Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 57 (8), 927-37

Use of Machine Learning to Improve Autism Screening and Diagnostic Instruments: Effectiveness, Efficiency, and Multi-Instrument Fusion

Affiliations

Use of Machine Learning to Improve Autism Screening and Diagnostic Instruments: Effectiveness, Efficiency, and Multi-Instrument Fusion

Daniel Bone et al. J Child Psychol Psychiatry.

Abstract

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.

Keywords: Autism; diagnosis; machine learning; screening.

Conflict of interest statement

Conflict of Interest Statement: See Acknowledgments for disclosures.

Figures

Figure 1
Figure 1
Flow diagram of ML-based algorithm development
Figure 2
Figure 2
Illustration of model training, tuning, and testing through ‘nested’ cross-validation (CV) as used in the ‘effective algorithms’ section
Figure 3
Figure 3
Receiver operating characteristic plots. The Equal Error Rate (EER) Line indicates the UAR optimization point, where sensitivity and specificity are weighted equally. Classifiers should perform above the Chance Line, where UAR equals 50%. Note that we plot sensitivity vs. specificity in order to aid interpretation relative to UAR
Figure 4
Figure 4
Optimization curves versus number of codes for Age 10- (top) and Age 10+ (bottom) screeners. Optimization is biased towards sensitivity (roughly 2:1). An elbow-point at 95% of maximum performance is marked for both age groups

Similar articles

See all similar articles

Cited by 9 PubMed Central articles

See all "Cited by" articles

Publication types

Feedback