Figure 3.

Construction and validation of a risk assessment model based on lncRNA pairs. (A, B) Volcano plots of differentially expressed genes (DEGs) related to recurrence and immune response. (C) Venn diagram showing the intersection of relapse- and immune-related DEGs. (D, E) Feature selection of lncRNA pairs using LASSO regression. (F) Random forest (RF) error rate analysis, with error rates for relapsed and non-relapsed groups shown in red and green, respectively. The overall model error rate is shown in black. (G) Selection of 10 lncRNA pairs identified by LASSO and RF algorithms. (H) Final risk assessment model comprising 7 key lncRNA pairs. (I) Receiver Operating Characteristic (ROC) curve, showing the optimal cutoff value determined at the maximum inflection point. (J) Kaplan-Meier survival curves comparing overall survival between high-risk and low-risk groups. (K) Validation of key model features using a backpropagation neural network (BPNN). (L, M) ROC curves for training and test sets, demonstrating the model's classification accuracy.