Cohort study
Prediction of Shock-Refractory Ventricular Fibrillation During Resuscitation of Out-of-Hospital Cardiac Arrest
- Cohort study (prediction model development)
- machine learning (random forest)
- 80%/20% training/test split
Authors' conclusion
A machine learning algorithm using ECGs surrounding the initial shock predicts patients likely to experience refractory VF
Study context
- Population
- Ventricular fibrillation (VF) OHCA patients with available defibrillator recordings275 test patients
- Participants
- 1,376 patients
- Outcome
- Prediction of refractory VF (≥3 shocks) by ECG-based algorithm, AUC
Main finding
As reported in the abstract
Prediction of refractory VF (test set)
AUC
0.85 (95% CI, 0.79-0.89)
Specificity
91%
Sensitivity
63%
Positive likelihood ratio
6.7
Observational data: associations, not effects.
Further findings
| Outcome | RR (95% CI) | P value | |
|---|---|---|---|
| Total shock count and functional neurological survival | 0.95 (0.93–0.97) | P < 0.001 | |
| 0.512Relative risk (log scale) |
Refractory VF incidence
- Refractory VF
- 864 (63%)
Functional neurological survival
- Functional neurological survival
- 591 (43%)
Original publication
Prediction of Shock-Refractory Ventricular Fibrillation During Resuscitation of Out-of-Hospital Cardiac Arrest. Circulation. Published 2 June 2023. doi:10.1161/circulationaha.122.063651
Summarised from the published abstract; every number is checked against it