Evaluating Autistic Children's Dynamic Balance Ability by an Azure Kinect Sensor

We suggested using an Azure Kinect sensor to evaluate the dynamic balance ability of children with autism spectrum disorder (ASD). Two metrics, i.e., body mass center fluctuation (BMCF) and upper limb swing (ULS), were extracted from the trajectories of 32 body joints' movement during the balance beam test. Findings showed that ASD children had lower BMCF and ULS than typically developing (TD) children. Furthermore, the prediction using BMCF achieved a maximal accuracy of 79% and an AUC value of 0.69.