Speech as a Biomarker for COVID-19 Detection Using Machine Learning

Table 3

Performance metrics for binary classification algorithms.

Classification algorithms	Optimal; Parameterization	Performance metrics; Mean value (standard deviation)
		ACC	PRE	REC	F1 score	AUC
BDT	No. of Leaves: 16; Learning rate: 0.05; No. of trees: 100	0.724 (0.048)	0.714 (0.037)	0.7037 (0.063)	0.7088 (0.052)	0.717 (0.053)
DF	Random split Count: 128; Maximum Depth: 32; No. of decision trees: 16	0.7317 (0.021)	0.7421 (0.017)	0.7892 (0.081)	0.7649 (0.025)	0.755 (0.017)
NN	Learning rate: 0.001; No. of hidden Nodes: 314	0.711 (0.031)	0.7271 (0.043)	0.7188 (0.018)	0.7229 (0.029)	0.7616 (0.095)
LoR	Optimization Tolerance: 1e-06; L1 regularization weight: 1; Memory size for L-BFGS: 18	0.6741 (0.019)	0.6805 (0.024)	0.6161 (0.027)	0.6467 (0.019)	0.6874 (0.065)
SVM	Lambda – 0.001	0.694 (0.017)	0.673 (0.074)	0.6027 (0.019)	0.6359 (0.011)	0.6619 (0.037)