世界中西医结合杂志
世界中西醫結閤雜誌
세계중서의결합잡지
WORLD JOURNAL OF TRADITIONAL CHINESE AND WESTERN MEDICINE
2014年
3期
297-299,318
,共4页
赵灵燕%毕力夫%张亚军%陈建新%赵慧辉%戴军有%王伟
趙靈燕%畢力伕%張亞軍%陳建新%趙慧輝%戴軍有%王偉
조령연%필력부%장아군%진건신%조혜휘%대군유%왕위
2型糖尿病%决策树%中医证候
2型糖尿病%決策樹%中醫證候
2형당뇨병%결책수%중의증후
Diabetes Type 2%Decision Tree%Syndrome of TCM
目的：采用决策树的数据挖掘方法建立2型糖尿病患者痰浊证与临床常规检测指标间的关联模式。方法采用多中心临床流行病学调查方法，在全国5家三级甲等医院共收集249例合格2型糖尿病病例，综合分析基本信息、中医四诊信息、临床常规检测指标。在t检验、非参数检验、Pearson相关分析基础上，进一步采用决策树的数据挖掘方法建立痰浊证与临床常规检测指标间的关联模式。结果249例患者中有106例为痰浊证，占42．57％。以尿素氮、白细胞、平均红细胞体积、超敏C反应蛋白、红细胞、甲状腺素6个核心指标建立了痰浊证决策树模型，10倍交叉验证得到模型的灵敏度为75．47％、特异度为76．22％，检测总正确率为75．90％。结论决策树模型可以清晰、直观的进行2型糖尿病患者痰浊证的判断，在证候客观化研究中显示了一定的优势。
目的：採用決策樹的數據挖掘方法建立2型糖尿病患者痰濁證與臨床常規檢測指標間的關聯模式。方法採用多中心臨床流行病學調查方法，在全國5傢三級甲等醫院共收集249例閤格2型糖尿病病例，綜閤分析基本信息、中醫四診信息、臨床常規檢測指標。在t檢驗、非參數檢驗、Pearson相關分析基礎上，進一步採用決策樹的數據挖掘方法建立痰濁證與臨床常規檢測指標間的關聯模式。結果249例患者中有106例為痰濁證，佔42．57％。以尿素氮、白細胞、平均紅細胞體積、超敏C反應蛋白、紅細胞、甲狀腺素6箇覈心指標建立瞭痰濁證決策樹模型，10倍交扠驗證得到模型的靈敏度為75．47％、特異度為76．22％，檢測總正確率為75．90％。結論決策樹模型可以清晰、直觀的進行2型糖尿病患者痰濁證的判斷，在證候客觀化研究中顯示瞭一定的優勢。
목적：채용결책수적수거알굴방법건립2형당뇨병환자담탁증여림상상규검측지표간적관련모식。방법채용다중심림상류행병학조사방법，재전국5가삼급갑등의원공수집249례합격2형당뇨병병례，종합분석기본신식、중의사진신식、림상상규검측지표。재t검험、비삼수검험、Pearson상관분석기출상，진일보채용결책수적수거알굴방법건립담탁증여림상상규검측지표간적관련모식。결과249례환자중유106례위담탁증，점42．57％。이뇨소담、백세포、평균홍세포체적、초민C반응단백、홍세포、갑상선소6개핵심지표건립료담탁증결책수모형，10배교차험증득도모형적령민도위75．47％、특이도위76．22％，검측총정학솔위75．90％。결론결책수모형가이청석、직관적진행2형당뇨병환자담탁증적판단，재증후객관화연구중현시료일정적우세。
Objective To establish the associated model between turbid phlegm syndrome and clini-cal routine indicators in the patients of diabetes type 2,using data-mining method of decision tree.Methods The multi-central clinical epidemiological investigation was adopted.Two hundred and forty-nine cases of diabetes type 2 were collected from 5 Three-A hospitals.The basic information,the information of four di-agnostic methods of TCM and clinical routine indicators were analyzed comprehensively.On the basis of the t test,nonparametric test and Pearson correlation analysis,the data-mining method of decision tree was adopt-ed further to set up the association model between turbid phlegm syndrome and clinical indicators.Results Of 249 cases,1 06 cases(42.57%)were differentiated as turbid phlegm syndrome.Six core indicators inclu-ding urea nitrogen,white blood cells,average red blood cell volume,hypersensitive C-reactive protein,eryth-rocyte and thyroxin were used to establish decision tree model of turbid phlegm syndrome.Ten times cross validation indicated that the model sensitivity was 75.47%,the specificity was 76.22%and the total accura-cy was 75 .90%.Conclusion The decision tree model clearly and intuitively determines turbid phlegm syn-drome of diabetes type 2 in the patients and displays a certain advantage in syndrome objectification study.
