Acute cardiac care & resuscitation
Pediatric cardiac arrest outcome prediction using data-driven machine learning of early quantitative electroencephalogram (quantitative EEG (qEEG)) features
Authors' conclusion
Data-driven models using 164 qEEG features accurately predicted neurologic outcomes after pediatric cardiac arrest, with stable performance over time.
Study context
- Population
- Children aged 3 months to 18 years with cardiac arrest and EEG monitoring within 24 h post-arrest; pre-arrest Pediatric Cerebral Performance Category (PCPC) >3 excluded53 % unfavorable outcome
- Participants
- 70 children
- Exposure
- 164 qEEG features vs 7-feature model
- Outcome
- Prediction of unfavorable neurologic outcome (death or PCPC ≥4), AUROC
- Follow-up
- Hospital discharge or 30 days post-arrest
Methodological detailssingle-centre
- Single-centre
Main finding
As reported in the abstract
AUROC for unfavorable neurologic outcome
| Group | AUC (LASSO) | AUC (Random Forest) |
|---|---|---|
| 164 qEEG features | 0.81 (95 % CI: 0.69-0.91) | 0.80 (0.67-0.90) |
| 7-feature model | 0.45 (0.31-0.58) | 0.65 (0.50-0.78) |
AUC (LASSO)
164 qEEG features0.81 (95 % CI: 0.69-0.91)
7-feature model0.45 (0.31-0.58)
AUC (Random Forest)
164 qEEG features0.80 (0.67-0.90)
7-feature model0.65 (0.50-0.78)
Observational data: associations, not effects.
Further findings
Added value of clinical variables
- Adding clinical variables did not improve performance
Performance stability over time · 6-h epochs from 6 to 24 h
- Area under the ROC curve (AUROCs) stable across 6-h epochs from 6 to 24 h
qEEG features associated with outcome
- Higher phase locking value, fractal exponent and coherence associated with better outcomes
- Higher delta power and suppression ratio variability associated with worse outcomes
Original publication
Pediatric cardiac arrest outcome prediction using data-driven machine learning of early quantitative electroencephalogram (qEEG) features. Resuscitation. Published 8 October 2025. doi:10.
Summarised from the published abstract; every number is checked against it