Muscle synergy analysis during badminton forehand overhead smash: integrating electromyography and musculoskeletal modeling

Abstract

Introduction: This study aimed to quantify shoulder muscle synergies during badminton forehand overhead smash (BFOS) via non-negative matrix factorization (NMF), validate musculoskeletal (MSK) models for high-speed movements by comparing electromyography (EMG)-derived synergies with simulation results, and explore the potential of NMF-based MSK models in advancing sports science.

Methods: Twenty elite badminton players (age: 24 ± 4 years; experience: 15 ± 4 years) performed maximal-effort BFOS while EMG signals from fifteen shoulder muscles were recorded. Three-dimensional motion analysis with a ten-camera Vicon system captured kinematic data at 100 Hz. A validated OpenSim upper extremity model was implemented to simulate muscle activations via static optimization. NMF extracted synergy vectors and activation coefficients from both experimental EMG and MSK modeling data.

Results: Three muscle synergies accounted for >90% variance in both analyses with no significant differences in global VAF (p = 0.12). The first synergy (trapezius-dominant) showed 95% EMG and 97% MSK variance; the second synergy (pectoralis/anterior deltoid) exhibited 97% EMG and 94% MSK variance; the third synergy (posterior muscles) demonstrated 95% EMG and 98% MSK variance. Strong agreement between approaches was observed for both weight vectors (W1:0.81 ± 0.04, W2:0.87 ± 0.01, W3:0.88 ± 0.03) and activation coefficients (C1:0.95 ± 0.02, C2:0.98 ± 0.01, C3:0.98 ± 0.01), with differences primarily in lower trapezius activation (similarity: 0.77 ± 0.05), likely due to challenges in recording deep muscle activity through surface electromyography. These findings validate the combined experimental-computational approach for analyzing complex, high-velocity movements.

Conclusion: The strong correspondence between experimental and computational synergies validates MSK modeling for analyzing neuromuscular control during high-velocity overhead movements. The identified synergies provide a framework for understanding muscle coordination during BFOS, with potential applications in targeted training program optimization and injury prevention strategies in overhead sports.

Keywords: biomechanical phenomena; elite athletes; joint instability; motor control; movement disorders; neuromuscular coordination; rotator cuff.

Figure 1

Experimental setup showing Vicon motion capture system with retroreflective markers and surface EMG electrodes monitoring 15 shoulder/arm muscles (anterior/middle/posterior deltoid, infraspinatus, pectoralis major, latissimus dorsi, triceps, biceps, trapezius, serratus anterior) during badminton overhead smash execution. [Reprinted from (25) with permission].

Figure 2

Sequential phases of the badminton forehand overhead smash (BFOS), preparation, acceleration, impact, and follow-through. Movement analysis was based on racquet velocity thresholds. [Reprinted from (25) under CC BY 4.0 license].

Figure 3

Opensim upper extremity musculoskeletal model shown from anterior, lateral, and posterior views, featuring 32 hill-type musculotendon actuators and 15 degrees of freedom across the shoulder, elbow, and wrist joints with subject-specific scaling. The views illustrate the complete muscle representation from different anatomical perspectives.

Figure 4

Methodological framework comparing muscle synergies derived from experimental EMG and musculoskeletal modeling during BFOS, including data acquisition, signal processing, modeling, non-negative matrix factorization, and quantitative comparison of synergy components.

Figure 5

Relationship between synergy number and global variance accounted for (VAF) from EMG (blue) and musculoskeletal model (red) during BFOS. Three synergies were optimal based on global VAF >90%, local VAF >75%, and <5% improvement with additional synergies (p = 0.12, MANOVA).

Figure 6

Comparison of (A) synergy weight vectors and (B) activation coefficients between EMG (blue) and musculoskeletal model (red) during BFOS. Three distinct functional synergies were identified: Synergy 1 (trapezius-dominant for scapular stabilization during preparation and early acceleration), Synergy 2 (pectoralis/anterior deltoid for shoulder flexion/internal rotation during late acceleration and impact), and Synergy 3 (posterior shoulder muscles for extension/external rotation during follow-through). The x-axis in panel A represents muscle weights ranging from 0-1 for each muscle (abbreviations defined in Table 2), while panel (C) shows temporal activation patterns across normalized movement cycle (0-100%). Shaded areas represent standard error (n = 20 participants).