An Embodied Sonification Model for Sit-to-Stand Transfers

Abstract

Interactive sonification of biomechanical quantities is gaining relevance as a motor learning aid in movement rehabilitation, as well as a monitoring tool. However, existing gaps in sonification research (issues related to meaning, aesthetics, and clinical effects) have prevented its widespread recognition and adoption in such applications. The incorporation of embodied principles and musical structures in sonification design has gradually become popular, particularly in applications related to human movement. In this study, we propose a general sonification model for the sit-to-stand (STS) transfer, an important activity of daily living. The model contains a fixed component independent of the use-case, which represents the rising motion of the body as an ascending melody using the physical model of a flute. In addition, a flexible component concurrently sonifies STS features of clinical interest in a particular rehabilitative/monitoring situation. Here, we chose to represent shank angular jerk and movement stoppages (freezes), through perceptually salient pitch modulations and bell sounds. We outline the details of our technical implementation of the model. We evaluated the model by means of a listening test experiment with 25 healthy participants, who were asked to identify six normal and simulated impaired STS patterns from sonified versions containing various combinations of the constituent mappings of the model. Overall, we found that the participants were able to classify the patterns accurately (86.67 ± 14.69% correct responses with the full model, 71.56% overall), confidently (64.95 ± 16.52% self-reported rating), and in a timely manner (response time: 4.28 ± 1.52 s). The amount of sonified kinematic information significantly impacted classification accuracy. The six STS patterns were also classified with significantly different accuracy depending on their kinematic characteristics. Learning effects were seen in the form of increased accuracy and confidence with repeated exposure to the sound sequences. We found no significant accuracy differences based on the participants' level of music training. Overall, we see our model as a concrete conceptual and technical starting point for STS sonification design catering to rehabilitative and clinical monitoring applications.