基于临床信息的小肠梗阻治疗方式预测模型的构建与验证

Construction and Validation of a Clinical Information-Based Prediction Model for Treatment Modality in Small Bowel Obstruction

摘要: 目的：基于临床信息构建并验证小肠梗阻治疗方式预测模型，评估其在手术与非手术分层中的应用价值。方法：采用单中心回顾性队列设计，纳入2022年1月至2025年12月青岛大学附属医院收治并确诊的小肠梗阻患者269例，按分层随机法划分为训练集188例和测试集81例。收集患者入院后首次一般资料、实验室指标及常规CT征象。以是否接受手术治疗为结局变量，在训练集中进行单因素和多因素Logistic回归分析，筛选独立相关因素并构建临床预测模型；采用受试者工作特征曲线及曲线下面积评价模型区分能力。结果：训练集与测试集在年龄、性别、BMI、既往腹部手术史、实验室指标及CT征象等方面差异均无统计学意义(均P > 0.05)。单因素Logistic回归分析显示，hematocrit、腹水、性别、小肠壁增厚、鸟嘴征、漩涡征、CRP及D-二聚体与手术治疗相关，既往腹部手术史呈边缘统计学意义。多因素Logistic回归分析显示，sex、CRP及hematocrit为预测手术治疗的独立相关因素，小肠壁增厚呈边缘统计学意义。基于sex、CRP、hematocrit及Small bowel wall thickening构建的临床模型在训练集中的AUC为0.884，在测试集中的AUC为0.892。结论：基于临床信息构建的预测模型对小肠梗阻治疗方式具有较好的区分能力和稳定性，其中性别、CRP及hematocrit为较关键的临床相关因素。

Abstract: Objective: To develop and validate a clinical prediction model for determining the treatment modality in patients with small bowel obstruction (SBO), and to evaluate its utility in stratifying patients for surgical versus non-surgical management. Methods: This single-center retrospective cohort study enrolled 269 patients with confirmed SBO admitted to the Affiliated Hospital of Qingdao University between January 2022 and December 2025. Patients were randomly allocated using stratified sampling into a training cohort (n = 188) and a testing cohort (n = 81). Baseline demographic data, laboratory parameters, and routine abdominal computed tomography (CT) findings obtained at admission were collected. The primary outcome was receipt of surgical treatment. Univariate and multivariate logistic regression analyses were performed in the training cohort to identify independent predictors and construct a clinical prediction model. Model discrimination was assessed using receiver operating characteristic (ROC) curve analysis and the area under the curve (AUC). Results: Baseline characteristics, including age, sex, body mass index (BMI), history of prior abdominal surgery, laboratory indices, and CT features, were comparable between the training and testing cohorts (all P > 0.05). Univariate Logistic regression identified hematocrit, ascites, sex, small bowel wall thickening, beak sign, whirl sign, C-reactive protein (CRP), and D-dimer as significantly associated with surgical intervention, while a history of prior abdominal surgery showed borderline significance. Multivariate analysis revealed that sex, CRP, and hematocrit were independent predictors of surgical management, with small bowel wall thickening demonstrating borderline significance. A clinical prediction model incorporating sex, CRP, hematocrit, and small bowel wall thickening yielded an AUC of 0.884 in the training cohort and 0.892 in the testing cohort. Conclusion: The developed clinical prediction model exhibits excellent discriminative ability and stability for guiding treatment decisions in SBO. Sex, CRP, and hematocrit emerged as key independent predictors.