Deep-Learning to predict outcome of CRT based on pulsed-wave Doppler, clinical biomarkers and pacemaker settings

Cardiac Resynchronization Therapy uses a biventricular pacemaker and is the recommended treatment for Heart Failure patients with Left Bundle Branch Block. The pacemaker settings are determined during the implantation procedure based on the patient’s acute response and can be altered during follow-up exams months after implantation. Despite the careful follow-up, only 65-70% of eligible patients respond to this (expensive) therapy and better selection of candidates for CRT as well as optimization of the pacemaker settings for an elected patient are required. Therefore, we developed a deep learning (DL) algorithm on commonly acquired echocardiographic recordings during the standard care path to predict how a patient will likely respond to certain pacemaker settings after 6 months.