Coronary disease & ACSDiagnostic or prognostic study
Merging machine learning and patient preference: a novel tool for risk prediction of percutaneous coronary interventions
- Prognostic model development and validation study
- multicentre registry
- external validation
- patient preference survey
Authors' conclusion
Using common pre-procedural risk factors, the BMC2 machine learning models accurately predict post-PCI outcomes.
Study context
- Population
- Patients undergoing percutaneous coronary intervention (PCI) at 48 Michigan hospitals (BMC2 registry), 1 April 2018–31 December 2021; external validation in 33 Washington hospitalsexternal validation 56,583 procedures
- Participants
- 107,793 patients
- Outcome
- Prediction of post-PCI complications (in-hospital mortality, AKI, new-onset dialysis, stroke, major bleeding, transfusion)
- Follow-up
- In-hospital
Performance
As reported in the abstract
Model discrimination (AUC)
- Mortality AUC
- 0.930 (95% CI 0.920-0.940)
- AKI AUC
- 0.893 (95% CI 0.883-0.903)
- Dialysis AUC
- 0.951 (95% CI 0.939-0.964)
- Stroke AUC
- 0.751 (95% CI 0.714-0.787)
- Transfusion AUC
- 0.917 (95% CI 0.907-0.925)
- Major bleeding AUC
- 0.887 (95% CI 0.870-0.905)
Further findings
Overall event rates · in-hospital
- In-hospital mortality 1.85% (n = 1999)
- AKI 2.51% (n = 2519)
- New-onset dialysis 0.44% (n = 462)
- Stroke 0.41% (n = 447)
- Major bleeding 0.89% (n = 942)
- Transfusion 2.41% (n = 2592)
External validation discrimination
- Similar discrimination in the external validation population
Patient preference survey (n = 66)
- Survey subjects preferred a comprehensive list of individually reported post-procedure outcomes
Original publication
Merging machine learning and patient preference: a novel tool for risk prediction of percutaneous coronary interventions. Eur Heart J. Published 17 January 2024. doi:10.1093/eurheartj/ehad836
Summarised from the published abstract; every number is checked against it