31890
Outcomes of a Robot-Assisted Social-Emotional Understanding Intervention for Young Children with Autism Spectrum Disorders
People in the autism spectrum show impairment in social emotional skills and these difficulties contribute to the impairments in initiating and sustaining typical peer relationships (Berenguer et al., 2018; Jones et al., 2018).
Social robots show great potential for boosting treatment outcomes in children with ASD and robot-assisted autism therapy (RAT) has become an emerging application area in the past few years (Scassellati et al., 2012).
Objectives:
This study is a randomized control trial aimed to test the role of a human-assisted social robot as a treatment mediator in a social emotional understanding protocol for children with autism spectrum disorders (ASD).
Hence, we aim to investigate the feasibility and efficacy of using a partially controlled social robot (NAO) in the context of a group-based CBT intervention trial focused on emotion comprehension and related mentalizing skills, knowledge of basic emotion lexicon and the ability to attribute appropriate emotions in relation to the context.
Methods:
A total of n= 14 high functioning ASD children (13M:2F) participated in the study and completed the intervention. Children were randomly assigned either to the robot-assisted intervention group (RG) (n= 7 children, 6M:1F, mean age in months= 73.3; SD = 16.1) or to the control group (CG) applying the same intervention protocol without the use of the assistive-robot (n= 7 children, 6M:1F, mean age in months=82.1; SD = 12.4).
The intervention consisted of 10 sessions (90 minutes each administered twice a week) of a group-based cognitive behavioural (CBT) emotional understanding skills training protocol based on the principles of the Rational Emotional Behaviour Therapy (REBT). The protocol was specifically designed for implementation with the assistance of a social robot actively interacting with the children and providing verbal antecedents, prompts and reinforcing consequences according to an Antecedent-Behaviour-Consequence (ABC) model.
Pre- and post-intervention assessments were conducted using the Test of Emotional Comprehension (TEC) (Pons & Harris, 2000) and the Emotional Lexicon Test (ELT) (Grazzani, Ornaghi, & Piralli, 2009).
Both the TEC and the ELT are objective measures of the child’s performance related to Emotion Comprehension (EC) and Expression in specific social contexts.
Results:
Pre-treatment outcome measures were comparable between the two groups.
The RG displayed a significant improvement on TEC (U=47.0, W=75.0, Z=2.9, p=.001), with a gain of 59% in the total score. Conversely the CG, although improving on the TEC total score by 21%, did not show a significant change (U=32.5, W=60.5, Z=1.04, p=.16).
Likewise, post-treatment total scores on ELT increased by 48% in the RG and by 12% in the CG, showing a significant improvement in the RG (U=49.0, W=77.0, Z=3.4, p=.001), but not in the CG (U=36.5, W=64.5, Z=1.9, p=.06). Furthermore, following the intervention, all children in the RG, but none in the CG, reached the top scores on ELT (Table 3).
Conclusions:
Substantial improvements in contextualized emotion recognition, comprehension and emotional perspective taking through the use of human-assisted social robots were obtained.