Acute cardiac care & resuscitationResearch article
Early stratification of risk for poor neurological outcome after cardiac arrest is improved with processed EEG data
- Machine learning models developed to predict hospital discharge and 6-month long-term outcome
- Six machine learning models
- Two additional operating points (high specificity for poor outcome and for good outcome)
- Feature importance analysis
Authors' conclusion
The combination of early processed EEG and clinical data were best able to stratify neurological risk early after cardiac arrest using machine learning algorithms, but external validation is needed.
Study context
- Population
- Comatose patients in the ICU after return of spontaneous circulation (ROSC) after cardiac arrest
- Participants
- 913 patients
- Comparison
- Combined clinical and EEG data vs clinical data alone and EEG (BIS-SR) data alone for predicting poor Cerebral Performance Category score (CPC 3-5)
- Outcome
- Poor Cerebral Performance Category score (CPC 3-5)
- Follow-up
- Hospital discharge and 6-month long-term outcome
Main finding
As reported in the abstract
| Group | AUC, poor long-term outcome (combined 6-h EEG and clinical data) | AUC, poor discharge outcome (combined 6-h EEG and clinical data) |
|---|---|---|
| Combined clinical and EEG data | 0.88 (0.87-0.90) | 0.86 (0.84-0.87) |
- Combined clinical and EEG data vs Clinical data alone or EEG (BIS-SR) data alone: P < 0.001 — significantly improved the AUC with combined data compared to either data set alone
Further findings
Baseline characteristics
- Median age
- 59 years
- Male
- 69%
- Initial shockable rhythm
- 44%
Poor outcome observed
- At discharge
- 70%
- At long-term assessment
- 72%
Original publication
Early stratification of risk for poor neurological outcome after cardiac arrest is improved with processed EEG data. Resuscitation. Published 14 August 2026. doi:10.1016/j.resuscitation.2026.111265
Summarised from the published abstract; every number is checked against it