| Objective:Chronic hepatitis B poses a persistent threat to human health.Infection with the hepatitis B virus may lead to cirrhosis and hepatocellular carcinoma,posing a serious threat to human health and safety.However,if the stage of liver fibrosis before the occurrence of cirrhosis and hepatocellular carcinoma is detected in a timely manner and corresponding treatment is given,the deterioration of liver fibrosis can be effectively controlled.Population heterogeneity has an important impact on the occurrence and prevention of liver fibrosis in patients with chronic hepatitis B.This paper mainly uses K-modes cluster analysis,hierarchical clustering analysis and potential category analysis to study the population heterogeneity of chronic hepatitis B patients.Combined with the occurrence of liver fibrosis in patients with different subtypes of chronic hepatitis B and its predictors,a prediction model of liver fibrosis was constructed,and the prediction effect and fitting effect of each model were compared to provide the basis for further selecting the optimal model of liver fibrosis and provide the necessary basis for the prevention strategy of liver fibrosis.Methods:For 848 hepatitis B patients included in the study,frequency and constituent ratio were used to describe discrete variables,and mean ± standard deviation was used to describe continuous variables.In this study,heterogeneous population classification combined with binary logistic regression was used to construct a prediction model for liver fibrosis,K-modes cluster analysis,latent class analysis and hierarchical cluster analysis were used to identify different subtypes of hepatitis B patients,and then different subtypes of hepatitis B patients were divided into two groups according to 7:3for the construction and validation of the model.AIC,SC and-2Log L were used to evaluate the fitting effect of the model,and ROC curve,AUC value,accuracy and other indicators were used to evaluate the prediction effect of the model.Results:The study collected information from 964 liver disease outpatient patients at Shengjing Hospital affiliated to China Medical University.After the inclusion and exclusion process,848 effective study subjects were included.There is obvious group heterogeneity among hepatitis B patients.The results of latent category analysis showed that there were two subtypes of hepatitis B,and the characteristics of patients with different subtypes were significantly different.The proportion of patients with good lifestyle habits,smoking and drinking,is less than 5%;A type of patients with poor lifestyle habits,with over 70% of them smoking and drinking alcohol.The influencing factors of liver fibrosis in patients with different subtypes of hepatitis B also have some differences.Among the asymptomatic patients with good living habits,AGE,APOA1,AFP,APTT,TT and MPV are the risk factors of liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the constructed liver fibrosis prediction model in the training group and the validation group were 0.816(0.770-0.856)and 0.708(0.629-0.780),respectively;In the group of male patients with poor lifestyle habits,A/G and PLT are protective factors for liver fibrosis(OR<1,P<0.05),while GGT is a risk factor for liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the constructed liver fibrosis prediction model in the training group and validation group were 0.789(0.711-0.853)and 0.772(0.654-0.865),respectively.Hierarchical clustering analysis showed that there were two subtypes of hepatitis B,666 patients with good living habits without signs and 182 patients with poor living habits.There were also some differences in the influencing factors of liver fibrosis between the two types of hepatitis B patients.Among the asymptomatic patients with good living habits,AGE,ALP,AFP,TT and MPV were the risk factors of liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the constructed prediction model of liver fibrosis in the training group and the validation group were 0.792(0.746-0.832)and 0.735(0.662-0.799),respectively;In the group of male patients with poor lifestyle habits,APOB is a protective factor for liver fibrosis(OR<1,P<0.05),while ALT and AFP are risk factors for liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the constructed liver fibrosis prediction model in the training group and validation group were 0.824(0.738-0.891)and 0.788(0.632-0.899),respectively.K-modes cluster analysis showed that there were two subtypes of hepatitis B,630 asymptomatic patients with good living habits and 218 male patients with poor living habits.The characteristics of the two subtypes of patients were significantly different,and there were also some differences in the influencing factors of liver fibrosis in hepatitis B patients.Among patients with good living habits and no signs,APOA1,AFP,TT and MPV were the risk factors of liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the constructed prediction model of liver fibrosis in the training group and the validation group were 0.820(0.776-0.858)and 0.682(0.601-0.755),respectively;In the group of male patients with poor lifestyle habits,A/G and PLT were protective factors for liver fibrosis(OR<1,P<0.05),while ALT and DBIL were risk factors for liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the constructed liver fibrosis prediction model in the training group and validation group were 0.808(0.730-0.872)and 0.668(0.532-0.786),respectively.The overall analysis results showed that A/G and APOB can be explained as protective factors for liver fibrosis(OR<1,P<0.05),while SMOKING,ALP,DBIL,AFP,TT,and MPV are risk factors for liver fibrosis(OR>1,P<0.05).The area under the ROC curve(AUC)of the model in the training and validation groups were 0.810(0.772 0.844)and 0.695(0.628 0.757),respectively.Conclusion: Chronic hepatitis B patients can be divided into two subtypes:asymptomatic patients with good living habits and male patients with poor living habits.There are significant differences in the characteristics between patients,and there are also some differences in the influencing factors of liver fibrosis.The results suggest that targeted prevention and intervention measures should be taken for hepatitis B patients with different subtypes;among the different prediction models of liver fibrosis constructed,the prediction model of liver fibrosis in hepatitis B patients constructed based on hierarchical cluster analysis had the best prediction effect in the validation group,and it was recommended that hierarchical cluster analysis could be used to classify hepatitis B patients before constructing the prediction model of liver fibrosis. |
PDF Full Text Request