University of Southern California, Los Angeles, CA

September 28, 2012

The Engineering and Autism Workshop was workshop at USC that brought together engineering researchers, especially in the broad domain of social/affective and communicative behavioral computing, with leading Autism researchers and clinicians as well as the broader community of stakeholders including families. It aimed to highlight and explore the opportunities and possibilities of engineering and computing advances in supporting research and translation in Autism by creating synergies and partnerships by initiating a multidisciplinary dialog.

Agenda

Welcome Remarks
Shrikanth S. (Shri) Narayanan
Professor of Electrical Engineering & Computer Science - University of Southern California

Computational behavioral sciences: From behavioral signal processing to behavioral informatics
Jim Rehg
Professor of Interactive Computing - Georgia Institute of Technology
David Forsyth
Professor of Computer Science - University of Illinois-Urbana Champaign
Shri Narayanan
Professor of Electrical Engineering & Computer Science - University of Southern California

Early detection of autism: Challenges and approaches
Wendy Stone
Ph.D., Susan and Richard Fade Endowed Chair and Director University of Washington Autism Center
Abstract Early detection of autism is the key to specialized services that can lead to significant gains in children’s social, language, and behavioral functioning. Yet early identification efforts are often hampered by great variability in behavioral expression, limited knowledge about underlying mechanisms, and inconsistent community screening practices. This presentation will describe the early behavioral features of autism, the challenges of early identification, and potential strategies for facilitating detection and improving outcomes in these children.

Three technology highlights (engineering PhD students)
Daniel Bone - University of Southern California
Yin Li - Georgia Institute of Technology
Javier Hernandez Rivera - Massachusetts Institute of Technology

Engaging autism: Developmental implications for intervention
Connie Kasari
Professor of Psychological Studies in Education & Psychiatry - UCLA
Abstract Autism is a serious, neurodevelopmental disorder that changes with development. Developmentally appropriate and behaviorally based interventions have yielded impressive outcomes for children’s social, communication and language development. In this talk, developmental change in core deficits of children with autism will be described and results presented for a series of targeted interventions for core deficits. Particular focus will be on underserved, under-represented and under-resourced children with autism and their families as well as randomized controlled trials implemented in natural environments of homes and schools.

Human-centered approaches to technologies for autism
Gillian Hayes
Assistant Professor in Informatics in the School of Information and Computer Sciences - UC Irvine
Abstract Interactive technologies have the potential to support both children and adults with ASD as well as friends, family, teachers, therapists, and researchers. However, appropriate design of these technologies requires considering the concerns of a wide variety of people impacted by them, the environments in which they will be used, and the potential to gather data about their efficacy. Human-centered approaches to design and development of novel interactive technologies can bridge the divides between engineering research and family and clinical practices. In this talk, I will describe a variety of research projects focused on interactive technologies for autism and outline some considerations for a human-centered approach for autism technology research.

Three technology highlights (engineering PhD students)
Rahul Gupta - University of Southern California
Denis Lantsman - Georgia Institute of Technology
Gabriela Marcu, Carnegie Mellon University

Strategies for tackling the challenges of autism heterogeneity
Pat Levitt
Provost Professor of Neuroscience, Pharmacy, Psychiatry, Pediatrics and Psychology - University of Southern California
Abstract A significant challenge in improving the clinical manifestations of autism spectrum disorder (ASD) has been to determine the underlying mechanisms that drive disorder heterogeneity. Heterogeneity in children can be defined on many different levels, including the core ASD symptoms, genetic and other biomarkers, and co-occurring conditions that impact mental and physical health. Two different approaches to examining the underlying features of ASD that may help stratify populations functionally will be presented. First, using neuroimaging and genetics methods, studies will be described that demonstrate a way forward in characterizing individuals with ASD who have similar functional and structural brain patterns that stratify with a genetic marker in the ASD risk gene MET. Second, co-occurring medical conditions often occur in children with ASD. Using hierarchical statistical clustering, subsets of more homogeneous groups of children can be identified with common patterns of medical symptoms. We have found that a certain subgroup, those with ASD and gastrointestinal disturbances, are also enriched in children with minimal verbal abilities compared to ASD alone. Our goal of stratification based on rich clinical and neurobiological characterization also is amendable to further investigation using signal processing methods of audio-visual data from child-clinician interactions. When applied to a sufficiently large cohort, the application of this engineering methodology may be able to detect subtle, yet consistent patterns of communication and social interaction that may be important for developing predictive models of best practices in treating children with ASD.

Automated monitoring of vocal development and early detection of autism
D. Kimbrough (Kim) Oller
Professor and Plough Chair of Excellence - University of Memphis
Abstract It has recently been demonstrated (Oller et al. 2010) that audio recordings of infants and young children in their natural environments can be analyzed with automated methods to determine the child’s stage of development. Using a model based on research in infant vocal development, we assessed 12 acoustic parameters with no human intervention and determined that we could predict infant age from 2-48 months with a high degree of accuracy (predicted age and real age correlated at about 0.8). Further we showed that children with autism or language delay could be differentiated from typically developing children with high accuracy. The research was a collaboration with the LENA Research Foundation, and much subsequent work is underway utilizing modifications on the same approach. The presentation for the conference will bring this line of research up to date by reviewing the new lines of effort extending the published findings.
Oller, D. K., Niyogi, P., S. Gray, J. A. Richards, J. Gilkerson, D. Xu, U. Yapanel, S. F. Warren (2010). Automated Vocal Analysis of Naturalistic Recordings from Children with Autism, Language Delay and Typical Development. Proceedings of the National Academy of Sciences, 107, 30, 13354-13359.

Proposed application of automated computational analysis to the early detection of autism spectrum disorders
Ted Hutman
Assistant Professor of Psychiatry & Biobehavioral Science David Geffen School of Medicine - UCLA
Abstract Because Autism Spectrum Disorders (ASDs) are generally not diagnosed until the fourth year, relatively little is known about their overt manifestations during infancy or the neural mechanisms that underlie them. In my lab, we are conducting longitudinal evaluations of infant siblings of children with ASD because these infants are roughly twenty times more likely to be diagnosed with ASD than low-risk controls. We are tracking the infants’ social, cognitive, motor, and physical development from 6 weeks to three years of age, at which time clinical diagnoses are considered stable. Studies in our lab and elsewhere find few behavioral signs of ASD at 6 months, but several robust markers by 12 months. Most of the atypical behaviors associated with ASD at 12 months are social in nature. These include imitation, orienting to faces, initiating and responding to attentiondirecting behaviors, and responding to distress. After I review the findings of this research program, I will describe the methodology and our conceptual approach to studying highrisk infants. I will propose that diagnostic groups can be differentiated earlier, group differences can be magnified, and more children can be screened by applying automated computational analysis to video footage of a wide range of infant behavior. Behavioral targets of interest include vocalization, gesture, gross motor activity, affective displays, and social orienting.

Discussion & wrap up
Panelists
Gregory Abowd - Georgia Institute of Technology
Portia Iversen - The Descartes Institute
Clara LaJonchere - Autism Speaks
Olga Solomon - University of Southern California
Facilitators
Matthew Goodwin - Northeastern University
Agata Rozga - Georgia Institute of Technology

Organized by:

Shri Narayanan University of Southern California

Matthew Goodwin Northeastern University

Gregory Abowd Georgia Institute of Technology

Agata Rozga Georgia Institute of Technology

Danielle Hamra University of Southern California