Table 2.

Classification performance of the supervised models with the classification threshold set at 95%.

		Classification performance (%)
Model and metrics		Ca vs Bb	C vs NBc	B vs NB	C vs ALSd
Random forest
	Accuracy	58.3	56.1	68.8	75.1
	Sensitivity	4.8	30.4	0.0	65.6
	Specificity	100.0	100.0	100.0	98.8
Naïve Bayes
	Accuracy	82.3	68.8	72.8	75.1
	Sensitivity	64.7	54.6	15.8	65.6
	Specificity	96.1	93.3	98.6	98.8
Logistic regression
	Accuracy	92.8	77.7	74.1	76.0
	Sensitivity	84.8	65.1	16.7	66.4
	Specificity	99.0	99.6	100.0	100.0
Linear discriminant analysis
	Accuracy	88.1	70.6	71.7	71.1
	Sensitivity	72.7	53.5	0.9	59.5
	Specificity	100.0	100.0	100.0	100.0
Neuronal network
	Accuracy	92.6	84.8	73.1	86.8
	Sensitivity	83.2	76.1	20.5	81.6
	Specificity	100.0	100.0	96.8	99.8
Support vector machine
	Accuracy	86.3	71.1	70.7	71.1
	Sensitivity	68.8	54.3	6.1	59.4
	Specificity	100.0	100.0	100.0	100.0

^aC: control group.

^bB: patients with amyotrophic lateral sclerosis (ALS) with bulbar involvement.

^cNB: patients with ALS without bulbar involvement.

^dALS: all patients with ALS.