The Model Construction Study On Evaluating The Imbalance Status Of Liver Injury And Liver Regeneration In Chronic Hepatitis B Patients

Objective This study aims to establish the biology index system e_valuating the imbalance status of liver injury and liver regeneration in chronic hepatitis B patients with no obvious symptoms in early phase.The aim of constructing the model of e_valuating the imbalances status of liver injury and liver regeneration is to provide the basis for clinical monitoring or intervention of Chinese medicine in chronic hepatitis B patients who can not be diagnosed without obvious symptoms and prevent the illness futher development.Methods Firstly,set "liver injury" and "liver regeneration" as keywords and search the database CNKI.To establish the biology index system e_valuating the imbalance status of chronic hepatitis B in liver injury and liver regeneration in using literature metrology method and Delphi expert consultation method.Secondly,collect and filter the data about chronic hepatitis B patients from some Hubei province 3 armour hospital according to inclusion and exclusion criteria.Fillter the characteristics of a variable by using single factor analysis of variance and Logistic regression methods.Choose two the most appropriate hierarchical algorithms to construct model after comparing several ones in consideration of the problems of data such as sample size and linear problem and so on.Then obtain the best one after comparing the forecast results according to the indicators of sensitivity, accuracy and specificity.Thirdly,to e_valuate the level of imbalance status about liver injury and liver regeneration with the data of liver puncture in chronic hepatitis b patients when they received treatment the first time,the second time?half a year?and the third time?a year latter?.During the one-year treatment and dynamic monitoring.Test the model's sensitivity in clinical three times respectively.Results An expert questionaire containing 25 items for choosen was formed on the basis of the measurement results in 546 clinical trials research literatures.After the questionnaire survey and late consultation of 5 liver experts,the biology index system e_valuating the imbalance status of chronic hepatitis b in liver injury and liver regeneration was formed.It was constructed by two index sets,one is about liver injury and the other is about liver regerneration,which contains total 16 indicators such as ALT,AST,TNF-?,IL-6 and so on.Six elements that the degree of liver tissue lesions,liver function,anti-oxidation ability,reducing enzyme action,anti-inflammatory action and liver regeneration were reflexed.In the study,114 patients with chronic hepatitis b were included,of which 28 patients was used for empirical study and 86 patients were used for model training.After comparing two model's forecast results constructed by using Fisher discriminant algorithm and multiple classification based on principal component analysis and orderly Logistic regression algorithm with eight risk factors?ALT?ALB?TBIL?PDGF-B?G-CSF?TNF-??IL-6 and IL-18?for model selection,we obtain the optimal one that Fisher discriminant model?sensitivity: 97.06%,specificity: 100%,accuracy: 97.67%?.The model's typical non-standard discriminant function of are as follows:?1?F1=-0.514*Ln(ALT_value)+4.826*Ln(ALB_value)-1.454*Ln(TBIL_value)-0.403*Ln(PDGF-B_value)+0.239*Ln(G-CSF_value)-0.118*SQRT(TNF-?_value)+0.549*Ln(IL-6_value)-0.121*Ln(IL-18_value)-15.556;?2?F2=-0.166*Ln(ALT_value)+0.391*Ln(ALT_value)+0.087*Ln(TBIL_value)+0.328*Ln(PDGF-B_value)-0.878*Ln(G-CSF_value)+0.021*SQRT(TNF-?_value)+0.934*Ln(IL-6_value)-0.077*Ln(IL-18_value)-5.152;During the empirical study,the forecast accuracy of the model for chronic hepatitis b patients with syndrome of “liver-kidney essence deficiency” was first: 82.14%?23/28?,half a year :85.71%?24/28?and a year later :89.29%?25/28?.Conclusion The biology index system e_valuating the imbalance status of chronic hepatitis B in liver injury and liver regeneration has several features:?1?simple construction,?2?high coverage,?3?low repetition rate.The indexs included are easy to get so that it is convenient for popularization and application in clinic.The model constructed by Fisher discriminant algorithm proves to own high diagnostic ability because of the high accuracy during the phases of model training and empirical study.But the stability of the model is not high on account of the small sample size.So we should expand the sample size and improve the model by changing the Fisher discriminant algorithm criterion or introducing weighting factor to enhance the stability and accuracy of the model in a follow-up study.