Related Experiment Video
Updated: Jul 28, 2025
05:19
Author Spotlight: Therapeutic Benefit of Closed-Loop Deep Brain Stimulation in Depression Treatment
Published on: July 7, 2023
2.4K
Machine learning-based classification using electroencephalographic multi-paradigms between drug-naïve patients with
Kuk-In Jang1, Sungkean Kim2, Jeong-Ho Chae3
1Cognitive Science Research Group, Korea Brain Research Institute (KBRI), Daegu, Republic of Korea.
Journal of Affective Disorders
|June 4, 2023
Summary
Combining multiple electroencephalography (EEG) methods significantly improves the classification of major depressive disorder (MDD). This approach offers a more accurate diagnostic tool for drug-naïve MDD patients compared to single EEG paradigms.
Area of Science:
- Neuroscience
- Psychiatry
- Biomedical Engineering
Background:
- Electroencephalography (EEG) is an underutilized diagnostic aid in psychiatry.
- Major depressive disorder (MDD) is heterogeneous, complicating EEG's diagnostic utility.
- Accurate MDD classification requires integrating multiple EEG paradigms.
Purpose of the Study:
- To evaluate the classification performance of multiple EEG paradigms in drug-naïve MDD patients versus healthy controls (HCs).
- To determine if combining EEG methods enhances diagnostic accuracy for MDD.
Main Methods:
- Recruited 31 drug-naïve MDD patients and 31 HCs.
- Recorded resting-state EEG (REEG), loudness dependence of auditory evoked potentials (LDAEP), and P300.
- Employed Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) with feature selection.
Main Results:
- A layered classifier using 14 features (12 P300 amplitudes and 2 LDAEP) achieved 94.52% accuracy.
- SVM with 30 features (P300, LDAEP, REEG) reached 90.32% accuracy.
- Individual EEG paradigms showed lower accuracies: REEG (71.57%), P300 (87.12%), LDAEP (83.87%).
Conclusions:
- Integrating multiple EEG paradigms is superior to single methods for classifying drug-naïve MDD patients.
- This multimodal EEG approach shows promise for improving MDD diagnosis.
