Table 4.

Summary of the methods reviewed in this paper.

Author (Year)	Database	Data Preprocessing	Machine Learning Model	Results
Álvarez-Estévez et al. (2010) [84]	SHHS	Temporal aggregation rules	Single hidden layer FFNN	Sensitivity = 0.86, Specificity = 0.76
Behera et al. (2014) [86]	SHHS	Hjorth, etc.	Single hidden layer FFNN	Sensitivity = 0.933, Specificity = 0.914
Liang et al. (2015) [88]	SHHS	Band-pass filter, FFT, 22 features	C-ELM	AUC = 0.85, ACC = 0.79
Macias Toro et al. (2018) [87]	PhysioNet	Average power, etc.	Fully connected network	AUPRC = 0.261
Olsen et al.(2018) [46]	Local Dataset	CWT	Single hidden layer FFNN	Precision = 0.72, Sensitivity = 0.63
Chazal et al. (2020) [85]	PhysioNet	59 combining features from adjacent epochs	FFNN	Specificity = 70%

FFNN = feed forward neural networks; C-ELM = curious extreme learning machine.