Related Experiment Video
Updated: Sep 1, 2025
09:55
Neuroimaging-Guided TMS–EEG for Real-Time Cortical Network Mapping
Published on: June 13, 2025
1.1K
Predicting treatment response using EEG in major depressive disorder: A machine-learning meta-analysis
Devon Watts1, Rafaela Fernandes Pulice2,3, Jim Reilly4
1Neuroscience Graduate Program, McMaster University, Hamilton, Canada.
Translational Psychiatry
|August 12, 2022
Summary
Electroencephalography (EEG) models show promise in predicting treatment response for major depressive disorder (MDD), achieving 83.93% accuracy. These models are more effective at identifying non-responders and show higher accuracy for transcranial magnetic stimulation (rTMS) than antidepressants.
Area of Science:
- Neuroscience
- Psychiatry
- Machine Learning
Background:
- Psychiatric treatment selection is often trial-and-error.
- Machine learning (ML) models are increasingly explored to predict treatment response.
- Electroencephalography (EEG) offers a scalable method for predicting treatment outcomes in major depressive disorder (MDD).
Purpose of the Study:
- To meta-analyze the efficacy of ML models using EEG to predict treatment response in MDD.
- To conduct subgroup analyses for response to transcranial magnetic stimulation (rTMS) and antidepressants.
- To identify key EEG features and assess model performance metrics.
Main Methods:
- Searched PubMed, Scopus, and Web of Science for studies from 1960-2022.
- Included 15 studies using ML for EEG-based treatment response prediction in MDD (758 patients).
- Performed random-effects meta-analyses and subgroup analyses for rTMS and antidepressant response.
Main Results:
- Pooled accuracy across studies was 83.93% (AUC: 0.850).
- Subgroup analysis showed higher accuracy for rTMS (85.70%) compared to antidepressants (81.41%).
- EEG models demonstrated higher specificity (identifying non-responders) than sensitivity (identifying responders).
Conclusions:
- EEG-based ML models show significant promise for predicting treatment response in MDD.
- Further research requires prospective validation, standardization of response criteria, and large consortium studies.
- EEG models may perform better for rTMS due to targeting of cortical neural networks.
