A Bath-and-Ulster EEG headset lifts covert-consciousness detection in brain-injury patients from 39 to 69 per cent

By Inside BCI · July 1, 2026

Some patients who survive a severe brain injury end up trapped in a diagnostic middle ground. The bedside clinical exam (squeeze my hand, follow this light, blink twice for yes) misses awareness that is still there, and the standard behavioural scales categorise them as unresponsive when the internal experience is not. Neurologists call the phenomenon cognitive-motor dissociation, and a 2019 New England Journal of Medicine study by Jan Claassen and colleagues estimated it occurs in roughly fifteen per cent of acute brain-injury patients.

A team led by Damien Coyle at Ulster University’s Intelligent Systems Research Centre in Derry and the University of Bath’s Institute for the Augmented Human published a paper in Nature Communications Medicine on 17 April 2026 showing that a wearable electroencephalography (EEG) headset, worn across multiple sessions with real-time audio neurofeedback, materially improves detection of covert awareness in brain-injury patients. The University of Bath press-released the study on 30 June 2026, approximately ten weeks after publication. Lead author is Dr Naomi du Bois. Senior author is Professor Damien Coyle, Director of Bath’s Institute for the Augmented Human, whose authorship on this paper carries dual affiliations at Ulster and Bath, reflecting his institutional transition during the study period. The paper appears in Volume 6 as Article 344 with digital object identifier 10.1038/s43856-026-01574-x, is registered with ClinicalTrials.gov as NCT03827187 (registered on 30 January 2019), and reports results across a cohort of 42 brain-injury patients (14 in unresponsive wakefulness syndrome, 17 in minimally conscious state, 10 in locked-in syndrome, and 1 with complete locked-in syndrome, giving a total locked-in group of 11) plus two able-bodied benchmark participants.

The headline result is that detection of the minimally conscious state, the sub-category where covert cognitive activity is most likely to be missed by standard behavioural scales, rose from 39 per cent to 69 per cent in the cohort when the motor-imagery BCI diagnostic was added to the standard clinical assessment. The improvement came with a trade-off: locked-in syndrome detection sensitivity decreased modestly, from 78 per cent to 67 per cent. Overall balanced diagnostic accuracy across all patient categories rose from 55 per cent to 62 per cent when the two assessments were combined. Thirty-one of 42 patients (73.8 per cent) showed reliable intentional neural modulation on the motor-imagery task, and approximately ninety per cent of those progressed to the yes-and-no communication phase of the protocol.

What the Bath-and-Ulster system does that earlier work did not

The system uses a wearable EEG headset and a motor-imagery brain-computer interface paradigm. Participants imagined one of three two-class movement combinations, tailored per patient with input from family or clinicians: left arm versus right arm, right arm versus feet, or left arm versus feet. The system read the resulting EEG signals across six standard frequency bands from 0.5 to 40 Hz (delta, theta, mu, low beta, high beta, and low gamma). When it detected reliable intentional modulation, it delivered real-time audio neurofeedback back to the participant, and the participant progressed through a staged protocol that ends with a binary yes-and-no communication interface. The whole assessment is repeated across multiple sessions rather than a single lab visit.

The methodology extends a two-decade research substrate. Adrian Owen’s 2006 Science paper first showed that patients diagnosed as being in a vegetative state could produce distinguishable neural responses to a “tennis versus spatial navigation” mental imagery task when scanned in a functional magnetic resonance imaging (fMRI) machine. Cruse and colleagues extended the technique to EEG in a 2011 Lancet study. Naci and colleagues developed a naturalistic movie-watching fMRI paradigm in a 2014 study. Claassen and colleagues’ 2019 NEJM paper established the epidemiology of the problem. All of that prior work relied on single-session fMRI or laboratory-EEG snapshots. The Bath-and-Ulster contribution is to move the assessment from a single-session snapshot into a longitudinal, feedback-driven, wearable-hardware protocol that can plausibly move out of the imaging suite into a clinical ward, a rehabilitation facility, or a home care setting.

The paper does not claim to be the first EEG-based covert-consciousness detector. Its own framing is that it is a “structured multi-session BCI framework” that extends the existing single-session work. Independent readers should apply the same discipline.

The clinical and institutional context

The work was funded by UK Research and Innovation via an EPSRC Turing AI Fellowship (grant EP/V025724/1) awarded to Coyle. The fellowship covered the period during which Coyle led the team at Ulster’s Intelligent Systems Research Centre, before he subsequently took up the Directorship of Bath’s Institute for the Augmented Human. That timing is why his authorship on this specific paper carries both institutional affiliations. UK ethical approval was granted by the NHS Health Research Authority (Integrated Research Application System Project IDs 136640 and 247815, Research Ethics Committee reference 18/WA/0186, approved 7 October 2020). Republic of Ireland ethical approval was granted by the National Rehabilitation University Hospital Research Ethics Committee.

The Institute for the Augmented Human is a Bath research centre spanning computer science, engineering, health, and psychology, aimed at wearable and human-in-the-loop systems research. Coyle’s research programme has focused on BCI and motor imagery for over a decade at Ulster and now at Bath. The move of covert-consciousness assessment from research-grade laboratory equipment onto wearable EEG hardware is a substantive step toward clinical deployability, though the paper does not itself claim a specific regulatory clearance path.

Where the Bath-and-Ulster work sits in the broader BCI map

The BCI category is dominated in commercial press coverage by systems that restore motor function or speech in paralysed patients. Neuralink, Synchron, Paradromics, Precision Neuroscience, Neuracle and BrainGate all sit in that lane, all covered separately by Inside BCI. Meta’s non-invasive Brain2Qwerty v2, released 29 June 2026, sits in the parallel lane of non-invasive typing decoding in healthy volunteers.

The Bath-and-Ulster programme opens a different clinical use case: diagnostic assessment of covert awareness in patients who cannot participate in standard bedside exams. The use case is smaller in patient count than paralysis or ALS but higher-value per patient in terms of care-planning consequences, since a positive covert-awareness result materially changes the treatment plan, the family conversation, and the resource allocation for a patient who might otherwise be classified as unresponsive.

Commercially, the work is at academic-paper stage. There is no named spin-out company, no regulatory-clearance path publicly disclosed, and no industry partner named. The direct commercial descendants of this line of research to date have been academic clinical adoption at rehabilitation centres rather than device companies. Whether the multi-session wearable-EEG methodology attracts commercial licensing or spin-out interest is one of the signals to watch.

What to watch

The first signal is whether an independent laboratory replicates the 39 per cent to 69 per cent minimally-conscious-state sensitivity gain. The paper reports results from a 42-patient cohort across UK and Irish sites; a second-group replication in a different clinical population would convert a single-paper result into a field consensus.

The second signal is whether the methodology attracts a commercial spin-out or licensing partner. Wearable EEG systems used for clinical diagnostics have a regulatory precedent in seizure-monitoring and consumer sleep applications; a covert-consciousness diagnostic use case would need a distinct regulatory pathway.

The third signal is whether the trade-off pattern (major minimally-conscious-state sensitivity gain, modest locked-in syndrome sensitivity loss) generalises or turns out to be cohort-specific. Locked-in syndrome patients typically have preserved cognition and simply cannot produce motor output; the modest sensitivity loss in the LIS cohort suggests the multi-session neurofeedback protocol may be optimised for the harder MCS diagnostic problem at the expense of easier LIS confirmation.

The fourth signal is whether a broader clinical-guideline body (the American Academy of Neurology, the European Academy of Neurology, the Royal College of Physicians in the UK) updates its disorders-of-consciousness assessment guidelines to include a multi-session BCI diagnostic as a recommended adjunct. Guideline adoption is a leading indicator of clinical deployment scale.

Sources

research University-of-Bath Ulster-University Institute-for-the-Augmented-Human Naomi-du-Bois Damien-Coyle covert-consciousness disorders-of-consciousness EEG-BCI motor-imagery minimally-conscious-state Nature-Communications-Medicine NCT03827187

What the Bath-and-Ulster system does that earlier work did not

The clinical and institutional context

Where the Bath-and-Ulster work sits in the broader BCI map

What to watch

Weekly BCI Brief in your inbox