Evaluation of serum markers in the diagnosis of liver fibrosis in patients with chronic hepatitis B and the development of an artificial nerual network diagnosis model for liver fibrosis

Objective To evaluate the diagnostic value of 17 fibrosis serum markers,and to develop an artificial nerual network(ANN) model for the diagnosis of liver fibrosis with screened meaningful markers.Methods Totally 254 serum samples from the patients with chronic hepatitis B(CHB),who were confirmed to have developed liver fibrosis with biopsy in our hospital,were collected,and 17 serum markers correlated with liver fibrosis were detected and analyzed statistically.Multivariable regression analysis was conducted to have 6 serum markers screened with high diagnostic value and then an ANN model with the 6 serum markers combined was developed.Results Thirteen serum markers were correlated with liver fibrosis stage(P0.05),but the diagnostic value of the single marker was not satisfactory [area under receiver characteristic curve(AUROC)0.75].The model of 6 combined serum markers such as age,GGT,PLT,PT,TIMP-1 and DBIL had better diagnostic value than any single serum marker(AUROC=0.834).The accuracy,the sensitivity and the specificity of the ANN model in predicting liver fibrosis were 77.9%,82.7% and 75.4%,respectively.Conclusions The combination of the 6 screened serum markers is more valuable in the diagnosis of liver fibrosis.The ANN model of the 6 combined serum markers demonstrates good sensitivity and specificity,and is expected to be applied to clinical practice in the future.