The difficulties with motor skills in children with autism spectrum disorders (ASD) has become a major focus of interest. Our objectives were to provide an overall profile of motor capacities in children with ASD compared to neurotypically developed children through specific tests, and to identify which motor tests best discriminate children with or without ASD. Twenty-two male children with ASD (ASD-10.7 ± 1.3 years) and twenty controls (CONT-10.0 ± 1.6 years) completed an evaluation with 42 motor tests from European Physical Fitness Test Battery (EUROFIT), the Physical and Neurological Exam for Subtle Signs (PANESS) and the Movement Assessment Battery for Children ( M-ABC). However, it was challenging to design a single global classifier to integrate all these features for effective classification due to the issue of small sample size. To this end, we proposed a hierarchical ensemble classification method to combine multilevel classifiers by gradually integrating a large number of features from different motor assessments. In the ASD group, flexibility, explosive power and strength scores (p < 0.01) were significantly lower compared to the control group. Our results also showed significant difficulties in children with ASD for dexterity and ball skills (p < 0.001). The principal component analysis and agglomerative hierarchical cluster analysis allowed for the classification of children based on motor tests, correctly distinguishing clusters between children with and without motor impairments.
Keywords: autism spectrum disorders; children; cluster analysis; motor impairments.