Electrocardiographic sex index applied to children: a proxy assessment for heart maturation

Sex is typically assigned at birth based on the presence of reproductive organs. However, males and females exhibit considerable similarity and overlap at cellular, metabolic and hormonal levels To be able to account for such similarities between female and male sex. We recently introduced the concept—the Electrocardiographic Sex Index (ESI)—which represents sex as a continuous variable derived from adult electrocardiogram (ECG) data. Given the dynamic changes in cardiac electrophysiology during different stages of childhood, and the physiological distinctions between children and adults, it remains unclear whether ESI can be applied to pediatric populations, and if so, from what age it becomes valid.

To evaluate the performance of the ESI in children across various age groups and determine at what age boys and girls begin to exhibit adult-like ECG patterns indicative of male or female sex.

The original ESI was developed using approximately 3.5 million ECGs from adults aged 18 years and older. In that population, the mean ESI was 0.72 (SD 0.18) for males and 0.27 (SD 0.18) for females. The ESI alone classified adult sex with an AUC of 0.947 (95% CI: 0.946–0.947), where an ESI of 0 indicates closest alignment with female sex and 1 with male sex.

In this study, without any modification or fine-tuning, we applied the adult-derived ESI to ECGs from children aged 0 to 18 years. We calculated sex classification accuracy and ESI scores for different age brackets, using adult ESI distributions as reference points.

The analytical cohort included 61,930 ECGs from 37,234 children under 18 years of age (46.7% male; 36.5% White, 57.2% Black; mean age 9.1 ± 6.0 years). ESI-based sex classification accuracy was significantly lower in newborns (DeLong test, p < 0.0001) but improved steadily with age (Figure 1). By age 14, classification accuracy was equivalent to that observed in adults.

The average ESI was 0.33±0.20 for female children and 0.61±0.23 for male children. As expected, ESI declined with age in females and increased in males (Figure 2). By age 12, the mean ESI in females approximated that of adult females. In contrast, males reached adult-like ESI levels by age 14. These trends were consistent across racial groups, with minor variations in AUC and ESI values.

By age 12 in females and age 14 in males, children's ECG-derived ESIs closely resemble those of adults, aligning with the typical end of puberty. Notably, ESI begins to decline in females as early as age 1, while in males, ESI increases begin near the onset of puberty. These findings suggest that the adult-derived ESI may be applicable for cardiovascular risk stratification in adolescents beginning at these age thresholds. Future studies are warranted to assess the validity of applying adult cardiovascular risk prediction models to children based on ESI maturation patterns.Figure 1.

Sex Classification AUC acros