Logistic regression analysis of multiple interosseous hand-muscle activities using surface electromyography during...

Area of Science:

• Biomedical Engineering
• Neuroscience
• Kinesiology

Background:

• Surface electromyography (sEMG) signals from intrinsic hand muscles are prone to interference.
• Accurate assessment of deep hand muscle activity is crucial for understanding motor control and diagnosing neuromuscular disorders.

Purpose of the Study:

• To evaluate the effectiveness of univariate and multivariate logistic regression models in analyzing sEMG signals from intrinsic hand muscles during finger-specific tasks.
• To determine the reliability of surface EMG for differentiating activity among densely located muscles.

Main Methods:

• Ten healthy subjects performed isometric exercises of individual fingers.
• Surface EMG signals were recorded and analyzed using univariate and multivariate logistic regression, incorporating time-domain (power, amplitude) and frequency-domain variables.
• Statistical significance was assessed using P-values and Nagelkerke's R-squared.

Main Results:

• Univariate analysis revealed significant consistency (P < 0.001) between activated finger and dorsal interosseous muscle activity.
• Multivariate analysis incorporating frequency-domain variables (median frequency) for the fourth dorsal interosseous muscle and ring finger action showed a higher model correlation (Nagelkerke's R² = 0.716) compared to models without frequency-domain variables (Nagelkerke's R² = 0.583).

Conclusions:

• Logistic regression models demonstrate significant potential for analyzing sEMG signals from densely located intrinsic hand muscles during functional tasks.
• The inclusion of frequency-domain variables enhances the accuracy of sEMG-based muscle activity analysis for fine motor tasks.