Performance measures of machine learning algorithms in discriminating cognitively normal (CDR 0) from very mildly/mildly demented (CDR 0.5 and 1) participants.

Ten statistical machine learning algorithms were used to determine groups of markers capable of distinguishing very mildly/mildly demented (CDR 0.5 and 1 combined) from cognitively normal participants (CDR 0). Models were fit with two sets of predictors: 1) traditional biomarkers, or 2) traditional biomarkers plus RBM analytes; additionally, age, gender, and ApoE4 allele status were included in all models. Model performance measures shown are based on cross-validation, in which the test set results were averaged from 200 splits of the data between training (80%) and test (20%).