Comparison of Bagging and Boosting Ensemble Machine Learning Methods for Automated EMG Signal Classification

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Machine Learning

Background:

• Neuromuscular disorders are diagnosed using electromyographic (EMG) signals.
• Machine learning (ML) algorithms serve as decision support systems for diagnosis.
• Limited research exists on ensemble learning for neuromuscular disorder diagnosis via EMG.

Purpose of the Study:

• To evaluate the feasibility of bagging and boosting ensemble classifiers for diagnosing neuromuscular disorders.
• To compare the performance of different ensemble methods in classifying EMG signals.

Main Methods:

• Feature extraction using wavelet packed coefficients (WPC) from EMG signals.
• Statistical analysis of WPC to represent wavelet coefficient distribution.
• Classification using ensemble learning algorithms (bagging and boosting).

Main Results:

• Ensemble classifiers demonstrated superior performance in diagnosing neuromuscular disorders.
• The AdaBoost algorithm combined with a random forest ensemble achieved high accuracy (99.08%).
• Excellent performance metrics were reported: F-measure 0.99, AUC 1, and kappa statistic 0.99.

Conclusions:

• Ensemble learning methods, particularly AdaBoost with random forest, are highly effective for diagnosing neuromuscular disorders using EMG signals.
• The proposed method shows significant promise for automated diagnosis and clinical decision support.