Improved early detection of autism using novel statistical methodology

Completed

Campbell, Daniel

Chawarska, Katarzyna

Yale University

$102,846.00

2 years

Meixner Translational Postdoctoral Fellowship

New Haven

CT

United States

2011

http://www.yale.edu

City: 
New Haven
State/Province: 
CT
State/Province Full: 
Connecticut
Country: 
United States

In this age of multidisciplinary research, a wide variety of different types of data is being collected to help understand the etiology and development of autism spectrum disorders. Data as diverse as behavioral and cognitive assessments, physical growth features, and gaze and attention patterns are frequently gathered, and at younger and younger ages. However, the interaction between all these types of data with clinical outcome and with each other is very complex and continuously changing during early development. This complexity resists many straightforward attempts to analyze this wealth of data, and this is why the use of complex statistical methods and techniques can be crucial to understanding the development of autism in young children, and to identify features that can predict the occurrence of autism at a later age. Building on previous work on novel statistical techniques, the fellowship will focus on new techniques to classify adults with autism from typically-developing peers using eye-tracking data from watching a movie have been successful. Analyzing eye-tracking data from static images in new ways also shows promise at identifying patterns and properties of gaze that will predict whether an individual will develop autism or not. Additionally, using sophisticated statistical methods creatively has helped show that children with autism do not just have larger head sizes by the first year of life, but also an increase in overall body size. In all these cases, a statistical background can help in devising new ways of analyzing rich, complex data, and to extract meaningful quantitative measures that can indicate the presence or absence of autism. This postdoctoral fellow will extend these new methods to eye-tracking data from infants and toddlers at high-risk for developing autism, combining this data with morphological trajectory information, refining these techniques when needed, in order to identify quantitative markers that can predict development of autism as early as possible. The fellowship will also integrate eye-tracking methods and techniques, the etiology of autism, and the behavioral and clinical assessment of autism. These skills will inform statistical thinking by providing insights into better analyses, and a wider breadth of knowledge and expertise with which to contribute to interdisciplinary autism research.