Clinical Outcomes vs Brain-Signal Changes

The core idea

A claim such as "our device changed your brain waves" can be true and still tell you little about whether you will sleep better, think more clearly or feel less depressed. A brain signal is something a machine measures. A clinical outcome is how a person feels, functions or survives. The FDA states that a biomarker, which includes physiologic and radiographic measures, "is not an assessment of how an individual feels, functions, or survives." [1]

This page is an explainer and uses the plain evidence labels in how we label evidence. It does not recommend any test or device.

Two kinds of measures

The FDA also notes that when a biomarker is formally qualified, it is qualified for a stated context of use, and it is the biomarker, not the measurement method, that is qualified. [1]

What an EEG can and cannot tell you

An EEG is a test of the electrical activity of the brain. [2] MedlinePlus notes that an EEG cannot be used to measure intelligence, and that a normal EEG does not mean a seizure did not occur. [2] In medicine it is used for specific questions, so a patterned "brain wave" score from a consumer headband should not be read as a diagnosis or a measure of mental performance. For related topics, see neurofeedback research context.

Why a signal change may not mean much

• Many signal measures are unreliable. A meta-analysis of 90 task-fMRI experiments (1,008 people) found poor overall reliability, with a mean test-retest correlation of .397. [3]
• Small studies of brain-behavior links often fail to replicate. An analysis of about 50,000 people found brain-wide association effects smaller than previously thought, leading to underpowered studies, inflated effect sizes and replication failures at typical sample sizes. [4]
• Stimulation can move one measure and not others. A systematic review of tDCS in healthy adults found a reliable effect on only one of 30 neurophysiological measures, motor-evoked potential amplitude. [5] A separate review found no significant effect of single-session tDCS on cognitive outcomes in 59 analyses. [6] Together these reviews show that a brain-level effect and a cognitive effect are separate questions.

What a good clinical outcome looks like

Take the tDCS headset the FDA approved for depression in December 2025. The evidence is a randomized, sham-controlled trial scored on a depression rating scale (HDRS-17), where mean change was -9.4 with active stimulation and -7.1 with sham. [7] Even there, the FDA noted uncertainty because participants could often tell which arm they were in: 77.6% of the active group guessed correctly, against 40.7% of the sham group. [7] A rating that can be shaped by expectation is one reason blinding and sham comparisons matter. See tDCS and tACS research for the full details.

A short checklist for claims

1. What was actually measured: a brain signal, a test score or a daily-life result?
2. Was there a comparison group, and could participants tell which group they were in?
3. How many people took part, and was the result repeated by others?
4. Did the study measure harm as well as benefit?
5. Does the claim name the exact device and use, or does it borrow credibility from a different product?

For early-stage studies, see reading early-stage research. Consumer EEG and wearable comparisons live in the Biohacking area.