| ObjectivesThe goal of this study is to analyze the clinical characterization of the blood stasis syndrome of chronic heart failure in coronary heart disease,and to explore the correlation between the blood stasis syndrome of chronic heart failure in coronary heart disease and the physical and chemical indicators,and further to establish the association model of the decision tree model and the neural network model to explore the blood stasis syndrome and its physical and chemical indicators.This study provides a foundation for the diagnosis of the blood stasis syndrome of chronic heart failure in coronary heart disease.MethodsRetrospective studies were used to study the datasets of 218 patients obtained from Department of Cardiology in Zhengzhou T.C.M Hospital from December of 2015 to December of 2017,including the information of four diagnostic consultations of Traditional Chinese Medicine,laboratory test results,and functional test results,etc.The patients were divided into two groups,blood stasis group and non-blood stasis group,according to the differentiation of syndrome.The?~2 test,and Multivariate Logistic regression were applied to compare the details of the two groups,including symptoms,signs,tongue diagnosis,and pulse diagnosis and other indicators,in order to analyze the main clinical characterization of the blood stasis syndrome.The t-test,rank sum test,and Logistic regression analysis were used to investigate the relationship between the blood stasis syndrome and the physical and chemical indicators.Weka3.8.3 was used to establish the decision tree model and Tensorflow to establish the neural network model exploring the association model between the combination of physical and chemical indicators and the blood stasis syndrome.Results(1)This study totally collected 218 cases that met the inclusion criteria,including125 male patients(57.3%),93 female patients(42.7%),37 Heart Function Class II cases(17.0%),and 181 Heart Function Class III cases(83.0%).This study found that on average the female patients were significantly older than the male patients(P<0.05).However,there was no statistically significant difference in the heart function classification between females and males(P>0.01).(2)Clinical Characterization of Blood Stasis SyndromeBasic information:there was no statistically significant difference in age,gender,nationality,BMI,and functional classification between the patients of the blood stasis syndrome and non-blood stasis syndrome(P>0.05).Symptoms and signs:fatigue,shortness of breath,difficulty in breathing,chest tightness,wheezing,lip purpura,loss of appetite,lower extremity edema,insomnia,difficulty in weight,heavy cumbersome limbs,dreaminess,chest pain,bruising,cold hand&foot,ventosity,irritability,nausea,frequent nocturia,urinary insufficiency,rib-side distention,and rib-side pain were more common in blood stasis syndrome(P<0.05).Tongue images:the dark purple tongue and the tongue with ecchymosis were generally found in the blood stasis syndrome group,while the red tongue was more common in the non-blood stasis syndrome group(P<0.05).Pulse:the unsmooth pulse was a common symptom in the blood stasis syndrome group(P<0.05).Logistic regression results:totally 6 indicators were screened in descending order of OR:lip purpura(OR=50.195,P<0.001),irritability(OR=10.196,P=0.006),chest pain(OR=8.164,P=0.004),insomnia(OR=6.596,P=0.003),tongue ecchymosis(OR=5.755,P=0.013),unsmooth pulse(OR=5.311,P=0.021).(3)Correlation Analysis between Blood Stasis Syndrome and the Physical and Chemical IndicatorDifference in comparison:PLT,Cr,Ca,TG,LVPW,LVEF and BNP of blood stasis group were higher than those of non-blood stasis group and the difference was statistically significant(P<0.05);AST,Na,CHOL and LDL of blood stasis group were lower than those of non-blood stasis group and the difference was statistically significant(P<0.05).Univariate factor analysis results:there was a correlation between the blood stasis syndrome and Na(OR=0.084,P=0.045)?Ca(OR=3.858,P<0.001)?TG(OR=1.399,P=0.015)?CHOL(OR=0.774,P=0.037)?LDL(OR=0.649,P=0.013)?LVPW(OR=0.692,P=0.004)?LVEF(OR=0.862,P=0.001)?BNP(OR=1.004,P=0.003).Logistic regression results:Totally six indicators were selected.According to the OR values from high to low,they were:Ca(OR=6.464,P<0.001),TG(OR=1.830,P=0.001)?BNP(OR=1.004,P=0.043),LVEF(OR=0.877,P=0.017),LVPW(OR=0.699,P=0.018),LDL(OR=0.661,P=0.049).(4)The Data Mining-Based Association Model of Physical and Chemical IndicatorsDecision tree mode results:there were 7 indicators picked up:PLT,Ca,LVOW,LVEDD,WBC,AST and HDL.Among them,LVPW was the best identification variable for blood stasis,and the accuracy rate of the model was 90.8%.Neural network model results:the combined model of the clinical performance of Lip purpura and chest pain,and the physical and chemical indicators of AST,PLT,LVPW,BNP,and Ca was the best,with an accuracy of 85.1%.Conclusions(1)Blood stasis syndrome is the main syndrome of chronic heart failure in coronary heart disease.(2)Lip purpura,irritability,chest pain,insomnia,tongue ecchymosis,and unsmooth pulse are main clinical traits associated with the blood stasis syndrome of chronic heart failure in coronary heart disease.(3)Ca,TG,LDL,BNP,LVEF and LVPW are the main the physical and chemical indicators associated with the blood stasis syndrome of chronic heart failure in coronary heart disease,and can be used as the objective index to determine of the blood stasis syndrome.(4)The association model of the core extracted physical and chemical indicators from the decision tree model and the clinical characterization from the neural network model can effectively identify the blood stasis symptom. |
PDF Full Text Request