Optimized joint analysis of multichannel electromyography spectrum and amplitude during dynamic muscle contraction

Abstract Muscular fatigue and recovery are crucial aspects for several applications such as sports medicine, neuromuscular rehabilitation, and prosthetics. The joint analysis of spectral and amplitude (JASA) techniques provide a global insight into the muscular condition by integrating both amplitude and frequency components of surface electromyography (sEMG) signals, providing a comprehensive evaluation of muscle fatigue and recovery. Initial studies using these methods showed more reliable results compared to separate parameters analyses; however, JASA methods have been limited by by static task constraints and single-channel configurations. In this study, a parameter optimization method for two JASA techniques –Evolutionary and Sequential– was introduced employing multichannel sEMG signals recorded from four healthy male participants during dynamic wrist contractions. The percentage of statistically significant regression slopes was used as the selection criterion. Five parameters were optimized. For both techniques, optimum computation window size (CWS) was found to be 1-sec. For Evolutionary JASA (EvoJASA), optimum values for both initial window size (IWS) and window incremental increase (WII) converged at 70 sec and 10 sec, respectively, which directly mirrored the optimum for Sequential JASA (SeqJASA) parameters, sliding window size (SWS) of 70 sec and shift interval (SI) of 10 sec. This study presents a systematic, statistically grounded procedure for selecting the JASA window parameters for dynamic, multichannel sEMG, demonstrated as a proof-of-concept in a small cohort of four healthy participants. The 96 JASA analyses performed (4 subjects $$\times$$ 8 channels $$\times$$ 3 DoFs) provide within-subject statistical depth but do not constitute population-level evidence; the reported parameter values are specific to this cohort, load, and task set, and validation across larger, sex-balanced, age-diverse, and independent populations, loads, and tasks remains necessary future work before the framework can be applied to real-world assistive technologies.