New research could potentially lead to shortened screening and diagnostic processes being used to assess people for autism according to a paper by Duda and colleagues* based at Stanford University in the United States. Based on the application of machine learning, the same technology which aids internet searches or flags spam email from genuine emails, researchers reported that a classifying computer program was able to quite reliably distinguish cases falling on the autism spectrum from non-spectrum controls.
Drawing on previous work carried out by the authors in this area, researchers analysed ADOS (Autism Diagnostic Observation Scale) scores for over 2600 participants derived from various datasets. Using eight specific behaviours connected to autism “including eye contact, imaginative play and reciprocal communication” they reported that their classifier – the observation based classifier (OBC) – was able to pick out cases of autism “with >95% statistical accuracy”. They concluded that their results “support the hypothesis that the OBC has both high recall and precision in a reasonably diverse sample of cases with and without autism, including children with other learning and developmental delays”.
This research group have previously talked about analysing home videos using non-clinical raters to enhance screening and detection rates for possible autism in light of the growing numbers of people being diagnosed as on the autism spectrum. Coupled to their machine learning program, they suggest that greater research efforts are needed to further examine such a strategy so “potentially enabling more families to receive care far earlier and during timeframes when interventions have the most positive benefits”.
* Duda A. et al. Testing the accuracy of an observation-based classifier for rapid detection of autism risk. Translational Psychiatry. 2014; 4: e424.
Read more about this research at: http://questioning-answers.blogspot.com/2014/09/an-observation-based-classifier-for-rapid-detection-of-autism-risk.html