Postoperative delirium risk in elderly high-risk patients undergoing hip and knee arthroplasty (ASA Ⅲ-Ⅳ): construction and validation of a nomogram model based on frail frailty index

Figures

• 

  Figure 1. Flowchart of patient selection for the study cohort. POD indicates Postoperative delirium; ASA, American Society of Anesthesiologists.

• 

  Figure 2. LASSO regression results for variable selection. (A) Coefficient path plot; (B) Cross-validation error plot. The right dotted line is Lambda at standard error (Lambda.1-SE). The left dotted line is Lambda at minimum error (Lambda.min). Our study selected the variables filtered by Lambda.1-SE, where nine non-zero coefficients were selected (λ=0.0333). LASSO, least absolute shrinkage and selection operator; SE, standard error.

• 

  Figure 3. Nomogram for predicting POD in elderly patients. POD, postoperative delirium; ASA, American Society of Anesthesiologists; ALB, albumin; CMMS, Chinese mini mental status; NRS, numerical rating scale.

• 

  Figure 4. ROC curves of individual predictors and the combined model for predicting POD. ROC, receiver operating characteristic; AUC, area under the curve; POD, Postoperative delirium; ASA, American Society of Anesthesiologists; ALB, albumin; CMMS, Chinese mini mental status; NRS, numerical rating scale.

• 

  Figure 5. ROC curves of the training cohort and the validation cohort. (A) training cohort, (B) validation cohort. ROC, receiver operating characteristic; AUC, area under the curve.

• 

  Figure 6. Calibration curves of the training cohort and the validation cohort. (A) training cohort, (B) validation cohort.

• 

  Figure 7. Decision curve analysis of the training cohort and the validation cohort. (A) training cohort, (B) validation cohort.

• 

  Figure 8. Model interpretability analysis based on SHAP. (A) Feature importance bar plot. The plot displays variables ranked by their mean absolute SHAP values. (B) Single-sample SHAP waterfall plot. The plot illustrates how the prediction [f(x) = 0.87] is decomposed relative to the baseline expected value {E[f(x)] = 0.19}. Red bars indicate positive contributions to POD risk; blue bars indicate negative contributions. The preoperative frailty index was the dominant positive contributor in this case. (C) SHAP beeswarm plot. The plot shows the distribution of SHAP values across all samples. The x-axis represents SHAP values; the y-axis lists features by importance. Color from blue to red reflects feature values from low to high. Higher frailty index, early postoperative NRS score, and age, as well as lower albumin and CMMS scores, were associated with increased POD risk. SHAP, SHapley Additive exPlanations; POD, postoperative delirium; ASA, American Society of Anesthesiologists; ALB, albumin; CMMS, Chinese mini mental status; NRS, numerical rating scale.