Statistical analysis
We used IBM SPSS Statistics 26 for statistics and analysis, and the
R4.1.2 software was used to construct and verify the CPMs. The R
packages used in this study included: “Rms 6.3.0 (Nomograms,
Calibration curve)”, “DescTools 0.99.46 (C-Index)”, “ROCit 2.1.1”
(ROC analysis), “ResourceSelection 0.3.5” (Hosmer-Lemeshow test),
“Rmda 1.6” (DCA analysis). In the randomized controlled trial, the
independent sample t-test was used to compare the data between the two
groups, and the change trend and difference of the two groups of
research data were compared by repeated measurement analysis of
variance. We used bootstrap resampling method, the ability of AUC and
C-Index evaluation models to distinguish patients with severe CRF from
patients with mild CRF. The accuracy of the model was evaluated with
Hosmer Lemeshow goodness of fit test and Calibration calibration curve,
and the clinical practicability of the model was evaluated with DCA
curve analysis results, so as to complete the internal evaluation of the
model. Finally we used the established prediction model for severe CRF
of CC patients to establish the prediction probability for each patient
in the validation group, and then draw ROC curve, Calibration
calibration curve and DCA curve according to the prediction probability
and actual probability to complete the validation of the model.
Inspection level: α=0.05 (bilateral), P <0.05.
The formula of CPMs is
Logit(P )=ln(P /1-P )=β0+β1 X1+β2 X2+β3 X3+⋯+βi Xi.