Multicomponent prediction of 2-year mortality and amputation in patients with diabetic foot using a random survival forest model: Uric acid, alanine transaminase, urine protein and platelet as important predictors

Abstract

The current methods for the prediction of mortality and amputation for inpatients with diabetic foot (DF) use only conventional, simple variables, which limits their performance. Here, we used a random survival forest (RSF) model and multicomponent variables to improve the prediction of mortality and amputation for these patients. We performed a retrospective cohort study of 175 inpatients with DF who were recruited between 2014 and 2021. Thirty-one predictors in six categories were considered as potential covariates. Seventy percent (n = 122) of the participants were randomly selected to constitute a training set, and 30% (n = 53) were assigned to a testing set. The RSF model was used to screen appropriate variables for their value as predictors of 2-year all-cause mortality and amputation, and a multicomponent prediction model was established. Model performance was evaluated using the area under the curve (AUC) and the Hosmer-Lemeshow test. The AUCs were compared using the Delong test. Seventeen variables were selected to predict mortality and 23 were selected to predict amputation. Uric acid and alanine transaminase were the top two most useful variables for the prediction of mortality, whereas urine protein and platelet were the top variables for the prediction of amputation. The AUCs were 0.913 and 0.851 for the prediction of mortality for the training and testing sets, respectively; and the equivalent AUCs were 0.963 and 0.893 for the prediction of amputation. There were no significant differences between the AUCs for the training and testing sets for both the mortality and amputation models. These models showed a good degree of fit. Thus, the RSF model can predict mortality and amputation in inpatients with DF. This multicomponent prediction model could help clinicians consider predictors of different dimensions to effectively prevent DF from clinical outcomes .

Keywords: diabetic foot inpatients; longitudinal cohort; mortality and amputation; multicomponent prediction model; random survival forest model.

FIGURE 1

Prediction error of RSF and VIMP of variables for mortality and amputation. (A) shows prediction error rate changing with the number of trees for mortality. (B) shows VIMP of variables predicting mortality. (C) shows prediction error rate changing with the number of trees for amputation. (D) shows VIMP of variables predicting amputation.

FIGURE 2

ROCs for predicting mortality and amputation. (A) shows ROCs for predicting mortality in the development cohort and validation cohort; (B) shows ROCs for predicting amputation in the development cohort and validation cohort. Legends, respectively, showed cut‐off, specificity, sensitivity and AUC of prediction model. Compared with the AUC of Model 1: *p < 0.05; **p < 0.01; ***p < 0.001.