作者：薛锦1，申晓梅2，游继军1，王熠1，徐跻京1

单位：

1.225500江苏省泰州市第二人民医院胸外科　2.225500江苏省泰州市第二人民医院肿瘤内科

Units：

1.Thoracic Surgery, Taizhou Second People's Hospital, Taizhou 225500, China2.Medical Oncology, Taizhou Second People's Hospital, Taizhou 225500, China

关键词：

癌，非小细胞肺；胸腔镜手术；心律失常；贝叶斯网络

Keywords：

Carcinoma, non-small-cell lung; Thoracoscopic surgery; Arrhythmia; Bayesian network

CLC：

DOI：

10.12114/j.issn.1008-5971.2022.00.168

Funds：

摘要：

目的　基于贝叶斯网络构建非小细胞肺癌（NSCLC）患者胸腔镜手术后并发心律失常风险预测模型。方法　选取2020—2021年在泰州市第二人民医院实施胸腔镜手术的213例NSCLC患者作为研究对象，将胸腔镜手术后住院期间并发心律失常的NSCLC患者纳入心律失常组，未并发心律失常的NSCLC患者纳入非心律失常组。比较两组患者基线资料及实验室检查指标；采用多因素Logistic回归分析探讨NSCLC患者胸腔镜手术后并发心律失常的影响因素；基于多因素Logistic回归分析结果，采用贝叶斯网络构建NSCLC患者胸腔镜手术后并发心律失常风险预测模型。结果　213例NSCLC患者完成胸腔镜手术后有34例并发心律失常，心律失常发生率为16.0%。心律失常组年龄＞75岁者占比、TNM分期为Ⅱ期者占比、合并高血压者占比、吸烟指数＞400年支者占比、第1秒用力呼气末容积占预计值的百分比（FEV1%）≤70%者占比、左心房前后径＞40 mm者占比、并发低氧血症者占比高于非心律失常组（P ＜0.05）。多因素Logistic回归分析结果显示，年龄＞75岁、TNM分期为Ⅱ期、合并高血压、吸烟指数＞400年支、FEV1%≤70%、左心房前后径＞40 mm、并发低氧血症是NSCLC患者胸腔镜手术后并发心律失常的危险因素（P ＜0.05）。基于多因素Logistic回归分析结果，采用贝叶斯网络构建NSCLC患者胸腔镜手术后并发心律失常风险预测模型，结果显示，年龄、TNM分期、合并高血压、吸烟指数、FEV1%、左心房前后径、并发低氧血症均与NSCLC患者胸腔镜手术后并发心律失常直接相关；基于条件规律分布基础，以术后并发心律失常为父节点，年龄＞75岁、吸烟指数＞400年支、TNM分期为Ⅱ期、FEV1%≤70%、左心房前后径＞40 mm、并发低氧血症、合并高血压的条件概率分别为47.4%、44.3%、42.5%、42.2%、40.4%、24.3%、21.9%；且年龄可通过影响肺功能间接影响心律失常的发生，肺功能可通过影响左心房前后径间接影响心律失常的发生，吸烟指数可通过影响TNM分期间接影响心律失常的发生。结论　年龄＞70岁、TNM分期为Ⅱ期、合并高血压、吸烟指数＞400年支、FEV1%≤70%、左心房前后径＞40 mm、并发低氧血症是NSCLC患者胸腔镜手术后并发心律失常的危险因素，基于上述危险因素及贝叶斯网络构建的NSCLC患者胸腔镜手术后并发心律失常风险预测模型能反映NSCLC患者胸腔镜手术后并发心律失常的风险。

Abstract：

Objective　To build a risk prediction model for arrhythmia in patients with non-small cell lung cancer(NSCLC) after thoracoscopic surgery based on Bayesian network. Methods　A total of 213 NSCLC patients who underwentthoracoscopic surgery in Taizhou Second People's Hospital from 2020 to 2021 were selected as the research subjects. NSCLCpatients with arrhythmia during hospitalization after thoracoscopic surgery were included in the arrhythmia group, and NSCLCpatients without arrhythmia were included in the non-arrhythmic group. The baseline data and laboratory indicators werecompared between the two groups. Multivariate Logistic regression analysis was used to explore the influencing factors ofarrhythmia in patients with NSCLC after thoracoscopic surgery. Based on the results of multivariate Logistic regression analysis,Bayesian network was used to construct a risk prediction model for arrhythmia in patients with NSCLC after thoracoscopicsurgery. Results　Among 213 NSCLC patients who completed thoracoscopic surgery, 34 were complicated with arrhythmia, andthe incidence of arrhythmia was 16.0%. The proportion of age > 75 years old, the proportion of TNM stage Ⅱ, the proportionof combined hypertension, the proportion of smoking index > 400, the proportion of FEV1% ≤ 70%, the proportion of left atrialanteroposterior diameter > 40 mm, the proportion of complicated with hypoxia and the proportion of hyperemia in the arrhythmiagroup were higher than those in the non-arrhythmic group (P < 0.05) . The results of multivariate Logistic regression analysisshowed that age > 75 years old, TNM stage Ⅱ, combined with hypertension, smoking index > 400, FEV1% ≤ 70% , left atrialanteroposterior diameter > 40 mm, complicated with hypoxemia were risk factors of arrhythmia in patients with NSCLC afterthoracoscopic surgery (P < 0.05) . Based on the above risk factors and Bayesian network construction of NSCLC patients afterthoracoscopic surgery for arrhythmia risk prediction model results show that age, TNM stage, combined hypertension, smokingindex, FEV1%, left atrial anteroposterior diameter, and concurrent hypoxemia were all directly related to arrhythmia in patientswith NSCLC after thoracoscopic surgery. Based on the distribution of conditional regularity, with postoperative concurrentarrhythmia as the parent node, the conditional probabilities of age > 75 years old, smoking index > 400, TNM stage Ⅱ, FEV1%≤ 70%, left atrial anteroposterior diameter > 40 mm, complicated with hypoxemia, combined hypertension were 47.4%, 44.3%,42.5%, 42.2%, 40.4%, 24.3%, and 21.9%, respectively. Age can indirectly affect the occurrence of arrhythmia by affecting thelung function, lung function can indirectly affect the occurrence of arrhythmia by affecting the anterior and posterior diameterof the left atrium, and smoking index can indirectly affect the occurrence of arrhythmia by affecting the TNM stage. Conclusion　Age > 70 years old, TNM stage Ⅱ, combined hypertension, smoking index > 400, FEV1% ≤ 70%, left atrial anteroposteriordiameter > 40 mm, complicated with hypoxemia are the risk factors of arrhythmia in patients with NSCLC after thoracoscopicsurgery. Based on the above risk factors and Bayesian network, the risk prediction model of arrhythmia in patients with NSCLCafter thoracoscopic surgery can reflect the risk of arrhythmia in patients with NSCLC after thoracoscopic surgery.

ReferenceList：