AI model tracks child maturity patterns through routine ECG data

A new study from Wake Forest University School of Medicine suggests a routine heart test – an electrocardiogram (ECG) – may offer researchers a new way to measure biological development in children and adolescents. 

The findings were published in a study in European Heart Journal - Digital Health. 

Pediatric researchers often lack reliable measures of pubertal stage or hormone levels in large datasets, forcing them to rely on broad sex-based categories that may not fully capture the gradual nature of biological development. To address this, researchers developed the Electrocardiographic Sex Index (ESI), an AI-based score derived from standard ECGs that reflect biological development on a spectrum rather than in fixed categories. 

In a study of more than 60,000 children's ECGs, researchers found the score changed in predictable way as kids grow. In early childhood, the values were very similar among children. But as they reach later childhood and the teen years, the values begin to separate in ways that reflect normal growth and hormonal changes. 

These findings suggest that ESI captures the step-by-step changes of normal development, rather than a simple split into categories. This approach could give researchers a more precise way to account for developmental stage especially when hormone or puberty data isn't available. 

One of the most exciting aspects of this work is it shows routine ECG data may contain meaningful information about biological maturation in children and adolescents. ESI offers a continuous measure that may help researchers account for developmental stage when Tanner staging, a standard medical system used to describe the five stages of puberty, or hormone data, are not available." 

Tolga Hayit, Ph.D., visiting researcher and study's co-lead author

"ECGs, traditionally underutilized for capturing developmental biology, can now, when coupled with state-of-the-art AI approaches, highlight their potential to uncover patterns of maturation and cardiovascular development at scale," said Ibrahim Karabayir, Ph.D., assistant professor of cardiology and at the Wake Forest Center for Artificial Intelligence Research and the study's other co-first author. 

Key findings 

• In early childhood, ESI values were tightly centered, showing little difference between children grouped by biological sex. 

• Beginning in late childhood and becoming more pronounced through adolescence, ESI values diverged in opposite directions, plateauing in mid‑to‑late adolescence. 

• The same age-related trends were observed similarly among all races. 

• As the children got older, the model's accuracy steadily improved with children approaching adult-level performance in older adolescents. 

To conduct the study, researchers applied the adult‑trained ESI model to 61,930 ECGs from children ages 0-18 years drawn from the clinical ECG archive at the University of Tennessee Health Science Center. The model was applied without retraining or recalibration, allowing investigators to directly observe how ECG features evolve relative to adult benchmarks. 

Why it matters 

Many studies involving children treat sex as simple category, which doesn't always fully capture how their bodies develop overtime. 

ESI may offer a better way to account for these gradual, stage-by-stage changes as children grow, the study's authors said. 

For researchers working on large studies, ESI could be helpful when information about puberty or hormone levels aren't available. 

Next steps 

The researchers highlight the need for longitudinal studies incorporating Tanner staging, a standard five stage clinical scale used to access physical development during puberty hormone measurements and outcomes to further evaluate the clinical and biological significance of ESI in pediatric populations. 

While the study does not assess clinical outcomes in children, it establishes a foundation for future research examining how developmental maturity influences cardiovascular risk, treatment response or long‑term outcomes - using ECGs already collected in routine care. 

Future studies should evaluate ESI longitudinally and incorporate Tanner staging, hormone measurements and clinical outcomes to better understand its biological and clinical significance in pediatric populations.