Cardiología de precisión: integración de la genética y la inteligencia artificial para la predicción del riesgo cardiovascular

Cardiovascular diseases (CVDs) remain the leading cause of death worldwide, yet conventional predictive models often fail to capture the complex interplay of genetic and physiological risk factors in diverse populations. This study evaluated a hybrid predictive model integrating polygenic risk scores (PRS), artificial intelligence (AI) applied to electrocardiography (ECG), and traditional clinical variables to enhance cardiovascular risk prediction across Latin American cohorts from Mexico, Colombia, and Ecuador. A total of 6,450 participants aged 30 to 75 years were analyzed. PRS were derived from genome-wide association data, and ECGs were processed using deep convolutional neural networks. Model performance was assessed using AUROC, F1-score, and calibration metrics, with interpretability achieved through SHAP (Shapley Additive Explanations) analysis. The hybrid model demonstrated superior predictive accuracy (AUROC 0.91; F1-score 0.87; calibration slope 0.97) compared with the clinical (0.72), PRS (0.78), and AI-ECG (0.86) models. The most influential predictors were PRS, systolic blood pressure, age, and AI-derived electrical age. Subgroup analyses revealed consistent performance across ancestries (Mestizo 0.91; Amerindian 0.89; European 0.93; Afro-descendant 0.87) and countries, confirming the model’s generalizability and fairness. These findings demonstrate that integrating genetic and AI-based physiological data significantly improves cardiovascular risk assessment, enabling early, equitable, and personalized prevention. The proposed hybrid framework provides a scalable foundation for implementing AI-driven genomic precision medicine in multi-ancestry populations, marking an essential step toward reducing cardiovascular disparities in Latin America.