Abstract:
Electrocardiogram (ECG) interpretation is a core competency in medical training, yet remains one of the most challenging skills for students and early-career clinicians to master. Traditional methods of teaching ECG analysis often rely on static examples, delayed feedback, and limited exposure to diverse pathological patterns. These limitations hinder the development of diagnostic confidence and critical thinking in real-world scenarios. To address this educational gap, we introduce ECGTwinMentor, a digital twin software platform designed to enhance cardiology education through interactive simulation, intelligent prediction, and personalized feedback. ECGTwinMentor combines a machine learning model trained on a curated ECG dataset with a user-friendly interface that enables learners to input key cardiac parameters. The system also generates synthetic ECG traces and provides contextual explanations, allowing the users to compare their inputs against established physiological norms. Built with scalability and flexibility, the platform supports cloud and edge deployments and incorporates robust security features such as AES encryption and API rate limiting. The software was validated through model performance analysis and feedback from cardiology students during diagnostic simulations. With a deep learning model accuracy of 95%, ECGTwinMentor effectively combines AI, digital twins, and medical education into an interactive and clinically grounded learning tool.
Autores: Daniel Flores-Martín, Francisco Díaz‐Barrancas, Pedro J. Pardo, Javier Berrocal, Juan M. Murillo
Publicación / Evento: SoftwareX