Assessing Inter-Day Reliability of Surface EMG Amplitude for Resistance Training Exercise Selection

Abstract

In resistance training, surface electromyography (SEMG) assesses muscle excitation and aids in exercise selection for strength, hypertrophy, and performance. However, SEMG amplitude variability may limit its predictive power for training adaptation. This study evaluated the reliability and exercise relationship of SEMG amplitudes of the biceps brachii (BB) during the concentration (CONC), dumbbell (DBELL), hammer (HAMM), and incline (INCL) resistance arm curl exercises over four separate sessions. Eight trained (5 males and 3 females) and 7 untrained (4 males and 3 females) participants (37 ± 15 years) performed 7 repetitions per exercise to a metronome set to 30 BPM (2 s each for concentric and eccentric movements followed by a 2 s pause) at an intensity of one-third of the elbow flexion maximum voluntary isometric contraction (MVIC) force. The exercise sequence was randomly selected, and rest periods between exercises lasted 4 (±1) min. SEMG signals were processed using the root mean square method with a 200 ms window. The mean SEMG amplitude of the last 4 repetitions was normalized to the peak value from an MVIC of the BB (MVCnorm) and to the peak value of the BB obtained during the last four repetitions of the DBELL (DYNnorm). A generalized linear mixed model was performed to examine the differences in normalized amplitude for the exercises, sessions, and the interaction between exercises and sessions. Intraclass correlation coefficients (ICC) were calculated to determine variability across the four sessions for the model and the individual exercises. Pearson correlations of contrast effect sizes between exercises were performed to assess if the magnitude of the effect sizes changed over multiple trials. The main effect of exercise was significant for both MVCnorm (p =.039) and DYNnorm (p =.049). There were no significant main effects of sessions or exercise:trial interaction for both normalization methods. Post hoc tests indicated CONC MVCnorm and DYNnorm were significantly higher than HAMM MVCnorm (p =.023, ES= 0.694 or large) and DYNnorm (p =.031, ES = 0.909 or large), respectively. Participant repeatability was higher in MVCnorm (R = 0.543) than in DYNnorm (R = 0.217). Overall, model-adjusted ICC values suggest moderate reliability for both normalization methods; however, individual exercises for MVCnorm ranged from moderate to good agreement, and DYNnorm showed poor to moderate agreement. In both normalization methods, CONC showed the highest reliability. In contrast, HAMM showed the lowest reliability, suggesting that the degree of contribution of the brachialis and brachioradialis in the HAMM may have affected reliability. Pearson correlations between the magnitude of the effect size across trials were strong to very strong regardless of the normalization method, indicating that the relationship between SEMG amplitude magnitudes between exercises remained throughout the trials. Clinicians and researchers should be aware that although SEMG amplitude shows high heterogeneity between individuals, reliability for resistance training exercise generally demonstrates favourable results with the rank relationship between exercises maintained across trials. Exercises involving multiple synergist muscles that significantly contribute to the motion may cause reliability to fluctuate and affect target muscle excitation. This variability should be considered during training program design and to investigate the influence of SEMG amplitude on training adaptation.