基于膈肌超声心脏参数血气指标构建重症肺炎脱机拔管失败的nomogram预测模型

Construction of a Nomogram Prediction Model for Failure of Extubation in Severe Pneumonia Based on Diaphragmatic Ultrasound Cardiac Parameters and Blood Gas Indices

目的:探究膈肌超声、心脏参数、血气指标联合预测重症肺炎脱机拔管失败价值,并构建nomogram预测模型,以期为临床早期针对性制定干预方案提供参考。方法:选取2021年3月至2023年12月贵州医科大学附属医院重症肺炎患者210例,作为研究对象,将研究对象随机分为70%(147例)训练集及30%(63例)验证集,统计患者一般资料,采用Lasso-Logistic回归方程筛选重症肺炎脱机拔管失败预测因子,构建nomogram预测模型,采用受试者工作特征曲线(ROC)、决策曲线(DCA)、校准曲线分析预测模型效能。结果:训练集、验证集中,病例组(脱机拔管失败)与对照组(脱机拔管成功)年龄、机械通气时间、APACHEⅡ评分、ICU入住时间、CRP/ALB、心脏参数、膈肌超声参数、PaO_(2)、PaCO_(2)、P/F、PA-aO-2、心肺基础疾病史比较差异有统计学意义(P<0.05);Logistic回归方程显示,年龄、DTF、DE、E/A、PaO_(2)、PaCO_(2)、CRP/ALB、心肺疾病基础史均是重症肺炎脱机拔管失败影响因素(P<0.05);经R语言软件可视化处理得到重症肺炎脱机拔管失败nomogram预测模型,nomogram预测模型在训练集、验证集中AUC分别为0.866(95%CI:0.801~0.930)、0.917(95%CI:0.853~0.982),校准曲线接近于48度参考线,预测点分布均匀,DCA曲线在0.35~0.8区间内,nomogram预测模型在训练集、验证集中能取得最大获益。结论:基于膈肌超声、心脏参数、血气指标构建重症肺炎脱机拔管失败的nomogram预测模型可用于临床早期预测脱机拔管风险,以针对性制定相应干预方案,改善预后。

Objective:To explore the value of combined phrenic ultrasonography,cardiac parameters,and blood gas indexes in predicting the failure of offline extubation of severe pneumonia,and to construct a nomogram prediction model for early clinical intervention.Methods:A total of 210 patients with severe pneumonia in the Affiliated Hospital of Guizhou Medical University from March 2021 to December 2023 were selected as the study subjects,and the study subjects were randomly divided into a training set(70%,147 cases)and a verification set(30%,63 cases).The general information of the patients was analyzed.The lasso-Logistic regression equation was used to screen the predictors of offline extubation failure for severe pneumonia,and a nomogram prediction model was constructed.The receiver operating characteristic curve(ROC),decision curve(DCA),and calibration curve were used to analyze the efficacy of the model.Results:In the training set and the validation set,the differences in age,mechanical ventilation time,APACHEⅡscore,ICU stay,CRP/ALB,cardiac parameters,diaphragm ultrasound parameters,PaO_(2),PaCO_(2),P/F,PA-aO-2,and history of underlying cardiopulmonary disease were statistically significant when comparing the case group(failed extubation off the machine)with the control group(successful extubation off the machine)(P<0.05);Logistic regression equation showed that age,DTF,DE,E/A,PaO_(2),PaCO_(2),CRP/ALB,and basic history of cardiopulmonary disease were all influencing factors for the failure of offline extubation of severe pneumonia(P<0.05).The nomogram prediction model for offline extubation failure for severe pneumonia was obtained by visualization using R language software.The AUC of the nomogram prediction model was 0.866(95%CI:0.801-0.930)in the training set and 0.917(95%CI:0.853-0.982)in the verification set,respectively.The calibration curve was close to the 48°reference line,the prediction points were evenly distributed,and the DCA curve was within the range of 0.35~0.8.The nomogram prediction model could obtain the greatest benefits in both the training set and the verification set.Conclusion:This nomogram prediction model based on diaphragm ultrasound,heart parameters,and blood gas indexes can be used to predict the risk of offline extubation of severe pneumonia at an early stage.Accordingly,appropriate intervention plans can be made and the prognosis is improved.