Arrhythmias & electrophysiology
Machine learning-based prediction of sudden cardiac death in the general population using electronic health record data
Authors' conclusion
We propose a prediction model designed to identify individuals at high risk of SCD among the general population.
Study context
- Population
- General population; Sudden cardiac death (SCD) cases and controls from electronic health record (EHR) data, Greater Paris and University of Washington (Seattle, USA)training 12,338 cases + 12,338 controls
- Outcome
- Discrimination of SCD (AUC)
- Follow-up
- 5 years preceding SCD
Methodological detailsmachine learning prediction model development and external validation, temporal external validation, geographical external validation
- Machine learning prediction model development and external validation
- Temporal external validation
- Geographical external validation
Main finding
As reported in the abstract
SCD prediction (AUC)
AUC, temporal validation cohort
0.81
95% CI 0.80–0.82
AUC, geographical cohort
0.66
95% CI 0.58–0.73
SCD detected in highest decile, Paris
26%
SCD detected in highest decile, Seattle
33%
Observational data: associations, not effects.
Further findings
No CV diagnoses in 5 years preceding SCD
- Subjects with no CV diagnosis
- 25.7%
MI prediction
- Model specific to SCD; not predictive of MI
Original publication
Machine learning-based prediction of sudden cardiac death in the general population using electronic health record data. Eur J Prev Cardiol. Published 6 March 2026. doi:10.
Summarised from the published abstract; every number is checked against it