Table 4.

Comparative study of methodologies and results between proposed study and previous studies.

Study	Study Sample	EMG Features	Findings	Application
Lu et al. [18]	Eight post-stroke subjects	Root mean square (RMS), 4th order auto regressive (AR) Coefficients, and waveform length (WL)	Mean classification accuracy, GNB): 84.8%; SVM: 83.3%; paired t-Test, p: 0.125	Classification of six hand motion patterns for controlling a robotic hand
Lee et al. [67]	Twenty stroke patients	Mean absolute value (MAV), the number of zero crossing (ZC), the slope sign change (SSC), and WL	Mean classification accuracy, LDA: = 71.3% for moderately impaired subjects.	Classification of task-specific hand movements
Castiblanco et al. [61]	Eighteen stroke patients and twenty-eight healthy control	MAV, RMS, SSC, MNF, mean power (MNP), MDF, and spectral moments (SM)	Accuracy classification of stroke and control group, KNN: 0.87; SVM: 0.82, and LDA: 0.74.	Identification of the fingers and hand motions for robotics-based rehabilitation.
Angelova et al. [62]	Ten stroke patients and fifteen healthy adults	Power spectrum features: MNF, MDF, Maximal power	MNF and MDF are lower for stroke patients compared with healthy control group.	Identification of changes in features during elbow flexion.
Proposed study	Forty-eight stroke patients and seventy-five healthy adults	MNF, MDF, PKF, TP, MP	Classification performance, neural network model: precision: 88%, specificity: 89%, accuracy: 80%.	Prediction of stroke-impaired myoelectrical changes through statistics and machine learning for understanding post-stroke impairment.