Usability and effects of text, image and audio feedback on exercise correction during augmented reality based motor rehabilitation

摘要

This paper aims to evaluate text, image and audio feedback regarding usability, exercise correction and user preferences when using Augmented Reality (AR) to guide movements execution in rehabilitation applications. For that, this study used an AR rehabilitation tool, named ARkanoidAR, which attempts to correct the wrong movement performed by a user through text, image and audio feedback, while he/she is playing a game controlled based on the therapy protocol defined by a physiotherapist for that particular user. The ARkanoidAR uses a Kinect sensor to track the human body and its movements, and it is completely configurable according to patient therapy needs. To evaluate feedback effects individually, this study collected quantitative data to measure success rate, which represents the amount of time that the user executed the movements correctly when using the application with each feedback individually activated. Experiments were composed of five phases: control, text, image, audio and control. The feedback sequence was randomized before the tests in order to avoid bias. Quantitative results show that users understood the instructions for performing the movements correctly since the success rate along time increased. Tests also included observation and qualitative analysis through questionnaires and a semi structured interview to evaluate the usability perception of users. The qualitative data helped to detect different types of problems for each feedback based on users' answers. This study was carried out with healthy people, and did not predict the particularities and limitations of physiotherapy patients. (C) 2019 Elsevier Ltd. All rights reserved.