Answers are grounded in this paper and cite their sources.
Machine learning identifies unique EEG features in insomnia, suggesting a need for personalized approaches to treatment.
Abstract Insomnia-specific sleep EEG features have remained elusive. Using machine learning, we analyzed two independent electroencephalogram (EEG) datasets spanning two nights (Nsubjects/nights=198/396), including individuals with insomnia disorder (ID) (ranging from mild to moderate/severe) and good sleeper controls (GSCs). Sleep EEG spectral features can differentiate ID from GSC only when using the same participants for both training and testing data. This shows that performance depends on recognizing individual EEG profiles instead of generalizable ID patterns. We also demonstrate that epoch-averaged EEG spectral features exhibit highly individual-specific signatures, identified using unsupervised learning, similarity matrix analyses and periodicity assessments. Our results further indicate that signatures are primarily driven by high frequency cortical activity, possibly reflecting cortical arousal during sleep. While ID may be characterized by EEG features beyond spectral power, our findings underscore the importance of a precision brain health framework that focuses on deviations from an individual’s own neural baseline rather than relying solely on group-level comparisons.
No takes yet. Share an insight, caveat, or question.
Kyllönen et al. (2025) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: