A data driven machine learning approach to differentiate between autism spectrum disorder and attention-deficit/hyperactivity disorder based on the best-practice diagnostic instruments for autism.

Researchers used computer analysis to find which parts of autism assessment tools are most important for diagnosis. They discovered that just 5 key questions from each of the main autism tests can accurately identify autism, even when ADHD is also present. This could make autism diagnosis faster and more focused, using 92% fewer questions while still being accurate.

This study used machine learning to identify which items from standard autism diagnostic tools (ADOS and ADI-R) best distinguish between autism spectrum disorder (ASD), ADHD, combined ASD+ADHD, and no diagnosis groups. Researchers found that just five features from each assessment tool could reliably differentiate ASD groups from non-ASD groups, representing a 92% reduction in required items while maintaining high diagnostic accuracy. The distinguishing features related to both social-communication difficulties and restrictive/repetitive behaviors. This approach could potentially streamline autism diagnosis, particularly in complex cases where ASD and ADHD co-occur, by focusing assessment on the most diagnostically relevant items.

Results suggest potential for developing streamlined autism diagnostic protocols that focus on most discriminating features. Could improve diagnostic efficiency in specialized clinics, particularly for complex cases with co-occurring ADHD. However, further validation studies needed before clinical implementation.

Sample size not reported in abstract. Study design unclear. Machine learning validation methods not specified. Generalizability to broader clinical populations uncertain. No information provided about demographic characteristics of participants or replication of findings.

Autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD) are two frequently co-occurring neurodevelopmental conditions that share certain symptomatology, including social difficulties. This presents practitioners with challenging (differential) diagnostic considerations, particularly in clinically more complex cases with co-occurring ASD and ADHD. Therefore, the primary aim of the current study was to apply a data-driven machine learning approach (support vector machine) to determine whether and which items from the best-practice clinical instruments for diagnosing ASD (ADOS, ADI-R) would best differentiate between four groups of individuals referred to specialized ASD clinics (i.e., ASD, ADHD, ASD + ADHD, ND = no diagnosis). We found that a subset of five features from both ADOS (clinical observation) and ADI-R (parental interview) reliably differentiated between ASD groups (ASD & ASD + ADHD) and non-ASD groups (ADHD & ND), and these features corresponded to the social-communication but also restrictive and repetitive behavior domains.

In conclusion, the results of the current study support the idea that detecting ASD in individuals with suspected signs of the diagnosis, including those with co-occurring ADHD, is possible with considerably fewer items relative to the original ADOS/2 and ADI-R algorithms (i.e., 92% item reduction) while preserving relatively high diagnostic accuracy. Clinical implications and study limitations are discussed.