AI-Driven ECG Analysis for Age and Gender Classification in Pediatrics

Overview

This study developed machine learning models to classify age groups and sex from ECG features in a large pediatric cohort. Using 29,408 ECGs from children aged 1 day to 21 years without heart disease, the models leveraged 177 ECG variables to establish age- and sex-specific ECG standards.

Background

Electrocardiograms (ECGs) have been a cornerstone in cardiac diagnostics since 1898, but pediatric ECG interpretation is challenging due to rapid physiological changes during development. Traditional ECG standards rely on limited healthy subjects and isolated variables, missing complex multivariate patterns. Recent advances in AI and machine learning enable comprehensive ECG analysis, but pediatric-specific AI models for age and sex classification have not been established. Developing such models is critical for improving automated screening and diagnosis of pediatric cardiac conditions.

Data Highlights

Key Findings

• A curated cohort of 29,408 ECGs from children and young adults without heart disease was analyzed.
• 177 ECG variables including wave amplitudes, intervals, axes, and integrals were used as features for machine learning.
• ECGs were stratified into 9 developmental age groups from newborn to young adult.
• Machine learning models were trained to classify both age group and sex based solely on ECG data.
• The study establishes foundational age- and sex-specific ECG standards for pediatric AI-enhanced ECG analysis.

Clinical Implications

AI models that accurately classify age and sex from pediatric ECGs enable development of adaptive, context-aware diagnostic tools tailored to developmental physiology. This can improve screening for rare congenital and inherited cardiac conditions by providing normative references that account for age- and sex-specific ECG variations. Ultimately, such AI-enhanced ECG analysis may enhance early detection and personalized management of pediatric heart disease.

Conclusion

This study demonstrates the feasibility of using machine learning to classify age and sex from pediatric ECGs, establishing critical normative standards. These findings pave the way for automated, AI-driven ECG interpretation tailored to the unique developmental changes in children and adolescents.

References

1. Einthoven 1898 -- Invention of the Electrocardiogram
2. Large Pediatric ECG Cohort Study 2012-2022 -- ECG Data Source and Standards
3. Recent Advances in AI-Enhanced ECG Analysis -- Machine Learning Models
4. AI Applications in Adult ECG Analysis -- Age and Sex Estimation
5. Pediatric AI-ECG Studies -- Detection of Cardiac Conditions and Sex Determination