Quantitative EEG in Depression: Predictors of Treatment Response
The central problem of major depressive disorder is not that we have no treatments — it is that we have too many, and no principled way to pick between them. Any given SSRI helps roughly half of the patients who try it; rTMS is similar. So the standard protocol is trial-and-error: try one, wait six weeks, try another. Meanwhile the patient stays depressed.
For forty years, quantitative EEG has been circling this problem. Resting scalp EEG is inexpensive, non-invasive, and captures millisecond-scale neural dynamics. It has also, for most of its history, had the reputation of a tool that overpromises. In one specific domain — predicting antidepressant and rTMS response — that reputation has quietly stopped being true.
From early local to network predictors
The story of qEEG response prediction is a story of successive generations of biomarker, each broader in scope than the last.
The early predictors were local, spectral, and single-electrode. Prefrontal theta cordance — a composite that z-scores absolute and relative theta power per electrode and sums them — predicted antidepressant outcomes, replicated in double-blind trials across two very different treatments. A 2024 meta-analysis of twenty studies confirmed it.
Alongside cordance sat individual alpha peak frequency — the frequency of the dominant posterior alpha rhythm at rest — which Martijn Arns showed to be a highly heritable trait present in roughly 17% of depressed patients that predicts non-response to antidepressants, stimulants, and rTMS alike. Frontal alpha asymmetry — reduced left-vs-right frontal alpha power at rest, one of the most well-known spectral signatures of the depressed brain — rounded out the early toolkit. Between them, these markers established that qEEG could identify likely responders and non-responders before treatment even began[3,4,5,6,1].
The newer network predictors step out of the surface electrode framing aligned with the broader neuroscience of depression as a default mode network dysconnectivity – the circuits underlying the sense of self.
Network qEEG predictors are rapidly evolving. Bailey and colleagues showed that fronto-midline theta connectivity predicted rTMS response at high sensitivity and specificity. Peng et al. moved from pairwise coupling to whole-brain network topology, classifying baseline with accuracy from graph metrics computed on the resting EEG. Choi et al. brought the analysis into source space, using degree centrality and clustering coefficient of a 31-node functional network to distinguish treatment-refractory from responding MDD. Shao et al. added the time dimension, using the dynamic modular flexibility of default-mode-network nodes to predict post-rTMS Hamilton scores[2,7,8,9,10]. Each successive generation reads a response prediction closer to the network structure that defines the disease.
The neurotransmitter frontier
It is now possible using source-localized resting EEG to read out which neurotransmitter system contributes to the qEEG dysregulation. Receptor-specific pharmaco-EEG has isolated time-frequency "fingerprints" for serotonin, dopamine, and α2-adrenergic transmission, and MEG has shown that resting networks in delta and gamma bands covary directly with cortical GABA, NMDA, dopamine, and serotonin receptor/transporter densities. Still investigational, but shows that a source-localized qEEG could steer prescribing toward the specific pharmacology the patient's brain actually needs rather than sampling the pharmacopoeia by trial and error.
Where we are, and where we're going
We are not yet at the point where a psychiatrist orders a qEEG before writing the first prescription. But the pieces are moving into place. In 2013 the FDA carved out a new device category — Neuropsychiatric Interpretive EEG-based Assessment Aids (NIEA) — to clear the NEBA System as a diagnostic aid for ADHD in children, demonstrating that the agency will build fresh regulatory pathways for qEEG-based clinical decision support when the evidence warrants it. For a decade, however, the complexity of the computations were prohibitive to advance the biomarker space. That is also changing. A growing ecosystem of modern qEEG platforms with rapid-turnaround analysis pipelines now support the kind of prospective, biomarker-stratified clinical trials the field has needed for decades. Each successive biomarker generation serves the broader drug-development ecosystem by giving trialists faster, more mechanistically grounded readouts of who is responding and why[11].
The regulatory path exists. The measurement tools are getting sharper and easier to generate. For four decades, the argument about qEEG was whether the signal even existed. The next decade is about turning a research tool with a real signal into a clinical instrument, and the momentum is finally moving in the right direction.
References
1. Heitmann H, Siani J-F, Zebhauser PT, Henningsen P, Leucht S, Priller J, Ploner M. Resting-state EEG activity as a Biomarker and Treatment Target in Depression: A Systematic Review and Meta-analysis. 2025. https://www.medrxiv.org/content/medrxiv/early/2025/10/24/2025.10.22.25338525.full.pdf
2. Kaiser RH, Andrews-Hanna JR, Wager TD, Pizzagalli DA. Large-scale network dysfunction in Major Depressive Disorder: Meta-analysis of resting-state functional connectivity. 2015. https://europepmc.org/articles/pmc4456260?pdf=render
3. Hunter A, Cook I, Leuchter A. The promise of the quantitative electroencephalogram as a predictor of antidepressant treatment outcomes in major depressive disorder. 2007. https://www.semanticscholar.org/paper/e80ea206ab34ea0f024ec8635feb6eb457906f40
4. Bareš M, Brunovský M, Novák T, Kopeček M, Stopková P, Šóš P, Höschl C. QEEG Theta Cordance in the Prediction of Treatment Outcome to Prefrontal Repetitive Transcranial Magnetic Stimulation or Venlafaxine ER in Patients With Major Depressive Disorder. 2015. https://www.semanticscholar.org/paper/27a5b5a6aecc950634e9659b2c3ab5ec10e9284a
5. Srivastava A, Sanyal S, Jaiswal S, Srivastava S. Meta-analysis on QEEG Changes to Antidepressant Treatment Among Patients with Depression. 2024. https://pmc.ncbi.nlm.nih.gov/articles/PMC11572393/pdf/10.1177_02537176241271716.pdf
6. Arns M. Personalized medicine in ADHD and depression: A quest for EEG treatment predictors. 2008. https://www.semanticscholar.org/paper/81c21096bb7f2c0d0936c7cc03ff124bb4461f2d
7. Choi K-M, Hwang H-H, Yang C, Jung B, Im C-H, Lee S-H. Association between the functional brain network and antidepressant responsiveness in patients with major depressive disorders: a resting-state EEG study. 2025. https://www.cambridge.org/core/services/aop-cambridge-core/content/view/8F6C20B0DE1A92EA507E255DD72CC07D/S0033291724003477a.pdf
8. Bailey NW, Hoy KE, Rogasch NC, Thomson RH, McQueen S, Elliot D, Sullivan C, Fulcher BD, Daskalakis ZJ, Fitzgerald PB. Responders to rTMS for depression show increased fronto-midline theta and theta connectivity compared to non-responders. 2018. https://www.semanticscholar.org/paper/3a2b4daba49fe2238e1363e4a762434596f3dd42
9. Peng Y, Huang Y, Chen B, He M, Jiang L, Li Y, Huang X, Pei C, Zhang S, Li C, Zhang X, Zhang T, Zheng Y, Yao D, Li F, Xu P. Electroencephalographic Network Topologies Predict Antidepressant Responses in Patients With Major Depressive Disorder. 2022. https://ieeexplore.ieee.org/ielx7/7333/4359219/09872032.pdf
10. Shao Y, Zhou Z, Mao J, Xu P, Luo Y, Yu F, Wang S. Predicting rTMS treatment efficacy in depression based on modular flexibility of functional connectivity. 2026. https://www.semanticscholar.org/paper/8b9bade30a9e75e6164f358d3832caad2f6c1a98
11. Medica Central. Quantitative Electroencephalogram (qEEG) and Referenced Electroencephalogram (rEEG). Medical Policy MP9622, effective September 2024. https://mo-central.medica.com/Document-Library/pdf/Medical-Policies/qeeg-9622