PROJECT SUMMARY Ongoing development and validation of screening and diagnostic tools has been a major focus of research in autism spectrum disorders (ASD) during the past 30 years. Diagnostic validity of several widely used tools has been established by showing that, in most cases, children with ASD score above established cut-offs, whereas children with non-ASD score below cut-offs. However, a growing body of literature indicates that sensitivity and specificity of ASD symptom measures varies significantly based on the characteristics of the study population. This means that the ability of a given tool to differentiate children with and without ASD is variable, and is affected by an individual's demographic (age, sex), developmental (cognitive ability, language level), and/or behavioral (clinically significant behavior problems) profile. The long-term goal of the proposed research is to transform screening and diagnostic practices through increased individualization of measure selection and interpretation. Phenotypic heterogeneity of youth referred for possible ASD is a well-recognized challenge that prevents one-size-fits-all assessment approaches from being validly employed. Yet, with ever-growing demands for such services, clinical and research entities increasingly rely on specified batteries to make decisions about triage and diagnosis. While use of standardized tools offers many advantages, uniform application or interpretation of specific instruments disregards the vast individual heterogeneity that is a hallmark of ASD. Thus, the field is in need of updated practices, wherein tools are selected, combined, and interpreted in the context of an individual child's presentation, with specific reference to how likely it is that scores on a given test or combination of tests will indicate ASD caseness for that child. Toward this end, the proposed secondary data analysis will identify which tools, and combinations of tools, work best for identifying ASD in sub-groups of youth with shared demographic, developmental, and/or behavioral phenotypes (e.g., in toddler girls with phrase speech vs. verbally fluent adolescent boys with clinically elevated behavior problems). We will analyze data from several widely-used ASD measures, aggregated from more than 17,500 children between the ages of 18 months and 17 years, 11 months. These youth were either clinically referred for ASD diagnostic assessment, assessed due to heightened risk for ASD or other developmental delays, or recruited for ASD-focused research projects. They were assigned a best-estimate diagnosis of ASD or non-ASD (e.g., intellectual disability, ADHD, language disorder) by expert clinicians or clinical-researchers following a comprehensive assessment. By considering the interplay between individual characteristics and instrument scores in the largest sample to date, the proposed study will move the field toward more individualized approaches for establishing ASD caseness. Findings from th...