FORCE VARIABILITY DURING FATIGUING SUBMAXIMAL ISOMETRIC ELBOW FLEXION EXERTIONS Tara Kajaks, Cassandra E. Petruzzi and Jim R. Potvin Department of Kinesiology, McMaster University, Hamilton, ON, Canada, kajakst@mcmaster.ca INTRODUCTION Understanding the patterns in force variability throughout the time-history of a fatiguing contraction may be useful in predicting instantaneous relative effort levels associated with an absolute known force demand, thus providing a ratio that could indicate fatigue level. Such a finding would provide a great deal of utility in ergonomic applications where fatigue can be a limiting factor in work task design. To date, studies investigating force variability in fatiguing contractions have shown that the standard deviation (SD) and coefficient of variation (CV) of the force output increase with fatigue [1], although few researchers have sought to use this measure to predict relative effort. Although unsuccessful, Robinson et al. (1994) did attempt to classify submaximal and maximal isometric efforts using torque CV. Their findings show good classification of maximal isometric efforts using CV, but poor classification of submaximal isometric efforts due to highly variable CVs under submaximal conditions. More recently, researchers have begun to study frequency domain changes in force variability, or physiological tremor, during fatiguing contractions. Huang et al. (2006) found a notable increase in the spectral peak, in the 8-12 Hz range, due to fatigue. However, they noted that these changes were most apparent due to fatigue induced from low-level contractions rather than high-exertion contractions. By way of explanation for the upwards shift in frequency content during low-level fatiguing contractions, they hypothesize a correlation between the recruitment of additional motor units. The fact that most, or all, motor units would have been initially recruited in the high level contractions may explain why the same phenomenon was not seen in high-level contractions. These results are supported by literature showing that mechanical- reflex properties are responsible for tremors below 7 Hz [4], the central nervous system mediates tremors in the 8-12 Hz range [5], and oscillations in exertion force signals between 16-30 Hz are due, in part, to cortical inputs [6]. Thus, it is clear that different physiological phenomenon may work to influence oscillations, or variability, within a force exertion signal. The purpose of this study was to quantify changes in force variability during fatiguing force-varying submaximal isometric exertions of the elbow flexors, using different filtering techniques to identify a method useful for predicting relative effort. Assuming a successful prediction of effort, then fatigue, defined as a decrease in force generating capacity, could likewise be estimated using the filtered data by determining the difference between relative exertion effort and force produced. It was hypothesized that, as local muscular fatigue increased during the protocol, there would be an increase in the variability of the total elbow joint flexor moment of force. It was further hypothesized that, should increased force variability be a consequence of fatigue, the magnitude of force variability at a given contraction level could be used to predict the net instantaneous fatigue status of the contracting muscles required to produce the joint moment. METHODS Seven healthy female subjects volunteered for this study (age: 21.1 years ± 0.4, height: 169.4cm ± 6.5, mass: 67.5 kg ± 8.8). Subjects sat in a height-adjustable chair alongside a custom- made aluminum structure (80/20 Inc., IN, USA) of fixed height. The right arm rested comfortably on a foam pad, elbow flexed to 90°, with the forearm supinated, and shoulder abducted to 90°. A reinforced cuff was placed around the subject’s distal forearm, and served to keep the wrist and fingers in a neutral posture. The cuff was fixed to a cable that was attached to an in-line uni-axial force transducer. Surface electromyography (SEMG) data were collected from the right biceps brachii, brachioradialis, brachialis and triceps brachii muscles using a Bortec AMT-8 amplifier (Bortec Biomedical Ltd., Calgary AB). Disposable bipolar Ag-AgCl surface EMG electrodes (MediTrace 133 adhesive electrodes), with an inter-electrode distance of 30mm, were placed over the muscle bellies on a line parallel to the orientation of the muscle fibers of the desired muscles. Subjects performed three different isometric elbow flexion exertion trials. The first trial was used to determine the maximal voluntary contraction (MVC) force in this posture. Subjects then traced a pattern on a computer monitor that consisted of 11 randomly selected 7-second plateaus at target force levels ranging from 10-95% MVC. This served as a rested force variability calibration trial. Next, subjects performed a fatiguing trial. They were required to trace a pyramid-like template that consisted of 15-second submaximal exertion plateaus of 20, 40, 60, 40, 20, and 0% MVC. A 5- second MVC and 5-second rest period followed each plateau. This exertion pattern was repeated until there was a 40% decrease in force generating capacity. The force transducer (MLP-300-C0, A-Tech Instruments, Scarborough, Canada) and differentially amplified SEMG (gain = 1000 and 5000, common mode rejection ratio = 115dB (at 60Hz), input impedance = 10GΩ) data were collected with LabVIEW software (National Instruments, Austin Tx.) using a PC compatible computer and converted by a 12-bit A/D card (National Instruments, Austin Tx.). Data were synchronously sampled at 1000 Hz. Post-processing of the force signal was performed using a custom-made LabView program. Post-processing included extracting windows of data from the rested calibration and fatiguing pyramid trials, followed by signal filtering. For the rested calibration trial, 3-second windows of data were extracted from the middle portion of each of the 7-second plateaus. For the fatiguing trial, 10- second windows of data were extracted from each of the 15- second submaximal plateaus, and 3-second windows were extracted from each of the 5-second post-plateau MVCs. To extract these data, an algorithm was developed that selected