Imitation of complex whole-body sequence in profoundly hearing-impaired adolescents

Many studies have identified specific visual advantages in deaf individuals. However, few studies have linked these advantages to motor learning. This study aimed to compare the differences in movement and learning processes between profoundly hearing-impaired (HI) and normal-hearing (NH) adolescents by capturing motions that imitate complex whole-body sequences. To analyze multiple parameters simultaneously in a time series and evaluate their similarity to the reference, we used a deep-learning strategy with an autoencoder for anomaly detection. In this study, anomaly detection indicated the difference between the reference and the participants. We set the threshold to 3σ. An analysis targeting the hands showed that HI had a significantly lower anomaly rate than NH in the first imitation; even when checking the trajectory, we found that all HI followed the same pattern as the reference dancer. Compared with NH, HI could capture and reproduce the motion characteristics of the reference in the first trial, especially in the hands, which broadens the perspective of motor instruction for HI adolescents.