New ECG biomarkers and sex-stratified models for the detection of Arrhythmogenic Cardiomyopathy with left ventricular involvement
[EN] Arrhythmogenic Cardiomyopathy (ACM) is a rare cardiac genetic disease that can lead to severe cardiac structural and electrical abnormalities. Diagnosing ACM includes several ECG biomarkers without sex distinction, and it remains challenging, particularly in cases involving the left ventricle (...
| Autores: | , , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | inglés |
| OAI Identifier: | oai:riunet.upv.es:10251/230537 |
| Acceso en línea: | https://riunet.upv.es/handle/10251/230537 |
| Access Level: | acceso abierto |
| Palabra clave: | Arrhythmogenic Cardiomyopathy diagnostic ECG Vectorcardiogram Principal Components Analysis Logistic Regression Feature selection Classification Sex stratification |
| Sumario: | [EN] Arrhythmogenic Cardiomyopathy (ACM) is a rare cardiac genetic disease that can lead to severe cardiac structural and electrical abnormalities. Diagnosing ACM includes several ECG biomarkers without sex distinction, and it remains challenging, particularly in cases involving the left ventricle (LV) and biventricular (biV) forms. Our study aims to improve the yield of current criteria in patients with LV and biV ACM involvement by introducing innovative ECG-based biomarkers that ultimately feed different Logistic Regression models to identify the disease. Additionally, we aim to find differences in the performance of the biomarkers and the classification models through a sex-focused stratification. In this work, 87 individuals (age 43 ± 18 years, 53 % female) underwent cardiological evaluations as part of family screening for ACM, resulting in 41 ACM cases and 46 healthy relatives. We extracted 87 parameters from each ECG, analyzed their statistical significance, and performed different feature selections to optimize each classification model, maximizing G-metric values. We identified 30 significant ECG biomarkers common to both sexes, 9 specific to women, and 15 specific to men. Classification results reached a maximum G value of 0.84 (sensitivity 0.85, specificity 0.83) for the whole cohort, 0.88 (sensitivity 0.87, specificity 0.89) for the male cohort, and 0.81 (sensitivity 0.83, specificity 0.79) for the female cohort. These findings suggest that specific disease ECG biomarkers and sex-stratified models may enhance ACM detection, particularly in males. After validation in larger cohorts, the proposed ECG biomarkers and machine learning approaches might improve diagnostic accuracy for ACM during family screening. |
|---|