摘要
目的利用快速因果推断(Fast causal inference,FCI)算法构建因果图模型,分析影响非小细胞肺癌治疗质量的直接和间接因素,为改善患者治疗质量提供依据。方法收集10家三甲医院的非小细胞肺癌患者病例信息;确定影响因素为研究变量,不良事件发生率为患者治疗质量评价指标,即结局变量;利用FCI算法挖掘病例数据,构建研究变量与结局的因果图模型,分析研究变量与结局变量及不同研究变量之间的因果关系。结果本研究共纳入2846例患者,平均年龄56.00±7.70岁,不良事件发生率为9.63%。因果图模型共包含24个节点,71条边,其中有向边54条,双向边7条。影响不良事件发生的直接因素包括医院类型、组织学分级、是否淋巴结清扫及住院天数;间接因素包括职业、医保类型、现病史、病理分期、综合治疗、手术性质及肺切除类型;因素间相互作用分析结果显示,现病史、组织学分型、综合治疗、手术性质、肺切除类型决定患者是否接受淋巴结清扫;手术性质、肺切除方式、综合治疗影响住院天数;既往史影响肺癌组织学分型;职业、医保类型影响患者就诊医院类型。结论在非小细胞肺癌治疗质量影响因素分析中,因果图模型能够获得影响不良事件发生的直接和间接因素,发现可干预的目标变量,为改善非小细胞治疗质量提供依据;医院可通过提高淋巴结清扫、综合治疗接受率,降低不良事件发生率。
Objective The aim of this study was to use the fast causal inference(FCI)algorithm to construct a causal graph model,analyze the direct and indirect factors that affect the quality of treatment for non-small cell lung cancer(NSCLC),and provide a basis for improving the quality of patient treatment.Methods Case information of NSCLC patients from 10 tertiary hospitals was collected;the influencing factors were determined as the research variable,and the incidence of adverse events was the evaluation indicator of patient treatment quality,i.e.the outcome variable;the FCI algorithm to mine case data were used to construct a causal diagram model of research variables and outcomes,and analyze causal relationships between research variables and outcome variables,as well as between different research variables.Results A total of 2,846 patients with an average age of 56.00±7.70 years were included in this study,and the incidence of adverse events was 9.63%.The causal diagram model consisted 24 nodes and 71 edges,including 54 directed edges and 7 bidirectional edges.The direct factors affecting the occurrence of adverse events included hospital type,histological grade,lymph node dissection,and length of hospitalization;indirect factors included occupation,medical insurance type,current medical history,pathological stage,comprehensive treatment,surgical nature,and type of lung resection;The analysis of the interaction between factors showed that the current medical history,histological classification,comprehensive treatment,surgical nature,and type of lung resection determined whether the patient received lymph node dissection;The nature of surgery,method of lung resection,and comprehensive treatment affected the length of hospitalization;Medical history affected the histological classification of lung cancer;The type of occupation and medical insurance affected the type of hospital where patients sought medical treatment.Conclusion In the analysis of factors affecting the quality of NSCLC treatment,the causal diagram model can obtain direct and indirect factors that affect the occurrence of adverse events,identify target variables that can be intervened,and provide a basis for improving the quality of NSCLC treatment;Hospitals can reduce the incidence of adverse events by increasing the acceptance rate of lymph node dissection and comprehensive treatment.
作者
姚雪佩
白山奇
刘美娜
YAO Xuepei;BAI Shanqi;LIU Meina(Department of Biostatistics,Public health college of Harbin Medical University,Harbin 150081,China)
出处
《实用肿瘤学杂志》
CAS
2024年第4期227-234,共8页
Practical Oncology Journal
基金
国家自然科学基金(编号:82173614)。
关键词
非小细胞肺癌
影响因素
因果图模型
快速因果推断算法
Non-small cell lung cancer
Influencing factors
Causal diagram model
Fast causal inference algorithm
