Chen et al reported models for predicting significant fibrosis and cirrhosis in a patient with chronic hepatitis B (HBV). The authors are from Zhejiang University in Hangzhou, China.
Patient selection: chronic hepatitis B (HBV)
Parameters:
(1) RDW in percent
(2) platelet count in 10^9/L
(3) hemoglobin in g/L
model for significant fibrosis =
= (0.357 * (RDW)) - (0.025 * (platelet count)) - 1.817
probability of significant fibrosis =
= 1 / ( 1 + EXP((-1) * (model for fibrosis))
model for cirrhosis =
= (0.181 * (RDW)) - (0.023 * (platelet count)) - (0.045 * (hemoglobin)) - 3.471
probability of cirrhosis =
= 1 / ( 1 + EXP((-1) * (model for cirrhosis))
NOTE: There may be a problem with model for cirrhosis. The predicted probability is very low for the inputs.
Performance:
• The area under the ROC curve for the first model is 0.83.
• The area under the ROC curve for the second model is 0.89.
