Table 6.

Comparison with similar works.

Author	Signals	Methodology	Accuracy
Katsis et al. (2008) [26]	EMG, ECG, EDA and Respiration	Two statistical features were extracted as mean value and root mean square. 10-fold-cross validation and SVM classifier was used.	79.3%
Fu and Wang (2014) [27]	EMG and ECG	A preprocessing step was headed with the Fast Independent Component Analysis from both signals. Two nonlinear measurements were obtained from the windows (peak factor and maximum of cross-relation curve). 10-fold-cross validation and Mahalanobis distance used as classifier.	86.7%
Wang and Guo (2020) [16]	EMG	Pseudoinverse Learning Algorithm based Autoencoder (PILAE) was used for the representation learning of signals and AdaBoost classifier was used as the final step. Leave-One-Out-cross validation was employed.	58%
Rastgoo et al. [30]	ECG, vehicle and environmental data	Fusion of the CNN and LSTM models to develop the classifier. CNN is used to fuse the information obtained from ECG, vehicle, and environmental data. LSTM is used as classifier.	92.8%
El Haouij et al. [40]	EDA	4-level Discrete Wavelet Decomposition is performed to the right-hand EDA signal. Haar Wavelet is used as mother wavelet. Random Forest classifier is used.	81%
Proposal	EMG	Statistical Properties are used as features. Support Vector Machine is used as the classifier. 10-fold-cross validation is employed.	96%