Attention deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder affecting various aspects of life. Some features of the mental disorders affect peo- ple’s movement patterns. In the recent decade, researchers have paid attention to the analysis of gait and balance pattern using new technological tools, as well as artifi- cial intelligence algorithms. Therefore, the present study aims to propose an intel- ligent method to identify ADHD in children using gait and balance pattern features extracted from the person’s movements obtained from the skeleton data. Given that designing and extracting effective motor features for diagnosing the aforementioned disorder is the main objective. In the present applied development experimental study, human movement samples related to the gait and balance were recorded in the standard test of perceptual-motor development, from healthy and ADHD-diag- nosed children. After preprocessing the data recorded by the Kinect device, effec- tive features for diagnosis are designed and extracted from the appropriate special movement tests. Comparing the features extracted from gait and balance tests by skeleton data, the results indicated that the data based on other types of methods for differentiation into healthy and ADHD groups are in line with those of the present study. The results of diagnosis and separation of healthy children from those with disorders in the different movement tests, standing on the ground with the superior foot, standing on a balance stick with the superior foot, and walking heel forward on a balance stick, to identify ADHD by SVM classification method are 86.4%, 90.2%, and 88.1%, respectively. The obtained significant results have been achieved relying on machine learning-based methods using the effective features obtained from skel- eton gait and balance data of children along with analyzing the descriptive statistics of the features of gait and balance tests.

Keywords