中国全科医学

• •

基于支持向量机的慢性心力衰竭患者社会衰弱现状及影响因素可解释性分析

卢静, 孙国珍, 王洁, 高敏, 于甜栖, 孙姝怡, 王琴, 温高芹   

  • 收稿日期:2023-12-21 修回日期:2024-02-12 接受日期:2024-03-15
  • 通讯作者: 孙国珍
  • 基金资助:
    国家自然科学基金面上项目(72074124); 江苏省高校优势学科建设工程“三期”护理学(苏政办发[2018]87号)

Interpretable Analysis of Influencing Factors of Social Frailty in Patients with Chronic Heart Failure Based on SVM

LU Jing,SUN Guozhen,WANG Jie,GAO Min,YU Tianxi,SUN Shuyi,WANG Qin,WEN Gaoqin   

  • Received:2023-12-21 Revised:2024-02-12 Accepted:2024-03-15
  • Contact: SUN Guozhen
分享到

摘要: 背景 心衰合并衰弱的综合管理对衰弱评估提出了多维需求,然而作为心衰患者不良健康结局的增量预测因子,衰弱的社会维度尚未得到充分关注。目的:分析慢性心力衰竭患者社会衰弱现状及其影响因素。方法 采用便利抽样法于2022年9月-2023年7月,选取南京市某三甲医院213例慢性心衰患者为研究对象,采用一般资料调查表、HALFT量表、孤独感量表、双向社会支持量表、掌控感量表、抑郁量表进行调查。采用单因素分析、支持向量机-特征递归消除进行特征筛选,构建支持向量机分类模型,引入SHAP(SHapley Additive exPlanations)值进行影响因素分析。结果 慢性心衰患者中社会衰弱前期及社会衰弱人数占比分别为46%,17.8%,支持向量机-特征递归消除模型性能最佳时,采用最优特征子集构建分类模型后进行SHAP可解释性分析,模型预测准确率在训练集和测试集中分别为0.73和0.63,此时特征重要性排序及影响方向从高到低为无运动习惯(+)、掌控感(-)、疾病经济负担重(+)、双向社会支持(-)、抑郁(+)、孤独感(+)、无业(+)。结论 慢性心衰患者社会衰弱问题严重,医护人员应对患者的社会功能予以重视,关注患者缺失的社会资源属性及其上游因素,通过强化外部支持系统、培养患者内在信念、克服负性情绪体验,统筹制定管理方案,整合医疗资源实施干预延缓或逆转患者社会衰弱进程,改善其预后及生活质量。

关键词: 慢性心力衰竭, 社会衰弱, 影响因素分析, SVM-RFE, SHAP

Abstract: Background The integrated management of heart failure with frailty poses a multidimensional need for frailty assessment, but as an incremental predictor of adverse health outcomes in patients with heart failure, the social dimension of frailty has not received sufficient attention. Objective To understand the status of social frailty in patients with chronic heart failure and analyze its influencing factors. Methods From September 2022 to July 2023, convenience sampling was used to select 213 patients with chronic heart failure from a ClassΙΙΙ Grade A hospital in Nanjing as the research objects, the general information questionnaire, HALFT Scale, Loneliness Scale, the Brief 2-Way Social Support Scale, Personal Mastery Scale, and the Patient Health Questionnaire were used to investigate. Univariate analysis and support vector machine-feature recursive elimination were used to filter the feature, the SVM classification model was constructed, and the SHAP value was introduced to analyze the influencing factors. Results The proportion of pre-social frailty and social frailty in patients with chronic heart failure was 46% and 17.8%, respectively. When the SVM-RFE model plays the best performance, the optimal feature subset was used to construct the SVM classification prediction model and perform SHAP interpretability analysis. The accuracy of the model was 0.73 in the training set and 0.63 in the test set, respectively. At this time, the ranking of feature importance from high to low was no exercise habit (+), personal mastery (-), heavy economic burden of disease (+), 2-way social support (-), depression (+), loneliness (+), unemployment (+). Conclusion Chronic heart failure patients with serious social frailty problems. Medical staff should pay attention to the missing resource attributes of patients and their upstream factors, take targeted interventions to delay or reverse the process of social frailty by strengthening the external support system, cultivating inner beliefs, overcoming negative emotional experiences, making overall management plans, integrating medical resources and implementing interventions to delay or reverse the process of social frailty in patients with heart failure, improving their prognosis and quality of life.

Key words: Chronic heart failure, Social frailty, Root cause analysis, SVM-RFE, SHAP