Identifying Unique Subgroups of Emotional and Behavioral Presentations in a Large Inpatient and Community Sample of Autistic Youth

Autistic youth exhibit wide variability in emotional and behavioral challenges, yet few studies have identified meaningful subgroups based on these profiles. This study applied a random forests ensemble clustering algorithm to item-level parent-report data from the Emotion Dysregulation Inventory (EDI) and the Child Behavior Checklist (CBCL) in a combined sample of 1311 autistic youth (ages 6-17), drawn from the Autism Inpatient Collection (n = 446) and the Interactive Autism Network (n = 865). Four distinct subgroups emerged: Global High (GH; 22%), characterized by elevated scores across all four subscales (EDI-Reactivity, EDI-Dysphoria, CBCL Internalizing, and CBCL Externalizing); High-Reactivity Dominant (H-RD; 41%), marked primarily by high emotional reactivity; Moderate-Internalizing Dominant (M-ID; 18%), with elevated internalizing scores; and Global Low (GL; 19%), showing uniformly low scores. Feature importance analyses identified EDI-Reactivity and CBCL Aggression items as the strongest drivers of subgroup membership. No significant differences were found across subgroups in age, sex, race, or ethnicity. However, subgroups with greater emotional and behavioral challenges were associated with lower household income, single-parent status, and higher rates of family psychiatric history. These findings suggest that emotional reactivity and aggression severity are key differentiating features among autistic youth, and that sociodemographic and family mental health factors meaningfully shape emotional and behavioral outcomes.