Wearable Sensors Detect Differences between the Sexes in Lower Limb Electromyographic Activity and Pelvis 3D Kinematics during Running

, 50% of runners su er from injuries. Consequently, more studies are being published about running biomechanics; these studies identify factors that can help prevent injuries. Scienti c evidence suggests that recreational runners should use personalized biomechanical training plans, not only to improve their performance, but also to prevent injuries caused by the inability of amateur athletes to tolerate increased loads, and/or because of poor form. This study provides an overview of the di erent normative patterns of lower limb muscle activation and articular ranges of the pelvis during running, at self-selected speeds, in men and women. Methods: 38 healthy runners aged 18 to 49 years were included in this work. We examined eight muscles by applying two wearable super cial electromyography sensors and an inertial sensor for three-dimensional (3D) pelvis kinematics. Results: the largest di erences were obtained for gluteus maximus activation in the rst double oat phase (p=0.013) and second stance phase (p=0.003), as well as in the gluteus medius in the second stance phase (p=0.028). In both cases, the activation distribution was more homogeneous in men and presented signi cantly lower values than those obtained for women. In addition, there was a signi cantly higher percentage of total vastus medialis activation in women throughout the running cycle with the median (25th–75th percentile) for women being 12.50% (9.25–14) and 10% (9–12) for men. Women also had a greater

cases, the activation distribution was more homogeneous in men and presented signi cantly lower values than those obtained for women. In addition, there was a signi cantly higher percentage of total vastus medialis activation in women throughout the running cycle with the median (25th–75th percentile) for women being 12.50% (9.25–14) and 10% (9–12) for men. Women also had a greater range of pelvis rotation during running at self-selected speeds (p=0.011). Conclusions: understanding the di erences between men and women, in terms of muscle activation and pelvic kinematic values, could be especially useful to allow health professionals detect athletes who may be at risk of injury. Keywords:running; kinematics; surface electromyography; wearables 1. Introduction Recreational running is becoming an increasingly popular pastime [1], with approximately 15% and 70% of amateur athletes currently engaging in this activity in the United Kingdom and the United States, respectively [2,3]. Various studies have shown that 50% of runners su er an injury each year [4], although there are discrepancies in the literature, due to incidence values that vary from 18.2% to 92.4% [5] and reported prevalence ranging from 46% to 90% among amateur runners [6,7]. In recent years, an increasing number of studies have been published in relation to the biomechanics of running, including factors that could help prevent and treat injuries in runners [8–11]. Running is a Sensors2020,20, 6478; doi:10.3390 /s20226478 /journal/sensors

Sensors2020,20, 6478 2 of 13 popular recreational activity, but a lack of adequate training in correct running techniques may account for the reported increase in injuries among these athletes [12]. Thus, in this work, we aimed to provide an overview of the di erent normative patterns of lower limb muscle activation and pelvic joint ranges during running at self-selected speeds in men and women. We analyzed the biomechanics of running by measuring the activation of the main muscles involved in this activity, as well as the dynamic ranges of joint movement, especially in the pelvis [13]. The choice of a preferred speed could be a ected by the level of performance and the intensity of the training habits [14]. It is reasonable to expect that amateur runners, with a higher level of performance, will train at higher intensities and, therefore, select a higher running speed for pleasure and metabolic cost [14,15]. Portable dynamic surface electromyography (sEMG) measurement devices, together with inertial sensor units (IMUs), are currently used for this type of analysis [16,17]. These systems provide information about muscle use intensity and activation time, and re ect the di erent contraction strategies, neuromuscular control systems, and three-dimensional (3D) pelvic kinematics used during running [18–20]. The use of wearable systems for these biomechanical measurements allow the data to be captured under more realistic conditions [21]. Given the intrinsic variability of these biomechanical values, the eld still lacks a set of reliable reference values for use when assessing both the status and evolution of injured individuals. Some studies have determined these values based on the dynamic range of the pelvis and level of muscle activation by using sEMG for the main muscles involved in running [21–27]. One study noted increased hamstring and hip exor tension in runners caused by excess anteroposterior pelvic movement or tilt [23], while in another, back pain was correlated with limited lower knee range [24]. Many studies, exploring the di erences in muscle activation in the stance and swing phases of running, are now available in the literature [19,28–30]. However, none have systematically categorized the values for muscle

hip exor tension in runners caused by excess anteroposterior pelvic movement or tilt [23], while in another, back pain was correlated with limited lower knee range [24]. Many studies, exploring the di erences in muscle activation in the stance and swing phases of running, are now available in the literature [19,28–30]. However, none have systematically categorized the values for muscle function and pelvic kinematics during the di erent phases of running. Moreover, running mechanics also di er between the sexes, but the di erences in the normative patterns of muscle activation, in di erent phases of running between male and female amateur runners, has not yet been determined [31]. Measuring and characterizing human movements during activity to evaluate athlete performance, improve technique, and prevent injuries is a crucial part of modern training programs [32]. Collecting these data will increase scienti c knowledge of kinematic patterns and the degree of muscle activation in runners. Therefore, the purpose of this study was to establish the di erences between the sexes in terms of lower limb sEMG activity and three-dimensional (3D) kinematics of the pelvis during running. 2. Materials and Methods 2.1. Participants Healthy participants were recruited, who typically engaged in at least 90 min of continuous running training per week, and who had not su ered any injury in the prior year that could have changed their movement patterns. In addition, we excluded individuals who reported having su ered an orthopedic, neurological, or surgical injury in the prior year that could have a ected their movement patterns. We explained the nature of the study to all of the participants and they signed their informed consent to participation prior to the start of the work. The entire study was carried out according to the principles of the Declaration of Helsinki, was approved by the ethics committee at CEU Cardenal Herrera University (reference number: CEI18/137), and was registered as a clinical trial (ClinicalTrials.gov ›: NCT04221698). 2.2. Procedure In this study, we measured the level of activation in the muscles of the dominant leg as well as the pelvic dynamic range of each participant. We

the principles of the Declaration of Helsinki, was approved by the ethics committee at CEU Cardenal Herrera University (reference number: CEI18/137), and was registered as a clinical trial (ClinicalTrials.gov ›: NCT04221698). 2.2. Procedure In this study, we measured the level of activation in the muscles of the dominant leg as well as the pelvic dynamic range of each participant. We used a treadmill (BH Fitness Columbia Pro

Sensors2020,20, 6478 3 of 13 130 cm 40 cm ) to establish standardized conditions under which the kinematic variables of running would be more reproducible. We set the incline to 1 and allowed each participant to select the speed [31,32] at which they regularly trained. The participants used their own shoes and were allowed a 15-min warm-up period in order to adjust to the treadmill. According to protocols used in previous running biomechanics studies [28,33–35], the initial speed was progressively increased over 2 min and was then maintained for 3 min while the data were collected. The dynamic range of the pelvis was assessed using an inertial sensor (BTS G-Sensor 2) with an ergonomic belt at the height of S1 to capture di erent kinematic and spatiotemporal variables. This IMU comprised a 16-axis triaxial accelerometer with multiple sensitivities ( 2, 4, 6, 8, and 16 g) with a frequency of 4 Hz to 1000 Hz, a triaxial gyroscope with multiple sensitivities ( 250, 500, 1000, 2000 o/s), with a frequency oscillating between 4 Hz to 8000 Hz, and a triaxial 13-bit magnetometer ( 1200 uT), with a frequency exceeding 100 Hz. Muscle activation was simultaneously studied by sEMG in eight muscles: the gluteus maximus, gluteus medius, rectus femoris, vastus medialis, biceps femoris, semitendinosus, medial gastrocnemius, and soleus. The skin was prepared according to SENIAM guidelines [36], and then two 20 mm pre-gelled self-adhesive bipolar Ag/AgCl disposable surface electrodes (Infant Electrode, Lessa, Barcelona) were placed on each muscle with a 20 mm interelectrode distance between them. A 10 g wireless probe (41.5 24, 8 14 mm) was placed on each pair of electrodes to capture the sEMG signal and send the information by Wi-Fi to the capture system (BTS FREEMG 1000, BTS Bioengineering, Milan, Italy) via a signal receiver (Wireless IEEE802.15.4) connected to a computer via USB [37]. The running phases analyzed by sEMG were the percentage of the stride cycle and percentage of each subphase. The start of the stride cycle corresponded to the initial contact and start of the contact of the same foot. The running subphases were: the

BTS Bioengineering, Milan, Italy) via a signal receiver (Wireless IEEE802.15.4) connected to a computer via USB [37]. The running phases analyzed by sEMG were the percentage of the stride cycle and percentage of each subphase. The start of the stride cycle corresponded to the initial contact and start of the contact of the same foot. The running subphases were: the rst stance, rst double oat, second stance, and second double oat (Figure). Thus, for the right leg, the rst stance occurred from the initial contact of the right foot to the take-o of the right toe. The rst double oat occurred from the initial oat phase of the right foot to the contact of the contralateral foot. This was then followed by the second stance, from the time of initial contact of the left foot to take-o of the left toe, and the second double oat from the initial oat phase of the left foot until contact of the contralateral foot. Figure 1. Figure of the running stride cycle sub-cycles: the rst stance (1st St), rst double oat (1st Sw), second stance (2nd St), and second double oat (2nd Sw). 2.3. Data Analysis The EMG signal was recorded simultaneously using a FREEEMG 1000 and EMG Analyzer (BTS Bioengineering, Milan, Italy) that was set to a sampling rate of 1000 Hz per channel, and the signals were band-pass ltered from 20 Hz to 450 Hz. The EMG signals were subsequently full-wave recti ed and low pass ltered using a bidirectional, 6th order Butterworth lter, with a cuto frequency of 5 Hz. The root mean square (RMS) in several subphases was detected. The IMU sensor detected every event performed, initial contact, and toe-o of each foot. Moreover, at the same time, the sEMG signal was recorded, so that the system selected the right and left strides and the di erent subphases ( rst stance phase, rst oat phase, second stance phase, second oat phase), as described in Figure. 2.4. Statistical Analysis To describe the demographic data of the population sample, descriptive statistics were calculated separately by sex for the participant age, height,

sEMG signal was recorded, so that the system selected the right and left strides and the di erent subphases ( rst stance phase, rst oat phase, second stance phase, second oat phase), as described in Figure. 2.4. Statistical Analysis To describe the demographic data of the population sample, descriptive statistics were calculated separately by sex for the participant age, height, weight, and training sessions performed during the chosen week and for the running dynamics data. The data from the study variables were analyzed to

Sensors2020,20, 6478 4 of 13 check for extreme outlying values using Chauvenet's criterion, because these may have represented abnormalities in the measurements, musculature, or nerve conduction of the participants. After testing compliance with the assumptions of normality (Shapiro–Wilk test) and homogeneity of variances (Levene's test), we decided to use non-parametric methods in our analyses. We used the Wilcoxon rank sum method (based on the Mann–Whitney U test) to compare the sex factor in the biomechanical patterns of pelvis use, muscular activation during the complete running cycle, and the mean activation between men and women at their self-selected speeds. G*Power software was used to calculate the sample size; to detect an e ect size of 0.8 with a statistical power of 0.8, we calculated that we would require at least 21 participants in each group. We nally obtained data from 22 men and 16 women, and post-hoc calculations gave us a statistical power of 0.75. RStudio Desktop software (version 1.2.5 for macOS; RStudio Inc., Boston, MA, USA) was used for all of our statistical analyses. 3. Results A total of 48 individuals initially participated in the study, of which eight were considered excluded values because of injury (n=7) or Bluetooth receiver failure (n=1). The demographics of these participants are described in Table. Table 1.Participant characteristics *. Value Female Male Participants,n 16 22 Age, years 27.07 9.16 26.39 6.61 Weight, Kg 58.31 7.06 70.14 8.3 Height, cm 166.3 0.06 177.5 0.07 Weekly number of training sessions 3.93 1.03 4.87 1.14 * Values represented as the mean and standard deviation (SD). Once the data from the 40 participants included in the trial had been analyzed, 2 participants were excluded because they were considered outliers, leaving a nal sample of 38 individuals. Regarding the self-selected speed, the mean for women was 9.22 ( 1.59) km/h, and for men it was 10.61 ( 1.56) km/h, with this di erence being statistically signi cant. Table p-value of the di erence between the speed and distance between the sexes, calculated using Mann–Whitney U tests. Table 2.Statistics and signi cance between sex and the speed and distance variables

self-selected speed, the mean for women was 9.22 ( 1.59) km/h, and for men it was 10.61 ( 1.56) km/h, with this di erence being statistically signi cant. Table p-value of the di erence between the speed and distance between the sexes, calculated using Mann–Whitney U tests. Table 2.Statistics and signi cance between sex and the speed and distance variables *. Female (avg) FemaleSD Male (avg) Male SD Wilcoxonp-Value Speed (km/h) 9.22 1.59 10.61 1.56 0.009 * Distance (Km) 0.79 0.13 0.9 0.14 0.02 * * Signi cant di erences atp<0.05. Speed expressed in kilometers/hour and distance measured in kilometers. 3.1. Kinematics of the Pelvis Signi cant di erences in the range of pelvic rotation (Figure) were observed between the sexes, with female runners presenting a greater range of rotation during running at their self-selected speed, but no signi cant di erences were observed in the tilt or obliquity between the sexes (Table).

Sensors2020,20, 6478 5 of 13 Table 3. Di erences between men and women in the kinematics of the pelvis during sprinting at a self-selected speed *. Variable Mean Men Mean Women p-Value Rotation 12.53 ( SD:3.2) 17.04 (SD:5.72) 0.011 * Obliquity 7.57 ( SD:1.99) 7.82 (SD:1.61) 0.391 Tilt 7.41 ( SD:1.68) 8.51 (SD:2.11) 0.086 * Mean values with their standard deviations (SD) are shown. * Statistically signi cant di erences atp<0.05. Figure 2. Variation of the rotation between women (F) and men (M), with each bar representing one participant. The lines summarize the distribution of the mean. 3.2. Mean Running Cycle Muscle Activation Table factor of total muscle activation during each running cycle. The vastus medialis showed a signi cantly higher percentage of activation in women throughout the running cycle (Figure) with a signi cantly di erent distribution between the sexes; there was greater muscle activation dispersion in women, indicating increased variability, while the vastus medialis activation homogeneity was reduced in men. Table 4.Statistics and signi cance of the percentage of total muscle activation during the running cycle *. Muscle % Activation Women % Activation Men Gluteus maximus 12 (11.25–15.50) 12 (11–13) Gluteus medius 12 (11–13) 11.50 (10.75–13) Femoral rectus 12 (11–14) 13.50 (12–15.25) Vastus medial 12.50 (9.25–14) * 10 (9–12) * Semitendinosus 14 (13–15.75) 13 (11.75–16) Femoral biceps 14.50 (13.25–17.30) 15.00 (13–15) Medial gastrocnemius 10.50 (9–12) 11.00 (10–12) Soleus 10.00 (10–11.75) 12 (11–13.70) * Percentage value of the median (25th–75th percentile). * Signi cant di erences atp<0.05.

Sensors2020,20, 6478 6 of 13 Figure 3. Percentage of the total activation of the vastus medialis during the running cycle distributed between women (F) and men (M). The distribution of the mean andSDwere more homogeneous in men. 3.3. Muscle Activation for Each of the Phases There were signi cant di erences in the muscle activation measurements for each of the phases in each of the main muscles (Table). Figure erence in the gluteus maximus muscle activation between women and men running at their self-selected speeds. The distribution of the muscle activation in men was more homogeneous and presented signi cantly lower values than for women. Figure erence in gluteus medius muscle activation between women and men during the second stance, showing lower homogeneity in women and greater activation than in men. Table 5.Thep-values of the mean in the muscles with signi cant di erences between the sexes in di erent phases. Muscle 1st Stance 1st Double Float 2nd Stance 2nd Double Float Gluteus maximus p=0.114 p=0.013 * p=0.003 * p=0.647 Gluteus medius p=0.198 p=0.057 p=0.028 * p=0.584 * Signi cant di erences atp<0.05. Figure 4.Variation by sex in the gluteus maximus in the rst double oat (A) and second stance (B).

The study analyzes sex differences in muscle activation and pelvic kinematics during running.