Experiment 1
In the experiment, the subjects performed three kinds of tasks in VEs—normal motion on a plane, mirrored motion on a plane, and grasping motions—with the VRRS and LP systems; i.e., six tasks were performed (Fig. 4b). The normal motion task and the grasping motion task are conventional motor tasks for rehabilitation [21] and the mirrored motion task which is an emerging therapy can promote brain function remodeling and induce motor function recovery [22]. Before each task started, the subjects were positioned in a chair which was approximately 1.6 m away from cameras with an initial upper body posture that was the same as that of the avatar. Their hands were stretched naturally, with palms facing down. Each task consisted of 30 trials, and the six tasks took approximately 45 min. After all tasks were implemented, the subjects were asked to fill in a questionnaire (see Additional file 1) to grade both the VRRS and LP systems.
Task 1/Task 3: Normal motion on a plane based on the VRRS/LP
The paradigms of one trial were based on the experimental procedure presented in Fig. 5a. Initially, the subject controlled the avatar to move its left upper limb/virtual left hand to the starting area, which was originally displayed in blue, by moving his or her left upper limb. The subject held this position for 1.5 s, after the area turned green. After that, a blue object of a constant size was generated and located in a random location in the reaching area, which was calibrated for each subject before the experiment began. The color of the starting area turned gray, and a blue line appeared between the object and the starting area. The subject was required to move his or her left upper limb along the blue line to reach the object and hold it for 1.5 s after the object and the line turned green. If the object could not be reached within 4 s, both it and the line disappeared, and the trial was regarded as unfinished. At that point, the subject had to move his or her left upper limb as quickly as possible back to the starting area. After he or she had returned to the starting area and the color of the starting area returned to blue, the next trial began. In contrast, if the subject was able to reach the object, a score was displayed on the top-middle of the screen for 1 s to provide encouraging visual feedback. The total score was shown in the upper-left corner of the screen throughout the task, which also served as visual feedback. Then, the subject moved the left upper limb back to the starting area as soon as possible and remained there until the starting area turned blue again, which meant that the trial had been completed and the next trial was beginning. The first 15 trials used the left upper limb; the remaining trials used the opposite limb, with the subject controlling the movements of the avatar's right upper limb/virtual right hand in the same manner (Fig. 5).
Task 2/Task 4: Mirrored motion on a plane based on the VRRS/LP
In these two tasks, the motion of the subject’s left upper limb was mapped to the avatar's right upper limb/virtual right hand. The avatar's left upper limb/virtual left hand was controlled by the subject’s opposite limb. Other procedures were the same as those used in task 1 and task 3.
Task 5/Task 6: Grasping motion based on the VRRS/LP
First, after the subjects moved their left upper limbs/virtual left hands to the starting area, blue object 1 was randomly generated in the reaching space. The reaching space was determined by the subject’s reach area and constant height. Then, the subject was required to grasp the object and hold it for 1.5 s after the object turned green. If the object could not be grasped within 4 s, it disappeared, and the task was regarded as unfinished. If the object was grasped, it turned yellow, and a new blue object 2 was randomly generated in the reaching area after 1.5 s. At the same time, a blue line appeared that connected object 2 and a point on the same plane that was based on the generated position of object 1. The subjects were asked to move their left upper limbs to reach object 2 along the blue line (Fig. 5). Other procedures were the same as those used in task 1 and task 3.
Experiment 2
Experiment 2 was a 5-day experiment that was completed by the subjects in all three groups with the same training time (350 min, 70 min per day). All patients were evaluated with FM on the first day and evaluated again with FM after they had completed the 5-day experiment. The patients in the experimental group and Control Group 2 were also asked to fill out a questionnaire after completing the experiment. Every stroke patient in the experimental group participated in the experiment on the feasibility of the VRRS (Fig. 4a), which included ten training sessions. These patients were required to participate in sessions twice a day with an interval of 2 min between sessions. Before the first session, the motor function of the stroke patients was evaluated with FM. Each session consisted of three tasks: normal motion on a plane, mirrored motion on a plane, and grasping motion. Based on the duration and intensity that the patients were able to adapt to, the number of trials for each task was set to 30, and the patients took 2-min breaks between the three tasks. In Control Group 1, the patients received conventional therapy for shoulder, elbow, wrist, and finger mobilization. In Control Group 2, the patients performed the same training procedure as the experimental group, but with LP (Fig. 6).
Outcome measures
In Experiment 1, the variables (unfinished rate, game performance, and velocity peaks of the wrists) were measured by Unity for each task. The unfinished rate was obtained by dividing the quantity of unfinished trials by the total number of trials in each task. The velocity peaks were measured as the subjects moved along the line in the VEs. Game performance was determined by the total score, which was the sum of the scores for 30 trials. The score for each trial was determined by the average distance between the line and the location based on the position of the center of the hand. The location is obtained by mapping the position of the hand to a plane at the same height as the line. The score for each trial was calculated as follows:
$${\text{Score}} = \left( {0.1 - {\text{average}}\,{\text{distance}}} \right)*1000,$$
where the maximum score is 100, and the average distance denotes the mean value of the sum of the distance between the location based on the hand position and the line in each frame during movement along the line. After the subjects completed all tasks, they were asked to complete the questionnaire, which integrated avatar embodiment [23] and a System Usability Scale [24]. The questionnaire was used to assess the sense of body ownership, agency, and location of the body and the usability of the VRRS and LP. In Experiment 2, the 24 patients underwent two FM evaluations. The patients in the experimental group and Control Group 2 were also asked to fill in the same questionnaire that they had completed in Experiment 1 after the second evaluation.
Statistical analysis
Analysis was performed using the SPSS statistical software system (version 25.0) and the Origin function draw tool (version 2018). First, we used a homogeneity of variance test to determine whether the overall variance was unequal. Only the unfinished rate for the normal motion and grasping motion tasks were not consistent with homogeneity of variance. Then, in Experiment 1, we used the analysis of variance of two-factor fixed design to compare the VRRS and LP in terms of game performance, velocity peaks, unfinished rate, and the participant responses to the questionnaire. In Experiment 2, the paired t-test was applied to compare the FM scores at the two evaluations for each group. For the participants’ responses to the questionnaire, we used an unpaired t-test to examine differences. We also used one-way repeated measure ANOVA to separately compare the FM scores at the first evaluation, those at the second evaluation, and the D-value of the FM scores at the two evaluations among the three groups. The significant values were considered as p < 0.05.