(HealthDay News) – The use of a short eight-item classifier identifies individuals with autism with high sensitivity and specificity.
Dennis P. Wall, PhD, from Harvard Medical School in Boston, and colleagues used a series of machine-learning algorithms to investigate the accuracy of autism classification. A complete set of scores from the 29-item Module 1 of the Autism Diagnostic Observation Schedule-Generic (ADOS) was compared for 612 individuals with a classification of autism and 15 non-spectrum individuals.
The researchers found that the use of eight of the 29 items from Module 1 was sufficient to classify autism with 100% accuracy. The accuracy of this classifier was validated against complete sets of scores from two cohorts: 110 individuals with autism from the Boston Autism Consortium and 336 individuals with autism from the Simons Foundation. The classifier performed with nearly 100% sensitivity, correctly classifying all but two individuals with autism, and with 94% specificity against observed and simulated non-spectrum controls. Several elements from the ADOS were found in the classifier, demonstrating high test validity and resulting in a quantitative score that measured classification confidence and severity.
“Using machine-learning algorithms, we found the alternating decision tree to perform with almost perfect sensitivity, specificity, and accuracy in distinguishing individuals with autism from individuals without autism,” the authors write. “Our findings may help to guide future efforts, chiefly including mobile health approaches, to shorten the evaluation and diagnosis process overall such that families can receive care earlier than under current diagnostic modalities.”