A predictive model for lateral cervical lymph node metastasis in thyroid cancer

Objective To investigate the predictive factors for lateral lymph node metastasis (LLNM) in patients with papillary thyroid carcinoma (PTC) and to develop an individualized prediction model. Methods Clinical data from 241 PTC patients who underwent lateral neck dissection were analyzed. Logistic regression and machine learning methods were employed to identify predictive factors and construct a model. The predictive value of three-dimensional morphological parameters (total tumor surface area and total tumor volume) was also evaluated. Results Maximum tumor diameter, total tumor surface area, total tumor volume, sphericity, central lymph node metastasis, postoperative TSH, and tumor location were identified as independent predictors of LLNM. A baseline combined model based on maximum tumor diameter showed good predictive performance (AUC 0.832). Furthermore, three-dimensional parameters (total surface area and total volume) demonstrated superior predictive ability compared to the baseline model. Conclusion An effective clinical prediction model for assessing LLNM risk was successfully developed. Three-dimensional morphological parameters represent more promising predictive indicators.