Use of machine learning to improve autism screening and diagnostic instruments: effectiveness, efficiency, and multi-instrument fusion

• Published in: Journal of child psychology and psychiatry Vol. 57; no. 8; pp. 927 - 937
• Main Authors: Bone, Daniel, Bishop, Somer L., Black, Matthew P., Goodwin, Matthew S., Lord, Catherine, Narayanan, Shrikanth S.
• Format: Journal Article
• Language: English
• Published: England Blackwell Publishing Ltd 01.08.2016
• Subjects: Adolescent; Adult; Algorithms; Artificial intelligence; Autism; Autism Spectrum Disorder - diagnosis; Autism Spectrum Disorders; Autistic children; Bones; Caregivers; Child; Child & adolescent psychiatry; Child, Preschool; Clinical assessment; Clinical Diagnosis; Clinical research; diagnosis; Female; Humans; Learning; machine learning; Male; Medical diagnosis; Medical screening; Mental disorders; Middle Aged; Psychiatric Status Rating Scales; Responsiveness; screening; Sensitivity; Support Vector Machine; Translation; Validity; Young Adult; Autism; diagnosis; screening; machine learning
• ISSN: 0021-9630; 1469-7610
• DOI: 10.1111/jcpp.12559

Background Machine learning (ML) provides novel opportunities for human behavior research and clinical translation, yet its application can have noted pitfalls (Bone et al., 2015). In this work, we fastidiously utilize ML to derive autism spectrum disorder (ASD) instrument algorithms in an attempt to improve upon widely used ASD screening and diagnostic tools. Methods The data consisted of Autism Diagnostic Interview‐Revised (ADI‐R) and Social Responsiveness Scale (SRS) scores for 1,264 verbal individuals with ASD and 462 verbal individuals with non‐ASD developmental or psychiatric disorders, split at age 10. Algorithms were created via a robust ML classifier, support vector machine, while targeting best‐estimate clinical diagnosis of ASD versus non‐ASD. Parameter settings were tuned in multiple levels of cross‐validation. Results The created algorithms were more effective (higher performing) than the current algorithms, were tunable (sensitivity and specificity can be differentially weighted), and were more efficient (achieving near‐peak performance with five or fewer codes). Results from ML‐based fusion of ADI‐R and SRS are reported. We present a screener algorithm for below (above) age 10 that reached 89.2% (86.7%) sensitivity and 59.0% (53.4%) specificity with only five behavioral codes. Conclusions ML is useful for creating robust, customizable instrument algorithms. In a unique dataset comprised of controls with other difficulties, our findings highlight the limitations of current caregiver‐report instruments and indicate possible avenues for improving ASD screening and diagnostic tools.