Muscle Engagement Monitoring Using Self-Adhesive Elastic Nanocomposite Fabrics

Insight into, and measurements of, muscle contraction during movement may help improve the assessment of muscle function, quantification of athletic performance, and understanding of muscle behavior, prior to and during rehabilitation following neuromusculoskeletal injury. A self-adhesive, elastic fabric, nanocomposite, skin-strain sensor was developed and validated for human movement monitoring. We hypothesized that skin-strain measurements from these wearables would reveal different degrees of muscle engagement during functional movements. To test this hypothesis, the strain sensing properties of the elastic fabric sensors, especially their linearity, stability, repeatability, and sensitivity, were first verified using load frame tests. Human subject tests conducted in parallel with optical motion capture confirmed that they can reliably measure tensile and compressive skin-strains across the calf and tibialis anterior. Then, a pilot study was conducted to assess the correlation of skin-strain measurements with surface electromyography (sEMG) signals. Subjects did biceps curls with different weights, and the responses of the elastic fabric sensors worn over the biceps brachii and flexor carpi radialis (i.e., forearm) were well-correlated with sEMG muscle engagement measures. These nanocomposite fabric sensors were validated for monitoring muscle engagement during functional activities and did not suffer from the motion artifacts typically observed when using sEMGs in free-living community settings.


Introduction
Quantitative measurements of how muscles are engaged to enable functional movements are invaluable for diagnosing, treating, and managing musculoskeletal disorders and injuries (e.g., during physical therapy and rehabilitation). Electromyography (EMG), especially surface electromyography (sEMG), is a well-established method for assessing the health and activation of muscle fibers and muscle groups. In principle, EMG signals stem from the depolarization and repolarization of the muscle fiber cell membrane during muscle contraction. The electrophysiological action potentials generated by the transmembrane flow of positively charged sodium, potassium, and calcium ions across the concentration gradient in muscle and nerves create an electrical voltage measurable using surface or indwelling electrodes [1,2]. A single motor unit is composed of a motor neuron and the muscle fibers it innervates. Depending on the function of the muscle, a motor unit may be small (fewer fibers per neuron, indicating finer control) or large (greater fibers per neuron,

Nanocomposite Fabric Sensor Fabrication
Motion Tape was fabricated according to the procedure illustrated in Figure 1 and outlined by Lin et al. [20]. In short, Motion Tape was prepared by spray-coating GNS and ethyl cellulose (EC) thin films onto unidirectionally stretchable K-Tape (Rock Tape ® , Durham, NC, USA). GNS was synthesized by water-assisted liquid-phase exfoliation [21]. First, a 2 wt.% EC solution was prepared by mixing EC in 200 proof ethyl alcohol (EtOH) and stirring the mixture for 24 h. GNS was added to the EC-EtOH solution at a concentration Sensors 2022, 22, 6768 3 of 12 of 15 mg/mL and subjected to 2 h of bath sonication (150 W, 22 kHz). The dispersed GNS/EC-EtOH solution was then heated at 60 • C for~12 min using a Thermo Fisher Scientific digital hotplate to evaporate some of the EtOH solvent while increasing viscosity. The sprayable GNS/EC-EtOH ink was obtained after it cooled to room temperature. First, a 2 wt.% EC solution was prepared by mixing EC in 200 proof ethyl alcohol (EtOH) and stirring the mixture for 24 h. GNS was added to the EC-EtOH solution at a concentration of 15 mg/mL and subjected to 2 h of bath sonication (150 W, 22 kHz). The dispersed GNS/EC-EtOH solution was then heated at 60 °C for ~ 12 min using a Thermo Fisher Scientific digital hotplate to evaporate some of the EtOH solvent while increasing viscosity. The sprayable GNS/EC-EtOH ink was obtained after it cooled to room temperature. Second, a Paasche VL-series airbrush (Kenosha, WI) was used to spray-coat the GNS/EC-EtOH ink onto K-Tape substrates that were masked, to expose only the rectangular region where the nanocomposite sensing element was desired. Spray-coating was repeated three times, while pausing ~2 min between each layer for the ink to dry after each deposition step. An additional and final layer of GNS/EC thin film was drop-casted before drying the Motion Tape specimen for at least 1 h. It was found that drop-casting enhanced the overall nanocomposite uniformity and electrical conductivity. Last, measurement electrodes were established at opposite ends of the GNS/EC sensing element by drying flexible conductive ink (Voltera, Kitchener, ON, Canada), followed by soldering of multi-strand wires (Digi-Key Electronics, Thief River Falls, Minnesota, MN, USA). The result is a self-adhesive, elastic fabric sensor that can be customized to be different sizes during fabrication and can be affixed practically anywhere on the skin.
It should be mentioned that many other studies have also proposed various skinmounted and fabric-based sensors for assessing human movements and muscle engagement. For example, Anaya and Yuce [22] developed a portable, flexible, triboelectric-nanogenerator-based device for forearm muscles/tendons motion measurement and for assessing the motor dysfunctions associated with Parkinson's disease. Yang et al. [23] developed a textile strain sensor by integrating carbonic ink with spandex/polyamide fabrics. The sensor successfully demonstrated highly linear, repeatable, and stable signals during various human motion monitoring events. Similarly, Reddy K et al. [24] fabricated a flexible strain sensor by coating reduced graphene oxide onto a polyester knitted elastic band. The resulting sensor showed high strain sensitivity and signal-to-noise ratios when measuring strains as small as wrist pulses and as large as knee bending movements. This work not only adds to the breadth of fabric-based skin-strain sensors already proposed but is also unique in investigating the use of a self-adhesive and unidirectionally stretchable fabric as the substrate.

Sensing Characterization
Motion Tape specimens were individually mounted in a TestResources 150R (Shakopee, MN, USA) load frame for monotonic, uniaxial, and tensile cyclic electromechanical testing, as shown in Figure 2. Peak strains of 2.0%, 4.0%, 6.0%, 8.0%, and 10.0% were applied at a constant rate of 0.1 mm/s, while electrical resistance was recorded using a Keysight 34465A digital multimeter (Santa Rosa, CA, USA) sampling at 2 Hz. The load frame's crosshead displacement and applied load were also recorded at 10 Hz. All the Second, a Paasche VL-series airbrush (Kenosha, WI, USA) was used to spray-coat the GNS/EC-EtOH ink onto K-Tape substrates that were masked, to expose only the rectangular region where the nanocomposite sensing element was desired. Spray-coating was repeated three times, while pausing~2 min between each layer for the ink to dry after each deposition step. An additional and final layer of GNS/EC thin film was drop-casted before drying the Motion Tape specimen for at least 1 h. It was found that drop-casting enhanced the overall nanocomposite uniformity and electrical conductivity. Last, measurement electrodes were established at opposite ends of the GNS/EC sensing element by drying flexible conductive ink (Voltera, Kitchener, ON, Canada), followed by soldering of multi-strand wires (Digi-Key Electronics, Thief River Falls, Minnesota, MN, USA). The result is a self-adhesive, elastic fabric sensor that can be customized to be different sizes during fabrication and can be affixed practically anywhere on the skin.
It should be mentioned that many other studies have also proposed various skinmounted and fabric-based sensors for assessing human movements and muscle engagement. For example, Anaya and Yuce [22] developed a portable, flexible, triboelectricnanogenerator-based device for forearm muscles/tendons motion measurement and for assessing the motor dysfunctions associated with Parkinson's disease. Yang et al. [23] developed a textile strain sensor by integrating carbonic ink with spandex/polyamide fabrics. The sensor successfully demonstrated highly linear, repeatable, and stable signals during various human motion monitoring events. Similarly, Reddy K et al. [24] fabricated a flexible strain sensor by coating reduced graphene oxide onto a polyester knitted elastic band. The resulting sensor showed high strain sensitivity and signal-to-noise ratios when measuring strains as small as wrist pulses and as large as knee bending movements. This work not only adds to the breadth of fabric-based skin-strain sensors already proposed but is also unique in investigating the use of a self-adhesive and unidirectionally stretchable fabric as the substrate.

Sensing Characterization
Motion Tape specimens were individually mounted in a TestResources 150R (Shakopee, MN, USA) load frame for monotonic, uniaxial, and tensile cyclic electromechanical testing, as shown in Figure 2. Peak strains of 2.0%, 4.0%, 6.0%, 8.0%, and 10.0% were applied at a constant rate of 0.1 mm/s, while electrical resistance was recorded using a Keysight 34465A digital multimeter (Santa Rosa, CA, USA) sampling at 2 Hz. The load frame's crosshead displacement and applied load were also recorded at 10 Hz. All the data were collected simultaneously using the Keysight BenchVue software for ease of time synchronization.

Human Subject Testing
Three different sets of human subject tests were conducted as part of this study. For all the tests, the adhesive backing was peeled off to directly affix, without any pre-stretching, the self-adhesive Motion Tape sensors onto skin that was pre-cleaned using an alcohol wipe. First, the sensor verification test was based on Motion Tape sensors mounted to a subject's calf and tibialis anterior, as shown in Figure 3. Squats were performed, and the subject's movements were captured using a 12-camera Vicon optical motion capture (mocap) system (Vicon Motion Systems Ltd., Oxford, UK), which recorded the 3D positions of all retroreflective markers at 120 Hz. The Motion Tape sensors were connected to the Vicon Lock Lab ® 64-channel analog interface, so that their time-synchronized electrical resistance was recorded simultaneously, at a sampling rate of 120 Hz, using the Vicon data collection software.  In addition to using mocap to measure subject movements, a pair of retroreflective markers were affixed adjacent to each Motion Tape, so that the 3D positional data of the markers could be used to estimate the linear strains induced during functional movements. It should be mentioned that the calf and tibialis anterior were selected for this sensor verification test because the skin in these regions remained relatively flat during squatting, which ensured that the mocap-estimated linear strains were comparable to the skinstrains measured by Motion Tapes. However, estimating skin-strains from the change in distance between two retroreflective markers would inevitably introduce intrinsic errors. Figure 4 depicts a case when Motion Tape is mounted on a slightly curved surface (e.g., skin). When the surface is strained (e.g., due to movement or muscle engagement), the

Human Subject Testing
Three different sets of human subject tests were conducted as part of this study. For all the tests, the adhesive backing was peeled off to directly affix, without any pre-stretching, the self-adhesive Motion Tape sensors onto skin that was pre-cleaned using an alcohol wipe. First, the sensor verification test was based on Motion Tape sensors mounted to a subject's calf and tibialis anterior, as shown in Figure 3. Squats were performed, and the subject's movements were captured using a 12-camera Vicon optical motion capture (mocap) system (Vicon Motion Systems Ltd., Oxford, UK), which recorded the 3D positions of all retroreflective markers at 120 Hz. The Motion Tape sensors were connected to the Vicon Lock Lab ® 64-channel analog interface, so that their time-synchronized electrical resistance was recorded simultaneously, at a sampling rate of 120 Hz, using the Vicon data collection software.

Human Subject Testing
Three different sets of human subject tests were conducted as part of this study. For all the tests, the adhesive backing was peeled off to directly affix, without any pre-stretching, the self-adhesive Motion Tape sensors onto skin that was pre-cleaned using an alcohol wipe. First, the sensor verification test was based on Motion Tape sensors mounted to a subject's calf and tibialis anterior, as shown in Figure 3. Squats were performed, and the subject's movements were captured using a 12-camera Vicon optical motion capture (mocap) system (Vicon Motion Systems Ltd., Oxford, UK), which recorded the 3D positions of all retroreflective markers at 120 Hz. The Motion Tape sensors were connected to the Vicon Lock Lab ® 64-channel analog interface, so that their time-synchronized electrical resistance was recorded simultaneously, at a sampling rate of 120 Hz, using the Vicon data collection software.  In addition to using mocap to measure subject movements, a pair of retroreflective markers were affixed adjacent to each Motion Tape, so that the 3D positional data of the markers could be used to estimate the linear strains induced during functional movements. It should be mentioned that the calf and tibialis anterior were selected for this sensor verification test because the skin in these regions remained relatively flat during squatting, which ensured that the mocap-estimated linear strains were comparable to the skinstrains measured by Motion Tapes. However, estimating skin-strains from the change in distance between two retroreflective markers would inevitably introduce intrinsic errors. Figure 4 depicts a case when Motion Tape is mounted on a slightly curved surface (e.g., skin). When the surface is strained (e.g., due to movement or muscle engagement), the In addition to using mocap to measure subject movements, a pair of retroreflective markers were affixed adjacent to each Motion Tape, so that the 3D positional data of the markers could be used to estimate the linear strains induced during functional movements. It should be mentioned that the calf and tibialis anterior were selected for this sensor verification test because the skin in these regions remained relatively flat during squatting, which ensured that the mocap-estimated linear strains were comparable to the skin-strains measured by Motion Tapes. However, estimating skin-strains from the change in distance between two retroreflective markers would inevitably introduce intrinsic errors. Figure 4 depicts a case when Motion Tape is mounted on a slightly curved surface (e.g., skin). When the surface is strained (e.g., due to movement or muscle engagement), the curvature of the surface would change. The actual surface strains are shown in red in Figure 4, but mocap estimates linear strains by only considering the line-of-sight change in distance between the two retroreflective markers. Strain is calculated using (L New − L Original )/L Original , which is only accurate if the strains are confined to a flat, rather than curved, surface. curvature of the surface would change. The actual surface strains are shown in red in Figure 4, but mocap estimates linear strains by only considering the line-of-sight change in distance between the two retroreflective markers. Strain is calculated using (LNew − LOriginal)/LOriginal, which is only accurate if the strains are confined to a flat, rather than curved, surface. The second muscle engagement human subject verification test involved participants performing biceps curls using different weights ( Figure 5). Following visual cues and instruction, subjects performed biceps curls to approximately the same angle (i.e., ~51°-52°). Elbow angles during biceps curls were measured using two Xsens DOT™ IMUs (Enschede, Netherlands) worn at the biceps and the forearm. Motion Tape was applied perpendicular to the biceps brachii, as shown in Figure 5, and its electrical resistance was recorded using a PXIe-4082 digital multimeter data acquisition (DAQ) system (National Instruments [NI], Austin, TX, USA) sampling at 296 Hz. A Delsys Trigno™ Avanti (Natick, MA, USA) wireless sEMG sensor (aluminum bar electrodes with 10-mm inter-electrode spacing) was also worn adjacent to the Motion Tape, but in parallel with the biceps brachii, to measure muscle engagement, as shown in Figure 5. The skin was also cleaned with EtOH prior to sEMG sensor attachment, which was secured on the skin using bare K-Tape. The sEMG data were recorded using the Delsys EMGworks ® analysis software running on the same personal computer (PC), to ensure time-synchronized measurements. The third test for validating muscle engagement monitoring followed a similar protocol as the aforementioned biceps curl verification tests. Here, Motion Tape was affixed The second muscle engagement human subject verification test involved participants performing biceps curls using different weights ( Figure 5). Following visual cues and instruction, subjects performed biceps curls to approximately the same angle (i.e.,~51 • -52 • ). Elbow angles during biceps curls were measured using two Xsens DOT™ IMUs (Enschede, The Netherlands) worn at the biceps and the forearm. Motion Tape was applied perpendicular to the biceps brachii, as shown in Figure 5, and its electrical resistance was recorded using a PXIe-4082 digital multimeter data acquisition (DAQ) system (National Instruments [NI], Austin, TX, USA) sampling at 296 Hz. A Delsys Trigno™ Avanti (Natick, MA, USA) wireless sEMG sensor (aluminum bar electrodes with 10-mm inter-electrode spacing) was also worn adjacent to the Motion Tape, but in parallel with the biceps brachii, to measure muscle engagement, as shown in Figure 5. The skin was also cleaned with EtOH prior to sEMG sensor attachment, which was secured on the skin using bare K-Tape. The sEMG data were recorded using the Delsys EMGworks ® analysis software running on the same personal computer (PC), to ensure time-synchronized measurements.
curvature of the surface would change. The actual surface strains are shown in red in Figure 4, but mocap estimates linear strains by only considering the line-of-sight change in distance between the two retroreflective markers. Strain is calculated using (LNew − LOriginal)/LOriginal, which is only accurate if the strains are confined to a flat, rather than curved, surface. The second muscle engagement human subject verification test involved participants performing biceps curls using different weights ( Figure 5). Following visual cues and instruction, subjects performed biceps curls to approximately the same angle (i.e., ~51°-52°). Elbow angles during biceps curls were measured using two Xsens DOT™ IMUs (Enschede, Netherlands) worn at the biceps and the forearm. Motion Tape was applied perpendicular to the biceps brachii, as shown in Figure 5, and its electrical resistance was recorded using a PXIe-4082 digital multimeter data acquisition (DAQ) system (National Instruments [NI], Austin, TX, USA) sampling at 296 Hz. A Delsys Trigno™ Avanti (Natick, MA, USA) wireless sEMG sensor (aluminum bar electrodes with 10-mm inter-electrode spacing) was also worn adjacent to the Motion Tape, but in parallel with the biceps brachii, to measure muscle engagement, as shown in Figure 5. The skin was also cleaned with EtOH prior to sEMG sensor attachment, which was secured on the skin using bare K-Tape. The sEMG data were recorded using the Delsys EMGworks ® analysis software running on the same personal computer (PC), to ensure time-synchronized measurements. The third test for validating muscle engagement monitoring followed a similar protocol as the aforementioned biceps curl verification tests. Here, Motion Tape was affixed The third test for validating muscle engagement monitoring followed a similar protocol as the aforementioned biceps curl verification tests. Here, Motion Tape was affixed to the forearm, perpendicular to the flexor carpi radialis, as shown in Figure 6, and its electrical resistance was also recorded using the NI DAQ system. A Delsys Trigno Avanti sEMG sensor was attached adjacent to the Motion Tape but in parallel with the flexor carpi radialis ( Figure 6). Biceps curls were performed using different weights, while the sensing streams from both the Motion Tape and sEMG were recorded simultaneously.
to the forearm, perpendicular to the flexor carpi radialis, as shown in Figure 6, and it electrical resistance was also recorded using the NI DAQ system. A Delsys Trigno Avant sEMG sensor was attached adjacent to the Motion Tape but in parallel with the flexor carp radialis ( Figure 6). Biceps curls were performed using different weights, while the sensing streams from both the Motion Tape and sEMG were recorded simultaneously.

Strain Sensing Properties
The hypothesis that Motion Tape could capture skin-strain measurements correlated with the degree of muscle engagement was tested in this study. First, the strain sensing properties of Motion Tape were verified by subjecting them to tensile cyclic loads to dif ferent peak strains while simultaneously measuring their electrical resistance (Figure 2) Here, a 10% strain limit was selected, because the expected skin-strains over major muscl groups should be below this threshold, unlike joints, which can be greater than 30%.
The electrical responses of tests loaded to different peak strains are overlaid in Figur 7a. Since all the tensile cyclic tests were conducted at the same loading rate of 0.1 mm/s the overlaid plots in Figure 7a were produced by normalizing the time scale (x-axis) with respect to the shortest duration test (i.e., 2% peak strain). This post-processing step allow one to directly compare Motion Tape sensing response corresponding to different peak strains. Overall, the set of representative resistance time histories acquired during tensil cyclic testing confirmed the Motion Tape's stable and repeatable strain sensing behavio (Figure 7a). In addition, sensor linearity was assessed by normalizing its change in re sistance (∆R) with respect to its unstrained baseline resistance (R0) and then plotting th data against applied strains, as shown in Figure 7b. Linear least-squares regression line fitted to the data confirmed sensor linearity (with correlation coefficients, ρ, that all ex ceeded 0.99) and consistent sensitivity, which ranged from 13.0 to 18.5. These results ar consistent with the findings presented by Lin et al. [20], where Motion Tape sensor line arity and sensitivity remained constant, with no baseline resistance drifts, even after >200 cycles of cyclic loading.

Strain Sensing Properties
The hypothesis that Motion Tape could capture skin-strain measurements correlated with the degree of muscle engagement was tested in this study. First, the strain sensing properties of Motion Tape were verified by subjecting them to tensile cyclic loads to different peak strains while simultaneously measuring their electrical resistance ( Figure 2). Here, a 10% strain limit was selected, because the expected skin-strains over major muscle groups should be below this threshold, unlike joints, which can be greater than 30%.
The electrical responses of tests loaded to different peak strains are overlaid in Figure 7a. Since all the tensile cyclic tests were conducted at the same loading rate of 0.1 mm/s, the overlaid plots in Figure 7a were produced by normalizing the time scale (x-axis) with respect to the shortest duration test (i.e., 2% peak strain). This post-processing step allows one to directly compare Motion Tape sensing response corresponding to different peak strains. Overall, the set of representative resistance time histories acquired during tensile cyclic testing confirmed the Motion Tape's stable and repeatable strain sensing behavior (Figure 7a). In addition, sensor linearity was assessed by normalizing its change in resistance (∆R) with respect to its unstrained baseline resistance (R 0 ) and then plotting the data against applied strains, as shown in Figure 7b. Linear least-squares regression lines fitted to the data confirmed sensor linearity (with correlation coefficients, ρ, that all exceeded 0.99) and consistent sensitivity, which ranged from 13.0 to 18.5. These results are consistent with the findings presented by Lin et al. [20], where Motion Tape sensor linearity and sensitivity remained constant, with no baseline resistance drifts, even after >200 cycles of cyclic loading.
to the forearm, perpendicular to the flexor carpi radialis, as shown in Figure 6, and its electrical resistance was also recorded using the NI DAQ system. A Delsys Trigno Avanti sEMG sensor was attached adjacent to the Motion Tape but in parallel with the flexor carpi radialis ( Figure 6). Biceps curls were performed using different weights, while the sensing streams from both the Motion Tape and sEMG were recorded simultaneously.

Strain Sensing Properties
The hypothesis that Motion Tape could capture skin-strain measurements correlated with the degree of muscle engagement was tested in this study. First, the strain sensing properties of Motion Tape were verified by subjecting them to tensile cyclic loads to different peak strains while simultaneously measuring their electrical resistance ( Figure 2). Here, a 10% strain limit was selected, because the expected skin-strains over major muscle groups should be below this threshold, unlike joints, which can be greater than 30%.
The electrical responses of tests loaded to different peak strains are overlaid in Figure  7a. Since all the tensile cyclic tests were conducted at the same loading rate of 0.1 mm/s, the overlaid plots in Figure 7a were produced by normalizing the time scale (x-axis) with respect to the shortest duration test (i.e., 2% peak strain). This post-processing step allows one to directly compare Motion Tape sensing response corresponding to different peak strains. Overall, the set of representative resistance time histories acquired during tensile cyclic testing confirmed the Motion Tape's stable and repeatable strain sensing behavior (Figure 7a). In addition, sensor linearity was assessed by normalizing its change in resistance (∆R) with respect to its unstrained baseline resistance (R0) and then plotting the data against applied strains, as shown in Figure 7b. Linear least-squares regression lines fitted to the data confirmed sensor linearity (with correlation coefficients, ρ, that all exceeded 0.99) and consistent sensitivity, which ranged from 13.0 to 18.5. These results are consistent with the findings presented by Lin et al. [20], where Motion Tape sensor linearity and sensitivity remained constant, with no baseline resistance drifts, even after >200 cycles of cyclic loading.

Verification of Motion Tape for Skin-Strain Monitoring
Verification tests of skin-strain monitoring during functional movements were performed with two participants wearing Motion Tapes at the calf and tibialis anterior, which was described in Section 2.3. The sensor verification tests were performed by affixing a pair of retroreflective markers adjacent to each Motion Tape placed on the calf and tibialis anterior. An optical motion capture system recorded their 3D positions during squatting motion. The linear strain of each Motion Tape was estimated from changes in marker pair distances with respect to their initial separation distance when the subject stood still (Figure 4) [25]. Muscle engagement verification tests were conducted by placing an sEMG electrode over the upper arm and forearm so that the Motion Tape electrical resistance could be compared against muscle engagement during biceps curls.
Motion Tape electrical resistance and mocap data were first collected when subjects performed squats. Figure 8a plots the normalized change in resistance time history of the Motion Tape on the calf (Figure 3a), and this dataset was also overlaid with the mocapestimated linear strains. Mocap confirmed that compressive skin-strains were induced due to contraction of the gastrocnemius medialis (i.e., calf muscle), which corresponded to decreasing the normalized change in resistance (R n = ∆R/R 0 ) of Motion Tape as the subject moved from a standing to a squatting posture. Plotting R n with respect to linear strains verified its linear and low-hysteresis sensing performance, which was supported by the high correlation coefficient (ρ = 0.98) of the fitted linear least-squares regression line in Figure 8b. No apparent baseline resistance drifts were observed.

Verification of Motion Tape for Skin-Strain Monitoring
Verification tests of skin-strain monitoring during functional movements were pe formed with two participants wearing Motion Tapes at the calf and tibialis anterior, whi was described in Section 2.3. The sensor verification tests were performed by affixing pair of retroreflective markers adjacent to each Motion Tape placed on the calf and tibia anterior. An optical motion capture system recorded their 3D positions during squattin motion. The linear strain of each Motion Tape was estimated from changes in marker pa distances with respect to their initial separation distance when the subject stood still (Fi ure 4) [25]. Muscle engagement verification tests were conducted by placing an sEM electrode over the upper arm and forearm so that the Motion Tape electrical resistan could be compared against muscle engagement during biceps curls.
Motion Tape electrical resistance and mocap data were first collected when subjec performed squats. Figure 8a plots the normalized change in resistance time history of th Motion Tape on the calf (Figure 3a), and this dataset was also overlaid with the moca estimated linear strains. Mocap confirmed that compressive skin-strains were induce due to contraction of the gastrocnemius medialis (i.e., calf muscle), which corresponde to decreasing the normalized change in resistance (Rn = ∆R/R0) of Motion Tape as the su ject moved from a standing to a squatting posture. Plotting Rn with respect to linear strai verified its linear and low-hysteresis sensing performance, which was supported by th high correlation coefficient (ρ = 0.98) of the fitted linear least-squares regression line The same conclusions could be drawn from the data collected from mocap and M tion Tape mounted on the tibialis anterior during squatting (Figure 3b). Tension was i duced in the tibialis anterior, which caused Rn to increase (as opposed to decreasing o the calf), as shown in Figure 9a. Strong linear skin-strain sensing response was also o served, where ρ = 0.96 based on the best-fit linear least-squares regression line shown Figure 9b. However, in both cases, Figures 8b and 9b show some hysteresis response. Th hysteresis may be an artifact from errors associated with estimating linear strains fro mocap, which was explained in Section 2.3 and Figure 4. While the Motion Tape chara terization tests did not reveal any hysteresis behavior (Figure 7), the sensors could pote tially exhibit hysteresis at high strain rates (such as the case during these movement Future tests will investigate if and how Motion Tape behavior changes during differe applied strain rates during intense physical activities. The same conclusions could be drawn from the data collected from mocap and Motion Tape mounted on the tibialis anterior during squatting (Figure 3b). Tension was induced in the tibialis anterior, which caused R n to increase (as opposed to decreasing on the calf), as shown in Figure 9a. Strong linear skin-strain sensing response was also observed, where ρ = 0.96 based on the best-fit linear least-squares regression line shown in Figure 9b. However, in both cases, Figures 8b and 9b show some hysteresis response. This hysteresis may be an artifact from errors associated with estimating linear strains from mocap, which was explained in Section 2.3 and Figure 4. While the Motion Tape characterization tests did not reveal any hysteresis behavior (Figure 7), the sensors could potentially exhibit hysteresis at high strain rates (such as the case during these movements). Future tests will investigate if and how Motion Tape behavior changes during different applied strain rates during intense physical activities. In fact, further analyses were also performed to examine the Motion Tape's sen response among eight different trials. Normalized cross-correlation was applied betw the Motion Tape Rn and mocap-estimated linear strain time histories. The results in Fi 10a show that the average time lag was 0.10 s, suggesting nearly instantaneous Mo Tape response to changes in skin-strains. The correlation coefficients for all eight trial also summarized in Figure 10b. The average ρ was 0.90, with a standard deviation of ± again confirming a strong sensor linearity and Motion Tape's ability to accurately qua skin-strains.

Verification of Motion Tape for Measuring Muscle Engagement
Building upon the skin-strain sensing verification test results presented in Figu and 9, biceps curl subject tests were performed to test the hypothesis that Motion could measure muscle engagement during functional movements (see Section 2.3). In dition to mounting Motion Tape over the biceps brachii, a wireless sEMG sensor was worn in parallel with the biceps brachii to measure muscle activation ( Figure 5). IMUs were also attached to the forearm and upper arm to measure the change in e angle during biceps curls. Each subject then performed biceps curls using three differe weighted dumbbells, while maintaining approximately the same range of movemen confirmed by IMU-derived angles of rotations). Although sEMGs and IMUs can both fer from movement artifacts, especially during high-intensity activities, the biceps were performed slowly to avoid such experimental errors. Figure 11a overlays Motion Tape and sEMG data for biceps curls performed usi 5, and 10 lb dumbbells, where the maximum angle of rotation was maintained at 51°  In fact, further analyses were also performed to examine the Motion Tape's sensing response among eight different trials. Normalized cross-correlation was applied between the Motion Tape R n and mocap-estimated linear strain time histories. The results in Figure 10a show that the average time lag was 0.10 s, suggesting nearly instantaneous Motion Tape response to changes in skin-strains. The correlation coefficients for all eight trials are also summarized in Figure 10b. The average ρ was 0.90, with a standard deviation of ±0.06, again confirming a strong sensor linearity and Motion Tape's ability to accurately quantify skin-strains. In fact, further analyses were also performed to examine the Motion Tape's sensing response among eight different trials. Normalized cross-correlation was applied between the Motion Tape Rn and mocap-estimated linear strain time histories. The results in Figure  10a show that the average time lag was 0.10 s, suggesting nearly instantaneous Motion Tape response to changes in skin-strains. The correlation coefficients for all eight trials are also summarized in Figure 10b. The average ρ was 0.90, with a standard deviation of ±0.06, again confirming a strong sensor linearity and Motion Tape's ability to accurately quantify skin-strains.

Verification of Motion Tape for Measuring Muscle Engagement
Building upon the skin-strain sensing verification test results presented in Figures 8  and 9, biceps curl subject tests were performed to test the hypothesis that Motion Tape could measure muscle engagement during functional movements (see Section 2.3). In addition to mounting Motion Tape over the biceps brachii, a wireless sEMG sensor was also worn in parallel with the biceps brachii to measure muscle activation ( Figure 5). Two IMUs were also attached to the forearm and upper arm to measure the change in elbow angle during biceps curls. Each subject then performed biceps curls using three differently weighted dumbbells, while maintaining approximately the same range of movement (as confirmed by IMU-derived angles of rotations). Although sEMGs and IMUs can both suffer from movement artifacts, especially during high-intensity activities, the biceps curls were performed slowly to avoid such experimental errors. Figure 11a overlays Motion Tape and sEMG data for biceps curls performed using 2, 5, and 10 lb dumbbells, where the maximum angle of rotation was maintained at 51°-52°.

Verification of Motion Tape for Measuring Muscle Engagement
Building upon the skin-strain sensing verification test results presented in Figures 8 and 9, biceps curl subject tests were performed to test the hypothesis that Motion Tape could measure muscle engagement during functional movements (see Section 2.3). In addition to mounting Motion Tape over the biceps brachii, a wireless sEMG sensor was also worn in parallel with the biceps brachii to measure muscle activation ( Figure 5). Two IMUs were also attached to the forearm and upper arm to measure the change in elbow angle during biceps curls. Each subject then performed biceps curls using three differently weighted dumbbells, while maintaining approximately the same range of movement (as confirmed by IMU-derived angles of rotations). Although sEMGs and IMUs can both suffer from movement artifacts, especially during high-intensity activities, the biceps curls were performed slowly to avoid such experimental errors. Figure 11a overlays Motion Tape and sEMG data for biceps curls performed using 2, 5, and 10 lb dumbbells, where the maximum angle of rotation was maintained at 51 • -52 • . The sEMG data shown were denoised using an RMS calculation based on a moving window (0.125 s window length and 0.0625 s window overlap) [26], and the data were also normalized with respect to the peak voltage amplitude (i.e., results are displayed from 0% to 100%). Figure 11a shows that Motion Tape measured larger skin-strains on the biceps as heavier weights were lifted. The same trend and greater muscle engagement were also recorded by sEMG. The sEMG data shown were denoised using an RMS calculation based on a moving window (0.125 s window length and 0.0625 s window overlap) [26], and the data were also normalized with respect to the peak voltage amplitude (i.e., results are displayed from 0% to 100%). Figure 11a shows that Motion Tape measured larger skin-strains on the biceps as heavier weights were lifted. The same trend and greater muscle engagement were also recorded by sEMG. To show that these trends are not unique for a particular subject, Figure 11b plots similar results when a different subject performed biceps curls using 0, 3, and 5 lb weights. Similar trends were observed, with both the Motion Tape skin-strains and sEMG signal peaks increasing when heavier weights were lifted. The 0 lb case also showed changes in skin-strains (i.e., much greater than the sEMG signal changes), but this is expected because Motion Tape is a skin surface measurement and does not solely measure muscle engagement. The correlation between Motion Tape skin-strain measurements and sEMG signals suggest that Motion Tape is sensitive enough to capture skin-strain features associated with different degrees of muscle engagement. Furthermore, a comparison of Figures 11a  and 11b shows that the magnitude of Rn varies between different subjects. This is expected, given that muscle properties, body fat, and other subject-specific parameters would influence Motion Tape outputs, as they would for sEMG signals as well.

Validation of Motion Tape for Measuring Muscle Engagement
Additional Motion Tape and sEMG tests were conducted on a different muscle group to show that the sensing principle applies elsewhere on the body (Section 2.3). Both types of sensors (i.e., Motion Tape and sEMG) were mounted on a subject's forearm (flexor carpi radialis), as shown in Figure 6. Biceps curls using 5, 10, and 20 lb dumbbells were performed while maintaining a similar range of movement. The sEMG signals of Figure 12 confirm a greater muscle activation as heavier weights were lifted, and the Motion Tape Rn time histories also showed greater amplitudes, following a similar trend. These results successfully validated that Motion Tape skin-strain measurements are well-correlated with sEMG muscle engagement measurements. However, it should be noted that the biceps brachii and flexor carpi radialis are major muscle groups. Other and smaller muscle groups will be considered in future tests, to further validate this technology, as well as during other types of functional movements. To show that these trends are not unique for a particular subject, Figure 11b plots similar results when a different subject performed biceps curls using 0, 3, and 5 lb weights. Similar trends were observed, with both the Motion Tape skin-strains and sEMG signal peaks increasing when heavier weights were lifted. The 0 lb case also showed changes in skin-strains (i.e., much greater than the sEMG signal changes), but this is expected because Motion Tape is a skin surface measurement and does not solely measure muscle engagement. The correlation between Motion Tape skin-strain measurements and sEMG signals suggest that Motion Tape is sensitive enough to capture skin-strain features associated with different degrees of muscle engagement. Furthermore, a comparison of Figure 11a,b shows that the magnitude of R n varies between different subjects. This is expected, given that muscle properties, body fat, and other subject-specific parameters would influence Motion Tape outputs, as they would for sEMG signals as well.

Validation of Motion Tape for Measuring Muscle Engagement
Additional Motion Tape and sEMG tests were conducted on a different muscle group to show that the sensing principle applies elsewhere on the body (Section 2.3). Both types of sensors (i.e., Motion Tape and sEMG) were mounted on a subject's forearm (flexor carpi radialis), as shown in Figure 6. Biceps curls using 5, 10, and 20 lb dumbbells were performed while maintaining a similar range of movement. The sEMG signals of Figure 12 confirm a greater muscle activation as heavier weights were lifted, and the Motion Tape R n time histories also showed greater amplitudes, following a similar trend. These results successfully validated that Motion Tape skin-strain measurements are well-correlated with sEMG muscle engagement measurements. However, it should be noted that the biceps brachii and flexor carpi radialis are major muscle groups. Other and smaller muscle groups will be considered in future tests, to further validate this technology, as well as during other types of functional movements.

Conclusions
In summary, a self-adhesive, low-profile, conformable, and disposable wea skin-strain sensor for muscle engagement monitoring was presented. Experiments o tion Tape conducted using a load frame and on human subjects successfully verified linear, stable, and repeatable strain sensing properties. To show that Motion Tape measure different degrees of muscle engagement during functional movements, b curl tests were performed with subjects also wearing an sEMG sensor. Sensor data directly mounting Motion Tape over the biceps brachii, as well as over the flexor radialis (i.e., forearm), showed that greater muscle activation (i.e., higher amplitude s signals) was correlated with greater changes in Motion Tape resistance. It is worth n that, while sEMG is considered the gold standard for muscle engagement monitor is relatively rigid and susceptible to motion artifacts when not secured properly skin, whereas Motion Tape can remain affixed firmly in place on the skin while defor freely. Overall, these wearable textile-based sensors and their continued testing an velopment could lead to their potential use for sports coaching, physical rehabilit and telemedicine, among many other healthcare, athletic, and military applications. W Motion Tape cannot replace sEMGs for quantifying muscle activation, these se streams can potentially provide insights about the degree of muscle engagement, cially during movement-intensive activities. Future work will focus on clinical studie rigorously test Motion Tape for muscle engagement monitoring in different body among diverse subjects, and with larger subject pools.

Conclusions
In summary, a self-adhesive, low-profile, conformable, and disposable wearable skinstrain sensor for muscle engagement monitoring was presented. Experiments on Motion Tape conducted using a load frame and on human subjects successfully verified their linear, stable, and repeatable strain sensing properties. To show that Motion Tape could measure different degrees of muscle engagement during functional movements, biceps curl tests were performed with subjects also wearing an sEMG sensor. Sensor data from directly mounting Motion Tape over the biceps brachii, as well as over the flexor carpi radialis (i.e., forearm), showed that greater muscle activation (i.e., higher amplitude sEMG signals) was correlated with greater changes in Motion Tape resistance. It is worth noting that, while sEMG is considered the gold standard for muscle engagement monitoring, it is relatively rigid and susceptible to motion artifacts when not secured properly to the skin, whereas Motion Tape can remain affixed firmly in place on the skin while deforming freely. Overall, these wearable textile-based sensors and their continued testing and development could lead to their potential use for sports coaching, physical rehabilitation, and telemedicine, among many other healthcare, athletic, and military applications. While Motion Tape cannot replace sEMGs for quantifying muscle activation, these sensing streams can potentially provide insights about the degree of muscle engagement, especially during movementintensive activities. Future work will focus on clinical studies that rigorously test Motion Tape for muscle engagement monitoring in different body areas, among diverse subjects, and with larger subject pools.