Reliability of absolute versus log-transformed regression models for examining the torque-related patterns of response for mechanomyographic amplitude

This study examined the test–retest reliability of the slopes (b) and y-intercepts (a) of the absolute and log-transformed regression models applied to the mechanomyographic amplitude (MMGRMS) versus torque (TQ) relationship. Fifteen participants (mean ± SD age = 23 ± 4 yrs) performed two isometric maximal voluntary contractions (MVCs) and ten randomly ordered isometric leg extensions from 5% to 95% of their MVC during three separate trials. MMGRMS was recorded from the vastus lateralis during each MVC. Intraclass correlation coefficients (ICCs) and standard errors of measurement (SEMs) were calculated for test–retest reliability. ICCs for the b and a terms were 0.89 and 0.90 for the log-transformed and 0.85 and 0.76 for the absolute relationships, respectively. The SEM values (expressed as a percentage of the mean) for the b and a terms were 9.7% and 16.4% for the log-transformed and 18.9% and 57.1% for the absolute relationships, respectively. These results indicated that the b and a terms from both the absolute linear and log-transformed MMGRMS versus TQ relationships were relatively reliable (ICCs), however, the SEMs for the log-transformed relationships were lower than the absolute linear models. Furthermore, the b term from the log-transformed relationships may provide unique information regarding the nonlinear characteristics (plateau points) of the MMGRMS versus TQ relationship, whereas the a term may indicate upward or downward shifts in MMGRMS across the TQ spectrum. Thus, the log-transformed MMGRMS versus TQ relationships may offer an attractive alternative method for reliably quantifying and tracking changes in the TQ-related patterns of response for MMGRMS on a subject-by-subject basis.