Low back pain (LBP) is a highly prevalent and burdensome health condition, with approximately 568.4 million existing cases, 223.5 million new cases, and 63.7 million cases involving years lived with disability reported worldwide in 2019 [1]. It is anticipated that approximately 70% to 85% of adults will experience at least 1 episode of LBP during their lifetime [2,3], and once susceptible to LBP, individuals face twice the likelihood of experiencing recurring episodes [4].

The costs of diagnosing and treating LBP in the United States are substantial, collectively amounting to US $12 billion annually [5,6] and an economic impact including 149 million missed workdays per year [7]. Worldwide, the total annual costs associated with LBP are nearly US $100 billion, including lost wages and diminished productivity within businesses [8]. Given the high prevalence and burden to the individual and society, LBP is an important health condition to address clinically and in research.

Physical therapy (PT) is effective for the conservative, nonpharmacologic, and nonsurgical management of LBP. Specifically, active interventions such as exercises prescribed by physical therapists are effective for both preventing and treating LBP [9,10]. In PT, a licensed physical therapist conducts a comprehensive initial examination to identify musculoskeletal and neuromuscular impairments associated with the LBP problem by closely observing the patient’s low back posture and movement. Subsequently, the physical therapist works with the patient to develop a plan of care for in-clinic sessions and with an assigned home exercise program based on the PT evaluation and patient goals to enhance strength, stability, and mobility [2,11,12]. These interventions collectively aim to alleviate pain and mitigate disability [13,14]. Monitoring the patient’s posture and movement, along with other patient outcomes, is an important component of the PT examination, evaluation, and intervention for LBP.

Traditional methods for assessing posture and movement in PT include visual assessments by clinicians or use of low-technology tools such as goniometers and inclinometers to measure gross range of motion [15]. However, advances in sensor technology allow for more detailed objective measures and enable remote monitoring [16,17]. Remote monitoring can be useful for patient assessment in free-living environments where people engage in diverse activities at home and work [18]. Quantifying the repetitive nature of specific movement patterns, whether at home or in the workplace, can help identify posture and movement factors that may be linked to the risk of developing and perpetuating LBP [19-21].

Remote monitoring of low back posture and movement can also be used to monitor patient performance of and adherence to their prescribed home exercise program. Customized by physical therapists, these home exercise programs offer practical and cost-effective management of LBP [2,7]. Adherence to and proper execution of home exercises correlate with better pain management, function, and self-perceived progress [12,22-25]. However, people with LBP have several obstacles that hinder exercise performance at home [7,11]. Impaired proprioception in patients with LBP limits their ability to sense whether they are performing home exercises accurately [15,26,27]. Moreover, the absence of clinician oversight affects patient engagement with exercises [28,29]. Previous investigators have identified that this lack of monitoring and engagement leads to diminished exercise accuracy and adherence [25].

Remote monitoring for the assessment of low back posture and movement and home exercise adherence also has the potential to enhance the emerging practice of PT via telehealth, or telerehabilitation [30,31]. Successfully implementing telerehabilitation remains challenging, primarily due to limitations in conducting movement assessments, evaluating exercise performance, and providing corrective guidance. Each of these components can be addressed using mobile sensor technologies.

The reference standard for objective measurement of low back posture and movement is marker-based optical motion capture [32,33]. These systems offer exceptional precision and accuracy, but their use is constrained by space requirements, cost, and the expertise needed to operate them.

Several wearable and minimally invasive devices have been developed to address these limitations. In a systematic review, authors reported on various devices for measuring low back movement, which use accelerometers, electrogoniometers, gyroscopes, and strain gauges [34]. Specifically, inertial measurement units (IMUs) are commonly used portable devices for measuring lumbar spine posture and movement that use a variety of sensors, including accelerometers, gyroscopes, and magnetometers, making them well suited for capturing acceleration and orientation in real-world settings [35]. However, challenges with IMUs include their rigid structure, susceptibility to soft-tissue artifacts, misalignment, misplacement, and reduced precision during slow movements [36,37]. In addition, IMUs are not able to account for factors such as skin deformation [38] and the complex multisegmental nature of the spine [34]. Multiple IMUs are needed to evaluate spine posture and movement, which can become burdensome to the wearer [39]. Recently, flexible or fabric-based devices using piezoresistive sensors or other types of strain sensors have been used to address some of these previous limitations [37,40,41]. Table 1 shows a summary of existing sensor types for measuring low back posture and movement, characteristics that are measured, and benefits and limitations.

Given the challenges with objective clinical assessment and the limitations of previous portable sensor systems, there is a need for an accurate, low-profile, wireless, wearable device that can be comfortably used both in the clinic and in an individual’s free-living environment to assess low back posture and movement. Motion Tape (MT) is a flexible, fabric-based sensor using commercial kinesiology tape designed to be self-adhesive and disposable [42-46]. MT has been tested on the shoulder and ankle joints in human participants [46,47] and has demonstrated the capability to measure skin strain and joint angles in the shoulder and ankle when compared to IMUs and optical motion capture systems [48]. MT has the potential to be applied to the low back and used to measure posture and movement both in the clinic and in a free-living environment.

However, the complexity in using MT for a low back use case is that the lumbar spine is multisegmental and exhibits multiplanar movements with substantial variability in skin stretch when compared to the other extremities tested previously. Therefore, these sensors must be validated for a low back use case.

The purpose of this study was to (1) validate MT for measuring low back posture and movement and (2) assess user acceptability of MT. This is the first step in developing a use case for MT for measuring low back posture and movement. A device that is valid and acceptable in the laboratory could then be tested for use in the clinic and free-living environment for LBP diagnosis, treatment, and prevention and to further improve patient engagement and adherence to a home exercise program.

The primary hypothesis of this study was that strain-derived measures from the MT will be correlated with low back kinematics derived from a reference-standard optical motion capture system. The secondary hypothesis of this study was that users would find MT acceptable in terms of usefulness, ease of use, and wearability for the low back use case.

This study had a cross-sectional, observational, mixed methods (quantitative and qualitative) design (Figure 1), which was used to (1) validate MT for measuring low back posture and movement and (2) evaluate user acceptability of MT using semistructured interviews. Findings from this study will provide a basis for future sensor improvements.

A total of 10 participants were recruited from a university campus using flyers emailed to students, faculty, and staff in the kinesiology and PT programs. A sample size of 10 participants was considered adequate for a preliminary validation and acceptability study to provide a basis for improvement of the prototype device for subsequent testing in larger samples of healthy controls and people with LBP.

People were eligible to participate if they were between the ages 18 and 65 years and reported no history of LBP within the last year. People were excluded from participation if they were (1) unable to follow instructions in English; (2) unable to perform movements such as walking, sitting, and bending of the low back; and (3) unwilling to wear tight-fitting shorts and a sports bra (women) or no shirt (men). Recruitment and testing took place from January 2023 to March 2023. All data collection was conducted in the Rehabilitation Biomechanics Laboratory at San Diego State University.

This study was approved by the San Diego State University Institutional Review Board (HS-2022-0269), and each participant provided written informed consent before taking part. All participant data were coded, and participants were provided US $50 in compensation for their participation time.

MT is made by spray coating commercially available kinesiology tape with a thin film of graphene nanosheets (GNS) and ethyl cellulose (EC) in an ethyl alcohol solution 3 times [49]. To improve overall nanocomposite uniformity and electrical conductivity, a final layer of GNS and EC thin film is added through drop casting [48-50]. A flexible conductive ink is used to cover the sensor, and multistrand wires are soldered on for measurement electrodes at opposite ends of the GNS and EC sensing element [48]. MT has strain sensing due to piezoresistive properties of the integrated nanosheets in the tape [46], described in equation 1, which gives the direct relationship between measured resistance and strain. From previous research, MT has shown stable performance under cyclic strains [46,47].

A total of 10 people participated in the study (n=5, 50% male and n=5, 50% female; mean age 22.4, SD 2.1 y). Participant ages and anthropometric measurements are presented in Table 3.

Values for cross-correlation coefficients at zero phase shift between MT and motion capture low back kinematic measurements across the 6 movements for all 10 participants are presented in Figure 6. Across movement trials, 13.9% (50/360) of MTs had missing data because resistance exceeded the threshold of 10 SDs. There are two potential explanations for why sensors exceeded the resistance threshold: (1) the level of strain for the low back region exceeded the capacity of the sensor and (2) sensor resistance increased across trials due to sensor fatigue, resulting in high resistance values even at lower strains. The former was most common during flexion movements, and the latter occurred more often in trials near the end of the testing protocol, such as rotation movements.

For forward flexion movements, cross-correlations were mostly positive (green) and moderate to high (median 0.62-0.93), indicating that MT sensors were in tension and closely paralleled motion capture kinematic measures during forward flexion (Figure 6). However, there was a high rate of sensor failure for flexion movements (45%; 27/60), particularly for lower lumbar MTs.

To illustrate the association between MT strain and motion capture kinematics during forward flexion, data for sensor 1, sensor 2, and the upper lumbar angle are shown in Figure 7 for a single participant. In this example, upper lumbar MT strains are highly correlated with the upper lumbar angle (R=0.94 for sensor 1 and R=0.95 for sensor 2). These positive correlations reflect MT tension, which was consistent for all forward flexion movements.

For extension movements, many cross-correlation coefficients were negative, indicating MT compression. However, there were also several positive cross-correlations that had varying magnitudes. This resulted in many cross-correlations that were high in magnitude for individual MTs and participants but median values that were low (median –0.55 to 0.06), indicating that strain measures closely paralleled kinematic measures but were sometimes in tension and sometimes in compression.

To illustrate varied patterns of MT strain when compared to motion capture kinematics during extension, Figure 8 shows sensor 3 strain data and upper lumbar angle measures for 2 different participants who performed extension. For the first participant (left), MT measured compression (negative deflection) during the extension movement, resulting in a high negative cross-correlation value (R=–0.88). For this participant, sensor 3 also appeared to show a limit in the ability to measure maximal compression values, as evidenced by a flattening of the strain curve at peak extension. The second participant (right) showed an unexpected pattern during extension, in which MT measured tension (positive deflection) during the extension movement, resulting in a high positive cross-correlation value (R=0.77). For both participants, an increase in MT strain was also evident when the participant was returning to an upright position from the extension movement, which did not appear to align with the decrease in the kinematic measures.

For right and left lateral bending movements, cross-correlation coefficients were positive and high (median 0.87-0.94) on the side opposite the direction of the lateral bend, indicating that MT was typically in tension and closely paralleled kinematic measures during the lateral bend movements. Cross-correlation coefficients for MT sensors on the side ipsilateral to the lateral bend movement were more variable (median –0.52 to 0.79). This illustrates that some ipsilateral sensors (upper) were in compression during the trunk lateral bending movement but correlations were low to moderate, whereas other sensors (lower) were in tension and showed high positive correlations. The middle sensors on the side ipsilateral to the lateral bending movement showed participant-to-participant variability in both direction and magnitude of cross-correlations.

Figure 9 illustrates a case example of the expected MT strains for right and left lateral bending, in which the ipsilateral MT strain is negatively correlated (compression) and the contralateral MT strain is positively correlated (tension) with the motion capture upper lumbar angle. In contrast, Figure 10 illustrates a case example of a positive correlation between MT strain and motion capture kinematics on both sides of the low back during left lateral bending, suggesting that both MTs were in tension during this movement. However, during right lateral bending for the same participant, the expected MT tension on the contralateral side and compression on the ipsilateral side were observed.

For right and left rotation movements, cross-correlation coefficients were positive and high (median 0.84-0.93) for both upper MTs for both movement directions, indicating tension and strong association with motion capture kinematics. Cross-correlation coefficients for middle and lower MTs were more variable (median –0.11 to 0.83). Most middle sensors were in tension on the side ipsilateral to the rotation movement, and MT strain was highly correlated with motion capture kinematics (median 0.80-0.83 for sensors 3 and 4). However, on the side contralateral to the rotation movement, the middle sensors showed wide participant-to-participant variability in both direction and magnitude of cross-correlations. Cross-correlations between lower sensors and motion capture kinematics varied widely on the sides both ipsilateral and contralateral to the rotation movement (median –0.32 to 0.70).

Figure 11 illustrates a case example for data from middle MTs (sensors 3 and 4) for right and left rotation for a participant, in which the ipsilateral MT exhibited a positive correlation (tension) and the contralateral MT exhibited a negative correlation (compression) with the upper lumbar angle for both rotation directions. In contrast, a second case example (Figure 12) illustrates middle MTs that were both positively correlated with the upper lumbar angle during both left and right rotation, suggesting that both sensors were in tension during these movements.

MT demonstrated the ability to measure low back movement in multiple directions and in a manner comparable to that of a reference-standard motion capture system. Cross-correlations between MT strain and motion capture kinematic measures were moderate to high for most movement directions and appeared to better reflect kinematics for movement directions in which MT was in tension (Figure 6). Patterns of MT strain appeared different for different low back movements (eg, flexion, extension, lateral bending, and rotation), suggesting that MT can distinguish between different movement directions (Figure 5).

However, for several movements and sensors, there was variability in magnitude and direction of association between MT strain and motion capture kinematics. Figures 8-12 show case examples that demonstrate variability in direction (positive vs negative) of cross-correlations during extension, lateral bending, and rotation movements. Variability in direction of association, which reflects MT tension (positive) versus compression (negative), may be the result of differing movement strategies performed by each participant. As an example, Figure 10 illustrates a positive correlation between MT strain on both sides of the low back and the lumbar angle during left lateral bending (tension bilaterally) but a positive correlation only on the contralateral side (tension) and a negative correlation on the ipsilateral side (compression) during right lateral bending. These data may suggest that lateral bending movements are performed, in some cases, by lengthening the spine (Figure 10; bilateral tension with left lateral bending) and, in other cases, by compressing or pivoting at spinal segments (Figure 10; tension and compression with right lateral bending). Low back kinematics during lateral bending were not different between sides for this case example, suggesting that MT was able to capture a level of data that is different from motion capture, which could be useful for identifying new impairments in people with LBP.

A limiting factor of the existing lower back sensing technologies summarized in Table 1 is that some can only capture movement in a single plane [59]. Other existing devices that can capture multiple planes of movement often rely on more rigid sensors [38,60]. Ensuring that the device seamlessly integrates with the wearer’s natural movements and environment is a common challenge faced for wearable sensor technology [61]. MT provides a cost-effective solution for capturing kinematics in multiple planes and that has the potential for longer-term use in a free-living environment [60]. MT’s capability to stretch and conform with the skin sets it apart from other fabric-based and flexible sensors, providing more comprehensive measurement of lumbar posture and movement [37,51,62,63]. Therefore, MT holds the potential to become a valuable tool for the assessment, treatment, and monitoring of LBP.

Several key themes emerged related to the wearability, usefulness, and ease of use of MT. Concerning wearability, participants observed that MT securely adhered to their backs during the validation testing, and they anticipated that it would stay in place for approximately 2 to 3 days, with some variation due to external factors. This aligns with the typical time frame of use for commercially available kinesiology tape, estimated to last for 2 to 3 days [57,64]. The flush-with-skin fit and feel were perceived as not likely to disrupt daily activities, but participants expressed concerns about the wired design, the chest band attachment, and potential friction between clothing and the sensors. The current MT system design, with wires and a chest band attachment, may not be optimal [61]. Previous research has highlighted the widespread adoption and use of wireless technologies in various fields, particularly in the domain of health care wearable devices [58]. Therefore, a future iteration of MT that minimizes the wires and chest band attachments would be ideal to improve user perceptions of wearability.

Regarding usefulness, participants believed that MT had the potential to enhance personalized PT treatment and, importantly, serve as a helpful reminder to engage in and adhere to prescribed exercises. Devices that allow for remote monitoring of patients have the potential to broaden the scope of assessments, enhance treatment outcomes, and enable physical therapists to make informed decisions for future patients [61]. Nevertheless, certain design limitations might hinder the usefulness of this device by older, less flexible, or larger individuals. Our findings align with earlier research studies emphasizing the need for wearable sensors to be not only useful and convenient but also inclusive and accessible to a diverse population [65]. Therefore, future iterations of MT should address inclusivity and accessibility concerns to enhance user perceptions of usefulness.

The ease of use of a wearable device is closely intertwined with its usability and the user’s confidence in its correct operation [62]. Regarding ease of use, participants acknowledged that applying MT might be challenging without assistance, but they anticipated that it would be straightforward if accompanied by detailed instructions. Providing comprehensive information about the device fosters confidence and competence in its correct use, leading to reduced errors and improved user acceptability [63,66]. Failing to provide adequate use instructions could result in the incorrect use of MT, potentially adversely affecting patient outcomes and decreasing user acceptability.

Overall, participants expressed enthusiasm and curiosity regarding the innovative nature of MT and believed that it could offer more personalized and insightful treatment for LBP, particularly due to its potential for remote monitoring. However, they also highlighted certain aspects that would require attention in future iterations to enhance user acceptability.

The participants in this study were primarily university students in exercise and nutritional science programs who were young and fit, may have more knowledge of low back anatomy and PT, and may be more inclined to accept new technologies than people from other demographics. Collectively, this negatively impacts the generalizability of the findings to other clinical populations. Older adults or individuals with obesity may display different skin strains due to differing characteristics of skin and subcutaneous fat, which could impact the validity of MT measurements. In addition, patients with LBP may display limited movement or different movement characteristics, which were not tested in this study. Furthermore, MT may be less acceptable to older patients, who may have a preference against use of technology as part of PT treatment. However, starting with validity testing in healthy young participants allowed for testing of the full range of movement for MT measures, which may not be possible in other populations. The standardized verification, analytical validation, and clinical validation for biometric monitoring technologies recommends conducting analytical validation in a healthy population first, followed by validation in a clinical population [67]. Future research is needed to test MT acceptability and validity for measuring low back movement in a more diverse patient population, including people of different ages, with a variety of body types, and with LBP. Comparing results between people with and without LBP will also help differentiate movement patterns between the 2 groups.

In addition to assessing patient user acceptability, it is important to evaluate provider acceptability for use of new technologies in clinical practice. While this study did not assess provider acceptability of MT, we conducted a preliminary study to evaluate physical therapist acceptability of MT, and these findings are reported elsewhere [68]. As a first step in validating MT for a low back use case, this study was limited to a laboratory environment. Future studies including sensor improvements and development of a mobile app will allow the MT system to be used and tested for acceptability and validity in a free-living environment.

Because of their standard size, the location of MTs relative to spine anatomy may be slightly different for each person. As previously mentioned, the inferior MTs (5 and 6) were placed below the middle tapes such that the inferior MTs ideally crossed the L4-to-L5 and L5-to-S1 junctions. However, for 1 taller participant, the inferior row was not long enough to span the L5-to-S1 junction. Therefore, an additional limitation may be variable strain readings due to variability in sensor placement relative to participant anatomy. Our study team is currently investigating the impact of variability in placement on skin-strain measurements.

MT performed well when measuring mid ranges of movement that resulted in tension on the sensor but was limited in its ability to measure maximal tension and compression. Daily tasks are rarely performed at end ranges of movement; thus, the ability to distinguish movement in mid ranges may be the most critical for ecological monitoring. However, future sensor iterations will focus on increasing limits for measuring maximal tension and compression. Second, there were some instances of increases in MT strain that do not appear to correspond with motion capture kinematics (eg, Figure 8; return from extension). It is possible that this increase in strain could reflect increased muscle engagement associated with the phase of movement. Because MT measures skin strain, it may have the potential to detect changes in skin strain as a result of muscle engagement. The instances of MT strain that did not correspond with motion capture kinematics may indicate that the MT detected another physiological phenomenon, such as muscle engagement. This has been demonstrated empirically in other areas of the body, including the biceps and gastrocnemius muscles [48]. Research is currently underway to investigate the extent to which MT has the capability to capture muscle engagement in the low back.

It is also possible that increases in MT strain that do not correspond with motion capture kinematics may be due to a sensor rebound effect. Preliminary laboratory tests conducted by investigators confirmed a rebound effect when a compressed MT returns to a neutral position (Wyckoff E, unpublished data, February 2024). Due to the piezoresistive property of MT, the resistance may momentarily increase due to a delayed mechanical relaxation of the GNS and EC ink matrix, causing a temporary increase in the distance between conductive pathways. This rebound effect has also been observed in piezoresistive carbon [69]. The rebound effect could explain some of the lower correlations observed for movements that result in compression of MT. Following compression movements, there were positive strain values that did not correspond to the kinematics but, rather, may reflect this rebound effect. Additional research is needed to investigate the extent and true nature of the rebound effect and determine how this effect can be accounted for in measures of low back strain.

In this study, MT demonstrated moderate to high association with most low back motion capture kinematic measurements and can distinguish among multiple directions of movement. The median cross-correlation values were highest for lateral bending (0.87-0.94) and rotation (0.84-0.93) but varied more during forward flexion (0.62-0.93). For movements with expected positive correlations (MT tension), the highest correlations were observed in the upper MTs, 1 and 2 (0.84 and 0.80, respectively). However, several measurement limitations exist for the current version of MT, including limited ability to measure compression as demonstrated by poor to moderate median cross-correlation values for extension movements (–0.55 to 0.06). The MT also demonstrated limited capacity for measuring maximal tension associated with end ranges of certain movements (eg, flexion). User acceptability assessment indicates primarily positive feedback in the domains of perceived wearability and usefulness but more equivocal feedback related to ease of use in its current form. Future sensor developments and testing will be focused on addressing these issues.