Intelligent monitoring to predict atrial fibrillation (NOTE-AF): clinical study 1 for the 'Health virtual twins for the personalised management of stroke related to atrial fibrillation (TARGET)' project – a protocol for a prospective cohort analysis

Introduction Atrial Fibrillation (AF) is the most common arrhythmia worldwide affecting an estimated 5% of people over the age of 65 and is a leading cause of stroke and heart failure. Identification of patients at risk allows preventative measures and treatment before these complications occur. Conventional risk prediction models are static, do not have flexibility to incorporate dynamic risk factors and possess only modest predictive value. Artificial intelligence and machine learning-powered health virtual twin technology offer transformative methods for risk prediction and guiding clinical decisions.

Methods and analysis In this prospective observational study, 1200 patients will be recruited in two tertiary centres. Patients hospitalised with acute illnesses (sepsis, heart failure, respiratory failure, stroke or critical illness) and patients having undergone high-risk surgery (major vascular surgery, upper gastrointestinal surgery and emergency surgery) will be monitored with a patch-based remote wireless monitoring system for up to 14 days. Clinical and electrocardiographic data will be used for modelling the risk of new-onset AF. The primary outcome is episodes of AF >30 s and will be described as ratio of episodes/patient and as percentage of patients having episodes of AF. Secondary outcomes include 30-day and 90-day readmission rates and complications of AF.

The aim of this study is to generate data for the development and validation of health virtual twins predicting onset of AF in an at-risk population. The intelligent monitoring to predict atrial fibrillation (NOTE-AF) study is part of the TARGET project, a Horizon Europe funded programme which includes risk prediction, diagnosis and management of AF-related stroke (https://target-horizon.eu/).