Related Experiment Video
Updated: Aug 6, 2026
Application of an Amplitude-integrated EEG Monitor (Cerebral Function Monitor) to Neonates
Published on: September 6, 2017
EEG criticality as a prognostic tool for functional outcomes in sedated pediatric intensive care patients
Derek Newman1,2, Mark Grinberg3, Kevin Jones3
1Integrated Program in Neuroscience, McGill University, Montreal, QC, Canada.
Insights
Electroencephalography (EEG) criticality features show promise for predicting recovery in sedated pediatric intensive care unit (PICU) patients. These EEG markers accurately forecast patient outcomes, offering a behavior-independent prognostic tool.
Area of Science:
- Neuroscience
- Critical Care Medicine
- Biomedical Engineering
Background:
- Predicting recovery in sedated pediatric intensive care unit (PICU) patients is challenging due to a lack of reliable, behavior-independent prognostic markers.
- Electroencephalography (EEG) signals reflect the brain's dynamic balance and information processing capacity, offering potential insights into patient status.
Purpose of the Study:
- To assess the association between criticality-related EEG features and functional outcomes in sedated PICU patients.
- To determine if EEG features can predict meaningful recovery in pediatric patients under sedation.
Main Methods:
- A multi-center retrospective cohort study of 32 sedated PICU patients (ages 5-18) was conducted.
- Criticality-related EEG features (entropy, fractal, complexity), spectral EEG features, and demographic data were analyzed.
- Machine learning models were trained to predict patient outcomes using these features, assessed via the Glasgow Outcome Scale-Extended (GOS-E).
Main Results:
- Criticality-related and spectral EEG features significantly differed between good and poor recovery groups (GOS-E=4 threshold).
- EEG features predicted patient recovery with a maximal accuracy of 87% and an AUC of 0.92, outperforming demographic data alone.
- Patients with good outcomes showed greater EEG complexity, entropy, and fractal patterns.
Conclusions:
- EEG spectral and criticality-related features hold significant prognostic potential for sedated PICU patients.
- These findings suggest EEG analysis can provide valuable, behavior-independent predictions of functional recovery in critically ill children.
Objective:
Predicting meaningful recovery in sedated patients remains a major challenge in the pediatric intensive care unit (PICU) due to the lack of reliable, behavior-independent prognostic markers for children. Criticality of electroencephalography (EEG) signals reflects the brain's dynamic balance between order and chaos and capacity for information processing. The objective of this study was to assess the association between criticality-related EEG features and the functional outcomes of sedated PICU patients.
Methods:
A multi-center, retrospective cohort observational study was conducted with 32 patients admitted to two PICUs in urban Canada between 2014 and 2024, 5-18 years of age, 14 females. Patients were admitted with mixed etiology: acute seizure (28%), acute brain injures (22%), and systemic illness (50%). All patients received a clinically indicated EEG while exposed to an inhibitory anesthetic (midazolam, propofol, dexmedetomidine with a GABAergic sedative). Nine EEG features were calculated from three categories of criticality-related measures (entropy, fractal and complexity); five spectral EEG features and three patient demographic features were extracted from the databases. Patient outcomes were assessed with the Glasgow Outcome Scale-Extended three months post-injury. All features were statistically compared between good and poor recovery groups, and several machine learning models were trained with different combinations of features to predict patient outcome.
Results:
A threshold between lower and upper severe disability (GOS-E = 4) optimized classification of recovery. Criticality-related and spectral EEG features differed significantly between good and poor recovery groups, and EEG features predicted patient recovery above and beyond patient demographics, with a maximal predictive accuracy of 87% and AUC of 0.92. Patients with good outcomes exhibited greater EEG complexity, entropy and fractal patterns compared to those with poor outcomes.
Conclusion:
This study demonstrates the prognostic potential of EEG spectral and criticality-related features in predicting outcomes for sedated PICU patients.
