The Feasibility and Usability of an Artificial Intelligence-Enabled Conversational Agent in Virtual Reality for Patients with Alcohol-Associated Cirrhosis: A Multi-Methods Study.

Over 7000 Canadians undergo lower limb amputations (LLAs) each year, of which approximately 70% will experience phantom limb pain (PLP)—a form of neuropathic pain perceived in a limb despite its physical absence following amputation. PLP onset can occur within the first few days to 1 month following amputation, with most people with LLAs experiencing it within 1 year. Recent studies also suggest PLP may be the most amenable to intervention within the first 6 months following amputation, theoretically by preventing the maladaptive reorganization of the sensorimotor cortex implicated in PLP. One such intervention, graded motor imagery (GMI), was originally developed by Moseley and colleagues as a structured, sequential protocol designed to progressively restore sensorimotor congruence and reduce pain in conditions such as PLP and complex regional pain syndrome. GMI comprises three distinct stages: (1) implicit motor imagery, achieved through a left/right limb discrimination task; (2) explicit motor imagery, involving the mental simulation of limb movements without overt execution; and (3) mirror therapy, which stimulates phantom motor imagery through visual illusion created by positioning the intact limb in front of a mirror. Meta-analyses indicate that GMI is an efficacious intervention compared with placebo or its individual components, such as mirror therapy or explicit motor imagery alone, and findings from an individual randomized controlled trial further suggest that adherence to the protocolized sequence of stages is essential for reducing PLP intensity according to validated self-report measures. Unfortunately, effective intervention is often delayed due to long outpatient wait times (>6 months in Canada), the intensive involvement of health care personnel required for longitudinal administration, limited patient motivation for or belief in the efficacy of GMI (also known as treatment buy-in), and an overreliance on pharmacological approaches that have demonstrated limited effectiveness for PLP. Given PLP has been independently associated with lower quality of life, as well as increased rates of depression, anxiety, and suicidality, improving the timeliness and accessibility of effective interventions is warranted. The feasibility of GMI may be improved through virtual reality (VR)—a three-dimensional computer-generated simulation in which a user is immersed in a virtual world through a headset. Recent systematic reviews suggest that virtual visual feedback of the phantom limb, mimicking mirror therapy, is an efficacious PLP treatment in community settings. Phantom motor execution—characterized by the control of a virtual phantom limb through electromyographic signals detected at the residual limb—has likewise demonstrated efficacy for PLP treatment and comparable outcomes to explicit motor imagery. Case series also indicate that VR-based treatments, incorporating both virtual visual feedback and phantom motor execution, can be successfully self-administered at home and show preliminary effectiveness. Studies further suggest that the immersion and gamification offered by VR make treatment more engaging and may enhance compliance, even in hospital settings. However, nearly all existing studies have been conducted months or years after amputation in chronic pain populations, leaving a critical gap in understanding how VR-based interventions could be implemented during the immediate postoperative recovery period—when cortical reorganization is most dynamic and potentially modifiable. While mirror therapy, explicit motor imagery, and phantom motor execution have each been tested through VR, the sequenced GMI protocol has never been implemented, despite its potential to improve accessibility through self-administration and gamified delivery. VR has shown feasibility in nonsurgical hospital settings, typically over one to five sessions for distractive pain relief, suggesting that inpatient delivery is both possible and acceptable. However, the feasibility of VR-based GMI as an early PLP intervention remains unknown. The present proof-of-concept study aims to evaluate the feasibility of administering a VR GMI program during the immediate postoperative period following LLA. It represents the next phase in a staged program of research guided by the Virtual Reality Clinical Outcomes Research Experts (VR-CORE) model and follows an earlier developmental phase in which the VR GMI program was refined through input from people with lived experience of LLAs who trialed the prototype and provided feedback to enhance its usability and interactivity. In accordance with the VR-CORE model, this phase represents an early feasibility evaluation intended to refine methodological procedures and inform the design of a subsequent, larger-scale investigation. The primary objective was to assess the feasibility of the VR GMI program in the acute postoperative setting via recruitment, eligibility, and consent rates; the secondary