Upper limb movement trajectories in virtual reality versus conventional occupational therapy: a within-subject study

Virtual reality (VR) is increasingly used as a complementary approach to conventional occupational therapy (OT), but it remains unclear whether corresponding VR and physical tasks elicit similar spatial movement characteristics. Evaluating these differences is clinically relevant because movement strategies practiced in VR may influence transfer to real-world activities. This study compared hand trajectory length, as a spatial kinematic indicator, between corresponding VR-based and physical OT tasks in participants with upper-limb motor impairments and healthy controls and evaluated the usability of the VR system. In this comparative within-subject study, 39 participants (20 with upper-limb motor impairments and 19 healthy controls) performed three corresponding OT tasks—range-of-motion (ROM) arc, cone stacking, and peg transfer—under both physical and VR conditions. The order of conditions and tasks was counterbalanced using a Latin square design. Task performances were video-recorded and analyzed using a video-based three-dimensional body-tracking system, with wrist coordinates used to calculate trajectory length. Trajectory length was compared between the physical and VR conditions within each group and between groups within each condition. System usability was assessed using the standard 10-item System Usability Scale (SUS). For participants with upper-limb motor impairments, trajectory length did not differ significantly between VR and physical conditions for the ROM arc ( p = .522) or cone task ( p = .294), whereas peg trajectories were significantly longer in VR ( p = .007). In healthy controls, ROM arc trajectory length did not differ between conditions ( p = .381), while cone and peg trajectories were significantly longer in VR (both p < .001). Between-group comparisons showed no significant differences for the ROM arc or cone tasks under either condition, whereas peg trajectory length differed significantly between groups in both VR ( p = .032) and physical conditions ( p < .001). Mean SUS scores were 76.6 ± 16.4 in healthy controls and 65.4 ± 15.2 in participants with upper-limb motor impairments. Corresponding VR and physical OT tasks showed similar trajectory lengths during the relatively constrained ROM arc task, whereas greater differences emerged during tasks requiring more complex or precise object manipulation. These findings suggest that VR may approximate the spatial extent of hand movement for relatively constrained tasks, but trajectory length alone does not establish equivalence of overall movement quality or motor strategy. Lower usability among participants with upper-limb motor impairments further indicates the need for improved interface design, device ergonomics, and interaction methods to support broader clinical implementation.