10/26/2010
Written by
Two computer science professors from the University of Illinois at Urbana-Champaign are part of a $10 million effort to develop new computing techniques and approaches for identifying children at risk for autism and other developmental delays. As part of the effort, Professors Karrie Karahalios and David Forsyth will develop novel computational methods to measure and analyze the behavior of children and adults during face-to-face social interactions. The two will be working with researchers from Georgia Tech , Carnegie Mellon, MIT, Boston University, the University of Pittsburgh, the University of South California as part of an NSF Expeditions grant led by Georgia Tech.
Autism affects 1 in 110 children in the U.S., with a lifetime cost of care of $3.2 million per person. Current best practices for evaluating behavior and assessing risk are based on direct observation by highly-trained specialists. It’s likely early ASD diagnoses lead to better treatment outcomes, but screening young children is difficult, and so many children are not screened.
Enter computing. The team hopes that by developing methods to automatically collect fine-grained behavioral data, this project will enable large-scale objective screening and more effective delivery and assessment of therapy. The project will develop multiple sensing technologies, including vision, speech, and wearable sensors, to obtain a comprehensive, integrated portrait of expressed behavior.
“People can show they’re engaged with gaze, gestures, expressions, body posture, and tone of voice,” said Forsyth, “so the project will use many different sensing modes to get a comprehensive portrait of what subjects are doing.”
The project will be developing unique capabilities for synchronizing multiple sensor streams including cameras, microphones, and wearable sensors, and then using these streams to measure behavior. “We’ll develop methods to model interactions between people,” said Forsyth, “to help screen for ASD and help therapists know the best time to encourage or discourage a behavior.”
Karahalios will expand her existing research into the use of speech visualization technologies and techniques to teach children with autism spectrum disorder how to create speech. To date, Karahalios and her team have visualized vocalization in the form of utterances, phonemes, and syllables in real time. Children and clinicians use the visual feedback to modify their vocal behavior, and to provide a metric for improvement.
The goal of her new effort is to move from utterances and syllables to words and phrases. Additionally, the team will develop visualizations to aggregate data from doctor’s visits for early diagnosis of Autism Spectrum Disorder (ASD) by visualizing performance of an evaluation metric called Rapid ABC.
“There are many ways that technology can augment diagnosis and therapy. One of the things we stress is that we do not want the technology to replace the clinicians or therapists, but rather to aid them in their work and provide a context. Visualizations of mass data sets may reveal patterns that are get unknown, provide new insights to parents and clinicians alike.”
The long-term goal of the overall project is the creation of a new scientific discipline of computational behavioral science, which draws equally from computer science and psychology in order to transform the study of human behavior.
“We hope to build technologies that will allow us to observe how people behave in a wide range of settings for long periods of time,” said Forsyth. “Data could help everyone from architects, who might build better buildings, to marketers, to therapists.”
To support their long term goal, the research team plans comprehensive educational activities, including a new interdisciplinary summer school for young researchers and the development of new courses in computational behavior. In addition, the team will be extending their on-going and significant collaborations with major autism research centers in Atlanta, Boston, Pittsburgh, Urbana-Champaign, and Los Angeles.