Clinical demographics of the training group of 284 patients related to HBsAg seroclearance and 137 patients related to HBsAg seroconversion.

Only variables, significantly related to HBsAg seroclearance and seroconversion in the univariate analysis, were used to build the neural network, entering in the multivariate analysis for the development of the logistic regression models. Constant coefficients of multivariate logistic regression for HBsAg seroclearance = 2.910, for HBsAg seroconversion = −1.371.*Time point 3 years.¶Time point 3 to 2 years. b, regression coefficients. CI, confidence interval.