How to read TMS research without a neuroscience degree

TMS generates a lot of headlines, and the gap between "study finds" and "this changes your treatment" is usually wide. Four questions that separate the two: **1. Who was in the study?** A protocol tested in people with a first depressive episode tells you little about treatment-resistant depression, and vice versa. Sample size matters too — striking results from a dozen participants are a reason for a larger trial, not a reason to change plans. **2. Compared to what?** TMS trials have an unusually strong sham-control problem: the coil makes a distinctive sound and sensation, so participants often guess correctly which arm they are in. Open-label results routinely look better than sham-controlled ones. Ask which this was. **3. Response or remission?** "Response" usually means symptoms dropped by at least half. "Remission" means you would no longer meet criteria for depression. A 70% response rate and a 70% remission rate are very different claims, and press releases blur them. **4. Real-world or trial conditions?** Registry and real-world data covers messier, more representative patients. Trial data is cleaner but narrower. Both are useful; they answer different questions. One recent example worth the read: University of Iowa researchers showed in *Nature Communications* that TMS can modulate hippocampal activity — but only when the stimulation site was personalised to each individual's connectivity. The finding is less "TMS reaches deeper now" than "how the target is chosen changes what the stimulation does." We wrote it up [in our news section](/news/), and the same theme runs through [AI-guided targeting](/blog/ai-guided-tms-targeting-2026/) and [the Stanford SNT protocol](/blog/stanford-neuromodulation-therapy-snt-2026/). If you come across a study you are trying to make sense of, post the link here.