Electrocardiograms (ECGs) are essential diagnostic tools in cardiology, yet access to practical ECG interpretation training remains limited, especially in resource-constrained educational environments. Addressing this gap, we present ECGTwinMentor, an intelligent digital twin system designed to enhance cardiology education through interactive and personalized learning. The system integrates a Deep Learning model capable of predicting potential cardiac abnormalities based on simulated ECG parameters and user interactions. ECGTwinMentor supports deployment across multiple platforms—including edge devices-offering students a versatile tool to explore various ECG configurations, visualize waveforms, and receive automated diagnostic feedback. The model was trained using synthetically generated ECG data to ensure controlled variability and pedagogical suitability. Preliminary results demonstrate accurate prediction performance and low-latency inference on low-power hardware. ECGTwinMentor contributes to closing the gap between theoretical knowledge and practical ECG interpretation skills, offering a scalable and accessible solution for modern cardiology training.