March 5, 2012
Internet
Researchers at Harvard Medical School was significantly reduced from hours to minutes the time required to accurately see autism in children.
The process of diagnosing autism complex, subjective, and often limited segment of the population in need. With the rising incidence of 1 in 88 children, the need for an accurate and widely deployable method for screening and diagnosis substantially. Dennis Wall, professor of pathology and director of computational biology initiative at the Center for Biomedical Informatics at Harvard Medical School, is working to resolve this issue and found a very accurate approach that can significantly reduce the complexity and time diagnostic process.
The wall developed algorithms and mechanisms to find autism-related spreading rapidly and with high accuracy. The algorithm is designed to work with mobile architecture, the integration of a small set of questions and a short video of the home of the subject, to allow rapid online analysis. This method can reduce the time to diagnosis of autism by nearly 95 percent, from hours to minutes, and can be easily integrated into routine practice of screening children for allowing a massive increase in the incidence of at-risk populations.
“We believe that this approach will allow more children to accurately assess during the early critical period when behavioral therapy is the most effective,” said Wall.
The research is published today in Nature: Translational Psychiatry.
Autism is diagnosed by a careful analysis of individual behavior. When children evaluated for autism, they usually take the Autism Diagnostic Interview, Revised, known as the ADI-R, 93-question questionnaire, and / or the Autism Diagnostic Observation Schedule, which is known as the ADOS test, which measures the behavior of some children. Altogether, the analysis can take up to three hours to complete and must be administered by a trained physician. Often, there is a delay of over a year between the early warning signs and diagnosis because of the waiting time to see a professional clinical tests can be administered and provide a formal diagnosis, said Wall .
Using machine learning approach, methods of artificial intelligence where machines are trained to make decisions, Wall and his team studied the results of the ADI-R Autism Research Genetics Stock Exchange for more than 800 people who was diagnosed with autism to find redundancies in the exam. They found that seven questions alone are enough to diagnose autism with nearly 100 percent accuracy, which is equivalent to a 93 question quiz over. They verified the accuracy of a survey of seven questions with responses from more than 1,600 people from the Simons Foundation and more than 300 people from Boston Autism Consortium.
Wall to apply the same approach to the ADOS test, this time in the classification of more than 1,050 people with near perfect sensitivity and slightly less than 95 percent specificity. The result of this work is not only a mechanism for evaluation of short children (8 of 29 measured), but also a roadmap for evaluating short video clip at home. Together these results have the potential to move the majority of the effort to mobilize health framework with a wide range of electronics and applications.
“This approach is a first attempt to retrospectively analyze large repositories of data in order to obtain a highly accurate tool for classification, but significantly shortened,” said Wall, who is also professor of pathology at Beth Israel Deaconess Medical Center. “This type of analysis should quickly make an important contribution to the diagnostic process to move forward and help lead to rapid screening and early treatment,” he said.
Survey traditional diagnostic for autism may be prohibitive for families and caregivers because they are long and should be administered by a licensed physician, often in a foreign environment for children, which can be a huge burden for families in remote areas, said Wall. “With this strategy mobilized, the parent or guardian is an important first step for diagnosis and treatment from the comfort of their own homes, and in just a few minutes.”
Today, the Wall has made a survey and video sites available to the public free to continue to analyze the effectiveness of the approach and shortened working on ways to mobilize the whole approach to enable a wide range of population needs. His team also launched Facebook page to spread the word and share the broader survey. To date, 2,500 people took the survey Autworks.