Assessment of the Effectiveness of an Adaptive Virtual Reality-Based Simulation for Motor Skill Training.

摘要

Adaptive training methods aim to accommodate individual differences in abilities while helping persons develop knowledge and skills required to perform a new task. Methods that adjust levels of training task difficulty as a function of how well a trainee performs, and that sustain trainee motivation, may be significant from a learning productivity perspective. In contemporary motor rehabilitation applications, therapists use robots or virtual reality (VR) systems to assist in delivery of various training conditions in real-time and to provide multimodal performance feedback to patients. Virtual reality can also allow for flexible and dynamic adaptation of task difficulty to facilitate specific learning algorithms. However, there is a lack of research on adaptive training techniques involving manipulation of task difficulty in VR for motor skill development. The integration of adaptive training methods in VR is expected to increase levels of patient learning and promote more efficient skill training.;The objective of this research was to implement and assess the effectiveness of an adaptive VR-based haptic simulation for motor skill training in a real-world psychomotor task, as compared with static and random training modes. The simulation was prototyped based on a block design (BD) pattern reconstruction test from a well-known commercial psychomotor test battery, the Wechsler Adult Intelligence Scale, and programmed to apply an adaptive training scheme. Twenty-four participants were recruited and received static, random or adaptive training in evaluation of the VR system. The adaptive training involved modifications in the level of task difficulty based on user performance. Otherwise, the level of difficulty held fixed across training trials or randomly changed.;Results revealed the adaptive condition to be superior to static training in terms of performance improvement between pre- and post-testing (considering both speed and accuracy). In addition, adaptive training was found to produce performance improvements comparable to a random condition but in significantly less training time. With respect to participant performance during training trials, the maximum level of difficulty achieved under the adaptive condition revealed a linear relationship with post-training retention test performance. In addition, results also indicated a linear association of the number of stimuli a trainee required to move to the first level of difficulty and pre-test performance. Results also revealed VR BD task performance to be significantly correlated with native BD performance in both pre- and post-testing.;The findings of this study support use of adaptive training schemes for improving motor skill learning and accelerating rate of skill development when using VR simulation, as compared with static and random training schemes. Findings also reveal adaptive training to yield diagnostic information on trainee motor ability extending beyond common task performance measures. The study also demonstrated VR simulations of standardized psychomotor tests can be designed and prototyped to support levels of user performance similar to observations on native tasks. VR systems also provide benefits over standardized clinical evaluation including automatic, real-time performance recording and task difficulty manipulation. The VR simulator developed for the present study could be extended to simulate other motor tasks, such as the Purdue Pegboard Test.