The incidence of autism has increased dramatically over recent years, making this mental disorder one of the greatest public health challenges of our time. The standard practice of diagnosis is strictly based on behavioral characteristics, as the genome has largely proved intractable for diagnostic purposes. Yet, the most commonly used behavioral instruments take as much as 3 hours to administer by a trained specialist, contributing to the substantial delays in diagnosis experienced by many children, who may go undiagnosed and untreated until ages beyond when behavioral therapy would have had more substantive positive impacts.
Objectives:
In the present study, our aim was to apply machine learning techniques to one of the most commonly used behavioral instruments, the Autism Diagnostic Interview-Revised (ADI-R), to determine if the exam could be shortened without loss of diagnostic accuracy.
Methods:
We used several machine-learning techniques to study the complete sets of answers to the ADI-R available at the Autism Genetic Research Exchange (AGRE) for 891 individuals diagnosed with autism and 75 individuals who did not meet the criteria for autism diagnosis. Through cross-validation we measured the sensitivity and specificity of the classifier and then further tested the accuracy against item-level data from two independent sources, a collection of 1654 autistic individuals from the Simons Simplex Collection and a collection of 322 autistic individuals from the Boston Autism Consortium.
Results:
Our analysis showed that 7 of the 152 items contained in the ADI-R were sufficient to diagnosis autism with 99.9% statistical accuracy. In both our external validation experiments, the 7-question classifier performed with nearly 100% statistical accuracy, properly categorizing all but one of the individuals from these two resources who previously had been diagnosed with autism through the standard ADI-R.
Conclusions:
With incidence rates rising, the capacity to diagnose autism quickly and effectively requires careful design of behavioral diagnostics. Our retrospective analysis yielded a highly accurate, but significantly abbreviated diagnostic instrument that appears to capture the key elements of the ADI-R while reducing the exam time from hours to minutes. Although more testing is required, this abbreviated approach may prove useful for initial screening and faster recognition in clinical settings as well as in mobile technologies to enable administration in remote areas.
See more of: Clinical Phenotype
See more of: Symptoms, Diagnosis & Phenotype