Table 5.

Selected features for each classifier.

classifiers	feature space based on feature selection
Initial	MMSE_b1, Left-VentralDC, lh_rostralanteriordngulate_thickness_std, lh_medialorbitofrontal_thickness, rh_entorhinal_volume, Left-Accumbens-area, lh_superiorfronta_meancurv, lh_parstriangularis_thicknessstd, right_subiculum, GENDER, lh_temporalpole_thicknessstd, lh_temporalpole_volume, Left-Hippocampus, rh_cuneus_area
AC	MMSE_bl, lh_parahippocampa_area, lh_rostralanteriorcingulate_volume, lh_transversetemporal_thickness, rh_fusiform_area, lh_middletemporal_thickness, lh_parahippocampa_meancurv,
HM	rh_precuneus_area, Right-Cerebellum-Cortex, lh_parahippocampal_volume, rh_medialorbitofrontal_thickness, left_CA1, 5th-Ventricle, lh_transversetemporal_area, MaskVol-to-eTIV, Right-Cerebellum-Cortex, rh_temporalpole_thicknessstd, lh_WhiteSurfArea_area, rh_precuneus_area, GENDER, Right-Hippocampus_hipposubfields, Right-VentralDC
AHM	MMSE_bl, rh_isthmuscingulate_volume, Left-Inf-Lat-Vent, lh_caudalanteriorcingulate_thicknessstd, rh_fusiform_thicknessstd, lh_parsopercularis_thicknessstd, lh_transversetemporal_thickness, CC.Mid.Posterior, lh_middletemporal_thickness, Right-Cerebellum-White-Matter rh_temporalpole_thickness, lh_entorhina_thicknessstd,
cHM	lh_rostralanteriorcingulate_thicknessstd, lh_precuneus_thicknessstd, GENDER, Left-Inf-Lat-Vent, MMSE_bl, rh_entorhinal_thickness, lh_bankssts_area, left_presubiculum, rh_superiorparietal_thickness, lh_entorhinal_volume, rh_parsorbitalis_area, rh_precentral_volume, lh_frontalpole_volume, lh_isthmuscingulate_volume, lh_xontalpole_thicknessstd, lh_isthmuscingulate.thicknessstd,

We use the same feature name as provided in the Kaggle data sets.