A Binary Classification Study of Alzheimer’s Disease Based on a Novel Subclass Weighted Logistic Regression Method

Based on proposed joint human connectome project multi-modal parcellation (JHCPMMP), the study binary classification of Alzheimer’s disease was conducted. We tried to build a novel model, which can be interpretative and have ability deal with complexity individual differences brain networks. The subclass weighted logistic regression (SWLR) based in this paper. conducted five groups experiments, accuracy HC vs. AD 95.8%, EMCI 91.6%, LMCI 93.7%, 89.5%, 91.6%. In addition, we follow-up analysis coefficient matrix found that distribution core deterioration regions different stages is development disease. located these two-dimensional images they generally show trend continuous counterclockwise migration.