# Advancing Identification of Late-Talking Children and Mapping their Developmental Trajectories Using Real World Data from Electronic Health Records

> **NIH NIH R21** · DUKE UNIVERSITY · 2024 · $442,750

## Abstract

Late language emergence (late talking), affects one in five children in the United States today. Therefore, a
question pediatricians often face at the 24-month well child visit is which late talking children will require specific
assessment and early intervention services to improve their language abilities and long-term outcomes. To
connect the right child with the right early intervention there is an urgent need to identify specific developmental
trajectories associated with late talking. To date, late talking research has mostly relied on data from relatively
small cohort and intervention studies, and included participants who may not be representative of broader
pediatric populations growing up in America today. The use of real world data from electronic health records
(EHR) offer a unique opportunity to address these gaps by studying large, representative cohorts of children,
over longer periods of time to better understand distinct late talking developmental trajectories. While EHR data
has been used to study neurodevelopmental and neurological conditions associated with late language
emergence, only two EHR studies have specifically focused on late talking. These studies have relied on ICD
diagnostic codes to identify the late talking phenotype in EHR records, an approach with inherent weaknesses.
Disparities related to child sex, race, ethnicity, primary home language, and insurance status may result in
delayed capture of ICD diagnostic codes in the medical record. Furthermore, delays may occur between when
parents’ first share concerns with a professional about their child’s development and documentation via an ICD
diagnosis of developmental conditions, including late talking. This study aims to improve how EHR data are used
to study late talking by employing novel machine learning approaches (natural language processing) to identify
late talkers within EHR databases, create open and shared data resources to identify late talking children within
EHR, and leverage inherent advantages of EHR data to delineate late talking developmental trajectories. Our
experienced research team of psychiatrists, language experts, pediatricians, informaticists, and data scientists
are well positioned to achieve our study aims. Our long-term goal is to delineate distant trajectories associated
with late talking, thus enabling a personalized intervention approach to improve child outcomes. In follow-up
work, we will seek to develop a multi-site consortia to study and intervene with late talkers in real world
environments. This will ultimately improve clinical decision-making and referral practices for late talking children.

## Key facts

- **NIH application ID:** 11031139
- **Project number:** 1R21DC022440-01
- **Recipient organization:** DUKE UNIVERSITY
- **Principal Investigator:** Lauren Franz
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $442,750
- **Award type:** 1
- **Project period:** 2024-09-01 → 2026-08-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/11031139

## Citation

> US National Institutes of Health, RePORTER application 11031139, Advancing Identification of Late-Talking Children and Mapping their Developmental Trajectories Using Real World Data from Electronic Health Records (1R21DC022440-01). Retrieved via AI Analytics 2026-07-25 from https://api.ai-analytics.org/grant/nih/11031139. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*