Evaluating the Usability of a Social Skills Training App for Children with ASD

Friday, May 13, 2016: 10:00 AM-1:30 PM
Hall A (Baltimore Convention Center)
B. G. Kinsella1, A. Kushki2 and S. Chow1, (1)Autism Research Centre, Holland Bloorview Kids Rehabilitation Hospital, Toronto, ON, Canada, (2)Bloorview Research Institute, Toronto, ON, Canada

Deficits in social communication are defining features of autism spectrum disorder (ASD). These deficits negatively have been associated with long-term negative outcomes such as unemployment and under-employment, low rates of independent living, and increased risk of psychiatric disorders. While prognosis can be significantly improved with intervention, few evidence-based interventions exist for social skill deficits in ASD. Existing interventions are resource-intensive, their outcomes vary widely for different individuals, and they often do not generalize to new contexts.

Technology-aided intervention is a motivating, low-cost, and versatile approach for social skills training in ASD. It can provide a safe and controlled venue for rehearsal of skills in a self-paced and personalized manner, allow for treatments to be implemented with high precision and fidelity, and reduce the cost, resource, and other accessibility barriers to existing treatments. Although early studies support the feasibility and potential effectiveness of technology-aided intervention for social skills training in ASD, existing approaches have been criticized for teaching skills based on human-computer interaction.  This can paradoxically lead to further social isolation and hinder skill generalizability.


A Google Glass-based application (“app”) was developed to help guide children with ASD through social interaction.  The Glass is a head-mounted wearable technology that uses a microphone to listen to its environment and an optical heads-up display (HUD) to prompt and interact with the user.  The app coaches a child through human-to-human interaction by listening to the conversation and providing appropriate responses as prompts on the HUD.  Before testing clinical efficacy, the objective of this study was to determine the technology’s viability as an appropriate medium.


10 children (mean age=13.2 σ=1.5, verbal IQ=104.4 σ=20.9, Social Communication Questionnaire score=19 σ=3.5) with ASD were recruited to be participants.  During each session, the participant used the app while engaging in a structured restaurant-themed interaction with a research assistant.  The app was evaluated on its effectiveness (i.e. how accurately the app responds), efficiency (i.e. how quickly the user and the app), and user satisfaction (based on a post-session questionnaire that includes a 5-point Likert scale).


The application’s detection accuracy (i.e. how often it detected a participants’ utterances) was 97%, and recognition accuracy (i.e. how accurately it recognized the participants’ phrases) was 89%.  The users’ average response time after being prompted was 2.5 seconds, and the system’s average response time to recognize the user’s speech was -0.25 seconds.  User feedback on perceived experience of system use was collected and compiled.  Notable responses include the participants believe the app can help them in daily life (4.0/5), the app was fun to use (4.4/5), and the app can help children communicate to people (4.0/5).


To the best of our knowledge, this system is the first technology-based intervention for ASD that employs human-to-human coaching in naturalistic settings, and the results show that the device is an appropriate medium for treatment delivery.  Future directions will include improvements to the prototype by incorporating relevant usability feedback, and a study to evaluate the clinical effectiveness of the app as a social skills intervention.