Systematic screening by a heart team and a machine learning approach contribute to unraveling novel risk factors in...

Area of Science:

• Cardiology
• Medical Informatics
• Health Services Research

Background:

• Real-world data on percutaneous or surgical revascularization for left main and/or 3-vessel coronary artery disease (CAD) show different baseline characteristics affecting mortality compared to randomized controlled trials (RCTs) due to stricter inclusion/exclusion criteria.
• Identifying these registry-specific factors is crucial for accurate risk stratification in diverse patient populations.

Purpose of the Study:

• To assess if systematic screening can identify novel, registry-specific baseline patient characteristics influencing long-term mortality in complex CAD.
• To compare the predictive performance of a machine learning-derived model with the SYNTAX Score II 2020 (SS2020) for 5-year mortality.

Main Methods:

• Least Absolute Shrinkage and Selection Operator (LASSO) regression was used to screen 42 baseline characteristics in 1035 complex CAD patients from a Polish registry.
• A Cox regression analysis was performed on selected factors to build a predictive linear model.
• The model's predictive accuracy for 5-year mortality was compared to the SS2020 using concordance index.

Main Results:

• The 5-year mortality rate in the registry was 12.3%.
• Pulmonary hypertension, chronic obstructive pulmonary disease, and insulin-dependent diabetes were the strongest predictors of mortality.
• The LASSO-derived linear model demonstrated superior prediction of 5-year mortality (concordance index 0.92) compared to SS2020 (concordance index 0.75).

Conclusions:

• Machine learning approaches enhance the identification of registry-specific risk factors in all-comer patients undergoing revascularization.
• Risk factors identified in RCTs may not fully represent those observed in real-world clinical practice.
• Systematic screening is valuable for uncovering unique predictors of mortality in complex CAD patients treated in routine care settings.