基于贝叶斯网络的机械通气患者误吸风险预测模型构建  

Construction of predictive model for the risk of aspiration in mechanically ventilated patients on the basis of Bayesian network

作　　者：贾佳妹 黄赣英 JIA Jiamei;HUANG Ganying(Fourth School of Clinical Medicine,Zhejiang Chinese Medicine University,Hangzhou 310003,China)

机构地区：[1]浙江中医药大学第四临床医学院,浙江杭州310003 [2]杭州市第一人民医院急诊科

出　　处：《全科医学临床与教育》2025年第4期311-314,341,共5页Clinical Education of General Practice

基　　金：杭州市医药卫生科技重点项目(ZD20230008);浙江省医药卫生科技计划项目(2022519502)。

摘　　要：目的利用贝叶斯网络构建机械通气患者误吸风险预测模型,并评估模型的预测性能。方法选取2021年1月至2023年12月在杭州市第一人民医院急诊科中符合纳入标准的机械通气患者262例。通过单因素分析和多因素logistic回归分析筛选出独立危险因素,并结合专家意见确定最终纳入模型的变量,构建贝叶斯网络模型。利用受试者工作特征(ROC)曲线评估模型的灵敏度、特异度及曲线下面积(AUC),并通过验证集进行外部验证。结果本次研究共纳入262例机械通气患者,其中56例发生误吸,发生率为21.37%。多因素logistic回归显示,体位角度、口腔分泌物量和声门下吸引是机械通气患者发生误吸的独立风险因素(OR分别=1.57、7.84、5.56,P均<0.05)。专家建议的胃肠道症状、胃残余量共同纳入风险模型。最终,模型AUC值为0.79,95%CI 0.74~0.85。结论机械通气患者发生误吸的独立风险因素包括胃肠道症状、胃残余量、体位角度、声门下吸引和口腔分泌物量。基于贝叶斯网络构建的机械通气患者误吸风险预测模型具有较好的预测价值。Objective To construct a predictive model for aspiration risk in mechanically ventilated patients using Bayesian network,and to assess its predictive performance.Methods A total of 262 mechanically ventilated patients who met the inclusion criteria were included in the emergency department of Hangzhou First People’s Hospital from January 2021 to December 2023.Independent risk factors were identified through univariate analysis and multivariate logistic regression,with final variables incorporated into the model based on expert consensus.Independent risk factors were screened through univariate analysis and multi-factor logistic regression analysis,and the final variables included in the model were determined by combining expert opinions to build a Bayesian network model.The sensitivity,specificity,and area under the curve(AUC)of the model were evaluated using receiver operating characteristic(ROC)curves and externally verified by validation sets.Results Among the 262 mechanically ventilated patients,56(21.37%)occurred aspiration.Multivariate logistic regression revealed that position angle,salivary secretion volume,and suction below the glottis were independent risk factors for aspiration in patients with mechanical ventilation(OR=1.57,7.84,5.56,P<0.05).Expert recommendations of gastrointestinal symptoms and gastric residual volume were included in the risk model.Finally,the model AUC value was 0.79,95%CI 0.74-0.85.Conclusion Independent risk factors for aspiration in patients with mechanical ventilation include gastrointestinal symptoms,gastric residual volume,postural angle,subglottic aspiration,and oral discharge volume.The risk prediction model of aspiration for patients with mechanical ventilation based on Bayesian network has good prediction value.

关 键 词：机械通气 误吸 贝叶斯网络 预测模型 

分 类 号：R459.7[医药卫生—急诊医学]