intraoperative VIS, absence of angiotensin-converting enzyme treatment, BMI, EuroSCORE II
Original publication
A new machine learning algorithm to predict veno-arterial ECMO implantation after post-cardiotomy low cardiac output syndrome. Anaesth Crit Care Pain Med. Published 11 November 2022. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →