Machine Learning Diagnostic Classifiers
We developed two classifiers using decision tree learning that performed optimally for classification of a wide range of individuals both on and off the autism spectrum. These classifiers, the Caregiver-Directed Classifier and the Video-Based Classifier are substantially shorter than the gold standard diagnostic instruments in the field today and both classifiers pinpoint several behavioral patterns that could guide future methods for expeditious observation-based screening and diagnosis in and out of clinical settings.
To evaluate the degree of redundancy of the ADOS and ADIR and if so determine whether a reduced set of uncorrelated features could correctly classify individuals with the same accuracy as the gold-standard diagnostic tests.
- DP Wall, Dally R, Luyster R, Jung JY, DeLuca TF. Use of Artificial Intelligence to Shorten the Behavioral Diagnosis of Autism. PLoS ONE. 2012 Aug 27.
- DP Wall, Kosmicki J, DeLuca TF, Harstad E, Fusaro VA. Use of machine learning to shorten observation-based screening and diagnosis of autism. Transl Psychiatry. 2012 Apr 10.
Related, we have developed machine learning classifiers to distinguish ASD children from typically-developing children, using feature extraction and sparsity-enforcing classifiers in order to find feature sets from ADOS (modules 2 and 3).
- J.A. Kosmicki, V. Sochat, M. Duda, and D.P. Wall. Searching for a minimal set of behaviors for autism detection through feature selection-based machine learning. Translational Psychiatry. February 2015; doi:10.1038/tp.2015.7.
Using archival data, we used feature selection and classification techniques to choose a subset of features on the Social Responsiveness Scale (SRS) from which we can distinguish ASD and ADHD. In work in submission, this method has been validated on data gathered through a mobile survey system.
- Duda, M., Ma, R., Haber, N., & Wall, D.P. (2016). Use of machine learning for behavioral distinction of autism and ADHD. Translational Psychiatry, 6(732), doi: 10.1038/tp2015.221. Published online 9 February 2016.