Eye-Tracking Assessment in Patients with Disorders of Consciousness: A Systematic Review

Disorders of consciousness (DOC), including vegetative state/unresponsive wakefulness syndrome (VS/UWS) and minimally conscious state (MCS), present significant diagnostic challenges. Misdiagnosis rates approach 40%, often due to limitations in detecting subtle behavioural responses through clinical observation alone. Eye-tracking technology offers objective, quantitative assessment of visual behaviours that may reveal covert signs of consciousness. This systematic review aimed to evaluate the diagnostic accuracy of eye-tracking technology compared to the Coma Recovery Scale-Revised (CRS-R) for detecting visual responses and consciousness signs in patients with DOC; to examine stimulus effects; and to assess prognostic value.

A systematic literature search was conducted across SciSpace, Google Scholar, PubMed, and institutional libraries following PRISMA 2020 guidelines. Eligibility criteria specified studies involving patients with prolonged DOC assessed using eye-tracking technology. Data extraction, risk of bias assessment, and GRADE certainty evaluation were conducted systematically.

Fifteen studies (n = 4-123 patients; published 2012-2025) were included. Eye-tracking detected visual responses in significantly more patients than clinical observation alone (46.2% vs. 18.1% in one study). Mirror stimuli demonstrated the highest detection sensitivity (97% vs. 69% for person and 57% for object). Affectively salient stimuli elicited stronger tracking responses in patients with MCS (37.3% vs. 29.9-30.6% neutral). Advanced VR-based systems achieved high diagnostic accuracy (sensitivity 100%, specificity 88.9%) with prognostic value (overt tracking predicting 62.5% command-following at one year). GRADE certainty was Low for detection rates and diagnostic discrimination, and Very Low for sensitivity, specificity, and prognostic outcomes.

Eye-tracking provides objective, sensitive assessment of visual behaviours in patients with DOC and may reduce misdiagnosis rates, supporting a conditional recommendation for its use as a supplementary assessment tool alongside CRS-R. Methodological heterogeneity, small sample sizes, and absence of blinding limit certainty. Adequately powered, multicentre prospective studies are urgently needed.