Accelerated cardiac ageing assessed by artificial intelligence in children and adolescents with central precocious puberty
A Vilar De Carvalho, B L Bandeira de Goes, E Araujo Bady Casseb, A L Cruz da Silva, D P de Araujo Melo, G de Azevedo Abrantes, W Meira Jr, G Pappa, A Xavier Nascimento, T Andrade Dantas, A Farrant Braz, G Miana de Matos Paixao, I Correia de Queiroz BarbosaAbstract
Introduction
Precocious puberty is associated with an increased risk of cardiometabolic diseases in adulthood, highlighting the necessity for early, non-invasive markers of cardiovascular risk. However, data regarding electrophysiological markers of premature cardiovascular ageing in paediatric populations remain scarce.
Purpose
To evaluate the artificial intelligence (AI)-estimated cardiac age derived from electrocardiography in children and adolescents with central precocious puberty, and to assess its association with bone age advancement and the duration of gonadotropin-releasing hormone (GnRH) analogue treatment.
Methods
An observational, cross-sectional, descriptive-analytical study was conducted including 20 patients followed at a specialized pediatric endocrinology outpatient clinic. The inclusion criteria were the onset of pubertal development before 8 years in girls and before 9 years in boys. Clinical data were collected, and standard 12-lead electrocardiograms were obtained and processed using a deep neural network trained to estimate cardiac biological age. Statistical analysis utilised Spearman’s coefficient for correlations.
Results
The cohort was predominantly female (95%), with a mean chronological age of 10.15 ± 1.66 years and a mean BMI Z-score of 1.35 ± 1.02. AI-estimated cardiac age was consistently higher (mean: 13.18 ± 4.44 years), establishing a statistically significant discrepancy of 3.02 ± 5.48 years in vascular ageing relative to chronological age (p = 0.023). In comparison, the mean bone age was 11.33 ± 1.61 years, as demonstrated in Figure 1. Bivariate analysis revealed a strong negative correlation between this cardiac age gap and chronological age (ρ = -0.72, p < 0.001), indicating that the predicted ageing advancement is significantly more pronounced in younger patients. Conversely, the heart age gap appeared independent of other clinical determinants, showing no significant correlation with BMI Z-score (p = 0.726), bone age advancement (p = 0.365), or the duration of GnRH analogue therapy (p = 0.234). These findings suggest a potential dissociation between electrophysiological cardiac ageing and conventional markers of somatic and pubertal maturation.
Conclusion
The present findings indicate that AI-enabled electrocardiography can identify patterns consistent with premature cardiovascular ageing in children and adolescents with central precocious puberty. Although promising as a non-invasive screening tool, further studies with larger cohorts and pediatric-specific model calibration are required before clinical implementation.