Extracorporeal membrane oxygenation (ECMO) PAL VV: using deep neural networks for survival prognostication in venovenous extracorporeal membrane oxygenation
Authors' conclusion
ECMO PAL VV demonstrated strong accuracy on contemporary international registry data (73%) with strong sensitivity (81%) and precision (77%) to predict survival to hospital discharge, outperforming existing published scores.
Study context
- Population
- Adults receiving VV ECMO, ELSO Registry, 660 hospitals
- Participants
- 39,501 adults · 660 hospitals
- Comparison
- ECMO PAL VV vs published VV ECMO outcome scores
- Outcome
- Prediction of survival to hospital discharge
Methodological detailsmulticentre, international registry, deep neural networks
- Multicentre
- International registry
- Deep neural networks
Performance
As reported in the abstract
Prediction of survival to hospital discharge, temporal validation (2024 data)
| Test or model | Value | Accuracy |
|---|---|---|
| RESP | 65% | 65% |
| Lazzeri score | 60% | 60% |
Value
RESP65%
Lazzeri score60%
Accuracy
RESP65%
Lazzeri score60%
- Temporal validation accuracy
- 73%
- Temporal validation AUC
- 0.78
Further findings
Prediction of survival to hospital discharge, internal training (2017–2023 data)
- Internal training accuracy
- 79%
- Internal training AUC
- 0.87
- Sensitivity to mortality prediction, internal training
- 71%
Original publication
ECMO PAL VV: using deep neural networks for survival prognostication in venovenous extracorporeal membrane oxygenation. Crit Care. Published 18 April 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →