A Randomized Controlled Trial of the Effectiveness of Pivotal Response Treatment with and without Use of a NAO Robot in Young Children with ASD

Background: Pivotal Response Treatment (PRT) is a promising intervention effective in promoting "pivotal skills" in children with autism spectrum disorder (ASD) (Bozkus-Genc & Yucesoy-Ozkan, 2016). However, evidence is lacking on which treatment components contribute to higher gains in generalized social communication skills and general clinical functioning and to diminished ASD symptoms. The use of robots within the treatment of ASD is hypothesized to contribute to higher treatment gains, since robots are intrinsically appealing to many children with ASD and robots may generate motivation for social interaction (Scassellati, 2007) and engagement in interventions (Boccanfuso et al., 2017). However, contemporary research on using robots in ASD interventions often lacks 1) methodological rigor (Begum, Serna, & Yanco, 2016), 2) evidence on generalization of treatment effects and follow-up (Pennisi et al., 2016) and 3) insight into how robots can be of additional value to established interventions (Diehl, Schmitt, Villano, & Crowell, 2012).

Objectives: The present study aims to examine the effectiveness of using a humanoid (NAO) robot within parent-mediated PRT for diminishing ASD-related symptoms and improving general clinical functioning in young children with ASD compared with PRT (without robot) and treatment-as-usual (TAU).

Methods: A randomized controlled trial (RCT) design was used to compare 1) parent-mediated PRT with using the NAO robot (PRT+robot), 2) parent-mediated PRT, and 3) TAU in 74 young children (3-8 years) with ASD. PRT consisted of 20 weekly therapy sessions, including 14 parent-child sessions and 6 parent sessions. In both PRT conditions, parents were trained in use of the PRT techniques and in the PRT+robot condition, the robot was added in all parent-child sessions to train the child with programmed scenarios incorporating PRT techniques. At baseline, week 10, 20 and 3-month follow-up, the child's ASD-related symptoms were assessed with the Social Responsiveness Scale (SRS, Roeyers, Thys, Druart, De Schryver, & Schittekatte, 2011) completed by parents and teachers. Clinically significant improvement was examined using the Clinical Global Impression-Improvement Scale (CGI-I, Guy, 1976) completed by a blinded child psychiatrist.

Results: Repeated measures analysis using the GLM approach indicated a significant main effect of time (F(3,147) = 14.53, p < .001) and a time x group interaction effect (F(6,147) = 2.78, p = .018) on the SRS completed by parents, reflecting steeper decrease of ASD-related symptoms over time in the PRT+robot condition (see Figure 1). Furthermore, the percentage clinical responders (much and very much improved on the CGI-I) was significantly higher in the PRT+robot condition (66.7%) compared with both the PRT (36.4%) and TAU condition (36.4%) (χ²(1) = 4.22, p = .040) at week 20, but not at follow-up (PRT+robot (63.2%) vs PRT (50.0%): χ²(1) = 0.67, p = .408; PRT+robot (63.2%) vs TAU (40.0%): χ²(1) = 2.09, p = .148).

Conclusions: Results of this study provide a first indication that the use of robotics in PRT may be an effective treatment component in diminishing ASD-related symptoms and improving general clinical functioning in young children with ASD. Implications for clinical use of robotics in PRT and future research will be discussed.