FIGURE 1.
Flowchart of the statistical analysis. For 10-fold cross validation, the training set was used to (1) fit the bi-factor item response theory (IRT) model, which was then used to estimate the overall IRT scores and the domain-specific IRT scores for all participants in the training set and the validation set, (2) find the optimal cutoffs for the IRT scores that can classify participants in the training set into three groups (0, 0.5, and 1 for Clinical Dementia Rating global; 0, 0.5 and ≥1 for box scores). Then the same cutoffs derived from the training set will be applied to the IRT scores of the participants in the validation set to evaluate the classify accuracy of the IRT scores