Multivariate linear regression analysis to evaluate multiple-set performance in active and inactive individuals

35 Objective: To determine how EMG, anthropometric, and psychological factors, and 36 physical activity levels affect isokinetic peak torque performance (IPT) of multiple set 37 exercise sessions. Methods: 20 participants (27±7 years old), classified as active 38 (A=10) and inactive (I=10), performed 10x10:40secs of maximal unilateral knee 39 flexions and extensions at 120o.s-1. The IPT, EMG, glucose, LDH and lactate 40 concentrations and perceptions of pain, effort, recovery. Results: Active volunteers 41 showed higher muscularity (52±5 vs 47±4 cm; p<0.05), PTI (262±4 vs 185±4 Nm; 42 p<0.05), relative lower drop in performance (14±2 vs 27±3% ; p<0.05), major MDF 43 (83±1 vs 76±1 Hz; p<0.05), lower log -Fins5 (-12.9±0.3 vs -12.7 ± 0.3 Hz; p<0.05), 44 smaller subjective perception of effort (14.8±0.3 vs 17.0±0.3) and higher subjective 45 perception of recovery (14.2±0.2 vs 12.3±0.3). There was a significant interaction 46 between relative fatigue and the number of sets (F=6.18; p<0.001). Stepwise multiple 47 regressions revealed that subjective perception of recovery best explained the fatigue 48 level generated in the active volunteers [fatigue level= 85.084–5255(SPR)] while for 49 body mass was the best determinant for the inactive group [fatigue level = -21.560 50 +1.828(BMI)]. Conclusion: Data from the present analysis suggest that physically 51 active individuals show higher torque development and a smaller fatigability index 52 when compared to inactive individuals. Among the fatigue models studied, it is possible 53 that alterations in biochemical components, psychophysiological and EMG are not 54 sensitive to the direct influence of the fatigue dynamics protocol, both in active or 55 inactive individuals. 56


Introduction
Therefore, the purpose of this study was to investigate via multivariate discriminant 104 analysis how electromyographical, metabolic, anthropometric, psychologic factors and 105 level of physical activity interact during isokinetic peak torque performance. activity per week were considered as active, while subjects below 150 minutes were 119 classified as inactive (insufficiently active) as per WHO (1998). After IPAQ analysis, 120 subjects were separated into two groups: active (n=10) and inactive (n=10).

Isokinetic parameter
To avoid unwanted body movements during the exercise, belts were fixed on the 139 thigh, pelvic area and torso. The dynamometer was aligned to the subject's right knee´s 140 axis and the arms were crossed across the torso. All subjects performed eight 141 submaximal repetitions, concentric actions for extension and eccentric actions at a speed 142 of 120 o s -1 . The Isokinetic Dynamometer was calibrated with the knee joint flexed to 90º.

143
The calibration of the isokinetic dynamometer was performed according to the 144 manufacturer specifications. The fatigue protocol was applied on the fourth day. All 145 tests and experimental processes were executed on the fourth meeting to avoid 146 variations and bias in our data.

147
Following the familiarization period, subjects performed the fatigue protocol, as 148 outlined in a previous study 6 . Briefly, all subjects performed a short warm-up using the provided between the sets. All test sessions were supervised by the same researcher and 156 all subjects were encouraged verbally during testing.

157
The isokinetic torque developed by the knee extensors and associated muscles 158 was continuously assessed as outlined previously. During the execution of the fatigue 159 protocol, the highest concentric peak torque in each of the 10 repetitions in each set was 160 recorded, and from this, the percentage decrement score (PDS) was calculated, as The surface EMG signal was recorded using an eight-channel module 166 conditioner (band pass filter at 20-500 H, amplifier gain of 1000 and a common mode 167 rejection ratio >120 dB). All data were acquired and processed using a 16-bit analog to 168 digital converter (EMG System do Brasil Ltda®), with a sampling frequency of 2 kHz.

169
The system included active bipolar electrodes with a pre-amplification gain of 20x. The 170 placement sites for the electrodes were cleaned with a cotton ball soaked in alcohol.
Afterward, disposable 10 mm-diameter Ag/AgCl surface electrodes (MediTrace®) were 172 attached to the belly of the muscle with inter electrode distance center-to-center of 2 cm 173 in the rectus femoris , vastus medialis oblique and vastus lateralis muscles.

174
EMG signals for evaluation of muscle fatigue were captured, including the root-

237
As illustrated in Figure 1, a significant main effect was noted for physical 238 activity level (F = 167.14; p < 0.001) and between sets (F = 13.05; p < 0.05). There were no significant interactions between the physical activity level and sets for absolute 240 torque (F = 1.193; p = 0.053). Active individuals showed higher absolute torque than 241 the inactive individuals. In relation to the absolute torque produced in the first sets, 242 volunteers began to fatigue (p<0.01) from the fourth set on.

243
There was a significant difference observed between groups for PDS (p < 0.05), 244 and ANOVA showed that there was a significant interaction between groups and sets 245 related to torque in the first set (F = 6.18; p < 0.001). The inactive group had a higher 246 decrease in performance both in PDS (Figure 2 A) and in torque in the first set ( Figure 1 247 B). Figure 1B shows that the inactive group begins to fatigue from the fourth set on.

248
The same observations occur in the active group from the sixth set on.

293
This study assessed the influence of various markers of fatigue on isokinetic 294 peak torque in active and inactive individuals. The primary finding of the study was an 295 impairment of peak torque in both groups associated with alterations in 296 electromyographical and psychophysiological signals, with no significant differences in 297 biochemical parameters. Also, considering that fatigue is a complex phenomenon, we 298 developed a predictive equation through multivariate regression analysis, taking into 299 account the parameters analyzed in this study.

300
Neither variable, biochemical nor electromyographical, obtained a significant 301 correlation. When groups were isolated, no significant correlation was found in the 302 active group, indicating that the fatigue process is different between active and inactive 303 subjects.

304
Taking into account the isokinetic peak torque´s values, the data of this analysis 305 differs from our previous study on the topic. One hypothesis for these differences could 306 be related to the level of trainability or physical activity level of the active group. The inactive group in this study presented a higher reduction in the isokinetic torque (-91 ± 308 12 %) when compared to a previous study (46 ± 5 %) 6 , although it was similar to other 309 studies 12 .

310
EMG analysis is a popular technique used by researchers to determine muscular 311 fatigue 1,2 . In the present study, a significant difference was found between groups only representing the onset of anaerobic metabolism. There is consistent evidence 19 that 334 intensities of exercise higher than 30% of 1RM incite the vascular beds via a vaso-335 compression mechanism resulting in a significant change of local blood perfusion. At 336 intensities above 85% of 1RM the vaso-compressive effect is total.

337
Alterations in lactate dehydrogenase enzyme concentration is another factor 338 believed to be associated with fatigue and musculoskeletal/metabolic adaptation.

339
Although there were no differences observed between groups in the present study, the 340 increase of enzyme concentrations points to the leakage of the enzyme through the sarcolemma; this has been postulated as a primary marker for muscle injury. Moreover, studies indicate that the LDH enzyme takes part in the metabolism of lactate 343 formation 20,21 , which is in agreement with our findings. This can be seen when we 344 consider the similarity between concentrations of lactate and lactate dehydrogenase.

345
This indicates that both groups completed the fatigue protocol with similar effort and 346 intensity, and were both subjected to the same metabolic overload.

347
The perception of effort has been widely used as an indicator of internal loading

363
Considering that fatigue is complex and multifactorial, with elements of neural, 364 metabolic, psychologic, mechanical and behavioral factors, its analysis previously has 365 been univariate or bi-variate. To our knowledge, this is the first study to analyze fatigue 366 via a multivariate approach, using parameters already outlined in the literature as 367 determinants observed in the process of performance reduction during exercise.

368
Consequently, the isolated interpretation of results in studies of fatigue is not erroneous,