| Objective(1)Adopt the decision tree classification prediction method of data mining to establish a learning model of the phlegm-heat excess syndrome-heat excess syndrome,and improve the objectivity of the empirical dialectic of it.(2)The phlegm-heat excess syndrome-heat excess syndrome characteristics of intestinal mucosal barrier and brain intestinal peptide were studied by case-control study.Methods(1)Data mining decision tree classification prediction method is adopted to establish the relieving phlegm-heat excess syndrome machine learning model:using encephalopathy stroke database established by the encephalopathy team of our hospital(1975-2005),by means of case retrospective studies within 3 days of stroke onset of 1018 cases of hospitalized patients with age,type of disease,illness weight,character,habits,past medical history,initial stroke state into four diagnostic information,and according to the patient from the hospital to the onset of 14 days course record,record whether there appear relieving phlegm-heat excess syndrome empirical,the presence of phlegm-heat excess syndrome relieving empirical two groups on the single factor analysis of exploration of relieving phlegm-heat excess syndrome empirical related influencing factors.Based on the method of decision tree classification prediction,an empirical prediction model was established.(2)Adopt the method of case-control study relieving phlegm-heat excess syndrome positive intestinal mucosal barrier and the characteristics of the brainintestinalpeptide:to system research of 61 patients with stroke(including age,gender,lifestyle,personality,past medical history,four diagnostic information,etc.)and collect blood samples within 7 days of stroke onset.By using the method of case-control study to distinguish the phlegm-heat excess syndrome and non phlegm-heat excess syndrome on the clinical features,intestinal mucosa damage degree,diamine oxidase(DAO),intestinal mucous membrane permeability(D-lactic acid,D-LAC),plasma levels of endotoxin lipopolysaccharide(LPS)and Ghrelin(auxin(Ghrelin),satiety molecule(Nesfatin-1),neuropeptide Y(NPY)and vasoactive intestinal peptide(VIP),calcitonin gene related peptide(CGRP)),analyze the intestinal mucosa function and brain intestinal peptide change of phlegm-heat excess syndrome.Results(1)Single factor analysis was carried out for two groups with or without.phlegm-heat excess syndrome,found the characteristics of phlegm-heat excess syndrome is age 70 or higher(P<0.05),NIHSS>6 points(P<0.01),(P<0.01)in bed,disturbance of consciousness(P<0.01),cerebral hemorrhage(P<0.01),the viscera(P<0.05),the type of stroke(P<0.05),the circle of completely before infarction(P<0.05).Has words in clinical four diagnostic information Jian acerbity(P<0.01),adverse comes(P<0.01),water choking cough(P<0.01),and it dried(P<0.01),pale tongue(P<0.01),tone stinking(P<0.01),be agitated or upset irritability(P<0.01).partial body numbness(P<0.01).urinary incontinence(P<0.05),slow(P<0.01),indifferent to expression(P<0.05),vague language(P<0.01),stay(P<0.05),pale tongue(P<0.01),red tongue(P<0.01),tongue is violet(P<0.05),string(P<0.05),white moss(P<0.01),yellow moss(P<0.01),thin moss(P<0.01),thick(P<0.01),yellow greasy moss(P<0.01),and is fine veins(P<0.05),(P<0.05)as the related influence factors,significant difference was found in two groups.(2)To make the decision tree classification prediction study of the above differences.and to obtain the classification rules of four clinical significance factors of phlegm-heat excess syndrome in clinical significance,as:Rule 1:apoplexy+bad breath(64.3%),that is,the incidence percentage of phlegm-heat excess syndrome in patients of stroke with is bad breath 64.3%.Rule 2:the stroke+in bed+NHISS score more than9points+red tongue(65.5%),namely the stroke patients lie in bed and the degree of nerve function defect is greater than 9 points,the red tongue,then the incidence percentage of phlegm-heat excess syndrome may was 65.5%.Rule 3:apoplexy+in bed+yellow moss+bad breath(90.9%),that is,patients with apoplexy who had been lying in bed with liver-yellow and smelly breath,then the incidence percentage of phlegm-heat excess syndrome may was 90.9%.Rule 4:stroke+yellow moss+in bed+tone stinking+agitated or upset irritability(90.1%),namely the stroke patients lie in bed and moss yellow,tone stinking with restless or irritable,then the incidence percentage of phlegm-heat excess syndrome may was 90.1%Using 85%data to train the classification prediction model of the decision tree,the accuracy rate of the phlegm-heat excess syndrome-fu-fu empirical prediction was 70.1%,and the remaining 15%data were used to predict the test with 72.8%accuracy.(3)According to the comparative study of with or without.phlegm-heat excess syndrome by neural function defect scale,hospitalization duration,pulmonary infection,Result that,the NIHSS score,average hosp-italization time,pulmonary infection of phlegm-heat excess syndrome group were higher than the non group,difference has statistical significance.(4)The differences of age,NIHSS score,consciousness disorder and dysphagia in the two groups were statistically significant.There was no significant difference between the two groups in the degree of intestinal mucosal injury,the level of brain intest-inal peptide and the plasma LPS level.There were statistically significant differences between the two groups in the plasma LPS level.Conclusion(1)Based on the difference analysis results between phlegm-heat excess syndrome and non-phlegm-heat excess syndrome,the results of decision tree classification and prediction are better to distinguish the phlegm-heat excess syndrome.(2)The greater the degree of neurological impairment,the greater the possibility of phlegm-heat excess syndrome,and the greater the possibility of pulmonary infection,the longer the hos-pital stay.(3)The LPS level of patients with phlegm-heat excess syndrome was significantly higher than that of non-phlegm-heat excess syndrome in the Primary stroke patient. |
PDF Full Text Request