Assessing the Feasibility of Using Apple Vision Pro While Performing Medical Precision Tasks: Controlled User Study.

Background: The emergence of next-generation video-see-through head-mounted displays, such as the Apple Vision Pro (AVP), has generated considerable interest in the medical field. While preliminary studies highlight AVP's potential, no controlled study has rigorously assessed its usability for precision-based medical tasks requiring fine motor control and real-world perception. Objective: This study aims to evaluate the feasibility of using AVP while performing real-world medical precision tasks. Methods: To assess AVP's feasibility, we conducted a controlled user study with 20 health care professionals, who performed 3 different suturing techniques across 3 intervention conditions. Participants completed the same tasks using AVP, the Microsoft HoloLens 2 (MHL2), and a baseline (without a head-mounted display). A within-subject design was used, ensuring that each participant experienced all intervention groups. We used a mixed methods research approach, incorporating both quantitative metrics, including task completion time, suturing performance, system usability score, cognitive load, virtual reality sickness, and presence score, as well as qualitative insights gathered through interviews. Results: Participants took significantly longer to complete the entire task using AVP (570.0, SD 192.0 s) compared with MHL2 (456.0, SD 120.0 s; P<.001) and baseline (472.0, SD 143.0 s; P<.001). The analysis on participants' average suture performance revealed no significant differences across interventions (P=.76). The total raw NASA Task Load Index score among participants was significantly higher for AVP (43.9, SD 15.9) compared with MHL2 (21.5, SD 13.8; P<.001) and baseline (19.1, SD 15.1; P<.001). The analysis of the presence questionnaire demonstrated a significantly higher presence score for MHL2 (115.0, SD 11.4) compared with AVP (93.7, SD 12.7; P<.001). The overall virtual reality sickness questionnaire score was significantly higher for AVP (66.9, SD 19.8) compared with MHL2 (41.1, SD 9.32; P<.001). Moreover, the calculated system usability score for MHL2 (72.7, SD 8.54) was significantly higher compared with AVP (50.3, SD 14.4; P<.001). Conclusions: In conclusion, AVP has potential for non-time-sensitive medical applications or those that emphasize digital elements over real-world interaction. Its current usability limitations, particularly increased cognitive load and prolonged task execution times, suggest that further optimizations are necessary before widespread clinical adoption is feasible.