Acute cardiac care & resuscitation
Electroencephalogram-based machine learning models to predict neurologic outcome after cardiac arrest: A systematic review
Authors' conclusion
RF and CNN were the two most common ML models with the highest AUCs for predicting the neurologic outcomes after cardiac arrest.
Study context
- Population
- Studies using EEG-based machine learning (ML) models to predict neurologic outcome after cardiac arrest
- Participants
- 17 studies
- Comparison
- EEG-based ML models: conventional ML vs Deep Neural Network (DNN)
- Random Forest (RF)n = 7
- Convolutional Neural Network (CNN)n = 4
- Outcome
- Most common EEG-based ML model with the highest AUC (conventional ML and DNN) for predicting neurologic outcome
Methodological detailsrisk of bias by Quality in Prognosis Studies guideline, Search of PubMed and Compendex up to June 2, 2023
- Risk of bias by Quality in Prognosis Studies guideline
- Search of PubMed and Compendex up to June 2, 2023
Pooled result
As reported in the abstract
AUC for neurologic outcome prediction
| Group | AUC range |
|---|---|
| RF | 0.8–0.97 |
| CNN | 0.7–0.92 |
AUC range
RF0.8–0.97
CNN0.7–0.92
- RF most common model in conventional ML category (n = 11)
- 7
- CNN most common model in DNN category (n = 6)
- 4
Further pooled results
Most common EEG features used in ML models
- Band power
- 12
- Shannon's Entropy
- 11
- Burst-suppression ratio
- 9
Original publication
Electroencephalogram-based machine learning models to predict neurologic outcome after cardiac arrest: A systematic review. Resuscitation. Published 14 November 2023. doi:10.
Summarised from the published abstract; every number is checked against it