Using Deep Learning to Improve Pulsed Electric Field Treatments in Cardiology and Oncology

Electroporation, based on pulsed electric fields, represents a promising approach for biomedical applications. Its effectiveness depends on optimal parameters, including waveforms, electrode characteristics, and evaluation of biological outcomes, making optimization complex and time-consuming. To address this, a deep learning framework was developed using literature and experimental data to predict key technological and biomedical parameters, reducing extensive testing. The approach was applied to cardiac pulsed field ablation for arrhythmias and to high-frequency irreversible electroporation for cancer ablation. For the first time, it also optimizes pulsed electric fields combined with ionizing radiation to target cancer stem cells, aiming to lower radiation doses.