目的 比较结直肠无蒂锯齿状病变（sessile serrated lesions， SSL）与增生性息肉（hyperplastic polyps，HP）内镜下特征，构建预测模型并加以验证。 方法 回顾性分析2019年6月至2022年6月于福建中医药大学附属福鼎医院行结肠镜下病变切除且术后病理证实是SSL或HP的连续病例，一共纳入116例SSL病变（94名患者）和208例HP病变（172名患者），按随机号码1∶1分为训练样本和验证样本。在训练样本中，通过单因素联合多因素logistic回归分析，确定有统计学意义的预测因子并赋值评分，以此构建SSL和HP内镜下鉴别的预测模型，最后在验证样本中验证其预测效能。 结果 与HP相比，SSL多位于右半结肠、长径> 5 mm多见、边界不清晰、形态不规则、有积云样外观、表面有黏液帽和曲张微血管，腺管开口呈Ⅱ-O型（均P < 0.001）。病变长径> 5 mm（OR = 7.717，95% CI：2.141 ~ 27.810，P = 0.002）、表面有黏液帽（OR = 12.063，95% CI：3.925 ~ 37.072，P < 0.001）和边界不清晰（OR = 7.020，95% CI：2.118 ~ 23.260，P = 0.001）是诊断SSL的独立预测因子。SSL和HP内镜下鉴别预测模型的受试者工作特征曲线下面积为0.924（95% CI：0.884 ~ 1.000，P < 0.001）。根据Youden指数确定最佳诊断阈值为11.90分，在验证样本中，预测模型的灵敏度、特异度、准确率、阳性预测值和阴性预测值分别为93.10%、94.23%、93.83%、90.00%和96.08%。 结论 病变长径> 5 mm、表面有黏液帽和边界不清晰是诊断SSL的独立预测因子，本研究构建的SSL与HP内镜下鉴别的预测模型具有较高的预测效能，可为内镜医师做出正确诊断提供重要的参考信息。

Objective To compare the endoscopic features between colorectal sessile serrated lesions and hyperplastic polyps， and to establish and assess a predictive model. Methods A total of 116 SSL lesions（94 patients） and 208 HP lesions（172 patients） which underwent colonoscopic lesion resection and confirmed by pathology from June 2019 to June 2022 at Fuding Hospital， Fujian University of Traditional Chinese Medicine were retrospective analyzed. The lesions were randomly divided into either a derivation sample or a validation sample using 1∶1 allocation according to the random number. In the derivation sample， SSL endoscopic features were used to establish a predictive model to distinguish between SSL and HP，in which predictors were identified by univariate and multivariate logistic regression analysis and predictive efficacy was evaluated by receiver operating characteristic （ROC） curve analysis. The predictive efficacy was assessed in the validation sample. Results Compared with HP，SSL was more common in the right colon with diameter > 5 mm and more likely to be manifested as： unclear boundary， irregular morphology， cumulus?like surface， surface mucus， varicose microvascular vessel and II?O pit pattern（P < 0.001）. Diameter > 5 mm （OR = 7.717，95% CI：2.141 ~ 27.810，P = 0.002），surface mucus （OR = 12.063，95% CI： 3.925 ~ 37.072，P < 0.001） and unclear boundary （OR = 7.020，95% CI：2.118 ~ 23.260， P = 0.001） were predictors to diagnose SSL. Predictive model was built in the derivation sample with area under the ROC curve of 0.924（95% CI：0.884 ~ 1.000，P < 0.001）. In the validation sample， the sensitivity， specificity， accuracy， positive predictive value and negative predictive value of the predictive model were 93.10%， 94.23%， 93.83%， 90.00% and 96.08%， respectively. Conclusion Diameter > 5 mm， surface mucus and unclear boundary were predictors to diagnose SSL. Our predictive model have good predictive value， and it provide an important reference information for the endoscopists to make the correct diagnosis.