Interrater Reliability of Electroencephalogram Interpretations in Febrile Comatose African Children: Implications for...

Area of Science:

• Clinical Neurophysiology
• Medical Informatics
• Global Health

Background:

• Electroencephalography (EEG) is crucial for comatose patient care but limited in resource-poor areas due to interpreter shortages.
• Machine learning can automate EEG interpretation, but requires extensive expert input for algorithm development.
• High interrater reliability (agreement between experts) can reduce the number of experts needed, accelerating AI development.

Purpose of the Study:

• To assess the interrater reliability of key EEG variables in febrile comatose African children.
• To determine if high agreement supports developing automated EEG interpretation for resource-limited settings.

Main Methods:

• Two electrophysiologists independently interpreted 171 EEGs from children with febrile coma.
• Standard interpretation methods were used, focusing on variables relevant to prognostication (e.g., cerebral malaria).

Main Results:

• High interrater reliability was observed for most EEG variables.
• Presence of electrographic seizures showed 96.5% agreement (kappa=0.781).
• Variables used for cerebral malaria prognostication also had high agreement rates.

Conclusions:

• EEG interpretation in Malawian children with febrile coma demonstrates high interrater reliability.
• This supports the feasibility of developing automated EEG interpretation algorithms with fewer expert trainers.
• Automated EEG tools can enhance clinical care and outcomes for febrile comatose children in low-resource settings.