中国全科医学

• •

基于Logistic回归与决策树模型的肺癌患者一级亲属筛查行为影响因素分析

张佳, 王海蓉, 赵婧, 苏怡帆   

  • 收稿日期:2023-12-28 修回日期:2024-02-24 接受日期:2024-03-20
  • 通讯作者: 王海蓉
  • 基金资助:
    山西省科学技术协会/山西省健康委员会项目(JKKP202124); 山西省科技厅科普项目(202204091003028)

Influencing Factors of Screening Behavior of First-degree Relatives of Lung Cancer Patients Based on Logistic Regression and Decision Tree Model

ZHANG Jia, WANG Hairong, ZHAO Jing, SU Yifan   

  • Received:2023-12-28 Revised:2024-02-24 Accepted:2024-03-20
  • Contact: WANG Hairong
分享到

摘要: 背景 肺癌是导致中国癌症死亡的首要原因。高危人群早期筛查是发现肺癌,改善预后最有效关键的方法。目前对于肺癌患者一级亲属这一集中高危群体的筛查行为影响因素研究少有报道。目的 采用Logistic回归和决策树模型分析肺癌患者一级亲属的肺癌早期筛查行为影响因素。方法 采用便利抽样法,选取2023年3月—6月山西省肿瘤医院呼吸内科与胸外科正在住医院治疗的310名肺癌患者一级亲属为调查对象。采用一般资料调查表、肺癌知识问卷、中文版肺癌筛查健康信念量表进行调查。分析影响肺癌筛查行为的影响因素,分别建立Logistic回归模型与决策树模型,并比较2种模型的差异性。结果 肺癌患者一级亲属接受过肺癌筛查率为23.9%(74/310);总体肺癌知识知晓率为75.2%(4 662/6 200);56.5%(175/310)愿意接受肺癌筛查。Logistic回归结果显示年龄、医疗保险类型、是否获得过肺癌筛查建议、筛查的意愿、感知障碍、感知效益、感知风险是肺癌患者一级亲属肺癌筛查行为的影响因素(P<0. 05);决策树模型结果显示感知障碍和年龄是一级亲属肺癌筛查行为的影响因素;Logistic 回归模型与决策树模型的预测效率比较差异有统计学意义(P<0. 05)。结论 肺癌患者一级亲属的整体肺癌认知水平较高但对筛查手段的认知较低,健康信念及筛查意愿处于中等水平,肺癌筛查率偏低。医务人员可联合应用2种模型,采取措施使一级亲属建立正确的筛查认知与健康信念,以期促进一级亲属的筛查行为。

关键词: 肺肿瘤, Logistic回归分析, 决策树, 一级亲属, 筛查行为, 影响因素

Abstract: Background Lung cancer is the leading cause of cancer death in China. Early screening of high-risk population is the most effective and key method to find lung cancer and improve prognosis. At present, there are few reports on the influencing factors of screening behavior of the first-degree relatives of lung cancer patients, a concentrated high-risk group. Objective To analyze the influencing factors of early screening behavior of lung cancer in first-degree relatives of lung cancer patients by Logistic regression and decision tree model. Methods 310 first-degree relatives of lung cancer patients who were hospitalized in the Department of Respiratory Medicine and Thoracic Surgery of Shanxi Cancer Hospital from March to June, 2023 were selected as the subjects of investigation. General information questionnaire, lung cancer knowledge questionnaire and Chinese version of lung cancer screening health belief scale were used to investigate. Analyze the influencing factors of lung cancer screening behavior, establish Logistic regression model and decision tree model respectively, and compare the differences between the two models. Results The rate of first-degree relatives of patients with lung cancer who had received lung cancer screening was 23.9% (74/310). The overall awareness rate of lung cancer knowledge was 75.2% (4 662/6 200). 56.5%(175/310) are willing to be screened for lung cancer. Logistic regression results showed age , medical insurance type , whether or not you have received advice on lung cancer screening, willingness to screen, perceived disorder, perceived benefit and perceived risk (P < 0.05). The results of decision tree model show that cognitive impairment and age are the influencing factors of lung cancer screening behavior of first-degree relatives. There was significant difference in the prediction efficiency between Logistic regression model and decision tree model (P < 0.05). Conclusion First-degree relatives of patients with lung cancer have a high level of overall lung cancer awareness but a low level of awareness of screening methods. Their health beliefs and willingness to screen are at a medium level, and the screening rate of lung cancer is low. Medical staff can jointly apply the two models and take measures to make the first-degree relatives establish correct screening cognition and health belief in order to promote the screening behavior of the first-degree relatives.

Key words: Lung neoplasms, Logistic regression analysis, Decision tree, First-degree relatives, Screening behavior, Influencing factor