Document resource
Background Physical activity (PA) derived from cardiac implantable electronic devices (CIEDs) has been demonstrated as a potential biomarker to predict risk of mortality in cardiac patients. However, the most effective predictive model to be implemented across patient cohorts is unclear.Aims To validate and evaluate predictive model approaches incorporating CIED-derived PA data for mortality risk stratification and clinical applicability. Secondarily, exploring how patient profiles, based on clinical and demographic characteristics, correlate with sedentary behaviour.Methods A retrospective observational analysis was conducted on 655 cardiac patients implanted with CIEDs between 2008 to 2025. Device-derived physiological data including PA levels, heart rate, atrial fibrillation (AF) burden, and device type was acquired from four CIED manufacturer databases (Medtronic, Biotronik, Boston Scientific, Abbott). Clinical variables including cardiac conditions, comorbidities (diabetes, stroke), age, BMI, LVEF, heart failure classification, and survival status was sourced from the NHS ‘Care Portal’ database. Statistical methods included replication of eight predictive models with varying model types which incorporate PA data such as, linear regression, Cox proportional hazards modelling, competing risk analysis and mediation. Group differences across physiological and clinical data was evaluated using chi-square tests and independent t-tests with statistical significance set at p < 0.05.Hypothesised Results Analysis currently ongoing but hypothesised results are; 1) Successful Validation of a robust, remote PA-based predictive model for mortality risk stratification. 2) Identification of clinical and demographic correlates of sedentary behaviour, potentially informing tailored rehabilitation strategies. 3) Evidence supporting the integration of CIED-derived PA data into standard cardiac care pathways.Conclusion This study will provide potential evidence which validates the best performing predictive modelling approach in a new cohort of patients to help stratify risk based on clinical and demographic covariates.