Table 3.

Random Forest performance testing for detecting AD patients ($MS2$ test dataset;10% of the original data).

Modalities	Precision (%)			Recall (%)			MCA (%)	MCF (%)
	CN	MCI	AD	CN	MCI	AD
All modalities	100.0	86.54	96.30	86.67	97.83	89.66	92.38	92.81
CS	100.0	71.43	93.75	86.61	97.83	51.72	81.90	83.29
NB	85.29	81.63	95.45	96.60	86.96	72.41	85.71	86.39
MRI	50.31	48.28	72.22	40.01	91.30	44.83	52.38	42.61
PET	50.00	55.22	73.53	36.67	80.43	86.21	60.95	58.66
Genetics	66.67	49.28	40.21	80.00	73.91	35.33	55.24	44.09
MH	33.33	52.54	38.46	36.67	67.39	17.24	44.76	40.93
NB + CS	100.0	84.91	96.15	86.67	97.83	86.21	91.43	91.93
NB + MRI	73.68	79.55	95.65	93.33	76.09	75.86	80.95	82.36
NB + PET	86.21	75.93	95.45	83.33	89.13	72.41	82.86	83.69
NB + Genetics	87.50	83.67	95.83	93.33	89.13	79.31	87.62	88.12
NB + MH	87.50	77.36	95.00	93.33	89.13	65.52	83.81	84.59
NB + CS + MRI	100.0	83.64	100.0	86.67	100.0	82.76	91.43	92.12
NB + CS + PET	100.0	82.14	100.0	86.67	100.0	79.31	90.48	91.27
NB + CS + Genetics*	100.0	86.79	100.0	86.67	100.0	89.66	93.33	93.82
NB + CS + MH	100.0	86.54	96.30	86.67	97.83	89.66	92.38	92.81
NB + CS + Genetics + MRI	100.0	85.19	100.0	86.67	100.0	86.21	92.38	92.97
NB + CS + Genetics + PET	100.0	85.19	100.0	86.67	100.0	86.21	92.38	92.97
NB + CS + Genetics + MH	100.0	86.79	100.0	86.67	100.0	89.66	93.33	93.82
NB + CS + Genetics + MH + MRI	100.0	86.54	96.30	86.67	97.83	89.66	92.38	92.81
NB + CS + Genetics + MH + PET	96.15	84.62	96.30	83.33	95.65	89.66	90.48	90.93

MCA: multiclass classification accuracy, MCF: multiclass F1 score; Asterisk ( ∗): is the subset of features with the best predictive performance; italic text is the best of single and pairs of modalities.