AutoDisorderID

PI: Kim Oller

Is it possible to use completely objective means to identify disorders such as autism or language delay? And especially, can such means be used for early identification in the first months of life?

In collaboration with a variety of universities and the LENA Research Foundation (Boulder, CO), we are developing strategies for automated analysis of all-day audio recordings conducted in infants' and children's homes. The method is totally automated and can discriminate groups of children with and without significant disorders such as autism, and the procedure requires no human intervention, making it a truly objective method of identification. The research considers vocal characteristics of the infant or child as well as patterns of interaction among parents, siblings and the infant or child wearing the battery-powered recorder. The research is also expected to inform research on vocal development more generally because the method allows the analysis of massive datasets of recording, including the 80,000 hour dataset of recordings from children's homes at the LENA Research Foundation, and many additional datasets under development currently.