Abstract
OBJECTIVES: To develop and validate a predictive model estimating the likelihood of pathological upgrading in patients with colorectal intraepithelial neoplasia (IN). METHODS: Using data from 158 patients with colorectal IN confirmed by endoscopic biopsy, we employed Least Absolute Shrinkage and Selection Operator (LASSO) regression followed by multivariate logistic regression to identify key predictive factors. A nomogram was constructed based on the selected variables. The performance of these models was assessed using calibration curves, the area under the receiver operating characteristic curve (AUC), and the Hosmer-Lemeshow goodness-of-fit test. Furthermore, decision curve analysis (DCA) was utilized to evaluate the practical utility of the models, thereby exploring their potential clinical applications. RESULTS: Four variables-rectal location, surface erosion, lesion size ≥ 30 mm, and villous histology-were incorporated into the nomogram. The model demonstrated strong discrimination (AUC = 0.822; 95% CI: 0.744-0.899) and good calibration (Hosmer-Lemeshow χ(2) = 1.731, p = 0.973). Internal validation yielded a consistent AUC of 0.813. DCA confirmed the model's broad clinical utility. CONCLUSION: This nomogram accurately predicts pathological upgrading in colorectal IN, allowing clinicians to identify high-risk patients early and tailor management accordingly.