Figure 2.

Flowchart of the study design. In step 1 of parameter set selection, stepwise logistic regression, receiver operating characteristic curve, and hypothesis testing were used jointly for preselection, and the final set was determined from all possible combinations. In step 2 of model construction and validation, the logistic regression (LR) model was constructed using the optimal parameter set and was compared with GIB-4 and non-alcoholic fatty liver disease fibrosis scores on the discovery set. Then, the LR model was validated on the validation sets. Its independence from possible confounders was evaluated. Its performances were compared to those of other machine learning methods. In step 3, we developed a web tool for fast applications.