The PICM Risk Prediction Study - Application of AI to Pacing

Retrospective review of patient characteristics including 12 lead resting electrocardiograms and imaging data (CMR, CT, echo, CXR and fluoroscopy of pacing leads) of patients with right sided ventricular pacing lead due to symptomatic bradycardia, who developed pacing induced cardiomyopathy (or need for CRT upgrade) versus patients who did not using supervised machine learning methods. Development of personalised predictive pacing algorithm to improve right ventricular lead placement, such as conduction system pacing or pre-emptive implantation of an additional left ventricular lead to prevent left ventricular dilatation and pacemaker-induced cardiomyopathy (PICM) with heart failure (left ventricular ejection fraction <50% by Simpson method), hospitalisation or death with the use of the retrospective patient data through machine learning.

Inclusion criteria

• All patients who received a pacing device (VVI, DDD, ICD, leadless pacemaker) from the GSTT/RBH/KCH/ICH database in the last 10 years (from 01/01/2014)
• All patients who are >18 years old.
• Male and Female

Exclusion criteria

• Patients who did not receive a pacing device (VVI, DDD, ICD, leadless pacemaker)
• All patients <18 years old
• Patients with congenital heart disease
• Patients who have received artificial heart valves or underwent cardiac bypass surgery
• Patients who did not have an echocardiogram after receiving a pacing device