Eye tracking in virtual reality for neurorehabilitation: a narrative perspective on needs, challenges, and pathways beyond game engines

Virtual reality (VR) systems with integrated eye tracking offer a powerful way to study and support sensorimotor and cognitive function in neurorehabilitation. Eye movements provide a high-bandwidth window onto information processing, visuomotor integration, cognitive load, and affect, while immersive VR enables more ecologically valid yet controllable tasks spanning visual exploration, movement execution, object interaction, and social exchange. This narrative review synthesizes recent work on eye tracking in VR for neurorehabilitation, focusing on three application domains: assessment, intervention, and supportive design, together with the technical and governance requirements needed to make these systems clinically meaningful and ethically responsible. We highlight how the dominant implementation pattern of integrated headsets streaming preprocessed gaze rays into game engines introduces black-box processing, frame-bound timing, and limited calibration control that pose threats to validity, reproducibility, and cross-site comparability. We review emerging workarounds, including modular architectures that decouple sensing and rendering, explicit latency benchmarking and cross-modal synchronization, adaptive and implicit calibration approaches, and privacy-by-design frameworks from digital phenotyping and metaverse healthcare. Taken together, the evidence suggests that eye-tracked VR is already capable of supporting informative assessments and promising interventions, but that realizing its full potential for neurorehabilitation will require a shift toward architectures that support transparent control over sampling, calibration, timing, and data governance, as well as handling eye tracking data as both a sensitive clinical signal and a protected form of personal data.