- 25 common EEG features categorised into time, frequency, complexity and connectivity domains as candidate biomarkers for affective disorders.
- Diagnostic utility validated on two resting state EEG datasets covering SZ, SAD, BD and MDD versus healthy controls, enabling feature based classification.
- Comprehensive discussion highlights challenges: multimodal integration, variability in classification, and factors influencing EEG feature extraction and reproducibility.
Hum Brain Mapp. 2026 Sep;47(13):e70628. doi: 10.1002/hbm.70628.
ABSTRACT
Electroencephalography (EEG) provides real-time, dynamic insights into brain function, making it a valuable tool for the screening, diagnosis, and treatment of affective disorders. The use of EEG-derived features as biomarkers for affective disorder diagnosis has attracted growing attention. In this work, we introduce 25 commonly used EEG features and categorize them into four domains: time-domain, frequency-domain, complexity, and connectivity. Each feature captures distinct aspects of emotional processing and serves as a potential biomarker for diagnosing affective disorders. To evaluate their diagnostic utility, we analyzed two independent resting-state EEG datasets. The first dataset comprised 84 healthy controls, 62 patients with schizophrenia (SZ), 43 patients with schizoaffective disorder (SAD), and 32 patients with bipolar disorder (BD), while the second included 28 healthy controls and 32 patients with major depressive disorder (MDD). These features are extracted and used for classification, allowing us to identify biomarkers with strong discriminative power. Finally, we provide a comprehensive discussion on key issues in the field, including multimodal biomarker integration, challenges in EEG-based diagnosis, variability in classification results, and factors influencing EEG feature extraction.
PMID:42702788 | DOI:10.1002/hbm.70628
Share Evidence Blueprint
Save to Google Notes
Search Google Scholar
Save as PDF
⭐ My Revision List
