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

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 (DD), split at age 10. Algorithms were created via a robust ML classifier, support vector machine (SVM), while targeting best-estimate clinical diagnosis of ASD vs. non-ASD. Parameter settings were tuned in multiple levels of cross-validation. Results The created algorithms were more effective (higher performing) than 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 limitations of current caregiver-report instruments and indicate possible avenues for improving ASD screening and diagnostic tools.

ASD Validity

“…Consequently, there is insufficient coverage for ASD construct validity based on unique ASD symptom co-expression. Some researchers have proposed to make ASD more homogeneous by refining ASD diagnostic criteria Sonuga-Barke 2016) and by developing more sensitive ASD diagnostic screening instruments (Bone et al 2016). However, it is unlikely that the many nondiagnostic symptoms such as ADHD, ID, epilepsy, and language impairment that occur with ASD that stem from varied ASD brain impairments caused by varied ASD risk factors (Kida and Kato 2015) will be eliminated by refinement of the ASD criteria or refinement of ASD screening measures.…”

Voice Patterns in Schizophrenia: A systematic Review and Bayesian Meta-Analysis

“…The results are promising (Bonneh, Levanon, Dean-Pardo, Lossos, & Adini, 2011;Faurholt-Jepsen et al, 2016;Martínez-Sánchez et al, 2015;Rapcan et al, 2010;Tsanas et al, 2011;7 VOICE IN SCHIZOPHRENIA: REVIEW AND META-ANALYSIS Williams et al, 2014), but a complete and comparative overview of the findings in schizophrenia is currently missing. Crucially, the reliability of ML results has been shown to be strongly dependent on the availability of large datasets and the validation of the findings across datasets (Bone et al, 2016;Chekroud, 2018;Foody, 2017;James et al, 2013;Van Der Ploeg et al, 2014), which presence we wanted to assess in the literature on voice in schizophrenia.…”

Meta-Learning for Robust Child-Adult Classification from Speech

“…Child-adult interactions have been used in the ASD domain primarily for diagnosis (ADOS [5]) and measuring intervention response (BOSCC [6]). Automated computational processing of the participants' audio [7] and language streams [8] has provided objective descriptions that characterize the session progress and understanding the relation with symptom severity.…”