[Analysis on factors associated with life quality of HIV-infected individuals in antiretroviral therapy in Wuxi based on three models].

内容摘要

Objective: To analyze factors associated with life quality of HIV-infected individuals receiving antiretroviral therapy by using three models and compare the prediction performance of different models in Wuxi. Methods: Using a cross-sectional study design, a questionnaire survey was conducted in HIV-infected individuals receiving antiretroviral therapy recruited through convenience sampling in Wuxi from June to September 2025 with a sample size of 346. The Chinese version of the World Health Organization Quality of Life Questionnaire for HIV was used to evaluate the life quality of HIV-infected individuals. The logistic regression model, decision tree model and random forest model were used to analyze factors associated with the life quality of HIV-infected individuals. The prediction performance of different models was compared by using confusion matrix and the area under receiver operating characteristic curve. Results: A total of 460 HIV-infected individuals were involved, with a total life quality score (79.49±8.46) lower than that in general population of national norm (80.28±17.46)(t=-2.01, P=0.045). The logistic regression analysis showed that risk factors associated with the life quality of HIV-infected individuals included informing no others of HIV infection status (aOR=1.72, 95%CI: 1.04-2.85), anxiety (aOR=2.23, 95%CI: 1.26-4.00), and depression (aOR=4.05, 95%CI: 2.42-6.87), while protective factor was better social support (aOR=0.60, 95%CI: 0.37-0.97). The decision tree model showed that depression, social support, and education level were factors associated with life quality of HIV-infected individuals. The random forest model indicated that main factors associated with life quality of HIV-infected individuals were depression and anxiety, followed by other chronic diseases, education level, and marital status. The accuracy rates of logistic regression, decision tree, and random forest models for the prediction of the life quality of HIV-infected individuals were 0.723, 0.715, and 0.752, respectively. Conclusions: The overall life quality of HIV-infected individuals was lower than that of general population, which was mainly associated with factors, such as depression, social support and education level. All the three models exhibited good accuracy in prediction of the life quality of HIV-infected individuals, but random forest model showed the better overall prediction performance. 目的： 运用3种模型分析无锡市抗病毒治疗HIV感染者生存质量的相关因素，并比较不同模型的预测性能。 方法： 采用横断面研究设计，2025年6-9月通过方便抽样方法在无锡市招募抗病毒治疗的HIV感染者开展问卷调查，样本量为346例。使用中文版WHO艾滋病生存质量简表评估调查对象的生存质量。采用logistic回归模型、决策树模型和随机森林模型分析HIV感染者生存质量的相关因素。通过混淆矩阵与受试者工作特征曲线下面积比较不同模型的预测性能。 结果： 共调查HIV感染者460例，生存质量总分（79.49±8.46）分低于全国常模的一般人群（80.28±17.46）分（t=-2.01，P=0.045）。Logistic回归模型分析结果显示，HIV感染者生存质量的风险因素包括HIV感染状况未告知他人（aOR=1.72，95%CI：1.04~2.85）、焦虑（aOR=2.23，95%CI：1.26~4.00）、抑郁（aOR=4.05，95%CI：2.42~6.87），保护因素为社会支持较好（aOR=0.60，95%CI：0.37~0.97）。决策树模型分析结果显示，抑郁、社会支持和文化程度是HIV感染者生存质量的相关因素；随机森林模型分析结果显示，HIV感染者生存质量的主要相关因素为抑郁和焦虑，次要相关因素为其他慢性病、文化程度和婚姻状况。Logistic回归模型、决策树模型和随机森林模型对HIV感染者生存质量预测的准确率分别为0.723、0.715和0.752。 结论： HIV感染者的整体生存质量低于一般人群，且主要与抑郁、社会支持和文化程度等因素有关。Logistic回归模型、决策树模型和随机森林模型在预测HIV感染者生存质量方面均有良好的准确性，但随机森林模型的综合预测性能较佳。.