基于Logistic回归模型的高血压、血脂异常、肥胖与糖尿病共病关联分析

Analysis of Comorbidity Associations among Hypertension, Dyslipidemia, Obesity, and Diabetes Based on a Logistic Regression Model

摘要: 本文研究了高血压、高血脂、肥胖与糖尿病这四种疾病的关联性。首先对数据进行预处理，用均值替代缺失值、删除异常值，对数据进行统计学分析、卡方检验说明研究对象之间存在显著性差异，并依据临床标准将连续指标转换为0/1二分类变量(1表示异常，0表示正常)；接着采用皮尔逊相关系数计算四种疾病间两两的线性相关系数，通过相关系数值衡量关联强度；最后以糖尿病为二分类因变量、其余三种疾病为自变量构建Logistic回归模型，分析各自变量对糖尿病的影响程度。结果显示，四种疾病均呈弱正相关，其中部分疾病间关联相对明显；Logistic回归模型中，部分自变量对糖尿病存在正向影响，且部分影响具有统计学显著性。

Abstract: This paper studies the relationships among four chronic diseases: hypertension, dyslipidemia, obesity, and diabetes. First, data pre-processing was conducted, involving imputation of missing values using the mean, removal of outliers, and statistical description. A chi-squared test was then performed, revealing significant differences in the distribution of the diseases among the study subjects. Continuous indicators were dichotomized into binary variables (1 for abnormal, 0 for normal) based on clinical standards. Then, Pearson correlation coefficients were calculated to quantify the linear correlations between the four diseases to assess the strength of their associations. Finally, a Logistic regression model was constructed with diabetes as the binary dependent variable and the other three diseases as the independent variables. This analyzed the degree of influence that each independent variable experts on diabetes. The results indicated weak positive correlations among all four diseases, with some pairwise associations being relatively more pronounced. Within the Logistic regression model, some independent variables exhibited a positive influence on diabetes, with part of these influences reaching statistical significance.