[Retracted] Review on Epileptic Seizure Prediction: Machine Learning and Deep Learning Approaches

Table 1

Review of existing surveys.

Author	Features	Dataset	Classifier	Performance
Rezvan [25]	Used maximum, minimum, standard deviation, and mean as evaluation parameters	Bonn	Multilayer perceptron	98.33
Sabrina et al. [28]	Intrinsic mode functions, Euclidean distance, Bhattacharya distance	CHB-MIT	PHA–unsupervised	98.84
Orellana et al. [29]	PCA, STF, moving maximum	CHB-MIT	Random forest	97.12
Datta Prasad et al. [30]	Incorporated Hilbert transform	Bonn	ANN	96
Birjandtalab et al. [31]	Spectral power estimation is used	CHB-MIT	Random forest + KNN	80.87
Mursalin et al. [18]	DWT and entropy methods are used	Bonn	Random forest	98.45
Raghu and Sriram [32]	28-statistical features	Bern-Barcelona	Random forest, SVM, KNN and Ada-boost	97.6 to 98.8
Subasi et al. [33]	Simple DWT is used for feature extraction	Bonn	SVM	98.83
Al Gahyab et al. [34]	Uses simple FFT-DWT for feature extraction	Bonn	LS-SVM	99
Chen. S. et al. [35]	Multiple types of entropies, spectral power	Bonn	LS-SVM	99.4
Tzimoutra et al. [36]	Use of DWT for feature extraction	Bonn and Freiburg	Random forest	99.74
Wang et al. [37]	STFT, mean, energy, and standard deviation	Bonn	Random forest	96.7
Fasil and Rajesh [38]	Total energy and power of the signal is used to estimate the seizures	Bonn and Barcelona	SVM	99.5
Andrzejak et al. [39]	Nonlinear deterministic dynamics	Real-time data	Random forest	98
Wu et al. [40]	HFO stacked denoising frequency autoencoder (SDAE)	CRCNS	SWAF-ABSVM	92.4%
Dedeo et al. [41]	Common frequency extremes (CFE)	CHB-MIT	Thresholding	—