<p>Personalized computational models of cardiac electrophysiology, referred to as cardiac digital twins, are considered a key technology for realizing precision cardiology by tailoring therapies to individual patients. However, the high computational demands of physically detailed simulations prohibit broader adoption of cardiac digital twins in practical applications. Here, we aim to address this challenge with a lightweight Physics-Informed Eikonal model that replicates the biophysical phenomena embedded within cardiac electrophysiology models along with clinical electrograms in near real-time, including arrhythmia scenarios. Validity of the proposed model is shown against a gold-standard cardiac electrophysiology model in complex arrhythmia simulations, including a scar-mediated ventricular tachycardia in a human whole-heart model. In this work, we show that the Physics-Informed Eikonal model delivers near-real time performance and retains high physical fidelity as required for the calibration of cardiac digital twins, thus providing a fundamental core technology to be leveraged in industrial and clinical applications.</p>