Predictive model development combining CT-FFR and SYNTAX score for major adverse cardiovascular events in complex coronary artery disease

Fig. 1. The study flowchart.

Fig. 2. (A) Illustration of the SYNTAX score. (B) Coefficient plot of the 9 predictors preliminarily selected by LASSO regression. The coefficient corresponding to a variable represents the effect of a unit change in that variable on the outcome event (MACE). A positive value indicates a positive correlation (risk factor), while a negative value indicates a negative correlation (protective factor). (C) LASSO coefficient profiles of the 56 candidate predictors. (D) A coefficient profile plot against the log (Lambda) sequence. A vertical line indicates the value selected using tenfold cross-validation, where the optimal Lambda resulted in 7 predictors with non-zero coefficients (Lambda = 0.0396).

Fig. 3. Predictive model nomogram: Based on the values of each predictor for a patient, the corresponding scores can be obtained from the nomogram. These scores are summed to calculate the total score. The total score is then used to determine the corresponding predicted probability, providing the 1-year MACE probability for the patient.

Fig. 4. (A) ROC curve of the nomogram. (B) Comparison of ROC curves among the established predictive model and other models and risk factors. “FSS” = Functional SYNTAX score, “ASS” = Anatomical SYNTAX score, "Mod-FSS" = The predictive model established in this study, "Mod-ASS" = The predictive model with “FSS” replaced by "ASS."

Fig. 5. Evaluation of the predictive model: (A) Calibration curve for the training cohort. The closer the “Apparent” curve and the “Bias-corrected” curve are to the “Ideal” curve, the better the predictive performance of the model. (B) Calibration curve for the validation cohort. (C) Clinical decision curve for the training cohort. The “Nomo” (red curve) represents clinical decisions made based on the predictive model. The “All” (gray line) reflects the scenario where all patients receive intervention, while the “None” (black line) represents the scenario where no patients receive intervention. The area where the red curve lies above both the “None” line and the “All” line indicates that the strategy using the predictive model provides greater net benefit compared to either intervening in all patients or withholding intervention altogether. (D) Clinical decision curve for the validation cohort. (E) Boxplot of group distributions based on Nomoscore in the training cohort. (F) Boxplot of group distributions based on Nomoscore in the validation cohort.