Machine learning and conventional Cox regression to predict target-lesion revascularization after percutaneous coronary intervention - Takeaways - MDSpire

Machine learning and conventional Cox regression to predict target-lesion revascularization after percutaneous coronary intervention

  • By

  • Mona El-Faramawi

  • Marco Busco

  • Sören Möller

  • Lisette Okkels Jensen

  • Jens Flensted Lassen

  • July 1, 2026

  • 0 min

Share

  • 1

    The study analyzed 24,360 patients with 34,149 lesions treated with PCI to predict target-lesion revascularization (TLR) using machine learning and Cox regression.

  • 2

    Machine learning-based Cox-LASSO showed a minor improvement in predicting short-term TLR risk compared to conventional Cox regression models.

  • 3

    At 1–5 years post-PCI, stepwise Cox regression demonstrated the best predictive performance for TLR, outperforming Cox-LASSO.

  • 4

    Most risk factors for TLR were consistent between conventional Cox models and the Cox-LASSO model, indicating similar predictive capabilities.

  • 5

    The study concluded that Cox-LASSO did not significantly enhance predictive performance over traditional Cox regression for TLR.

Original Source(s)

Machine learning and conventional Cox regression to predict target-lesion revascularization after percutaneous coronary intervention

  • by Mona El-Faramawi, Marco Busco, Sören Möller, Lisette Okkels Jensen, Jens Flensted Lassen

  • July 1, 2026

Related Content

Shorter DAPT Tested Following CABG

TOP-CABG compared 3 months vs 12 months of ticagrelor plus aspirin following bypass surgery.

Complete revascularization and heart failure risk in acute coronary syndrome across the ejection fraction spectrum: focus on LVEF-dependent effects

Case Report: A case of a giant right ventricular wall hematoma caused by coronary artery perforation during percutaneous coronary intervention