Data-driven pediatric ECG reference intervals with VSD-based validation

Area of Science:

• Cardiology
• Pediatrics
• Biostatistics

Background:

• Conventional pediatric electrocardiographic (ECG) reference ranges often use arbitrary age groupings, limiting their accuracy.
• Establishing precise, population-specific ECG norms is crucial for accurate diagnosis in children and adolescents.

Purpose of the Study:

• To create data-driven, age- and sex-stratified ECG reference ranges for Chinese pediatric populations.
• To address limitations of traditional, empirically defined age intervals in pediatric ECG interpretation.
• To validate the clinical utility of new reference ranges in identifying cardiac abnormalities.

Main Methods:

• Analysis of 35,088 ECG recordings from individuals under 18 years old.
• Application of unsupervised machine learning to identify natural developmental patterns in 149 ECG parameters.
• Derivation of data-driven age intervals and sex-specific stratification.

Main Results:

• Identification of four distinct age-dependent variation patterns across ECG parameters.
• Observed sex-related differences in most ECG measurements.
• Demonstrated higher sensitivity of data-driven intervals in detecting ECG deviations in children with VSD compared to existing standards.

Conclusions:

• Introduction of a machine learning-based approach for pediatric ECG reference values.
• New age- and sex-specific thresholds offer improved accuracy reflecting physiological changes.
• Enhanced clinical relevance for pediatric ECG interpretation and diagnosis.