Development of a prediction model for postoperative complications and economic burden analysis in older patients with hip fractures

Figures

• 

  Fig. 1. LASSO regression 10-fold cross-validation and screening of clinical variables. (A) The dashed line on the left represents the mean-squared error, indicating that the model has the best prediction performance (λ = 0.00503); The dashed line on the right represents the cross-validated mean-squared error within the range of 1 standard error from the minimum value, which further reduces the number of variables due to its stricter penalization (λ = 0.08040). (B) With the increase of the penalization coefficient (λ), the regression coefficient of the prediction variable gradually decreases to 0.

• 

  Fig. 2. The clinical nomogram predicting the probability of postoperative complications. (A) The nomogram was established using the training cohort by incorporating five parameters (ASA classification, anemia, male sex, diabetes, and the number of medical comorbidities), and the probability of postoperative complications was finally determined by reading the score above each risk factor. (B) The dynamic online nomogram is accessible at https://shi-haoning.shinyapps.io/Complications_DynNomapp/.

• 

  Fig. 3. Evaluation of discrimination and consistency of the prediction model for postoperative complications after hip fracture. (A) ROC curves of the prediction model; (B) calibration curves of the prediction model for the training cohort; (C) calibration curves of the prediction model for the validation cohort.

• 

  Fig. 4. Decision curve analysis of the nomogram using the training and validation cohorts. The threshold probability and net benefit are represented on the X and Y axes, respectively. The black line and the grey line represent the two extreme conditions, with the black line indicating that no patients developed postoperative complications and none of them had received any intervention and the grey line indicating that all the patients developed postoperative complications and had received interventions.