Latent class modeling of repetitive transcranial magnetic stimulation response trajectories in treatment-resistant depression

First published online June 24, 2026. In records from 308 people receiving rTMS for treatment-resistant depression, symptom patterns separated into groups with minimal, moderate or large improvement. A history of suicidal thoughts was more common in the high-severity group that improved least. These patterns may support closer monitoring, but they do not yet provide a validated tool for predicting an individual patient's outcome or prove that rTMS caused the observed changes.

Weissman CR, Cui B, Benster L, Stapper N, Miron JP, Daskalakis ZJ, Appelbaum LG, Kohn JN.
Journal of Affective Disorders · Publication date: 2026-06-24
Added to site: 2026-11-01
DOI: 10.1016/j.jad.2026.122167

First published online June 24, 2026 and assigned to the November 2026 issue. Added to Neurofeedback.com November 1, 2026 after comparison with current and reserve candidates. This is a historical addition, not a newly published November 1 study.

Modalities

Repetitive transcranial magnetic stimulation, Clinical outcome monitoring, Predictive modeling

Clinical focus

Treatment-resistant depression, Response trajectories, Early clinical monitoring

Population

Adults with treatment-resistant depression receiving routine acute rTMS at the UC San Diego Interventional Psychiatry Program

Sample

308 patients meeting the analysis criteria

Intervention / method

Once-daily acute rTMS prescribed in routine care, with depressive symptoms assessed at least weekly using the PHQ-9; treatment protocols and parameters varied across the cohort

Comparator

No randomized or untreated comparison group; the analysis compared data-derived symptom-trajectory classes within the treated cohort

Clinician summary

The raw PHQ-9 model identified three classes, while baseline-adjusted analysis produced two improvement classes. The high-baseline, large-improvement group had a 65.5% mean symptom reduction and 84% response rate. A LASSO multinomial model using nested cross-validation discriminated the three raw-score classes with an AUC of 0.82. Because class membership was strongly influenced by baseline severity and validation was internal, the model requires external replication before clinical use.

How to interpret the findings

This was a retrospective single-program cohort without randomization or an untreated comparator, so it cannot establish treatment efficacy or causal predictors. Protocols and parameters varied, confounding by indication remains possible, and class solutions depend on modeling choices and baseline severity. The authors state that performance may be optimistic despite cost-sensitive weighting and nested cross-validation; no external validation cohort was used. Safety and adverse events were not evaluated in this secondary electronic-record analysis. Associations with suicidal ideation should support careful assessment and monitoring, not automated treatment denial or substitution.

Equipment and disclosures

The cohort included heterogeneous routine-care rTMS protocols. This analysis does not establish superiority for a device, coil, targeting method or commercial system, and no equipment is attributed to BrainMaster.

Supported by NIMH K23 MH138746 and Wellcome Trust funding. Lawrence G. Appelbaum reported equity in Hybrid Harmony, consulting for Wave Neuroscience, and grants from Eiro and the Wu-Tsai Human Performance Alliance and Wellcome Foundation. Zafiris J. Daskalakis reported BrainsWay board membership and grants from the Kreutzkamp Family Foundation and Wellcome Foundation; Cory R. Weissman reported NIMH and Wellcome support. Other authors declared no known competing interests. No correction, retraction or expression-of-concern relation appeared in the checked PubMed, Crossref, PMC and bounded DOI/title records on November 1, 2026. AI-assisted source review, not independent human clinical review.

Related evidence

Compare findings from the same clinical area, including mixed results.

• Effectiveness of rTMS vs. Pharmacotherapy in Late-Life Treatment-Resistant Depression: An Observational Study of Individual Participant Data From Two Randomized Clinical Trials · Positive with limitations
• Right putaminal gray matter change tracks psychomotor recovery after accelerated iTBS in melancholic treatment-resistant depression · Mixed
• Clinical outcomes following five-day accelerated intermittent theta burst stimulation for treatment-resistant depression in routine clinical practice: A prospective naturalistic cohort study · Positive with limitations
• Excitation-Inhibition Balance and Fronto-Limbic Connectivity Distinguish TMS Treatment Outcomes in Refractory Depression · Positive with limitations
• Use of neurostimulation in patients with treatment-resistant depression: current options and future directions. · Positive with limitations
• Pilot randomized trial of intermittent theta-burst stimulation versus H-Coil transcranial magnetic stimulation for treatment-resistant depression. · Positive with limitations