Design and Implementation of an Interactive VR-Reconstructed ICU System for Psychological Recovery: A Mixed-Methods Pilot Study (Preprint)

<sec> <title>BACKGROUND</title> More than one-third of intensive care unit (ICU) survivors experience significant psychological impairment, including anxiety, depression, and post-traumatic stress disorder (PTSD). Delusional memories due to unpleasant experiences and lack of contextual understanding are major modifiable risk. We developed an interactive, VR-reconstructed ICU environment integrating multimodal psychological and informational support to facilitate patient-directed exposure therapy and mitigate trauma. </sec> <sec> <title>OBJECTIVE</title> This pilot mixed-methods study aimed to develop a novel multimodal VR intervention for post-extubation ICU patients and evaluate its usability, feasibility, and preliminary efficacy in improving psychological outcomes and ICU-related memories. </sec> <sec> <title>METHODS</title> We enrolled critically ill adults who were first-time ICU admissions, received mechanical ventilation for &gt;48 hours, and were successfully extubated. Participants received a phased, multimodal VR intervention based on the Timing-It-Right framework, tailored to their recovery stage (Stress: &lt;24h, Buffer: 24-48h, Acceptance: 48-72h post-extubation). Outcomes including anxiety (HADS-A), depression (HADS-D), PTSD symptoms (IES-R), and system usability (SUS) were assessed before and after the intervention. ICU memories (ICUMT) were evaluated one week after ICU discharge. Semi-structured interviews were conducted with a subset of participants (n=12) to explore user experience. </sec> <sec> <title>RESULTS</title> Thirty patients completed the study. The VR intervention demonstrated high feasibility, with 185 sessions completed (mean 6.17 sessions/patient, mean duration 7.08 minutes/session). The mean System Usability Scale score was 79.42±7.09, indicating ‘good’ to ‘excellent’ usability. Compared to baseline, post-intervention scores showed statistically significant and large reductions in anxiety (mean difference 4.43, p=.032, Cohen’s d=1.60), depression (mean difference 5.10, p&lt;.001, Cohen’s d=1.90), and PTSD symptoms (mean difference 14.90, p=.013, Cohen’s d=1.80). At one-week follow-up, factual memories (48.57±4.72) predominated over emotional (8.97±2.36) and delusional memories (4.70±0.88). Qualitative analysis revealed high acceptance, perceived usefulness in enhancing treatment understanding, and overall positive experiences, despite suggestions for technical refinements. </sec> <sec> <title>CONCLUSIONS</title> A theory-driven, co-designed multimodal VR intervention is feasible, acceptable, and shows promising preliminary efficacy for reducing acute psychological distress and promoting adaptive memory recall in ICU survivors. These findings support the conduct of a definitive randomized controlled trial to establish efficacy. </sec> <sec> <title>CLINICALTRIAL</title> Registered with the Chinese Clinical Trial Registry on 1 March 2023 under registration number ChiCTR2300068874. </sec>