邮件订阅 RSS 中文 | English
高级检索

贝叶斯网络模型在心脏手术相关急性肾损伤影响因素分析中的应用

Application of Bayesian network model in the study of influencing factors of acute renal injury related to cardiac surgery

    摘要
  • 摘要:
    目的 利用贝叶斯网络（Bayesian network，BN）模型，分析影响心脏手术相关急性肾损伤（cardiac surgery associated acute kidney injury，CSA-AKI）发病的相关危险因素及变量间交互作用，探讨BN在病因分析和疾病预测中的临床适用性。
    方法 选取2015年5月至12月在复旦大学附属中山医院接受心脏手术的1 778例住院患者，年龄、性别、体质量指数（body mass index，BMI）、既往病史、心脏手术相关信息等资料摘录于电子病历系统和实验室检测数据库。通过BN分析构建CSA-AKI发病因素网络并评价模型预测效能。
    结果 CSA-AKI发生率为34.6%（615/1 778）。BN分析发现年龄、术前估算肾小球滤过率（estimated glomerular filtration rate，eGFR）、左室射血分数、心脏手术类型、体外循环（cardiac pulmonary bypass，CPB）时间和术中输血量与CSA-AKI发生直接相关。术前血尿酸、糖尿病和冠脉造影剂量等通过节点肾小球滤过率间接地影响CSA-AKI的发生；心功能分级则通过体外循环时间和左室射血分数与CSA-AKI间接相关。风险预测模型的分类准确率为73.1%，受试者工作特征曲线下面积（area under curve，AUC）为0.758，优于logistic回归和logistic评分模型。
    结论 BN在描述AKI发病危险因素间交互作用和发病风险预测方面具有较好的适用性，有助于临床实践中早期发现CSA-AKI高风险人群，以早期预防疾病发生并改善患者预后。

     

    Abstract:
    Objective To analyze the relevant risk factors and the interactions between variables affecting the incidence of cardiac surgery associated acute kidney injury (CSA-AKI), and explore the Bayesian network in clinical applicability in etiology analysis and disease prediction by Bayesian network (BN) model.
    Methods 1 778 inpatients who underwent cardiac surgery at Zhongshan Hospital, Fudan University from May 2015 to December 2015 were recruited. The age, sex, body mass index, previous medical history, and information related to cardiac surgery were extracted from the electronic medical record system and laboratory testing database. BN analysis was used to construct the CSA-AKI incidence factor network and evaluate the model prediction effectiveness.
    Results The incidence of CSA-AKI was 34.6% (615/1 778). BN revealed that age, estimated glomerular filtration rate (eGFR), left ventricular ejection fraction (LVEF), type of cardiac surgery, estimated circulation time and intraoperative blood transfusion were directly related to the occurrence of CSA-AKI. Preoperative serum uric acid, diabetes and angiography dosage had indirect connections with CSA-AKI through eGFR; New York Heart Association(NYHA)classification grade was linked with CSA-AKI by affecting CPB time and LVEF. The AUC value of BNs model was 0.758, higher than that in logistic model and logistic score model.
    Conclusions BN has good applicability in describing the interaction among risk factors and predicting the risk of AKI, which is helpful for early detection of high-risk population of CSA-AKI in clinical practice, so as to prevent the occurrence of disease and improve the prognosis of patients.

     

    • HTML全文
    • 参考文献(17)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回