Acute cardiac care & resuscitationCohort study
Prediction of neurological outcome after pediatric cardiac arrest using heart rate variability and machine learning
- Retrospective study
- Extreme Gradient Boosting (XGB) models
- cross-validation
Authors' conclusion
Using machine learning, HRV features within 24 h after pediatric cardiac arrest predict unfavorable outcome with AUROC 0.8. Adding clinical variables did not improve model performance.
Study context
- Population
- Children who received post-arrest care in the PICU at the Children's Hospital of Philadelphia from 2020 to 2023patients who met inclusion criteria (median age 6.8 [IQR 10.4] years)
- Participants
- 75 children
- Exposure
- Adding cardiac arrest characteristics did not improve model performance
Main finding
As reported in the abstract
Prediction of unfavorable neurological outcome (Pediatric Cerebral Performance Category 4-6 and an increase >1 from baseline), AUROC · first 24 h after return of spontaneous circulation
| Group | AUROC |
|---|---|
| Heart rate variability (HRV) features and age | 0.80 (95% CI: 0.68-0.88) |
Observational data: associations, not effects.
Further findings
| Outcome | HRV features and age | Adding cardiac arrest characteristics | P value |
|---|---|---|---|
| Model performance with cardiac arrest characteristics added · did not improve model performance | P > 0.41 | ||
| AUROC | Not reported in the abstract | 0.83 [0.73-0.92] |
Top HRV predictors
- Top HRV predictors included standard deviation (SDNN), power at very low and low frequency bands, entropy, and fractal scaling
Performance across sequential 6-h epochs
- Performance was similar across the 6-h epochs
- p's > 0.1
Unfavorable outcome
- unfavorable outcome
- 51%
Original publication
Prediction of neurological outcome after pediatric cardiac arrest using heart rate variability and machine learning. Resuscitation. Published 5 January 2026. doi:10.1016/j.resuscitation.2026.110961
Summarised from the published abstract; every number is checked against it