广西医学

目的探讨视频辅助胸腔镜手术（VATS）肺大疱切除术（PBR）术后患者发生低氧血症的风险因素，并建立预测模型。方法回顾性分析616例接受VATS PBR的患者的临床资料，将患者按照7∶3的比例分为模型组（n=431）和验证组（n=185）。在模型组中，通过单因素Logistic回归、LASSO回归、多因素Logistic回归分析，筛选VATS PBR术后患者发生低氧血症的风险因素。基于风险因素和临床公认变量构建临床预测模型，利用列线图进行可视化展示。采用受试者工作特征曲线、校准曲线、决策曲线评估模型的区分度、校准度和临床适用性，并通过Bootstrap方法进行内部验证。结果肺部感染史、术前血清白蛋白水平、吸烟史、慢性阻塞性肺疾病（COPD）病史、年龄为VATS PBR术后患者发生低氧血症的影响因素（P<0.05）。基于上述因素和肺大疱数量构建的预测模型在模型组的曲线下面积为0.854，而在验证组为0.845，显示出良好的区分能力。校准曲线显示，校准曲线均与理想曲线拟合良好。决策曲线分析结果显示，该模型在不同风险阈值下提供了更高的净收益，其中，模型组的风险阈值范围为2.5%~97.5%，验证组的风险阈值范围为1.5%~98.0%。临床效用分析结果显示，该模型的预测能力及临床应用价值优于传统指标。结论肺部感染史、术前血清白蛋白水平、吸烟史、年龄、COPD病史是VATS PBR术后患者发生低氧血症的影响因素，基于上述因素及肺大疱数量构建的低氧血症预测模型具有良好的预测效能。

Objective To explore risk factors for the occurrence of hypoxemia in patients after video⁃assisted thoracoscopic surgery (VATS) pulmonary bulla resection (PBR), and to construct a prediction model. Methods Clinical data of 616 patients who underwent VATS PBR were retrospectively analyzed. Patients were divided into model group (n=431) or validation group (n=185) according to the ratio of 7:3. In the model group, univariate Logistic regression, LASSO regression and multivariate Logistic regression analyses were used to screen the risk factors for the occurrence of hypoxemia in patients after VATS PBR. A clinical prediction model was constructed based on risk factors and clinically recognized variables, and a nomogram was used for visualization. Receiver operating characteristic curve, calibration curve and decision curve were used to evaluate the discrimination, calibration and clinical applicability of the model, and the Bootstrap method was used for internal validation. Results The history of pulmonary infection, preoperative serum albumin level, smoking history, chronic obstructive pulmonary distress (COPD) history, and age were the influencing factors for the occurrence of hypoxemia in patients after VATS PBR (P<0.05). The prediction model based on the above factors and the number of pulmonary bullae had areas under the curve of 0.854 in the model group and 0.845 in the validation group, showing good discrimination ability. The calibration curve revealed that the calibration curve was well fitted to the ideal curve. The results of decision curve analysis indicated that the model provided higher net benefits under different risk thresholds. The risk threshold range of the model group was 2.5%-97.5%, and the risk threshold range of the validation group was 1.5%-98.0%. The results of clinical applicability analysis exhibited that the predictive ability and clinical application value of the model were superior to those of the traditional indicators. Conclusion History of pulmonary infection, preoperative serum albumin level, smoking history, age, and history of COPD are the influence factors for the occurrence of hypoxemia in patients after VATS PBR. The hypoxemia prediction model based on the above factors and the number of pulmonary bullae has a good prediction performance.