Development and validation of a nomogram for predicting 1-year mortality in infective endocarditis patients

Figures

• 

  Figure 1. Study enrollment flowchart. Flowchart showing the patient selection process and final study population.

• 

  Figure 2. Lasso regression coefficient paths. The coefficient paths of all variables in LASSO regression analysis. Each colored line represents the trajectory of each variable's coefficient as the lambda penalty parameter changes.

• 

  Figure 3. Lasso regression cross validation. Ten-fold cross-validation plot for LASSO regression. The optimal lambda value (lambda min = 0.0421) corresponds to the minimum cross-validation error.

• 

  Figure 4. Nomogram for predicting 1-year mortality in IE patients. The nomogram integrates five independent prognostic factors to predict 1-year mortality probability. Each variable is assigned points based on its regression coefficient, and the total points correspond to the predicted survival probability.

• 

  Figure 5. ROC curve of training set. Receiver operating characteristic curves for predicting 90-day, 180-day, and 1-year mortality in the training cohort. AUC values demonstrate excellent discriminative ability.

• 

  Figure 6. ROC curve of validation Set. Receiver operating characteristic curves for predicting mortality in the validation cohort, demonstrating consistent performance with the training set.

• 

  Figure 7. Calibration curve of training set. Calibration plot comparing predicted vs. observed mortality rates in the training cohort.

• 

  Figure 8. Calibration curve of validation set. Calibration plot for the validation cohort showing good agreement between predicted and observed mortality rates.

• 

  Figure 9. Decision curve analysis (DCA). Decision curve analysis showing the clinical utility of the nomogram across different threshold probabilities. The model provides superior net benefit compared to treat-all or treat-none strategies.

• 

  Figure 10. Clinical impact curve (CIC). Clinical impact curve illustrating the number of high-risk patients identified and the number of true positives at different risk thresholds.

• 

  Figure 11. Kaplan–Meier survival curves stratified by risk groups in the training cohort. Kaplan–Meier survival analysis demonstrating distinct survival trajectories across three risk stratification groups over a 1-year follow-up period. Patients in the training cohort (n = 268) were stratified into low-risk (n = 90, green), medium-risk (n = 89, blue), and high-risk (n = 89, red) groups based on tertiles of the Cox proportional hazards model-derived risk scores. The survival probability at 365 days was approximately 100% for the low-risk group, 90% for the medium-risk group, and 20% for the high-risk group. Shaded regions indicate 95% confidence intervals. The number of patients at risk at 60-day intervals is shown in the risk table below. Vertical tick marks represent censored observations. The log-rank test revealed statistically significant differences in survival among the three groups (p < 0.0001), demonstrating excellent risk stratification ability of the prediction model.

• 

  Figure 12. Kaplan–Meier survival curves stratified by surgery status. Kaplan–Meier survival curves comparing 1-year overall survival between IE patients who received surgical treatment (green line, n = 164) vs. those managed conservatively (red line, n = 104). Shaded areas represent 95% confidence intervals. Log-rank test shows highly significant survival difference (p < 0.0001), with 1-year survival rates of approximately 88% in the surgery group vs. 45% in the non-surgery group. The number at risk table displays patients remaining under observation at 60-day intervals. This figure intuitively demonstrates the substantial clinical benefits of surgical treatment in IE patients.