基于疗效影响因素对加味清络饮治疗类风湿性关节炎临床预测模型与列线图绘制的初步探索

Exploration of a Clinical Prediction Model and Nomogram Based on the Impact Factors of Modified Qingluo Yin in Treating Rheumatoid Arthritis

目的:通过研究加味清络饮治疗类风湿性关节炎预后的相关因素,构建临床预测模型并绘制列线图。方法:收集2018年1月—2019年9月间参与“加味清络饮治疗活动期类风湿性关节炎”临床试验的41例病例。收集48项相关指标,如基础信息、中医症状评分、舌脉之象、生化指标、情志评分及影像学检查结果。采用ACR50疗效标准对患者进行分组后,采用多因素逐步Logistics回归分析疗效的影响因素并构建临床预测模型,内部验证方法为5折交叉验证。使用R语言绘制列线图以构建可视化评分系统。结果:多因素逐步Logistics回归分析显示,粉尘接触史、BMI、中医关节压痛评分、小关节肿胀计数、中医体质、焦虑评分SAS、C-反应蛋白等7项指标,是加味清络饮治疗类风湿性关节炎预后是否能达到ACR50疗效标准的主要影响因素。根据以上因素构建临床预测模型的正确预测率为87.8%;5折交叉验证显示模型预测正确率为75.6%。结论:根据疗效影响因素构建的临床预测模型预测正确率较高。绘制的列线图可根据患者的评分,快速计算出患者服药后达到ACR50疗效标准的概率。

Objective:To study the impact factors of modified Qingluo Yin in treating rheumatoid arthritis and to construct clinical prediction model and draw nomogram.Methods:48 kinds of clinic data were collected from the clinical trial of 41 patients who had accepted the treatment of modified Qingluo Yin.Analysis was conducted on the data including basic information,disease related information,TCM symptom scores,information of tongue and pulse,biomedical indexes,emotion scores and ultrasound imaging results.ACR50 standard was used to evaluate patients and group them.Multivariate Logistics regression analysis was used to screen impact factors and to construct clinical prediction model.Nomogram was made according to the factors involved in Logistics model.Results:Multivariate Logistics regression model showed that dust exposure,BMI,TCM evaluation of pain,TJC of small joints,SAS,physique of TCM and CRP were impact factors in whether one patient could meet ACR50 after the treatment with modified Qingluo Yin.The accurate prediction rate was 87.8%in terms of clinical model prediction.The accurate prediction rate was 75.6%in terms of model prediction showed by cross validation.Conclusion:The accurate rate of clinical model prediction was high based on impact factors.The nomogram developed by this study could predict the possibility of outcome of ACR50.