Clinical Characteristics Of Qi Deficiency Syndrome In The Initial Stage Of Ischemic Stroke, Construction Of Discrimination Model And Biological Research Based On Metabolomic

Ischemic stroke belongs to the category of "stroke" in traditional Chinese medicine.It has the characteristics of rapid onset,rapid change,high incidence rate,and high rate of death and disability.TCM treatment based on syndrome differentiation has unique characteristics and advantages for the diagnosis and treatment of this disease,especially the study of initial syndrome is of great significance for the diagnosis and treatment of this disease.Qi deficiency syndrome is one of the important syndromes of ischemic stroke.The current research on this syndrome focuses on the recovery period of stroke,while the understanding of stroke with Qi deficiency syndrome as the initial syndrome is relatively insufficient.The diagnosis of Qi deficiency syndrome is mainly based on four diagnostic information,lacking the joint diagnosis of objective physical and chemical indicators.The research on the internal mechanism of the initial Qi deficiency syndrome is also relatively scarce.This study takes this as the starting point,The system presents the syndrome research form with the core of clinical characteristics of initial Qi deficiency syndrome-discriminant model-biological basis,which provides reference for clinical syndrome research,and improves the clinical ability to identify Qi deficiency syndrome,thereby increasing clinical efficacy.Objective1.Based on the cross-sectional study,the clinical characteristics of the initial qi deficiency syndrome were observed,and the influencing factors of the initial qi deficiency syndrome were explored by using Logistics regression analysis.2.Using the method of multi classification machine learning fusion BP neural network,we initially constructed the discriminant model of the fusion of traditional Chinese and western medicine features of the initial state of qi deficiency syndrome of ischemic stroke,and improved the objective standard of clinical discrimination of the initial state of qi deficiency syndrome.3.Based on metabonomics,the biological research of qi deficiency syndrome in the initial state of ischemic stroke was explored.Methods1.To explore the clinical characteristics and influencing factors of the initial qi deficiency syndrome:a cross-sectional study was conducted to collect 361 ischemic stroke patients who met the inclusion and exclusion criteria from January 2020 to April 2022.While collecting the patient’s age,gender,personal history,past history,four diagnostic information,laboratory indicators and other information,According to the score of Diagnostic Scale of Ischemic Stroke Syndrome Elements,the patients were divided into Qi deficiency syndrome group(Q group)and non Qi deficiency syndrome group(T group),including 61 cases in Qi deficiency group and 300 cases in non Qi deficiency group.To compare the clinical information and biochemical indicators between Qi deficiency syndrome and non Qi deficiency syndrome,and summarize the clinical characteristics of Qi deficiency syndrome.Subsequently,the indexes with statistical differences were included in the multivariate regression analysis to explore the relevant influencing factors of Qi deficiency syndrome in the initial state of ischemic stroke.2.Based on multi-classification machine learning and BP neural network,a discriminant model for the initial state of Qi deficiency syndrome was constructed.A total of 56 items with a frequency of more than 5%in the four diagnostic information of traditional Chinese medicine and objective physical and chemical indicators of Western medicine were included in the study.Firstly,LASSO regression algorithm was used to reduce the dimension of features,and a total of 26 features with coefficients not equal to 0 were selected,which were then entered into the subsequent supervised machine learning algorithm,All algorithms are built in the Python language environment.The data is divided into training set and test set at a ratio of 8:2,and then seven traditional machine learning methods,including decision tree,K-nearest neighbor,support vector machine,random gradient descent method,random forest,extreme random tree,and extreme gradient enhancement algorithm,are used to fuse BP artificial neural network to build a discrimination model of qi deficiency syndrome in the initial state of ischemic stroke,and the performance of the model is evaluated.3.Based on UPLC-Q-TOF/MS serum metabolomics technology,the micromechanism of Qi deficiency syndrome in the initial state of ischemic stroke was studied.Sixty patients with ischemic stroke were included,including 34 patients in the initial state of Qi deficiency group and 26 patients in the non Qi deficiency group.The serum metabolites of the two groups were compared using multi variate statistical methods including principal component analysis(PCA)and ort hogonal partial least squares discriminant analysis(OPLS-DA),To explore the differential metabolites and pathways in patients with ischemic stroke onset Qi deficiency syndrome and explore the microbiological basis of ischemic stroke onset Qi deficiency syndrome.Result1.Analysis of related factors of qi deficiency syndrome in the initial state of ischemic stroke based on binary logistic regression analysis.In this study,patients with initial Qi deficiency syndrome were mainly characterized by mental fatigue,pale tongue,tooth marks on the edge of the tongue,white tongue coating,fine veins,and deep veins.Their objective physical and chemical indicators were as follows:red blood cell count,hemoglobin content,white blood cell count,absolute value of lymph nodes,albumin specific globulin,total bilirubin,triglycerides,and homocysteine,which were lower than those of patients with non Qi deficiency syndrome The relative prolongation of activated partial prothrombin time and the relative shortening of thrombin coagulation time.Through single factor analysis,we can find that the inducement of fatigue,continuous drinking history,previous stroke history,previous history of diabetes,red blood cell count,hemoglobin content,white blood cell count,absolute value of median granule,absolute value of lymph,albumin biglobulin,total bilirubin,triglyceride,homocysteine,activated partial prothrombin time,thrombin coagulation time meet P<0.1,It is believed that the above factors are the difference factors affecting the initial state of ischemic stroke Qi deficiency syndrome and non Qi deficiency syndrome;All the above difference variables were included in the binary logistic regression analysis,and finally five influencing factors of Qi deficiency syndrome were obtained:fatigue inducement(OR=3.032,95%CI 1.279～7.183,P<0.05),white blood cell count(OR=0.858,95%CI 0.746～0.987,P<0.05),albumin specific globulin(OR=0.295,95%CI 0.104～0.843,P<0.05),total bilirubin(OR=0.924,95%CI 0.879～0.972,P<0.01)Triglyceride(OR=0.613,95%CI 0.393～0.955,P<0.05).2.Construction of discriminant model of qi deficiency syndrome in the initial state of ischemic stroke.After Lasso regression algorithm,26 representative features were screened,including fatigue,dizziness,purple lips,sputum in the throat,dry mouth,white fur,thick fur,light tongue,red tongue,tooth mark on the tongue,sinking pulse,fine pulse,slippery pulse,white blood cell count,red egg white content,absolute value of lymph,glucose,total cholesterol,triglyceride,low-density lipoprotein cholesterol,homocysteine Activate partial prothrombin time and thrombin clotting time,input all features of patients with and without Qi deficiency syndrome into machine learning and use 8:2 ratio for supervised model training.The machine learning model f1 values obtained in turn are support vector machine 0.625,random gradient descent method 0.422535,K nearest neighbor 0.438356,decision tree 0.62963,random forest 0.804348,extreme random tree 0.140351,and extreme gradient enhancement algorithm 0.744186,The performance of the model was evaluated by combining model accuracy(P),sensitivity(R)and AUC.Among them,random forest was the best,P=0.948718,R=0.698113,AUC=0.959.To use the stacking back-end fusion mode to fuse the trained classifiers with probability decision and BP artificial neural network to construct a discrimination model of qi deficiency syndrome in the initial state of ischemic stroke.The model has good classification ability,and the accuracy rate of the training concentration model is P=0.96,sensitivity R=0.92,F1 score=0.94,AUC=0.95;Test set P=0.92,R=1.0,flscore=0.96,AUC=0.99.4.Study on the Biological Basis of Qi Deficiency Syndrome in the Initial State of Ischemic Stroke Based on Metabonomics.The results of serum metabolomics by UPLC-Q-TOF/MS showed that there were 12 different metabolites between the initial stage of ischemic stroke with Qi deficiency syndrome and the non Qi deficiency group,which were LysoPC(18:2/0:0)、Choline、LysoPC(18:1/0:0)、LysoPC(18:0/0:0)、2-Methyl-1,3-cyclohexadieneAcetic acid、16-Hydroxyhexadecanoic acid、Palmitoleic acid、O-Phosphoethanolamine、Palmitic Acid、L-Glutamic acid、beta-Glycerophosphoric acid.MetPA analysis was conducted using the identified differential metabolites as the research object,and differential metabolic pathways were analyzed according to the screening conditions of P<0.05 and Impact value>0.1.A metabolic pathway that met the conditions was obtained,namely,D-glutamine and D-glutamic acid metabolism.Conclusion1.The main clinical manifestations of patients with Qi deficiency syndrome at the onset of ischemic stroke are mental fatigue,pale tongue,tooth marks on the edge of the tongue,white tongue coating,fine veins,and deep veins.Their physical and chemical indicators reflect the presence of low-level inflammation,poor immune status,reduced metabolic ability,and more obvious coagulation abnormalities in the body.Its formation is related to many factors,among which fatigue,white blood cell count,albumin globulin,total bilirubin and triglyceride are important factors for the formation of qi deficiency at the onset of ischemic stroke.2.The model based on multi classification machine learning fusion BP neural network has good classification ability for ischemic stroke initial state Qi deficiency syndrome,which can be further verified and used to guide clinical practice.3.The results of a non targeted metabonomic study on the serum of patients with initial Qi deficiency syndrome and non Qi deficiency group of ischemic stroke show that there are 12 different metabolites and one main metabolic pathway in the serum of the two groups of patients,indicating that there is metabolic disorder in the body of patients with initial Qi deficiency syndrome of ischemic stroke,mainly manifested in the disorder of lipid metabolism and amino acid metabolism.