Cardiac arrest & resuscitation
EEG-based machine-learning prognostication in comatose patients with indeterminate outcome after cardiac arrest
Authors' conclusion
In comatose patients after cardiac arrest with ERC/ESICM guideline-defined indeterminate prognosis, an interpretable EEG-based prognostic model provides clinically relevant information.
Study context
- Population
- Adult comatose patients after cardiac arrest in the Intensive Care Unit (ICU) of the University Hospitals of Geneva with fewer than two ERC/ESICM poor outcome factors (indeterminate prognosis), 2021–2025of 313 screened
- Participants
- 77 patients
- Outcome
- Neurological outcome, good (Cerebral Performance Category 1-2) vs poor (3-5)
- Follow-up
- 3-6 months
Methodological detailssingle-centre, machine learning (random forest classifier), repeated stratified cross-validation, Shapley value analysis
- Single-centre
- Machine learning (random forest classifier)
- Repeated stratified cross-validation
- Shapley value analysis
Main finding
As reported in the abstract
Discriminative performance for neurological outcome at 3-6 months
Mean Receiver Operating Characteristic - Area Under the Curve (ROC-AUC)
0.80
PR-AUC
0.83
Observational data: associations, not effects.
Further findings
Good neurological outcome (CPC 1-2)
- Patients with good neurological outcome
- 35 (45%)
Feature contributions (Shapley value analysis)
- Delta-band EEG features accounted for most of the predictive contribution
- Higher delta functional connectivity, higher delta relative power and higher probability of periodic discharges associated with poor outcome
- Preserved relative alpha power and higher probability of rhythmic delta activity associated with good outcome
Original publication
EEG-based machine-learning prognostication in comatose patients with indeterminate outcome after cardiac arrest. Crit Care. Published 27 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →