中国中西医结合杂志2025，Vol.45Issue(9)：1050-1058,9.DOI:10.7661/j.cjim.20250430.257
基于免疫代谢综合因素建立活动期类风湿关节炎寒热证型客观化诊断模型
Establishment of Objective Diagnostic Model for Active Rheumatoid Arthritis Patients with Cold and Heat Syndrome Based on Comprehensive Factors of Immune Metabolism
摘要
Abstract
Objective To establish a discriminant model for cold-heat Bi syndrome classification in active rheumatic arthritis(RA)patients by differentiating between cold and heat subtypes in active RA,analyzing their metabolic and immunological differences,and employing machine learning methods to incorporate multi-dimensional indicators on the basis of"cold-heat three phases"theory for treating Bi syndrome.Methods A total of 453 RA patients treated at Yijishan Hospital,Wannan Medical College from July 2022 to June 2023 were screened,and 60 of them met the inclusion criteria including 26 with damp-heat arthralgia syndrome and 34 with cold-dampness blocking syndrome.Twenty healthy volunteers were recruited as control group.Clinical blood routines,biochemical indictors,rheumatoid three-item tests,immunoglobulin(Ig)were detected.Serum levels of lactate(LAC),ketone bodies(KET),free fatty acids(FFA),glucose/lipid metabolism markers,and white adipose tissues(WAT)-related inflammatory/immune indicators in serum were measured.Intergroup differences were comparatively analyzed.Ten machine learning algorithms including logistic regression(LR),and linear discriminant analysis(LDA)were used to screen feature indicators and develop prediction models,with ten-fold cross-validation used to evaluate model stability,while the model performance was evaluated using F1 score and the area under the receiver operating characteristic curve(AUC).Results Compared with the control group,in glucose and lipid metabolism profiles of RA patients,triglycerides(TG)levels significantly decreased(P＜0.01),while fasting blood glucose(GLU)levels showed an upward trend(P＜0.01),along with increased levels of carnitine palmitoyltransferase-1(CPT-1),a key enzyme of fatty acid oxidation increased(P＜0.05),and KET accumulation.At the same time,the concentration of fatty acid transporter 1(FABP1)in the patients with heat Bi syndrome was up-regulated(P＜0.01).Moreover,FFA such as palmitic acid,linoleic acid,and stearic acid in blood of RA patients significantly decreased(P＜0.05),especially in the case of heat Bi syndrome.There were no significant differences in immune indices between cold Bi syndrome and heat Bi syndrome patients(P＞0.05).Ten commonly used machine learning models better predicted the subtypes of cold and heat Bi syndromes using 22 parameters including interleukin-17A(IL-17A),leptin(LEP),3-phosphoglycerate isomerase(3-PGI),matrix metalloproteinase 3(MMP3),among which the LDA model demonstrated superior performance as evidenced by an average AUC of 0.74 through ten-fold cross-validation.Conclusion The above results indicated that differences in immune and metabolic phenotypes were the bases for the development of cold and heat Bi syndromes,and integration of multi-modal indicators significantly enhanced the capabilities of subtype discrimination.关键词
类风湿关节炎/免疫/代谢/疾病分型/机器学习/诊断模型Key words
rheumatoid arthritis/immune/metabolism/disease classification/machine learning/diagnostic model引用本文复制引用
廖强,冯丹丹,黎青青,程修平,李国豪,李艳,左坚..基于免疫代谢综合因素建立活动期类风湿关节炎寒热证型客观化诊断模型[J].中国中西医结合杂志,2025,45(9):1050-1058,9.基金项目
国家自然科学基金资助项目(No.81973828,No.82274465) （No.81973828,No.82274465）
安徽省科技重大专项计划项目(No.202303a07020001) （No.202303a07020001）
