Background: Clinical interventions for autism spectrum disorder (ASD), which is defined by impaired social communication skills and by restricted and repetitive behaviors, typically include therapists to guide skill development and often in the context of early targeted interventions. While standard human treatments (SHT) can include toys and activities in general attention contexts, robot enhanced therapy (RET) explicitly centers on the clinical use of supervised therapeutic robots, such as the robot NAO. These RETs focus on facilitating specific targets such social interaction and social attention of children with ASD, and can also serve as a diagnostic tool by collecting interactive data during therapy. However, the effectiveness and efficacy of RET compared to Standard Human Treatment (SHT) remains unclear.

Objectives: The aim of this study was to assess the differences between the impact of SHT and RET on the attention stability of children with autism.

Methods: This study utilized the Development of Robot-Enhanced therapy for Children with spectrum disorders (DREAM) project, which included 3,100 therapy sessions. The DREAM project compared SHT and RET interventions with the same structure and skill focus on imitation, joint attention, and turn taking. The sample consisted of 61 children with autism (female=9), aged 3 to 6 and analyzed differences in attention stability across time between the conditions of SHT (n=31) and RET (n=30). Session-level variables included gaze variance, head variance, skeleton motion as a mean frame-frame displacement, with lower variance indicating more body stability. These variables were combined to create a composite attention stability index, which refers to the consistency of the child’s sustained attentional focus. A Welch’s t-test was implemented to compare the mean of each metric between RET and SHT, allowing for unequal variances and sample sizes. The effect size was calculated using Cohen’s d.

Results: The RET group showed higher gaze variance (Mean±SD=0.100) compared to the SHT (Mean±SD=0.077) group; t=24.74, p<.001, d=0.878. Similarly, the RET group demonstrated higher head variance (Mean±SD=0.067) compared to the SHT (Mean±SD=0.052) group; t=19.90, p<.001; d=0.709. This indicates less steady eye fixation and head positioning was present in the RET group. In contrast, skeletal motion was not significantly different between the groups (t=-0.36, p=0.716; d=?0.011). Overall, the composite index of attention stability was significantly higher with SHT (Mean±SD=?0.228;) than RET(Mean±SD=0.258); t=-18.37, p<.001; d=?0.655.

Conclusions: While there were no differences in gross body movement, RET is associated with lower micro-stability in the eyes and head, and overall lower attention stability during treatment sessions in comparison to SHT, indicating significant differences in child attention between human-robot and child-therapist interventions. While both interventions employed the same structure and goals, the lower attentional stability may be due to factors intrinsic to RET, such as participants focusing on various novel stimuli of the NAO robot or splitting attention between the NAO robot and supervising therapist. Subsequently, there may be potential areas in RET in need of refinement to increase attentional stability. Further analyses are required to explicate differences between children in terms of age, initial symptoms, and intervention targets.

Notes:

(1)ISOF, Consiglio Nazionale delle Ricerche, Bologna, Italy, (2)Lipidomic Laboratory, Lipinutragen srl, Bologna, Italy, (3)Fondazione Danelli ONLUS, Lodi, Italy, (4)Child Psychopathology Unit, Scientific Institute and IRCCS E. Medea, Bosisio, Italy, (5)IRCCS Istituto delle Scienze Neurologiche di Bologna, UOC Neuropsichiatria dell’Età Pediatrica, Bologna, Italy, (6)Pediatric Surgery Department, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Dipartimento di Scienze Mediche e Chirurgiche (DIMEC), Bologna, Italy, (7)IRCCS Istituto delle Scienze Neurologiche di Bologna, UOC Neuropsichiatria dell’Età Pediatrica, Centro Regionale per i Disturbi della Nutrizione e dell’Alimentazione in Età Evolutiva, Bologna, Emilia-Romagna, Italy, (8)Dipartimento di Scienze Mediche e Chirurgiche (DIMEC), Università di Bologna, Bologna, Emilia-Romagna, Italy, (9)Department of Applied Medical Physics, Medical School, National and Kapodistrian University, Athens, Greece, (10)Autism Research Unit, Villa Santa Maria SCS, Tavernerio, Como, Italy