Fig. 3

Cox survival model, roc, and calibration curves. (a, b) In the nomogram of T1 LABC COX survival prediction model, the line corresponding to each variable is marked with a scale, representing the value range of the variable, and the length of the line segment reflects the contribution of the factor to clinical outcome events. Points: Represents the corresponding single-item scores for each variable at different values. Total points: Represents the total value of the sum of the corresponding single item scores after the values of all variables. (c) The ROC curves of BCS and TM in the modeling group were calculated. (d) The ROC curves of BCS and TM in the validation group were calculated. (e–h) Calibration curve: Apparent: the probability of direct prediction according to the model, that is, the original prediction probability of the model output; Ideal: a perfect prediction in which the predicted probability is exactly the same as the observed probability; Bias-corrected: predicted probability corrected by bootstrap calibration method. The Bootstrap repeated self-sampling method was used to internally verify the nomogram model. The calibration curve obtained by repeated Bootstrap self-sampling for 1000 times showed that the trend trajectories of the simulated curve and the actual curve were basically the same, which had a strong consistency, indicating that the calibration effect of the model was good. (e) OS at 5 years of BCS in the modeling group. (f) OS at 5 years of TM in the modeling group. (g) OS at 5 years of BCS in the validation group. (h) OS at 5 years of TM in the validation group.