Artificial Intelligence-Driven Echocardiographic Prediction Model for Early Detection of Pediatric Cardiomyopathy in East Mediterranean Populations

Abstract

Pediatric cardiomyopathy is a leading cause of sudden cardiac death and heart transplantation in children, yet early detection remains limited in low-resource settings. This prospective study enrolled 1,026 children (ages 1–17 years) across six major pediatric cardiac centers in the East Mediterranean between March 2022 and December 2024 to develop and validate an AI-assisted echocardiographic prediction tool.

Left ventricular strain, ejection fraction, septal wall thickness, NT-proBNP levels, and genetic risk scores were collected at baseline. Of the total cohort, 184 children (17.9%) were diagnosed with early-stage cardiomyopathy within 12 months. The AI model was trained on 70% of the dataset and validated on the remaining 30%.

The final predictive model demonstrated a sensitivity of 88.7%, specificity of 91.4%, and an AUC of 0.94 (95% CI: 0.91–0.96). Among the top predictors were global longitudinal strain (GLS < -17%), NT-proBNP > 450 pg/mL, and specific MYH7 gene variants. The model flagged 92 of the 184 cardiomyopathy cases at least 4 months before clinical diagnosis, based on standard guidelines.

This AI-integrated echocardiographic tool reduced diagnostic delay by an average of 3.6 months (p < 0.001), particularly improving detection in asymptomatic children. Implementation in resource-limited cardiology units could significantly enhance early intervention, reduce long-term cardiac complications, and support telecardiology programs in underserved regions.

This is the first study to validate a region-specific AI model tailored to East Mediterranean pediatric populations, providing a scalable and low-cost solution for cardiomyopathy screening and triage.