The movements of children with autism can be faster or slower than their typically developing counterparts, depending on the task

Background: Atypical movement patterns in autism spectrum disorders (ASD) have been reported. Compared with typical developing (TD) children, children with ASD took more time to complete a point-to-point movement (Dowd et al., 2012), but adults with ASD performed faster horizontal arm swings than their typical counterparts (Cook et al., 2013). Incongruent kinematic results are common in the literature, which may imply that the kinematic features in ASD are task-dependent, but this is yet not well understood. Smart tablet gameplay has been proposed as a new paradigm to measure the movement features of ASD in young children (Anzulewicz et al., 2016). In this study, smart tablet games were employed to test for kinematic differences in autism, and the effect of the task. Objectives: The study aims to compute the swipe kinematics during smart tablet gameplay, and to compare these characteristic movements between ASD and TD children within different gameplay contexts. Methods: 37 ASD children (mean age: 4.5 years) and 45 age-matched TD children were recruited in the study. The children were shown two smart tablet games: "sharing" and "creativity" games. In the sharing game, the children were tasked to share the food pieces to four characters; in the creativity game, the children were tasked to select an object, trace the lines, and colour the object. Their touch trajectories on the smart tablet (iPad mini, Apple Inc.) were recorded during gameplay. The food-to-target swipes in the sharing game and the swipe gestures in the creativity game were identified using a customized MATLAB script. The travelled distance, duration, and speed of each swipe were calculated. For the sharing game, the difference between the travelled distance and the optimal distance (i.e. the straight line) was also calculated. Mann-Whitney U tests were used to determine kinematic differences between ASD and TD groups. Results: A total of 4785 food-to-target swipes were identified in the sharing game (ASD: 1585 swipes; TD: 3200 swipes) while 6178 swipes were identified in the creativity game (ASD: 2793 swipes; TD: 3385 swipes). Significant differences between ASD and TD were observed in the sharing game that ASD demonstrated slower food-to-target swipes than TD (median of 50.12 mm/s vs. 58.84 mm/s), and that they deviated from the optimal distance more than TD (median of 3.9 mm vs. 2.59 mm). There was no significant difference in the optimal distance. By contrast, ASD showed significantly faster gestures than TD (median of 81.77 mm/s vs. 60 mm/s) in the creativity game. Conclusions: The study compared the swipe kinematics between ASD and TD children in two smart tablet gameplay contexts. ASD demonstrated slower movement than TD in a goal-oriented food-to-target task, deviating more from the optimal trajectory. In contrast, ASD performed faster swipe gestures than TD in a relatively unconstrained creativity game. These data are the foundations to allow an understanding of how movement is controlled in autism within different contexts. Further, characterising movement features in ASD during smart tablet gameplay supports the development of algorithms that enable the early identification of ASD in serious game paradigms