Using continuous intra-cardiac electrogram monitoring to understand cardiovascular outcomes in patients with pacemakers and defibrillators

Pacemakers and implantable cardioverter defibrillators (ICDs) are amongst the most advanced medical implants. These are very good at treating life-threatening heart rhythms, but the challenge in caring for patients extends beyond the initial procedure to supporting a lifelong commitment. Every patient does not follow the same life-course - underlying heart disease can get worse, causing heart failure or recurrent, painful ICD shocks, and there is high risk of developing other conditions, such as stroke. We do not know why and when these events happen. The latest implants continuously monitor a range of health measurements, including detailed electrical waveforms recorded from wires in the heart. Similar measures recorded from body surface ECGs (electrocardiograms) show promise in detecting heart disease and estimating risk of death. Uniquely, the signals measured by these implants are ten times as strong and provide a continuous assessment which can be linked through remote monitoring technology directly to clinicians. This has the opportunity to provide truly remote, personal care for each patient. As doctors, we do not yet know how to use this wealth of information to help people live longer or feel better. This is a challenge and opportunity of the digital age. It requires analysing this information and linking it to other objective measures of health. Three patient charities, three industry partners, three hospitals, computer scientists and clinicians have helped to develop and will deliver this proposal, from design to dissemination. We will use detailed implant monitoring records from 20,000 patients and study these records across each individual's lifetime. We will use machine learning techniques ('artificial intelligence') to identify hidden patterns in the electrical activity of the heart. There are four broad opportunities where these data could be used: improving risk, diagnosis, informing treatment, and helping monitoring. We will therefore help to (1) identify low-risk groups who may not benefit from an ICD implant or ongoing function (2) inform timing of treatment if patients are at risk of recurrent ICD shocks (3) decide when to start anticoagulation to reduce stroke risk, and (4) understand why heart failure occurs in some patients but not others. To make sure that any findings can be applied generally, we will test models in different datasets from two other hospitals. Often patients from different ethnicities, female sex and older individuals have not been well represented in clinical trials, so we will analyse how well predictive models work for these groups to minimise the risk of bias and health inequality. A cardiac device registry will provide insight into trajectories of recipients for personalised care. Because the technology is already available to monitor a lot of information remotely, this study will help to find ways to make this useful. Success will be being able to use the most detailed monitoring clinically for patients who are at high risk of getting ill. The data and findings will be easily accessible to aid further research and monitoring algorithms can be used in future trials to assess the outcome of new treatments.