Based On Data Mining, Acute Ischemic Stroke Disease, Phlegm And Blood Stasis, Clinical Evidence Research

Objective:Through comparison between subjects in the syndrome of phlegm-heat and fu-organ excess(study group)and other syndromes(control group),and comparison within the study group before and after the treatment of relaxing bowels,clarify the pathological changes occurring in acute ischemic stroke patients of the study syndrome,from symptoms,syndromes,disease to laboratory indexes.Making use of various data mining algorithms,we set up data models to predict the estabilishemnt and severity of the syndrome of phlegm-heat and fu-organ excess in acute ischemic stroke patients.Methods:We enrolled qualified acute ischemic stroke patients,and then allocated them to the study group or the control group according to syndrome differentiation criteria of traditional Chinese medicine(TCM).Assessments were conducted once they were recruited,involving clinical scales such as TCM syndrome factor assessments and National Institutes of Health Stroke Scale(NIHSS),and laboratory indexes mainly covering inflammation and endocrine system.Through horizontal comparison between the study group and control group,from demographic information,symptoms,syndrome factors,patients' condition to laboratory indexes,we got the characteristics of the subjects in the study group.Stepwise logistic regression analysis was used to screen out the clinical characteristics which influenced the establishment of the study syndrome.Then we applied gradient boosted decision tree(GBDT)to set up data model for predicting the establishment of the study syndrome.For subjects in the study group,we conducted the assessments once more time after treatment of relaxing bowels.Through vertical comparison within the study group before and after treatment,we got the signicantly changed clinical characteristics.Then we set up artificial neural network(ANN)model to predict the severity of the study syndrome using these characteristics.By combining of the horizontal and vertical comparisons together,we could understand the clinical characteristics of the study syndrome in acute ischemic stroke better.Results:A total of 267 qualified subjects were enrolled,117 in the study group and 150 in the control group.Compared with the control group,subjects in the study group had higher body mass index(P<0.05),longer course of disease(P<0.05)and higher systolic pressure(P<0.01).Subjects in the two groups shared many high-frequency symptoms,tongue signs and plus conditions.However,manifestations of inner fire and phlegm-dampness were more extensive and prominent in the study group.Inner fire,phlegm-dampness and yin-deficiency showed up more frequently in the study group than the control(P<0.01,0.05),and the combinations of syndrome factors were more complicated in the study group(P<0.01).The percentages of subjects showed aphasia,facioplegia or hemiplegia were higher in the study group(P<0.05,0.01)and their overall conditions were worse than the control group(P<0.01).Subjects in the study group showed obvious differences in the function of hypothalamic-pituitary-thyroid/adrenal axis when compared with the control(P<0.01).Besides,subjects in the study group had higher level of lipoprotein(a)(P<0.01)and lower neuron-specific enolase(P<0.05),but no difference in the inflanmatory indicators or renin-angiotensin-aldosterone system(P>0.05).Inputting the differences mentioned above into logistic regression analysis,results showed that the higher the NIHSS score and the lower levels of free triiodothyronine and cortisol,the more likely that syndrome of phlegm-heat and fu-organ excess to establish.Making use of GBDT to predict the establishment of the study syndrome,the accuracy of the data model reached an average of 0.704 and a maximum of 0.84.After receiving comprehensive treatment,manifestations of inner wind,inner fire and phlegm-dampness reduced markedly in the study group.Subjects also improved greatly in the consistence of stools,concomitant symptoms,and frequency,difficulty,speed and awareness of passing stools.Except qi-deficiency,all other syndrome factors attenuated dramatically(P<0.01).The combinations of syndrome factors were not as complicated as before(P<0.01)and the overall conditions improved greatly(P<0.01,0.05).The NIHSS score pre-treatment correlated positively with the average constipation score(P<0.05).The degree of syndrome improvement correlated positively with the condition improvement(P<0.01).After treatment,the levels of free triiodothyronine,plasma renin activity and homocysteine decreased significantly(P<0.01,0.05),while the level of thyroid stimulating hormone increased(P<0.05).Using ANN,we set up a data model to estimate the severity of the study syndrome.Plasma renin activity contributed the most prominently for the output,followed by TCM condition assessment score,NIHSS,levels of homocysteine,free triiodothyronine and thyroid stimulating hormone.Conclusion:Syndrome of phlegm-heat and fu-organ excess in stroke patients reflects the core pathogenesis of stroke.The extensive phlegm-fire gives rise to a series of clinical symptoms and signs.The evolution of inner syndrome decides the changes of manifestations and prognosis.The formation and improvement of the syndrome of phlegm-heat and fu-organ excess is the results of complicated interaction of multiple systems,especially the neuroendocrine system.Making use of data mining algorithms for predicting the establishment and severity of TCM syndrome can help to authenticate the scientificity of TCM syndrome differentiation,providing new approach for further TCM research.