Study On The Quantized Diagnosis Of Syndromes In TCM Of UC Based On Latent Variable Model

ObjectiveBased on the literature search,expert consultation and other methodologies to establish TCM symptoms and signs library of ulcerative colitis,application of structural equation model theory and item response theory and other modern latent variables method,combining with the Chinese medicine theory,thus the syndrome diagnostic model was constructed.Based on the clinical data collected in Guangdong area,for example,it demonstrated how to set up a quantized diagnosis of syndromes model or scoring system which is in accordance with TCM syndrome differentiation and objective,explored the TCM syndrome classification quantitative methodology,and determined the differentiation entry and diagnostic threshold of ulcerative colitis with large intestinal damp-heat.MethodsBy the use of epidemiology,statistics and the Chinese medicine theory,we are here to study quantitative research of TCM syndrome types of ulcerative colitis.Firstly,according to the literature research and expert consultant method,we made symptom information collection table to carry out the clinical epidemiological investigation to collect data.Then,we explored the methodology of dialectical diagnosis model,introduced the latent variable model(structural equation model and item response theory),filter entries and built the latent variable model and established a scoring system to calculate the score.Later,ROC curve method is used to determine the type of syndrome diagnosis threshold.Finally,a diagnostic test design scheme is further tested and verified for the authenticity and reliability.Results1.Entry library and clinical investigation:the formation and type of syndrome and syndrome library entries are derived from each group to develop consensus,standards and the internal medicine of traditional Chinese medicine teaching material,based on the qualitative optimization library entry form e.hot and humid,wet aggregates,fever,the mixed pixu(spleen deficient)change,spleen and kidney Yang deficiency,Yin anemia empty,weakspleen and stomach,qi and blood stasis type,hot and humid nine pixu(spleen deficiency)certificate entries,81 syndrome entries,forming the ulcerative colitis often witness syndrome type information collection form.Clinical survey collect 770 cases,249 cases(32.3%),followed by e.humid card wet aggregates 123 cases(16.0%),followed by wet and heat syndrome,117 cases(15.2%),fever syndrome,106 cases of mixed(13.8%),liver depression,spleen deficiency 38 cases(4.9%),spleen weakness syndrome,19 cases(2.5%),spleen and kidney Yang deficiency syndrome,17 cases(2.2%),Yin anemia deficiency syndrome in 4 cases(0.5%),blood stasis,intestinal collaterals syndrome type in 3 patients(0.4%),other certificate,94 cases(12.2%).All of these above should meet the requirements of sample size,eliminate the evidence that does not meet the minimum sample size model,this study which chooses e.wet and heat syndrome in patients with UC as further research object is feasible.2.Theory model and SEM analysis results:filter entries,combined with the diagnostic standard of integral method and expert consultation method,selection of e.hot and humid card entry;The e.hot and humid card into "common(SR)module","hot(R)module" and "wet(S)module",including "the universal(SR)module" positioning for entries are:the generality of the large intestine diarrhea,mucus,and purulent blood,abdominal pain,tenesmus,and as a "wet(S)" and "hot"(R)personality module,respectively "anus heat,red tongue,yellow moss,fever,dry mouth,mouth bitter,bad breath,urine,pulse number slip" arid "moss greasy,poor appetite,eat less and less god" defecate,get absolute fitting model is:the ?2/df = 863.57/132 = 6.54,GFI = 0.89,AGFI=0.86,RMSEA = 0.09;Relative fitting index:CFI = 0.63,TLI = 0.57,NFI =0.60.Use frequency method,crowns Bach coefficient method,correlation coefficient method,factor analysis method and combining with TCM clinical practice,adjusting model is as follows:"the universal(SR)module" reserved"diarrhea,mucus,purulent blood,abdominal pain," four entries;"Hot(R))module" reserved "anus heat,red tongue,yellow moss,fever,dry mouth,"five entries;"Wet module(S)" reserved "moss greasy and defecation" two entries.Model inspection again,to get absolute fitting model is:the ?2/df =153.71/41 = 3.75,GFI = 0.97,AGFI = 0.94,RMSEA = 0.06;Relative fitting index:CFI = 0.90,TLI = 0.86,NFI = 0.87.The result shows that the model fitting has improved than before,and the model is acceptable.3.IRT analysis results:of the above analyses IRT entries in the SEM model,selection of single parameter Logistic model(IPLM),degree of differentiation of constant 1.7,difficulty parameter interval in[3],the SEM to establish the model of 11 items overall ideal quality,and there is no entry being cut.Each patient' s has reached the standardization of syndrome score and the subjects ability is valued on 11 projects,and the distribution is between[1.81,1.09]as a result,the standard deviation between[0.40,0.40].Considering the clinical operation convenience,turn the capacity value weights,get 11 items into a weight points:diarrhea(15),mucous stool(13),bloody purulent stool(12),abdominal pain(15),the anus heat(5),red tongue(14),yellow moss(13),fever(1),dry mouth(6),moss greasy(6)and defecation not great(1).4.ROC analysis results:expert dialectic as a gold standard,we score as a diagnostic scoring model,build the ROC curve,according to the degree of sensitivity and specific diagnostic threshold;Transformation area under the ROC curve results:weights of integral method conversion weighted integral take 81.5 points,Youden index,the largest is 0.723,the area under the curve is 0.915,P<0.01,suggesting the scoring system,in our study of e.hot and humid card has high diagnostic significance and value.5.Clinical validation results:144 cases with input validation cases to scale the authenticity and reliability of the test sample data,the conversion of weighted integral method sensitivity of 0.923,0.867 with specific degree,Youden index was 0.790,compared with other three integral methods are all high,prompt authenticity;Kappa value was 0.725,tips and jinbiao accord with high degree,good reliability.ConclusionThrough the collection of TCM syndrome and symptoms entries,we establish entry library of UC in a feasible way and formulate a library based on the basic theory of TCM and the theoretical model of expert opinion in accordance to the qualitative items libraries combined with the literature search,the book browsing and expert discussion,successful development of UC syndrome differentiation questionnaire clinical investigation.By the use of structural equation model and IRT,we establish model?screening of entries.Then,the diagnosis is established by using ROC curve.The diagnosis scale that we made is simple and easy to operate.Through the clinical verification,the accuracy and reliability,the results are higher,showing good clinical diagnostic value,and it can provide beneficial reference for physician's clinical syndrome differentiation.Idea and method of this study can prove to be useful for other disease syndromes quantitative study of the demonstration.This research adopts the theory of latent variables(including SEM and IRT)combined with the TCM theory to quantitatively studied UC TCM syndromes.By the use of the visual graphics,clear semantics,fine data,we can provide the basis of syndrome differentiation and treatment of UC,then establish a set of card type of quantitative research methods based on a certain method innovation.Since the establishment conforms to the principle of traditional Chinese medicine,its features and advantages of UC quantitative indicators are to determine the type of diagnostic threshold,make the diagnosis objective and standardized.The disadvantage of the study is the limitation of time and sample size,we need to increase sample size and extend the type of syndrome to verify the reliability,authenticity and the practicability of the scoring system.