Abstract
Objective
Remote assessment and diagnosis of functional impairment caused by osteoarthritis (OA) of the knee can achieve early intervention of patients’ functional impairment, prevent the deterioration of OA of the knee, and provide functional remote screening for patients with knee OA. This study introduced an inertial measurement unit (IMU) sensor‐based system to assess lower extremity function and perform gait analysis. Then, we compared its accuracy to gold‐standard motion capture and gait measurement systems.
Methods
Nine adults were selected to participate in a comparative study of gait assessment outcomes using an IMU sensor‐based wearable system, a gold‐standard motion capture system, and a pressure‐based gait analysis system. The subject walked on a path that incorporated all three systems. Data analysis was performed on spatiotemporal gait parameters, including velocity, cycle time, cadence, and stride length. This was followed by gait phases, including stance, swing, double stance, and single limb support phases. Data were processed using the data processing software of each system. An independent sample t‐test was conducted for inter‐group comparison to analyze the data.
Results
The spatiotemporal gait parameters of the systems demonstrated excellent consistency, and the gait phases showed high consistency. Compared to the gold‐standard pressure‐based gait analysis system (the GATERite system), the mean gait cycle time results were 1.124 s vs. 1.127 s (p = 0.404); cadence was 93.333 steps/min vs. 94.189 steps/min (p = 0.482); stance phase was 60.89% vs. 63.26% (p < 0.001); swing phase was 39.11% vs. 36.74% (p < 0.001); stride length was 1.404 m vs. 1.420 m (p = 0.743); speed was 1.093 m/s vs. 1.110 m/s (p = 0.725). Compared to the gold‐standard video‐based motion capture system, the root mean square error was 2.7° for the hip angle and 2.6° for the knee angle.
Conclusions
This IMU‐based wearable system delivered precise measuring results to evaluate patients with knee OA. This technology can also be used to guide rehabilitation exercises for patients with knee OA.
Keywords: Gait Analysis, Knee; Osteoarthritis, Remote Monitor
This study compares the spatiotemporal gait parameters of the wearable system, the gait analysis system, and the motion capture system. The IMU‐based wearable system demonstrated excellent consistency with gold‐standard motion capture and gait analysis systems in measuring spatiotemporal parameters and gait.
Introduction
Knee osteoarthritis (OA), a joint disorder marked by the deterioration of cartilage, is common, particularly in older people. 1 With an increasing elderly population, the incidence of knee OA has increased dramatically. According to previous research, one‐third of people over 60 years old in China will be affected by knee OA. 2 , 3 Furthermore, abnormal gait and restricted function are clinical manifestations of knee OA. Delayed diagnosis and intervention will lead to reduced mobility and lower quality of life. Advanced OA patients require total knee arthroplasty, placing a hefty financial strain on individuals and the healthcare system. 4
There are several limitations to current screening and monitoring technologies for knee OA. Due to China's large patient population and limited resources, most patients do not receive adequate early screening. 5 Additionally, elderly patients with limited mobility face increased difficulties traveling to the hospital for diagnosis and treatment. A previous study used a questionnaire to screen the knee functionality of adults over 40 years old in the community for the rapid diagnosis of knee OA. 6 Although questionnaires lack objective measures of knee function and gait abnormalities, early screening can help patients become aware of their condition. This can lead to earlier intervention and slow joint degeneration.
Early screening and evaluation of knee OA are essential to prevent the exacerbation of symptoms. Currently, no single screening and evaluation modality can be scaled to large populations. Medical imaging requires expensive machinery, and the interpretation of results is highly dependent on practitioners’ experience. Three‐dimensional motion capture devices and plantar pressure devices (e.g., video‐based motion capture and pressure‐based gait analysis systems) are time‐consuming, labor‐intensive, and inefficient to use. In addition, the abovementioned modalities cannot remotely measure and transmit data; patients must visit a hospital for assessment, which decreases compliance. 7 With the development of wearable sensors and artificial intelligence, remote assessment tools have been used for early screening. Wearable sensors are useful for remotely monitoring diseases in the community and home settings. Previous studies used video calls to monitor older adults’ physical activity. 8 Other studies have also used smart phones’ acceleration sensors to assess gait. 9 However, these sensors lack comprehensive measurement data and exhibit lower precision and accuracy. Technology will not only improve the screening efficiency of patients but also distribute uneven medical resources and alleviate the shortage of clinical staff. This study designed a remote assessment system based on inertial measurement unit (IMU) sensors to assess impairment of the lower extremities. An IMU is a device that integrates accelerometers, gyroscopes, and magnetometers to measure and track the acceleration, angular velocity, and orientation of objects. The objective of this study is to validate the accuracy of the IMU sensor‐based wearable system (wearable system) for gait assessment in patients with knee OA by comparing it with two gold‐standard gait analysis systems.
Materials and Methods
Materials
The materials used include the wearable system (Jiakang Zhongzhi Technology, China), a pressure‐based gait analysis system (GAITRite, CIR Systems, USA), the Vicon video‐based motion capture system (Vicon, OML, UK), and SPSS software (IBM, USA). The GAITRite electronic path comprises pressure‐activated sensors. Vicon's optical motion capture system includes eight cameras. The wearable system consists of data acquisition sensors, a mobile application, a cloud server, and a physician portal (Figure 1). Algorithms were used to calculate gait parameters and lower extremity range of motion (ROM), and the results were transmitted to a cloud server. The main purpose of the mobile app is to obtain data and analyze ROM and gait. Measurement data and patient information are saved and sent to the cloud server. The physician portal retrieves patient assessment data from a cloud server and evaluates the patient's condition.
FIGURE 1.
(A) The wearable system and included software; (B) the GAITRite system; and (C) the Vicon system. IMU, inertial measurement unit.
Methods
This study was approved by Beijing Jishuitan Hospital's ethics committee (Ethics Committee of Beijing Jishuitan Hospital with Approval Number 202107–26.). The inclusion criteria for patients were: (i) fulfilling the diagnostic criteria for knee OA as proposed by the Orthopedic Branch of the Chinese Medical Association; (ii) aged 60 years or above; (iii) being classified between grade I and grade III of knee OA based on X‐ray imaging examination; and (iv) being willing to voluntarily participate in this experiment and provide a signed informed consent form. The exclusion criteria were (i) patients with impaired cognitive function and (ii) those suffering from other musculoskeletal or neurological diseases. Nine subjects (five men and four women) aged 61–73 years (mean age: 65.6 ± 2.8 years; mean disease duration: 1.1 ± 1.6 years) were enrolled in this experiment. All patients had a K–L classification of grade II or less. We collected informed consent from all participants.
The experiment was conducted in a confined space, with participants traversing from one end to the other along a stretch of footpath measuring 8 m in length. A pressure‐based gait analysis system was installed. At the same time, the video‐based motion capture system was set up. We placed three IMU sensors on the patient's anterior mid‐thigh, anterior tibial region, and dorsum of the foot and recorded their walk (Figure 2). In a room, we placed the GAITRite system on 10 m of walkway. The Vicon system was set up in the same room. Subjects stood still for 3 s at the starting point to calibrate the GAITRite system, the Vicon system, and the wearable system. Afterward, the subject walked at a comfortable speed, and all three systems measured the subject's walking simultaneously, stopping at the end of the walkway.
FIGURE 2.
Wearing configuration of inertial measurement unit sensors
Data Analysis
Data was processed using each system's data processing software. The data collected included spatiotemporal parameters and joint mobility parameters, and representative points from the gait cycle were selected for analysis (Figure 3). Cycle time is the time required to complete a single walking cycle, which refers to the time from when one leg steps forward and the foot follows to when the heel touches the ground again. Cadence is defined as the number of times two legs alternate within a minute. Stride length is the distance between two heel strikes in meters. Speed is the distance per unit of time in meters per second. During the stance phase, the foot is in contact with the ground, expressed as a percentage of the cycle. Meanwhile, the swing phase is when the foot is off the ground, also expressed as a percentage of the cycle. Knee ROM is the angle of knee flexion during walking in degrees. Hip ROM is the angle at which the hip moves during walking in degrees. Negative values of the hip angle represent the posterior extension of the hip joint.
FIGURE 3.
Diagram demonstrating a gait cycle.
This study aims to investigate the following parameters: cycle time, cadence, stance phase, swing phase, stride length, speed, angle of the knee, and angle of the hip.
Statistical Analysis
We used IBM SPSS software (IBM Corporation, USA) to analyze the data analysis and independent sample t‐tests for inter‐group comparison. A p‐value of ≤0.05 indicates statistical significance, whereas a p‐value ≤0.01 indicates a statistically significant difference.
Results
Motion Parameter Results
Figure 4 shows the different results of the wearable system and gait analysis system for each step, including basic spatiotemporal parameters and phase parameters. Lower extremity ROM is also compared between the wearable and the motion capture systems.
FIGURE 4.
Blue color represents the wearable system, and gray color represents the GAITRite system. By comparing the data from both systems, the accuracy can be observed.
Temporal Parameters of Gait
The results indicate that the basic spatiotemporal parameters collected by the wearable system are similar to those for the gait analysis system (cycle time, 1.124 s vs. 1.127 s, p = 0.404; cadence, 93.333 step/min vs. 94.189 step/min, p = 0.482; stride length, 1.404 m vs. 1.420 m, p = 0.743; speed, 1.093 M/s vs. 1.110 M/s, p = 0.725). However, there are some differences in the results of the phase parameters between the two systems (stance phase, 60.89% vs. 63.26%, p < 0.001; swing phase, 39.11% vs. 36.74%, p < 0.001) (Table 1).
TABLE 1.
Comparison of gait parameters
| Wearable system, Mean (SD) | Gait analysis system, Mean (SD) | Interaction p value | Mean difference (95% CI) | |
|---|---|---|---|---|
| Cycle time (s) | 1.124 (0.013) | 1.127 (0.015) | 0.404 | 0.016 (0.004,0.028) |
| Cadence (steps/min) | 93.333 (2.449) | 94.189 (2.304) | 0.482 | 0.046 (0.008,0.012) |
| Stride length (m) | 1.404 (0.038) | 1.420 (0.027) | 0.743 | −0.02 (−0.13, 0.09) |
| Speed (m/s) | 1.093 (0.096) | 1.110 (0.090) | 0.725 | 0.126 (0.03,0.222) |
| Stance phase (% cycle time) | 60.89 (1.100) | 63.26 (0.724) | <0.001 | 3.43 (2.56, 4.3) |
| Swing phase (% cycle time) | 39.11 (1.100) | 36.74 (0.724) | <0.001 | 2.95 (2.66,3.24) |
Abbreviations: CI, confidence interval; SD, standard deviation.
Range of Motion
Wearable systems can accurately measure joint ROM. Using the motion capture system as a reference, the root mean square error (RMSE) is 2.7° for the hip and 2.6° for the knee (Table 2).
TABLE 2.
Comparison of hip and knee range of motion during walking
| Wearable system hip ROM | Motion capture system hip ROM | Wearable system knee ROM | Motion capture system knee ROM | |
|---|---|---|---|---|
| Max (°) | 26.9 (3.3) | 28.1 (8.9) | 52.9 (5.1) | 53.8 (12.6) |
| Min (°) | −14.6 (3.4) | −15.9 (9.4) | 1.1 (2.8) | 1.3 (6.6) |
| RMSE (°) | 2.7 (1.3) | 2.6 (1.2) | ||
Abbreviations: RMSE, root mean square error; ROM, range of motion.
Discussion
This study compared the spatiotemporal gait parameters between a wearable system, a gait analysis system, and a motion capture system. The experimental results demonstrated a high degree of consistency between the wearable system and the gait analysis system in speed, cycle time, cadence, and stride length. It confirmed the accuracy of the wearable system for monitoring gait. However, the measurements of the gait phase, including the percentage of time spent in the swing phase and stance phase, exhibited slight differences.
Precision of Measurement
Gait phase variations are due to the slightly different initial heel strike and final toe off measurements. In the gait analysis and the motion capture systems, precision is affected by spatiotemporal resolutions. Therefore, different systems exhibited slightly different measurements. Previous studies have shown that the principles of measuring spatiotemporal parameters are consistent across different systems by detecting the timing of two consecutive heel strikes. We have shown that the IMU sensor‐based wearable system is also consistent in measuring gait parameters.
Furthermore, the wearable system and motion capture system showed excellent consistency for lower extremity ROM. This demonstrated that the wearable system could accurately monitor changes in hip and knee angles during walking. 10 Previous studies have shown that the IMU can accurately evaluate knee ROM with a low RMSE of 3.0 to 4.8°. 11 This study demonstrated that the wearable system had a low RMSE of 2.6° during walking. We have shown that the IMU‐based system is accurate and automated. In the future, it will improve the efficiency of the clinical workflow and enable the remote evaluation of lower extremity functions. 12 , 13 , 14
Clinical Implications
The wearable system is not limited by environmental conditions, and it provides a reliable and convenient way to evaluate knee function. It can detect problems such as limitations of flexion–extension and gait changes due to early OA. The wearable system is consistent with the results of gold‐standard measurement, and it is more convenient than previous technologies. For example, the subject does not need to wear markers, and the subject is not limited to a physical walkway. This system is also useful for assessing postoperative rehabilitation progress following total knee arthroplasty.
Advantages and Disadvantages
As previously mentioned, the wearable system enables accurate assessment of knee function compared to the gold‐standard video‐based motion capture and plantar pressure systems. First, the wearable system is cost‐effective compared to traditional large‐scale motion capture and plantar pressure systems. Second, the wearable system is easy to operate, it can be turned on in a few simple steps, and it is suitable for the elderly population. The subject simply wears the sensor on the limb, turns on the power, and connects to the software via Bluetooth to measure the joint angle and gait pattern. The system's measurement data is analyzed using an artificial intelligence algorithm to generate the assessment results. Finally, this system does not have environmental restrictions; it can be used outdoors and in poor lighting. In comparison, the motion capture and pressure‐based gait analysis systems’ operational steps are more tedious and require the placement of large equipment. Thus, wearable systems are advantageous in clinical settings.
However, the wearable system has several limitations. This study used an inertial measurement unit for joint angle and gait measurements. The gyroscope may produce drift errors during long periods of use. 15 An algorithm was added to the analysis software to reduce the effects of drift errors. The wearable system may also be affected by the incorrect placement of sensors. To account for this error, we provided a tutorial on sensor placement and suggested tightening the elastic band around the limb to prevent a large error during walking. Additionally, we utilized a calibration algorithm to prevent errors from sensor shifts.
Limitations and Strengths
There are certain limitations to this study. We only compared the angle in the right leg to the video‐based motion capture system to ensure the visibility of the target joint. 16 Future experiments could capture and compare data from both legs. Additionally, our study had a limited sample size; future research could include larger sample sizes and individuals with pathological gaits to assess gait asymmetry.
The advantage of this paper lies in the gait of knee OA being measured by multiple systems simultaneously, including the gold‐standard Vicon system for gait assessment. This greatly improves the reliability of the evaluation results. Furthermore, the simultaneous measurement of multiple systems greatly reduces the measurement error.
Conclusion
The IMU‐based wearable system demonstrated excellent consistency with gold‐standard motion capture and gait analysis systems in measuring spatiotemporal parameters and gait. It also accurately tracked hip and knee ROM during walking. This system presents a cost‐effective and simple method of mobility assessment without environmental restrictions. It will be a valuable tool to improve clinical efficiency and promote remote assessment and monitoring of patients with knee OA. Future directions could incorporate remote interventions for patients with functional disabilities.
Author Contributions
Haohua Zhang: Writing, Original Draft, Investigation Methodology; Yang Song: Writing Review & Editing Data Curation; Cheng Li and Yong Dou: Supervision, Validation; Dacheng Wang: Formal analysis, Data Curation; Yinyue Wu and Xiaoyi Chen: Project administration; Di Liu: Conceptualization. All authors read and approved the final manuscript.
Funding Information
This research was supported by Beijing Natural Science Foundation—Haidian Original Innovation Joint Fund Project Ethics Approval Number: Jilunke Review Letter No. 202107‐26.
Conflicts of Interest
The authors declare no conflicts of interest that could have influenced the outcome or interpretation of this research.
Acknowledgments
We thank Beijing Jiakang Zhongzhi Technology for providing the wearable devices.
References
- 1. García‐Coronado JM, Martínez‐Olvera L, Elizondo‐Omaña RE, Acosta‐Olivo CA, Vilchez‐Cavazos F, Simental‐Mendía LE, et al. Effect of collagen supplementation on osteoarthritis symptoms: a meta‐analysis of randomized placebo‐controlled trials. Int Orthop. 2019. Mar;43(3):531–8. [DOI] [PubMed] [Google Scholar]
- 2. McAlindon TE, Bannuru RR, Sullivan MC, et al. OARSI guidelines for the non‐surgical management of knee osteoarthritis. OsteoarthrCartil. 2014;22:363–88. [DOI] [PubMed] [Google Scholar]
- 3. Chen H, Zheng X, Huang H, Liu C, Wan Q, Shang S. The effects of a home‐based exercise intervention on elderly patients with knee osteoarthritis: a quasi‐experimental study. BMC MusculoskeletDisord. 2019;20(1):160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Marsh J, Joshi I, Somerville L, Vasarhelyi E, Lanting B. Health care costs after total knee arthroplasty for satisfied and dissatisfied patients. Can J Surg. 2022;65(5):E562–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Alabajos‐Cea A, Herrero‐Manley L, Suso‐Martí L, Sempere‐Rubio N, Cuenca‐Martínez F, Muñoz‐Alarcos V, et al. Screening clinical changes for the diagnosis of early knee osteoarthritis: a cross‐sectional observational study. Diagnostics (Basel). 2022;12(11):2631. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Kumar H, Pal CP, Sharma YK, Kumar S, Uppal A. Epidemiology of knee osteoarthritis using Kellgren and Lawrence scale in Indian population. J Clin Orthop Trauma. 2020;11(Suppl 1):S125–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Gandsas A, Montgomery K, McKenas D, Altrudi R, Silva Y. In‐flight continuous vital signs telemetry via the internet. Aviat Space Environ Med. 2000;71(1):68–71. [PubMed] [Google Scholar]
- 8. Kanai A, Kiyama S, Goto H, Tomita H, Tanaka A, Kunimi M, et al. Use of the sit‐to‐stand task to evaluate motor function of older adults using telemetry. BMC Geriatr. 2016;16:121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Bakhshi S, Mahoor MH, Davidson BS. Development of a body joint angle measurement system using IMU sensors. Annu Int Conf IEEE Eng Med Biol Soc. 2011;2011:6923–6. [DOI] [PubMed] [Google Scholar]
- 10. Bonnet V, Joukov V, Kulic' D, et al. Monitoring of hip and knee joint angles using a single inertial measurement unit during lower‐limb rehabilitation. IEEE Sens J. 2016;16:1557–64. [Google Scholar]
- 11. Seel T, Raisch J, Schauer T. IMU‐based joint angle measurement for gait analysis. Sensors. 2014;14:6891–909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Cooper G, Sheret I, McMillian L, Siliverdis K, Sha N, Hodgins D, et al. Inertial sensorbased knee flexion/extension angle estimation. J Biomech. 2009;42:2678–85. [DOI] [PubMed] [Google Scholar]
- 13. Favre J, Aissaoui R, Jolles BM, de Guise JA, Aminian K. Functional calibration procedure for 3D knee joint angle description using inertial sensors. J Biomech. 2009;42:2330–5. [DOI] [PubMed] [Google Scholar]
- 14. Nicolucci A, Cercone S, Chiriatti A, Muscas F, Gensini G, REMOTE Study Group . A randomized trial on home telemonitoring for the Management of Metabolic and Cardiovascular Risk in patients with type 2 diabetes. Diabetes Technol Ther. 2015;17(8):563–70. [DOI] [PubMed] [Google Scholar]
- 15. Giggins O, Sweeney KT, Caulfield B. The use of inertial sensors for the classification of rehabilitation exercises. Annu Int Conf IEEE Eng Med Biol Soc. 2014;2014:2965–8. [DOI] [PubMed] [Google Scholar]
- 16. Eltoukhy M, Kuenze C, Oh J, Jacopetti M, Wooten S, Signorile J. Microsoft Kinect can distinguish differences in over‐ground gait between older persons with and without Parkinson's disease. Med Eng Phys. 2017;44:1–7. [DOI] [PubMed] [Google Scholar]