Ensemble Merit Merge Feature Selection for Enhanced Multinomial Classification in Alzheimer’s Dementia

Table 5

Description of the J48 ensemble model used for the multiclass classification.

Details	Value
Split method	Binary split
Cross validation accuracy	0.976
AUC with CV	0.971
Train and test accuracy	0.986
AUC with train and test	0.987
Common features selected by all methods	MMSE, CDR, hippocampus volume, and everyday cognition measures
Features added by CPEMM	Entorhinal measures, CDRSB, and Ray Auditory Verbal Learning Test-immediate