Table 4.

Accuracy of the ML classifier for unimodal IR, unimodal HRV and multimodal IR + HRV features-based models. The models with the best accuracy are highlighted in bold.

	Unimodal IR Features		Unimodal HRV Features		Multimodal IR + HRV Features
	DST	RAVLT	DST	RAVLT	DST	RAVLT
Decision Tree
Simple	46.2	59.2	46.2	44.7	48.1	56.6
Medium	50.0	59.2	50.0	39.5	48.1	53.9
Complex	50.0	59.2	50.	39.5	48.1	53.9
Discriminant Analysis
Linear	69.2	56.6	59.6	32.9	63.5	55.3
Quadratic	63.5	53.9	44.2	28.9	59.6	57.9
Logistic Regression	69.2	-	50.0	-	73.1	-
Support Vector Machine
Linear	73.1	65.8	50.0	28.9	73.1	63.2
Quadratic	65.4	53.9	42.3	26.3	61.5	51.3
Cubic	51.9	56.6	51.9	35.5	61.5	51.3
K Nearest Neighbor
Coarse	48.1	34.2	48.1	34.2	41.8	34.2
Medium	55.8	47.4	50.0	31.6	57.7	44.7
Fine	59.6	55.3	50.0	44.7	55.8	50.0
Ensemble
Bagged trees	53.8	71.1	53.8	44.7	59.6	75.0
Subspace discriminant	63.5	56.6	50	32.9	59.6	56.6
Subspace kNN	55.8	56.6	48.1	44.7	57.7	56.6
RUSboosted trees	55.8	47.4	48.1	47.4	46.2	35.5