Unsupervised Phenomapping of Perioperative Risk in Stable Coronary Artery Disease: Revealing Limitations of the Revised Cardiac Risk Index - PubMed
3 hours ago
- #Perioperative Risk
- #Risk Stratification
- #Stable Coronary Artery Disease
- Unsupervised machine learning identified six clinical phenotypes in patients with stable coronary artery disease (CAD) undergoing elective non-cardiac surgery.
- Perioperative major adverse cardiovascular events (MACE) incidence varied significantly across phenotypes, ranging from 1.0% to 16.4%.
- The Revised Cardiac Risk Index (RCRI) showed poor discrimination in three high-risk phenotypes accounting for 60% of MACE burden.
- The CHALSA score demonstrated higher and more stable discrimination across all phenotypes compared to RCRI.
- RCRI misclassified risk in high-risk phenotypes, particularly those with occult ischemia or metabolic dysregulation.
- CHALSA may offer better perioperative risk stratification, but external validation is needed before broader clinical adoption.