Development and validation of a nomogram for preoperative prediction of tumor deposits in colorectal cancer

Figure 1 Hematoxylin-eosin staining of tumor deposits. A: Magnification (× 20); B: Magnification (× 40).

Figure 2 The Lasso regression method is used to screen for predictable variables. A: The coefficient distribution of 17 baseline features; B: In Lasso regression, 10-fold cross-validation was used to select predictive variables using the minimum criterion (dashed line on the left).

Figure 3 A nomogram was used to predict the risk of tumor deposits in colorectal cancer patients. The predictors included sex, tumor position, preoperative intestinal obstruction, and lymph node metastasis.

Figure 4 The receiver-operating characteristic curves of the nomogram for predicting tumor deposits. A: Receiver operating characteristic (ROC) curve for the nomogram based on the training cohort. The area under the curve (AUC) is 0.839; B: ROC curve for the nomogram based on the validation cohort. The AUC is 0.853. ROC: Receiver operating characteristic; AUC: Area under the curve.

Figure 5 Calibration curve for the nomogram in the cohort. A: Calibration curve of the training cohort; B: Calibration curve of the validation cohort. ROC: Receiver operating characteristic.

Figure 6 Decision curve analysis of the nomogram. The Y-axis represents net income, and the red line represents the risk nomogram. When the threshold probability is > 7% and < 78%, this predictive model can achieve net clinical benefits.