Skip to main content

Spectrum: Autism Research News

Database groups common concepts in autism tests

by  /  27 October 2010

This article is more than five years old. Autism research — and science in general — is constantly evolving, so older articles may contain information or theories that have been reevaluated since their original publication date.

A. McCray Connect four: Categorizing questions from different autism tests may help researchers and ease the burden on families of children with autism.

A. McCray

Connect four: Categorizing questions from different autism tests may help researchers and ease the burden on families of children with autism.

A searchable new database will greatly ease the task of comparing results from more than 25 diagnostic tests for autism, by creating clusters of the various symptoms measured.

Alexa McCray, co-director of the Center for Biomedical Informatics at Harvard Medical School, introduced the tool yesterday at the Autism Consortium 2010 Symposium in Boston.

“What this does is give us a common language, where no common language has existed,” says Gregory D. Abowd, professor of interactive computing at Georgia Institute of Technology, and the father of two children with autism. Abowd is not involved with the database and was initially skeptical about its utility.

As a parent, he says, “my frustration is some of these [tests] don’t need to be as exhaustive as they are. There has to be a more cost-effective way to get the information.”

The new Autism Phenotype Ontology aggregates data from various tests that measure the cognitive, behavioral and biological features of autism. Combining these clinical features into one database allows researchers to share their data and potentially ease the burden of participating families.

Families often have to answer a dozen versions of the same question — ‘Does your child bang his head?’ for example — posed on numerous tests. At the same time, researchers must comb through thousands of questions on various instruments — tests such as the Autism Diagnostic Observation Scale and the Pediatric Quality of Life Inventory — to extract data.

To create the database, the researchers used information from more than 425 families who contributed medical histories, test results and blood samples.

They gathered more than 5,000 questions from different diagnostic tests, clustering questions together into three main concepts — personal traits, social competence and medical history — and several subgroups of these categories.

Personal traits, for example, include cognitive ability, executive functioning, language ability, behavior regulation and emotional traits. Each of those subgroups branches into more specific terms.

For example, executive functioning includes emotional regulation and control, inhibition, mental flexibility, planning, task performance and working memory. At the next level, emotional regulation and control branches into emotional reactions, and impulse control and regulation. At each level, there are questions on various testing instruments that measure that concept.

Quick queries:

Researchers can search categories by instrument — for example, all questions on the Child Behavior Checklist about aggression — or by concept, for instance, all questions on every test about emotional regulation and control.

The query tool permits scientists to select the information they need without having to dig through the results of each test.

“For people conducting studies looking at the correlation between genetic variance and particular features of autism, it will be very helpful,” says David T. Miller, clinical geneticist at Children’s Hospital in Boston.

Say, for example, a geneticist is interested in which tests measure language development and how an individual with a particular genetic profile scores on various measures. The researcher can either conduct a general query on that individual’s verbal ability or a more specific one on different aspects of language development.

Clinical researchers, by contrast, might be more interested in the overlap among the instruments to create a more efficient battery of tests.

Knowing where overlaps occur might eventually winnow down the number of tests, says McCray, who also created, a comprehensive database of clinical trials worldwide.

“The ontology is a very good starting point for asking if we need as many instruments as we have and, if not, how do we get to a smaller number,” she says.

A linguist and computer scientist, McCray was chief of the research and development laboratory at the National Library of Medicine before moving to Harvard. She has been a member of the Autism Consortium since it was founded in 2006 and chairs the committee on bioinformatics.

McCray plans to make the ontology freely available when it is completed next year. “I’m all about open access,” she says. The National Database for Autism Research is likely to make the ontology available on its website.

For all reports from the Autism Consortium Symposium 2010, click here.