机构地区:[1]杭州市萧山区中医院,311200 [2]杭州市萧山区第二人民医院,311241
出 处:《浙江临床医学》2025年第3期353-355,共3页Zhejiang Clinical Medical Journal
基 金:杭州市萧山区重大科技项目(2022207);浙江中医药大学附属医院科研专项资助计划(2022FSYYZY35);浙江省中医药科技计划项目(2024ZL787);国家中医药管理局第五批中医临床优秀人才研修项目(2022-1)。
摘 要:目的评价两种28个关节疾病活动评分(DAS28)在评估类风湿关节炎(RA)活动度的作用效果及差异,并应用机器算法构建新的活动性评估模型。方法回顾性分析2022年1月至2023年12月132例RA患者的临床资料,收集患者一般资料,包括年龄、性别、病程、关节压痛数、关节肿胀数、患者对自身病情的总体评价(GH)、CRP、ESR、类风湿因子(RF)和抗环瓜胺酸抗体(抗-ccp抗体)等指标,比较DAS28-ESR和DAS28-CRP两种评分在类风湿关节炎疾病活动度的作用及差异。通过R软件执行XGBoost算法构建一个机器模型评价RA的活动性,并采用SHAP方法对该模型特征因素进行重要性排名。结果DAS28-ESR和DAS28-CRP评分分别为3.68(3.42,3.95)分和3.21(2.96,3.47)分,两组比较差异有统计学意义(P<0.01)。相关分析显示,DAS28-ESR和DAS28-CRP存在正相关(r=0.908,P<0.01)。两者在低中度活动期评分差异率分别为52.78%和52.33%。机器算法构建的新模型特征因素包括关节肿痛数、关节压痛数、ESR、CRP、年龄及VAS评分。采用SHAP对新模型的特征因素进行重要性排序,结果显示关节压痛数贡献最高。结论DAS28-ESR和DAS28-CRP两种评分对RA活动度的判断存在差异,这种差异在低中度活动期RA患者中更明显,临床医师在实践中应基于这种差异作临床决策。新模型中关节压痛数对活动性评价贡献最高,可帮助指导临床实践活动。Objective To evaluate the effectiveness and differences between two 28-joint disease activity scores(DAS28)in assessing rheumatoid arthritis(RA)activity,and to construct a new activity assessment model using machine learning algorithms.Methods A retrospective analysis was conducted on the clinical data of 132 RA patients from January 2022 to December 2023.General patient information was collected,including age,gender,disease duration,tender joint count,swollen joint count,patient global assessment of disease activity(GH),C-reactive protein(CRP),erythrocyte sedimentation rate(ESR),rheumatoid factor(RF),and anti-cyclic citrullinated peptide antibody(anti-CCP antibody).The roles and differences between DAS28-ESR and DAS28-CRP in assessing RA disease activity were compared.The XGBoost algorithm was implemented using R software to build a machine learning model for evaluating RA activity,and the SHAP method was used to rank the importance of the model's feature factors.Results The DAS28-ESR and DAS28-CRP scores were 3.68(3.42,3.95)and 3.21(2.96,3.47),respectively,with a statistically significant difference between the two groups(P<0.01).Correlation analysis showed a linear relationship between DAS28-ESR and DAS28-CRP(r=0.908,P<0.01).The difference rates between the two scores in low-to-moderate disease activity were 52.78%and 52.33%,respectively.The new model constructed by the machine learning algorithm included the following feature factors:swollen joint count,tender joint count,ESR,CRP,age,and VAS score.Using SHAP to rank the importance of the feature factors in the new model,the results showed that tender joint count contributed the most.Conclusion There are differences between DAS28-ESR and DAS28-CRP in assessing RA activity,and these differences are more pronounced in patients with low-to-moderate disease activity.Clinicians should consider these differences when making clinical decisions.In the new model,tender joint count contributed the most to the assessment of disease activity,which can help guide clinical
关 键 词:关节炎 类风湿 28个关节疾病活动评分 疾病活动度
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...