Assessing Social Communication and Collaboration in Autism Spectrum Disorder Using Intelligent Collaborative Virtual Environments.
Lian Zhang, Amy S Weitlauf, Ashwaq Zaini Amat and 3 others
PMID 31583625WHAT IT FOUND
A virtual game system with an AI agent tracked children's speech and teamwork.
It matched human ratings for some behaviors but missed others, like directing movement, and failed to capture rare requests.
Key findings
01The system accurately classified dialogue acts at rates much higher than random chance across both human-agent and human-human interactions.
02System-generated failure frequency showed a strong negative correlation with human ratings of collaboration and communication for children with ASD.
03Features related to requesting color or object were too infrequent to be accurately measured and were dropped from the analysis.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study was very small, with only 20 pairs of children, and participants were not fully characterized by current cognitive or language levels. The system missed or inaccurately measured several key communication behaviors, particularly directing movement and making requests. The study used only one session and a limited set of goal-focused puzzle games, which may not reflect natural social interaction. Inter-rater reliability between the human judges was only moderate.
Declared interests
All authors declared no conflicts of interest.
The easy way to misread this
Do not assume this system can replace clinical assessment of social communication. It failed to accurately measure several important pragmatic language behaviors, such as directing movement and making requests, and its overall accuracy varied significantly depending on the specific skill being measured.