Brain biomarkers of late talking: a longitudinal study of early identification and risk of persistence

Florida State University received a $453,893 Project Grant from the National Institutes of Health (NIH) Office of the Director through the Trans-NIH Research Support program (CFDA 93.310), awarded on September 1, 2025, with completion targeted for August 31, 2027. This grant funds a longitudinal neuroimaging study designed to identify brain biomarkers of late talking in infants and young children and distinguish between transient late talkers who naturally recover and those who experience persistent expressive language delays. The project leverages the Early Brain Development Study, a dataset of over 700 children with structural and functional neuroimaging data collected from ages 0–2 years, combined with expressive language assessments at ages 2 and 6 years. Using advanced graph theoretical analysis techniques, the research will prospectively identify neural predictors of late talking before behavioral diagnosis occurs, addressing a critical gap in understanding the etiology of expressive language delays and their relationship to developmental language disorder, dyslexia, and autism spectrum disorder.

The project includes two sub-awardees who contribute specialized expertise to the research execution. The University of North Carolina at Chapel Hill provides access to imaging and behavioral databases through Dr. Gilmore and offers statistical analysis and data management services under Dr. Cornea's leadership, who has extensive experience with the Early Brain Development Study database. The University of California, Los Angeles contributes developmental cognitive neuroscience expertise through Dr. Schneider, who oversees pediatric structural and functional neuroimaging data processing, establishes clinical cutoffs for late talking risk using standardized developmental assessment tools, and ensures compliance with current best practices in neuroimaging data organization and processing. Together, these collaborative efforts aim to generate actionable biomarkers that could improve early identification and prognostication of persistent language delays in vulnerable populations.

Abstract

Children typically have a vocabulary of 150-450 words by 2 years old. However, ~15% of children exhibit expressive language delays, which is referred to as late talking. Late talking can be a precursor to neurodevelopmental disorders, such as developmental language disorder, developmental dyslexia, or autism spectrum disorder, and comes with long-term educational and social consequences. However, not all children who are late talkers experience persistent delays, with some late talkers recovering by early childhood; these children are referred to as transient late talkers. IT is difficult to prospectively know which late talkers will persist into future delays, thus studies of the etiology of late talking and persistent language delays are needed. Neuroimaging studies have identified differences in brain structure and function between late talkers and their non-late talking peers. However, these neural differences have only been studied after late talking is diagnosed, thus there is a critical need to prospectively identify late talkers by examining early neural development before observable behavioral differences manifest. Taken together, the current proposal seeks to address this need by prospectively identifying biomarkers of late talking before diagnosis and examining biomarkers that can separate transient late talkers from those who persist into future delays. This proposal aims to address this need by leveraging a large, rich, longitudinal, pediatric neuroimaging dataset collected from over 700 children, the early brain development study. The proposed study will use measures of structural and functional brain development from 0-2 years of age and measures of expressive language at ages 2 and 6 years to examine biomarkers of late talking and future delays. Cutting edge graph theoretical techniques will be used to estimate functional connectivity of the language and subcortical networks at rest. Trajectories of structure (aim 1A) and functional connectivity (aim 1B) of the language network will be estimated across ages 0-2 years. These trajectories will be tested as early life biomarkers that can predict language delays at 2 years. In aim 2 a hierarchical clustering algorithm will be used to identify brain-based subgroups of children who meet criteria for late talking at age 2. These subgroups will be examined for differences in verbal language skills at age 6 years to test whether biomarkers that parse heterogeneity in late talkers at age 2 years predict persistence, or alternatively transience, of language delays. The findings from this proposal stand to inform our understanding of the underlying biological mechanisms and developmental pathways of late talking, prior to diagnosis, allowing for timely identification of children at risk of language delays, resulting in better outcomes for these children. In accordance with the goals of the funding mechanism, all data derivatives used for the proposed studies will be curated on open science framework to create a novel shared resource for studying late talking.