Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness

Naomi Du Bois; Attila Korik; Stephanie Hodge; Leah Hudson; Ainjila Shakeb Elahi; Alain Desire Bigirimana; Natalie Dayan; Jose Miguel Sanchez-Bornot; Alison McCann; Kudret Yelden; Lloyd Bradley; Krishnan Padmakumari Sivaraman Nair; Simon Judge; Damon Hoad; Emma Vines; Venu Harilal; Sheryl Parke; Paul Johnson; Jacqueline Pogue; Emma Dodds; Abayomi Salawu; Raymond Carson; Karl McCreadie; Jacqueline Stow; Jacinta McElligott; Aine Carroll; Damien Coyle

doi:10.1038/s43856-026-01574-x

Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness

Naomi Du Bois, Attila Korik, Stephanie Hodge, Leah Hudson, Ainjila Shakeb Elahi, Alain Desire Bigirimana, Natalie Dayan, Jose Miguel Sanchez-Bornot, Alison McCann, Kudret Yelden, Lloyd Bradley, Krishnan Padmakumari Sivaraman Nair, Simon Judge, Damon Hoad, Emma Vines, Venu Harilal, Sheryl Parke, Paul Johnson, Jacqueline Pogue, Emma DoddsAbayomi Salawu, Raymond Carson, Karl McCreadie, Jacqueline Stow, Jacinta McElligott, Aine Carroll, Damien Coyle

Research output: Contribution to journal › Article › peer-review

Abstract

Background: Accurate assessment of residual awareness in patients with Prolonged Disorders of Consciousness (PDoC) remains a major clinical challenge, as conventional behavioural tools can underestimate covert cognition. This study evaluates whether a structured, multi-phase motor imagery Brain–Computer Interface (MI-BCI) protocol provides objective electroencephalography (EEG)-based indicators of awareness that complement behavioural assessments.

Methods: Forty-four participants (N = 44) completed repeated imagined-movement tasks using wearable EEG (PDoC: Unresponsive Wakefulness Syndrome (UWS, n = 14), Minimally Conscious State (MCS, n = 17), Locked-In Syndrome (LIS, n = 11); two able-bodied participants as benchmarks; ClinicalTrials.gov: NCT03827187; 30-01-2019). The protocol assessed sensorimotor rhythm modulation, training with and without neurofeedback, and binary question answering across phases. Standard behavioural assessments (CRS-R and WHIM) were administered at each session.

Results: Significant MI-BCI decoding accuracy (DA) is achieved by 73.8% of patients, of whom 90% progress to Q&A testing and frequently exceed the 70% usability threshold, revealing marked inter-individual heterogeneity. For significant MI-BCI runs, LIS outperform MCS (p = 0.007) and UWS (p = 0.048), while UWS exceed MCS during Q&A (p = 0.049), driven by familiar-voice stimuli. Using leave-one-subject-out cross-validation, combining predictions from DA and behavioural assessments improves balanced diagnostic accuracy to 62% (from 55%), increasing sensitivity to MCS (39% to 69%), with a modest reduction in LIS sensitivity (78% to 67%). Task-related activity over sensorimotor and parietal cortices differentiate diagnostic groups.

Conclusions: The structured MI-BCI protocol demonstrates potential as a movement-independent, EEG-based tool for distinguishing UWS, MCS and LIS. Integrating DA and spatial patterns yields diagnostic information that may augment behavioural assessment and advance objective tools for evaluating awareness in PDoC.

Original language	English
Journal	Communications Medicine
Early online date	17 Apr 2026
DOIs	https://doi.org/10.1038/s43856-026-01574-x
Publication status	E-pub ahead of print - 17 Apr 2026

Data Availability Statement

The data that support the findings of this study are available here: https://doi.org/10.15125/BATH-01632116. This anonymised dataset includes electroencephalography (EEG) recordings, descriptive variables, and session-level Coma Recovery Scale-Revised (CRS-R) and Wessex Head Injury Matrix (WHIM) scores. All data contained within the repository are provided under controlled access. Access will be granted upon reasonable request, which should include a brief description of the proposed research use, confirmation of relevant ethical approval where applicable, and agreement to data use conditions that prohibit data redistribution or use beyond the approved scope. Requests should be directed to the corresponding author (Prof D. Coyle, [email protected]) and will be reviewed on a case-by-case basis. A response will typically be provided within two weeks of receipt of a complete request. The source data graphed in Figures 4-9 and Supplementary Figure 7 are provided in Supplementary Data 1 (worksheets Supplementary Data A-E, H and I, respectively).

Acknowledgements

We would like to thank the clinical teams at the participating UK NHS hospitals and the National Rehabilitation Hospital (NRH) of Ireland for their dedication to patient recruitment. We are also grateful to the care teams and staff at the hospitals and care homes where participants resided, whose support was essential to the success of the trials. To the participants and their families, we extend our deepest thanks for their generous participation, trust, and commitment to this research.

Funding

This work was supported by internal funds from the University of Bath and Ulster University; by access to the Tier 2 High Performance Computing resources provided by the Northern Ireland High Performance Computing (NI-HPC) facility, funded by the UK Engineering and Physical Sciences Research Council (EPSRC) under Grant number EP/T022175; and the UK Research and Innovation (UKRI) Turing AI Fellowship 2021-2025 funded by the EPSRC under Grant number EP/V025724/1. Author K. P. S. N. was supported by the NIHR Sheffield Biomedical Research Centre/NIHR Sheffield Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, the NRH or the Department of Health and Social Care.

Access to Document

10.1038/s43856-026-01574-xLicence: CC BY

Cite this

Du Bois, N., Korik, A., Hodge, S., Hudson, L., Elahi, A. S., Bigirimana, A. D., Dayan, N., Sanchez-Bornot, J. M., McCann, A., Yelden, K., Bradley, L., Nair, K. P. S., Judge, S., Hoad, D., Vines, E., Harilal, V., Parke, S., Johnson, P., Pogue, J., ... Coyle, D. (2026). Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness. Communications Medicine. Advance online publication. https://doi.org/10.1038/s43856-026-01574-x

Du Bois, N , Korik, A, Hodge, S, Hudson, L, Elahi, AS, Bigirimana, AD, Dayan, N, Sanchez-Bornot, JM, McCann, A, Yelden, K, Bradley, L, Nair, KPS, Judge, S, Hoad, D, Vines, E, Harilal, V, Parke, S, Johnson, P, Pogue, J, Dodds, E, Salawu, A, Carson, R, McCreadie, K, Stow, J, McElligott, J, Carroll, A & Coyle, D 2026, 'Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness', Communications Medicine. https://doi.org/10.1038/s43856-026-01574-x

@article{1a4de97263dc4f61aca1e08ee4300e87,

title = "Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness",

abstract = "Background: Accurate assessment of residual awareness in patients with Prolonged Disorders of Consciousness (PDoC) remains a major clinical challenge, as conventional behavioural tools can underestimate covert cognition. This study evaluates whether a structured, multi-phase motor imagery Brain–Computer Interface (MI-BCI) protocol provides objective electroencephalography (EEG)-based indicators of awareness that complement behavioural assessments. Methods: Forty-four participants (N = 44) completed repeated imagined-movement tasks using wearable EEG (PDoC: Unresponsive Wakefulness Syndrome (UWS, n = 14), Minimally Conscious State (MCS, n = 17), Locked-In Syndrome (LIS, n = 11); two able-bodied participants as benchmarks; ClinicalTrials.gov: NCT03827187; 30-01-2019). The protocol assessed sensorimotor rhythm modulation, training with and without neurofeedback, and binary question answering across phases. Standard behavioural assessments (CRS-R and WHIM) were administered at each session. Results: Significant MI-BCI decoding accuracy (DA) is achieved by 73.8\% of patients, of whom 90\% progress to Q\&A testing and frequently exceed the 70\% usability threshold, revealing marked inter-individual heterogeneity. For significant MI-BCI runs, LIS outperform MCS (p = 0.007) and UWS (p = 0.048), while UWS exceed MCS during Q\&A (p = 0.049), driven by familiar-voice stimuli. Using leave-one-subject-out cross-validation, combining predictions from DA and behavioural assessments improves balanced diagnostic accuracy to 62\% (from 55\%), increasing sensitivity to MCS (39\% to 69\%), with a modest reduction in LIS sensitivity (78\% to 67\%). Task-related activity over sensorimotor and parietal cortices differentiate diagnostic groups. Conclusions: The structured MI-BCI protocol demonstrates potential as a movement-independent, EEG-based tool for distinguishing UWS, MCS and LIS. Integrating DA and spatial patterns yields diagnostic information that may augment behavioural assessment and advance objective tools for evaluating awareness in PDoC.",

author = "\{Du Bois\}, Naomi and Attila Korik and Stephanie Hodge and Leah Hudson and Elahi, \{Ainjila Shakeb\} and Bigirimana, \{Alain Desire\} and Natalie Dayan and Sanchez-Bornot, \{Jose Miguel\} and Alison McCann and Kudret Yelden and Lloyd Bradley and Nair, \{Krishnan Padmakumari Sivaraman\} and Simon Judge and Damon Hoad and Emma Vines and Venu Harilal and Sheryl Parke and Paul Johnson and Jacqueline Pogue and Emma Dodds and Abayomi Salawu and Raymond Carson and Karl McCreadie and Jacqueline Stow and Jacinta McElligott and Aine Carroll and Damien Coyle",

year = "2026",

month = apr,

day = "17",

doi = "10.1038/s43856-026-01574-x",

language = "English",

journal = "Communications Medicine",

issn = "2730-664X",

publisher = "Springer Nature",

}

TY - JOUR

T1 - Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness

AU - Du Bois, Naomi

AU - Korik, Attila

AU - Hodge, Stephanie

AU - Hudson, Leah

AU - Elahi, Ainjila Shakeb

AU - Bigirimana, Alain Desire

AU - Dayan, Natalie

AU - Sanchez-Bornot, Jose Miguel

AU - McCann, Alison

AU - Yelden, Kudret

AU - Bradley, Lloyd

AU - Nair, Krishnan Padmakumari Sivaraman

AU - Judge, Simon

AU - Hoad, Damon

AU - Vines, Emma

AU - Harilal, Venu

AU - Parke, Sheryl

AU - Johnson, Paul

AU - Pogue, Jacqueline

AU - Dodds, Emma

AU - Salawu, Abayomi

AU - Carson, Raymond

AU - McCreadie, Karl

AU - Stow, Jacqueline

AU - McElligott, Jacinta

AU - Carroll, Aine

AU - Coyle, Damien

PY - 2026/4/17

Y1 - 2026/4/17

N2 - Background: Accurate assessment of residual awareness in patients with Prolonged Disorders of Consciousness (PDoC) remains a major clinical challenge, as conventional behavioural tools can underestimate covert cognition. This study evaluates whether a structured, multi-phase motor imagery Brain–Computer Interface (MI-BCI) protocol provides objective electroencephalography (EEG)-based indicators of awareness that complement behavioural assessments. Methods: Forty-four participants (N = 44) completed repeated imagined-movement tasks using wearable EEG (PDoC: Unresponsive Wakefulness Syndrome (UWS, n = 14), Minimally Conscious State (MCS, n = 17), Locked-In Syndrome (LIS, n = 11); two able-bodied participants as benchmarks; ClinicalTrials.gov: NCT03827187; 30-01-2019). The protocol assessed sensorimotor rhythm modulation, training with and without neurofeedback, and binary question answering across phases. Standard behavioural assessments (CRS-R and WHIM) were administered at each session. Results: Significant MI-BCI decoding accuracy (DA) is achieved by 73.8% of patients, of whom 90% progress to Q&A testing and frequently exceed the 70% usability threshold, revealing marked inter-individual heterogeneity. For significant MI-BCI runs, LIS outperform MCS (p = 0.007) and UWS (p = 0.048), while UWS exceed MCS during Q&A (p = 0.049), driven by familiar-voice stimuli. Using leave-one-subject-out cross-validation, combining predictions from DA and behavioural assessments improves balanced diagnostic accuracy to 62% (from 55%), increasing sensitivity to MCS (39% to 69%), with a modest reduction in LIS sensitivity (78% to 67%). Task-related activity over sensorimotor and parietal cortices differentiate diagnostic groups. Conclusions: The structured MI-BCI protocol demonstrates potential as a movement-independent, EEG-based tool for distinguishing UWS, MCS and LIS. Integrating DA and spatial patterns yields diagnostic information that may augment behavioural assessment and advance objective tools for evaluating awareness in PDoC.

AB - Background: Accurate assessment of residual awareness in patients with Prolonged Disorders of Consciousness (PDoC) remains a major clinical challenge, as conventional behavioural tools can underestimate covert cognition. This study evaluates whether a structured, multi-phase motor imagery Brain–Computer Interface (MI-BCI) protocol provides objective electroencephalography (EEG)-based indicators of awareness that complement behavioural assessments. Methods: Forty-four participants (N = 44) completed repeated imagined-movement tasks using wearable EEG (PDoC: Unresponsive Wakefulness Syndrome (UWS, n = 14), Minimally Conscious State (MCS, n = 17), Locked-In Syndrome (LIS, n = 11); two able-bodied participants as benchmarks; ClinicalTrials.gov: NCT03827187; 30-01-2019). The protocol assessed sensorimotor rhythm modulation, training with and without neurofeedback, and binary question answering across phases. Standard behavioural assessments (CRS-R and WHIM) were administered at each session. Results: Significant MI-BCI decoding accuracy (DA) is achieved by 73.8% of patients, of whom 90% progress to Q&A testing and frequently exceed the 70% usability threshold, revealing marked inter-individual heterogeneity. For significant MI-BCI runs, LIS outperform MCS (p = 0.007) and UWS (p = 0.048), while UWS exceed MCS during Q&A (p = 0.049), driven by familiar-voice stimuli. Using leave-one-subject-out cross-validation, combining predictions from DA and behavioural assessments improves balanced diagnostic accuracy to 62% (from 55%), increasing sensitivity to MCS (39% to 69%), with a modest reduction in LIS sensitivity (78% to 67%). Task-related activity over sensorimotor and parietal cortices differentiate diagnostic groups. Conclusions: The structured MI-BCI protocol demonstrates potential as a movement-independent, EEG-based tool for distinguishing UWS, MCS and LIS. Integrating DA and spatial patterns yields diagnostic information that may augment behavioural assessment and advance objective tools for evaluating awareness in PDoC.

U2 - 10.1038/s43856-026-01574-x

DO - 10.1038/s43856-026-01574-x

M3 - Article

SN - 2730-664X

JO - Communications Medicine

JF - Communications Medicine

ER -

Advancing EEG-Based Assessment of Consciousness and Cognition in Prolonged Disorders of Consciousness

Abstract

Data Availability Statement

Acknowledgements

Funding

Access to Document

Fingerprint

Cite this