Autism Spectrum Disorder Screening: Machine Learning Adaptation and DSM-5 Fulfillment

Abstract

One of the primary psychiatric disorders is Autistic Spectrum Disorder (ASD). ASD is a mental disorder that limits the use of linguistic, communicative, cognitive, skills as well as social skills and abilities. Recently, ASD has been studied in the behavioural sciences using intelligent methods based around machine learning to speed up the screening time or to improve sensitivity, specificity or accuracy of the diagnosis process. Machine learning considers the ASD diagnosis problem as a classification task in which predictive models are built based on historical cases and controls. These models are supposed to be plugged into a screening tool to accomplish one or more of the aforementioned goals. In this paper, we shed light on recent studies that employ machine learning in ASD classification in order to discuss their pros and cons. Moreover, we highlight a noticeable problem associated with current ASD screening tools; the reliability of these tools using the DSM-IV rather than the DSM-5 manual. Hence the necessity to amend current screening tools to reflect the new imposed criteria of ASD classification in the DSM-5 particularly the diagnostic algorithms embedded within these methods.