Researchers evaluated functional brain wiring in 150 autistic and 159 non-autistic individuals aged 6 to 31 years, discovering that network metrics gathered from simple, non-invasive scalp recordings predicted how adaptive behaviors changed over an average 19-month follow-up period. While overall adaptive trajectories varied substantially across the cohort, the prognostic capacity of cortico-cortical functional connectivity proved strongest during the transitional period of late adolescence and early adulthood, between ages 15 and 24. Within this demographic group, network organization metrics explained up to 30 percent of the longitudinal variance in adaptive outcomes, significantly outperforming traditional behavioral assessments such as full-scale intelligence quotients and diagnostic trait scores.
Adaptive function represents a foundational neurocognitive construct encompassing the practical communication, daily living, and socialization skills necessary to navigate everyday environments self-sufficiently. In community and clinical surveys, only a minority of autistic adults attain full-time employment, complete postsecondary education, or live independently. Although supporting these real-world functional capabilities remains an essential clinical and community priority, trajectories of adaptive development are widely heterogeneous across autistic individuals. Standard behavioral indicators, such as baseline cognitive ability, existing functional scores, and early language metrics, leave the vast majority of developmental change unexplained. Earlier neurobiological investigations assessing the latency of the facial-processing N170 event-related potential or macroscale anatomical brain structure accounted for merely 6 percent of additional variance in subsequent outcomes, underscoring an acute clinical need for reliable, scalable prognostic biomarkers to assist personalized care and trial design.
To establish whether functional neural coordination could yield greater predictive power, investigators analyzed participants recruited across five European academic research centers participating in the Longitudinal European Autism Project within the AIMS-2-TRIALS consortium. The study analyzed electroencephalography data from individuals aged 6 to 31 years who had at least one usable 20-second epoch of resting-state recordings acquired while resting with their eyes closed, alongside high-resolution anatomical structural magnetic resonance imaging. Autistic participants held verified clinician diagnoses and were not excluded for co-occurring mental health conditions or attention-deficit hyperactivity disorder, ensuring realistic representation of the broader population, while individuals with intellectual disability, gross brain lesions, full-scale intelligence quotients below 75, or severe systemic diseases were excluded from the analysis cohort.
The research team measured electrophysiological signals across 60 to 70 scalp electrodes and reconstructed neural source activity within 68 distinct cortical regions mapped to the Desikan-Killiany anatomical atlas using a linearly constrained minimum variance beamformer algorithm. Functional connectivity between all pairwise cortical regions was quantified using the phase-locking value, a mathematically robust index that assesses the phase consistency of electrical oscillations across time. From these pairwise matrices, the researchers calculated two complementary graph theoretical metrics: mean degree, which reflects the average total density and number of functional connections across nodes, and small-world index, a normalized ratio capturing the delicate equilibrium between local specialized clustering and global network integration.
Adaptive functioning was formally evaluated at baseline and at a follow-up assessment roughly 19.7 months later using the Vineland Adaptive Behavior Scales second edition. Conducted as semi-structured clinical interviews with parents or primary caregivers, the assessments yielded standardized composite scores summarizing age-normed functioning across communication, daily life, and social domains. In parallel, trait autistic features were measured through parent-reported Social Responsiveness Scale scores, and cognitive capabilities were determined using standard age-appropriate Wechsler intelligence scales across all five participating European sites.
Cross-sectional evaluations across the broader cohort revealed marked developmental variations in functional connectivity between diagnostic groups. Although mean degree generally increased and small-world network organization decreased as participants matured from childhood into adulthood, these age-related trajectories progressed at significantly slower developmental rates in autistic individuals compared to non-autistic peers. When dividing participants into developmental brackets, autistic youth aged 15 to 24 demonstrated markedly lower mean degree compared to neurotypical peers, whereas child and adult brackets displayed no significant group differences. Furthermore, parent-reported autistic behaviors showed statistically significant cross-sectional correlations across the full age span, whereby lower overall network connectivity and higher small-world index values correlated with greater severities of social-communicative and repetitive features.
Longitudinal predictive modeling revealed that initial functional connectivity metrics significantly anticipated future adaptive skill development, particularly during late adolescence and emerging adulthood. Among autistic individuals aged 15 to 24 years, higher baseline mean degree strongly predicted improvements in adaptive behavior composite scores over the subsequent 19 months, explaining an additional 21 percent of variance after controlling for baseline adaptive function, intelligence quotient, age, and sex. Conversely, lower small-world index scores within this same youth cohort predicted greater functional gains, adding 30 percent of explained variance over conventional predictors. When analyzing the entire age range spanning 6 to 31 years, small-world index retained statistically significant predictive capacity across all participants, though it contributed a modest 4 percent of additional explained variance across the wider age span.
The investigative team subsequently tested how effectively these resting-state metrics could categorize individual patient trajectories based on minimal clinically important difference thresholds, defined as a shift of six points or more on the Vineland composite scale. Using leave-one-out cross-validation, models incorporating mean degree among 15-to-24-year-olds achieved an overall categorical accuracy of 63.2 percent, while small-world index models achieved 65.8 percent accuracy, both performing significantly higher than chance permutation thresholds. In binary classifications differentiating individuals demonstrating meaningful functional gains from those who remained static or declined, the mean degree model in youth achieved an area under the receiver operating characteristic curve of 0.84, with sensitivity and specificity reaching 0.73 and 0.74, while the small-world model achieved an area under the curve of 0.78.
Both mean degree and small-world index exhibited high test-retest reliability across repeated intrasession recording epochs, yielding intraclass correlation coefficients between 0.77 and 0.88. Furthermore, blood-derived polygenic likelihood scores revealed that variations in functional connectivity metrics significantly correlated with genetic variants influencing brain volume, but only within the autistic cohort.
The study authors highlighted several methodological considerations and clinical boundaries when contextualizing their discoveries. Longitudinal Vineland tracking was available for 38 autistic youth and only 3 autistic adults, meaning findings cannot be extrapolated to older adulthood without larger validation studies, while the smaller number of usable recording epochs obtained in younger children reduced reliability in the youngest age bracket. Despite these limitations, the authors emphasize that non-invasive, cost-effective electroencephalography provides substantial clinical utility for precision medicine. If successfully replicated across independent cohorts, these neurobiological markers could help clinicians identify adolescents at heightened risk of functional stagnation or decline, while allowing clinical trial sponsors to stratify participants effectively based on expected developmental trajectories.
Subject of Research: Prognostic value of electroencephalography functional connectivity for adaptive function trajectories in autistic youth
Article Title: EEG functional connectivity as a prognostic biomarker of adaptive function in autistic people
Article References: Mehra, C., Laiou, P., Garcés, P., Ewen, J. B., Loth, E., Johnson, M. H., Mason, L., Jones, E. J., Charman, T., Bourgeron, T., Buitelaar, J., Absoud, M., Richardson, M. P., Murphy, D., & O’Muircheartaigh, J. (2026). EEG functional connectivity as a prognostic biomarker of adaptive function in autistic people. Translational Psychiatry, 16(1), Article 625. https://doi.org/10.1038/s41398-026-04422-7
Image Credits: AI Generated
DOI: 10.1038/s41398-026-04422-7
Keywords: autism spectrum disorder, electroencephalography, functional connectivity, adaptive function, prognostic biomarkers, mean degree, small-world index, Vineland Adaptive Behavior Scales, longitudinal study, neurodevelopment, brain networks, precision medicine, phase locking value, cortico-cortical connectivity, polygenic scores, clinical stratification
News Source: Cassandra Pierce. (November 16, 2026). Resting Brain Wave Patterns Predict Adaptive Skill Shifts in Autistic Youth. Scienmag.



