Wearable knee joint fatigue estimating system based on curvature and pressure sensing

Abstract

BACKGROUND:

The injury of the knee joint is found to be directly related to the fatigue caused by excessive exercise. Many previous studies used wearable devices to measure the angle of knee joint during activities, but did not pay enough attention to the load of knee joint related to the fatigue degree of it.

OBJECTIVE:

A wearable embedded system was designed to sense the motion state and load of knee joint and uses the sensoring data to estimate and predict the fatigue degree of knee joint during exercise in real time, so as to prevent it from being injured.

METHODS:

An economical wearable system is designed to measure the parameters of the knee joint during exercises. Then the warning message and recommended healthy lasting time are able to be sent to users to avoid excessive exercise. 24 healthy volunteers aged 20–25 years were involved in the experiments. Two famous evaluation scales for knee joint from Department of Orthopedics (Lysholm score and IKDC score) were adopted to evaluate the protective effect.

RESULTS:

After 14 days of the first stage testing, all the participants with wearable devices reported healthy knee joint state to verify the effectiveness of the system. For the second stage, the testing group equipped with wearable warning devices did not receive obvious change in the two scales. However, Lysholm score of control group dropped by at least 7.4 and IKDC score dropped by at least 11.1 which were significantly reduced.

CONCLUSION:

Only using human perception to prevent knee joint fatigue had a risk of failure while the designed wearable system could protect the knee successfully from injuries during exercises, such as running, badminton, table tennis and basketball. Moreover, female gender and a high BMI value may be two factors that increase the risk of knee injuries during sports.

Keywords

Knee joint fatigue injury wearable device exercise

1. Introduction

Knee joint injuries often occur during exercise and cause more than four weeks’ absence from jobs or sports competitions [1, 2]. The rate of anterior cruciate ligament (ACL) lesions even reaches 2.8 and 3.2 injury per 10,000 h of exposure among college basketball and soccer players, respectively [1] which may cast a shadow on the career of an athlete or disrupt the work of an engineer [3]. People’s daily strenuous physical exercise without restriction may also cause acute or chronic injury to knee joints, resulting in physical injury and economic losses. Studies found that the knee injury from sports is closely related to knee joint fatigue caused by excessive exercise [4, 5, 6], in which the joint fatigue refers to the comprehensive fatigue state of muscles around knee joint and the premonitory damage of the passive structures (tendons and ligaments) caused by excessive exercise. Thus, if the knee joint fatigue can be estimated and predicted during sports activities, early warning can be made to reduce the intensity or duration of the exercise.

Previous researches have focused on monitoring the knee joint motion [7]. A motion capture system based on 17 inertial units and 53 opto-reflective markers is proposed to measure the knee adduction and joint contact force during daily living activities of elderly people with knee osteoarthritis [8]. This research proves the feasibility of the inertial motion detecting system but lacks the data analysis for pratical applications in daily life. Furthermore, a large number of sensors are applied in the system which increases its cost and structural complexity and may make it inconvenient to adopt the system outdoors in daily life. In order to improve the detecting accuracy and portability of the devices, a new structural sensor fixed with ordinary fabrics and conductive yarns [9] is also involved. The ground reaction forces during ski jump landing related to knees are studied using wearable sensors [10]. In that study, the plantar force insoles are combined with inertial motion units to determine the possible relationship between those forces. Moreover, there are many researches on the foot injury prevention during exercise [11, 12], but few systems are available to prevent the knee injury during sports. A motion capture system called Qualisys is introduced for kinematic detection of knees [13]. Four inertial measurement units are organized to measure the knee flexion-extension movement and the Zigbee wireless communication is adopted to transfer data to a laptop. This wearable system is able to acquire data on the moving angle of the knee joint but lacks the fatigue estimation and prediction. Another system uses e-textile and tri-axial accelerometers to measure the knee joint angle in daily life [14]. The Kalman filter is also taken for data fusion and the final value shows the flexion-extension angle of knees. In evaluation, the system prototype is able to acquire reliable data on dynamic knee movements, but it has not provided an economical embedded solution of portable devices for daily use, and lacks the testing dataset on people during sports activities. A three-dimensional knee kinematics measurement system based on vision detection is also used to acquire parameters of knee kinematics during marathon running [15]. It can detect the rotation degree and translation distance successfully during running but is unable to relate these parameters to the fatigue degree of knees. Furthermore, it requires bulky and expensive infrared and high-speed cameras and is not suitable to use as sport protection equipment in daily life. None of these studies mentioned above link the movement and stress of the knee joint with the judgement of the joint fatigue, which is key to knee protection. Estimation and prediction of knee fatigue based on the sensoring data often not taken seriously before are also meaningful for this study. Portability and low cost of the system are also issues that call for more attention. As far as we know, there is no research before linking the knee joint motion angle and load status during sports with the knee joint fatigue estimation to prevent injuries.

2. Methods

Because the fatigue state of the knee joint is affected by the exercise load and the degree of flexion and extension, it can be estimated and predicted according to the moving parameters. In this paper, exercise load and flexion angle of the knee joint are taken into consideration, and an acceleration-weighted curve fitting method is introduced to estimate and predict the percentage of knee fatigue. Then the excessive load during exercise can be avoided so as to protect the knee joint.

2.1 Mechanism and characterization of knee injury

The knee joint is composed of the lower end of the femur, the upper end of the tibia and the patella with many ligaments and muscles around it. The main ligaments are medial collateral ligament (MCL), lateral collateral ligament (LCL), anterior cruciate ligament (ACL), and posterior cruciate ligament (PCL). Although the fibula does not participate in the formation of the knee joint, its role in maintaining the stability of the knee joint cannot be ignored because of the origin of the lateral collateral ligament [16]. Stability maintaining mechanism of the knee joint includes the active stabilizing structure, the passive stabilizing structure and innervation [17]. The active stabilization structure includes the muscles and skin around the knee joint; the passive stabilization structure includes the knee ligament, the joint capsule, the meniscus, the articular cartilage, etc. Both the active and passive stabilization structure are controlled by the nervous system to jointly maintain the stability of the knee joint. The passive stabilization structure is the basis for maintaining the knee joint stability, so that the structure of each part of the knee joint moves towards a certain direction under the combined effects of the active stabilization structure and innervation.

The knee joint is an approximate ginglymus, whose range of flexion and extension in the sagittal plane is larger, while the range of motion in the other two planes is smaller [18]. The range of motion of a normal knee joint is defined as the steady range of knee, that is, the range of motion of normal adduction, abduction, internal rotation, external rotation, and hyperextension. When the range of motion exceeds a normal one, the knee joint at this time is considered to be in an unstable state called instability. Injuries to the knee include bone, cartilage, and ligament injuries. There are many ligament injuries, and bone or cartilage injuries generally combined with one or several ligament injuries [1, 2].

Common knee injuries include the ligament injury, the meniscus injury, and the patella strain [17]. Ligament injuries mostly occur in basketball, football and other ball sports, as well as gymnastics and running [19]. In these sports, the knee joint is often in a state of flexion. The calf suddenly abducts and externally rotates, or the foot and calf are fixed, and the thigh suddenly rotates and adducts. This is very likely to cause damage to the medial collateral ligament of the knee joint. Most of the collateral ligaments are damaged at the same time as the joint capsule, the cruciate ligament, or the meniscus. After the injury, the knee joint was partially swollen, and the flexion and extension function were limited. The meniscus injury is due to the inconsistent movement of the medial and lateral meniscus when the knee is suddenly extended from the flexed position. The misalignment of the thigh and calf positions will squeeze the meniscus, causing tearing or abrasive chronic injury of the meniscus. Furthermore, the patella strain is mainly due to the repeated flexion and extension of the knee joint, which causes the corresponding joint surface of the patella and femur to be abnormally misaligned, and the impact of the twisting and friction causes local tissue and cell metabolism abnormalities.

Excessive exercise and fatigue of the knee joint have been shown to significantly increase the risk of a knee injury [5, 15]. While the knee joint is gradually fatigued, the flexion angle and frequency of its movement will slightly decrease [1]. By observing the changes of these external parameters, combined with the corresponding exercise data, the progressive degree of knee fatigue can be analysed and predicted, and then an exercise plan can be adjusted in time to reduce the risk of knee injuries.

2.2 Comprehensive load for the knee joint

The knee joint fatigue is closely related to the contact force, the flexion angle and moving frequency of the joint. Thus, a comprehensive load formula is proposed to estimate the total burden of the knee joint during sports involving the three factors as shown in Eq. (1).

$\displaystyle l^{t}=\varepsilon\cdot k^{\rho\cdot\Delta t_{nr}/\Delta t_{g}}% \cdot b^{(\textit{BMI}-\delta)}\cdot(P-P_{bm})\cdot(C-C_{bm})$ (1)

where $l^{t}$ is comprehensive load at time slice $t$ ; $P$ represents the maximum plantar pressure acquired by sensors on the insoles in a small sampling time window around the time slice $t$ , which reflects the contact force of the knee joint; $P_{bm}$ denotes the benchmark value of the plantar pressure which is sampled when the user is standing still; $C$ is the maximum radian of flexion angle obtained by flex sensors in a small sampling time window around the time slice $t$ ; $C_{bm}$ is the benchmark value of flexion angle sampled when the user is standing still; The cut-off of overweight status of BMI, which represents body mass index, is usually 25 in the world and 24 in China; $\delta$ is an adjustment parameter specified by the user; $k$ denotes the base of an exponential function used to adjust the shape of the curve; $\Delta$ t ${}_{nr}$ represents the interval time of knee joint flexion sampled when the user is taking a normal walk; $\Delta$ t ${}_{g}$ is the interval time between the last knee flexion and the current one, used to describe the frequency of the knee motion; $\rho$ is an adjustment factor, specified by the user; $b$ denotes the base of another exponential function used to adjust the shape of the curve; $\varepsilon$ is also an adjustment factor, specified by the user. Before each experimental test, there should be a calibration procedure to acquire the parameter $\Delta$ t ${}_{nr}$ , $P_{bm}$ and $C_{bm}$ automatically from the user for the system during the user’s standstill and normal walking.

$\displaystyle s^{t}=\int_{0}^{t}{l^{t}}dt$ (2)

As shown in Eq. (2), the comprehensive load $l^{t}$ accumulates over time to cause total load $s^{t}$ , and the increase of flexion angle, plantar pressure and motion frequency will accelerate the accumulation of the value.

2.3 Acceleration-weighted curve fitting estimation

Because the knee fatigue is closely related to the exercise intensity of the muscles around the joint, the acceleration of the end of the tibial can be taken as a weight in the fitting estimation. A proportionalintegral (PI) controller [20] is introduced to involve the influence of the strenuous muscle. In addition to affecting curve fitting through proportional term, continuous intense muscle movements also produce cumulative effects through integral term. As shown in Eq. (3), $\Delta a(t)$ represents the acceleration redundancy at time $t$ which exceeds the threshold $a_{thr}$ , and it can be obtained in the calibration procedure during the user’s normal walking. Equation (4) shows the computational formula of the PI controller.

$\displaystyle\Delta a(t)=a(t)-a_{thr}$ (3) $\displaystyle w(t)=K_{P}\left[{\Delta a(t)+\frac{1}{T_{I}}\int_{0}^{t}{\Delta a% (\tau)d\tau}}\right]$ (4)

where $K_{P}$ represents the proportional gain, $T_{I}$ denotes the integral time and $w(t)$ is the acceleration redundancy at time $t$ which can be used as weight in the curve fitting. Because the Eq. (4) can only be used in analogue systems, the integral component should be discretized for the digital equipment. Equation (5) shows the formula for conversion from the integral term to the sum of discrete errors:

$\displaystyle\int_{0}^{t}{\Delta a(\tau)d\tau}=\sum\limits_{j=0}^{n}{\Delta a(% j)\cdot\Delta t}=T\sum\limits_{j=0}^{n}{\Delta a(j)}$ (5)

where $\Delta t=T$ represents the sampling period. In our experiments, the value of $T$ is set to 1. Equation (5) can then be rewritten in discrete form as Eq. (6):

$\displaystyle w(k)=K_{P}\left[{\Delta a(k)+\frac{T}{T_{I}}\sum\limits_{j=0}^{n% }{\Delta a(j)}}\right]$ (6)

According to Eq. (7), Eq.(6) can be further converted to the incremental form as Eq. (8). That formula can simplify the calculation and save storage space.

$\displaystyle w(k)-w(k-1)=K_{P}\left[{\Delta a(k)-\Delta a(k-1)+\frac{T}{T_{I}% }\Delta a(k)}\right]$ (7) $\displaystyle w(k)=w(k-1)+K_{P}\left[{\Delta a(k)-\Delta a(k-1)}\right]+K_{I}% \Delta a(k)$ (8)

where $K_{I}=K_{P}(T/T_{I})$ is the integral coefficient. Then each $w(k)$ should be standardized within the array $W$ which consists of all the $w(k)$ values following Eq. (9), where Max(W) represents the maximum element inside the array and Min(W) is the minimum one. Furthermore, the array $W_{\textit{norm}}$ is made up of all the $w_{\textit{norm}}(k)$ values and can be used for the next weighted curve fitting step.

$\displaystyle w_{\textit{norm}}(k)=\frac{w(k)}{\textit{Max(W)}-\textit{Min(W)}}$ (9)

After obtaining the normalized acceleration redundancy $w_{\textit{norm}}(i)$ , the weighted curve fitting is able to be run on the database to figure out the prediction of knee fatigue using Eq. (10).

$\displaystyle\hat{w}=(S^{T}W_{\textit{norm}}S)^{-1}S^{T}W_{\textit{norm}}F$ (10) $\displaystyle W_{\textit{norm}}=\left[\begin{array}[]{cccc}w_{\textit{norm}(1)% }&0&\cdots&0\\ 0&w_{\textit{norm}}(2)&&\\ \vdots&&\ddots&\\ 0&&&w_{\textit{norm}}(i)\\ \end{array}\right]$ (11)

where $w$ represents the fitting coefficient; $S$ denotes the vector of total load $s^{i}$ at each point; $F$ is the vectors of knee fatigue percentage $f^{i}$ ; $W_{\textit{norm}}$ is a weight diagonal matrix as shown in Eq. (11) containing acceleration redundancy $w_{\textit{norm}}(i)$ as its elements.

Figure 1.

The sensing and predicting procedure for knee fatigue judgment.

2.4 Fatigue judgement and predication

Due to the fact that knee fatigue will be reflected in the peak knee flexion angle [1], the angle can be used to help estimate the degree of knee fatigue. However, a key function of the system is to calculate the exercise load on the knee joint and evaluate the suitable duration for different exercises. Thus, the system needs to find the mapping from the total load $s^{t}$ to the percentage of knee fatigue $f^{t}$ at time slice $t$ for different exercises using the peak flexion angle as a tool. From the initial moment to the latest moment at which the knee joint is fatigued, the peak knee flexion reduces nearly 10 degree [1]. According to this basis, the mapping from the total load to the percentage of knee fatigue can be built in a calibration procedure. In that procedure, let the participant do sports until he feels his knee joint is tired. During that process, with every 1 degree of reduction from the peak knee flexion, the corresponding total load $s^{i}$ is calculated, where $i$ is a serial number. Then the percentage of knee fatigue $f^{i}$ is estimated by the peak knee flexion degree and the pair ( $s^{i},f^{i})$ can be viewed as a point in the mapping relationship. After acquiring these ( $s^{i},f^{i}$ ) key points, the cubic polynomial fitting can be performed to finish the whole curve of the mapping from $s^{t}$ to $f^{t}$ .

Figure 2.

(a) Illustration of the three kinds of sensors used in the sensoring device; (b) Illustration of the hardware of the wearable device; (c) Illustration of the wearable accessories embedded with sensors of the device; (d) Illustration of the shoes deployed with sensors of the device; (e) Illustration of the wearing situation of the sensoring device.

The whole sensing and predicting procedure is shown in Fig. 1 for getting the fitting coefficient $\hat{w}$ . After the calibration procedure and mapping building, $C$ , $P$ and $w(t)$ are sensed by the device to evaluate the current percentage of knee fatigue. For that purpose, a dealing time window, which can be set to 1, 2, 3 or 5 minutes under normal conditions, is needed for finding the key points in the estimation. At the end of each window, the data acquired before will be calculated to find the fitting coefficient $\hat{w}$ and the predicting curve will be refreshed. Then the healthy remaining time $t_{\textit{rem}}$ which indicates the suitable time left for the current exercise state to last without knee injury can be estimated by Eq. (12), where $f^{\textit{end}}$ represents the final fatigue state of knee decided by user; the $\hat{w}$ is the fitting coefficient; the $s^{\textit{now}}$ is the total load at this moment and $l^{\textit{now}}$ denotes the comprehensive load value at current time. The healthy remaining time $t_{\textit{rem}}$ can be used to inform the user how much time is left before a knee injury under the current exercise load.

$\displaystyle t_{\textit{rem}}=(f^{\textit{end}}/\hat{w}-s^{\textit{now}})/l^{% \textit{now}}$ (12)

2.5 Wearable sensoring device for knee joint

In order to acquire the exercise data in real time, a wearable sensoring device is implemented on the STM32F103C8T6 microcontroller platform from ST Microelectronics Corporation (Geneva, Switzerland). As shown in Fig. 2a three kinds of sensors are adopted in the device. The Flexiforce A301 is a pressure sensor from Tekscan Corporation (Boston, USA) with only 0.203 mm thickness and less than 5 ms response time, which can be deployed on the insole of the shoe to measure the plantar pressure as shown in Fig. 2c and d. A Flex Sensor 4.5" from Spectra-Symbol Corporation (Salt Lake City, USA) is used to reflect the flexion angle of the knee joint. An MPU-6050 module from InvenSense Corporation (Sunnyvale, USA) containing a 3-axis gyroscope and a 3-axis accelerometer is adopted to acquire the acceleration of the end of the tibial for weighted curve fitting. The hardware of the sensoring device is shown in Fig. 2b, in which the integrated operational amplifier OPA333 from Texas Instrument (Dallas, USA) is designed to regulate the input signal. A BC417143 Bluetooth module from Cambridge Silicon Radio (Cambridge, UK) is used to send messages out through wireless communication. Furthermore, an LP903158 3.7V lithium battery manufactured by ZONCELL Corporation (Shenzhen, China) with a storage capacity of 2500 mAh is adopted for power supply. A DC-DC converter chip TPS62203 from Texas Instrument (Dallas, USA) is also used to convert the voltage to 3.3 V for microcontroller. The Fig. 2e shows the wearing situation of the device and with the help of it the parameters of the knee joint can be measured during sports and the knee fatigue state can be estimated.

2.6 Statistical analysis

All results were presented as the mean $\pm$ S.E.M. All analyses were performed with IBM SPSS Statistics (version 21.0). The data were analyzed by one-way analysis of variance (one-way ANOVA). When assessing statistical effects was required by the experiment, one-way ANOVA with repeated measures was performed. Statistical significance was set at $p$ -value $<$ 0.05.

3. Experiment and participants

The experiments were performed involving 24 healthy volunteers aged 20–25 (12 males, 12 females; for male, age $=$ 22.9 $\pm$ 0.8 years, height 174.5 $\pm$ 3.3 cm, body mass 67.1 $\pm$ 6.8 kg; for female, age $=$ 23.2 $\pm$ 0.9 years, height 162.9 $\pm$ 2.9 cm, body mass 55.4 $\pm$ 6.9 kg) without heart disease, nervous system disease, unsound limbs, etc. The volunteers were recruited through advertising and participated in our project free of charge. The study was approved by the Ethics Committee of Fujian Medical University Union Hospital (NO: 2015KY016). Four popular exercises in daily life are adopted for the tests, including running, badminton, table tennis and basketball. Participants were excluded if they experienced any pain (acute or chronic) during sports. Flexion angle of knee, plantar pressure and the acceleration of the end of the tibial were sensed by the device and involved in the estimation after mean filtering. For the experiments, the parameters of the system were set as follows: $\varepsilon=\rho=$ 1, $k=b=e$ (the natural logarithm), $\delta=$ 24, $K_{P}=$ 3, $T_{I}=$ 6, $T=$ 1 for the $w(k)$ calculation, the sampling time window was set to 1 second and the dealing time window was set to 2 minutes. A calibration procedure was performed before testing to get parameter $\Delta t_{nr}$ , $P_{bm}$ , $C_{bm}$ , $a_{thr}$ and build the mapping from total load to knee fatigue percentage. The system showed warnings to participants when the fatigue degree reached 60 percent, provided the estimated remaining time for healthy exercise, and the participants stopped the exercise at 80 percent of the knee fatigue. There are two kinds of test modes: the test without a break and with a break. In the mode without a break, the participants do continuous exercise without a break until the end of the test, whereas in the mode with a break, the participants make a 5-minute break after 20 minutes of exercise and continue to do the exercise. The exercising time for these separate parts is added up to the total duration.

The whole procedure of experiment was divided into two stages. At the first stage, all participants were organized into the same group and equipped with fatigue predicting devices for warning to test their effectiveness. All participants were asked to take those four kinds of exercises every day in the two test modes with full recovery in between, and the whole testing procedure lasted two weeks. The participants were asked to run at 2.3 m/s. The predicting capability of the device, the gender-comparable situation, and the influence of BMI and break during exercises were studied at this stage.

At the second stage, the participants were divided into two groups, the testing group (6 males, 6 females; age $=$ 23.2 $\pm$ 0.7 years, height 168.9 $\pm$ 6.7 cm, body mass 61.0 $\pm$ 9.1 kg) and the control group (6 males, 6 females; age $=$ 22.9 $\pm$ 0.9 years, height 168.5 $\pm$ 6.3 cm, body mass 61.5 $\pm$ 8.8 kg). The participants in the testing group were equipped with the sensoring devices and stopped the exercise at 80 percent of the knee fatigue. However, the participants in the control group did not stop their exercise until they felt tired in the leg or the knee by themselves. All participants were asked to play basketball at least once every day and that procedure lasted two weeks with the same parameters of the first stage but without a break. Data were acquired through the designed hardware and the average values were collected for analysis. Moreover, the Lysholm score [21, 22] and the International Knee Documentation Committee (IKDC) score [23, 24] of the participants were judged by two independent experienced orthopedists from the Department of Orthopedics, Fujian Medical University Union Hospital. After that, the participants were given a three-month rest for the recovery. Before the next running test, these two scales of the participants were evaluated to ensure a complete recovery of their knee joints. All the setting of the running test was just the same as of the playing basketball. In addition, Lysholm score is a condition-specific score for evaluating the knee ligament injury and widely used in diagnosing various knee diseases. It is simple, non-traumatic and easy to apply. IKDC has relatively high reliability, effectiveness and sensitivity for the assessment of ligament injuries, especially anterior cruciate ligament injuries and defects. Thus, it is suitable for evaluating post-exercise knee joint injuries.

Figure 3.

(a) The healthy time of knee-joint exercise for different genders during the four exercises without a break at the first stage of test; (b) The healthy time for participants with different BMI scores for the knee joint during the four exercises without a break at the first stage of test; (c) The healthy time for the knee joint during the four exercises with and without a break at the first stage of test. ( ${}^{*}P<$ 0.05, ${}^{**}P<$ 0.01, ${}^{***}P<$ 0.001).

Figure 4.

(a) Lysholm sores of the testing and control groups before and after playing basketball and running at the second stage of test; (b) IKDC sores of the testing and control groups before and after playing basketball and running at the second stage of test; (c) Lysholm sores of the control group for different genders before and after playing basketball and running at the second stage of test; (d) IKDC sores of the control group for different genders before and after playing basketball and running at the second stage of test; (e) Lysholm sores of the control group for participants with different BMI scores before and after playing basketball and running at the second stage of test; (f) IKDC sores of the control group for participants with different BMI scores before and after playing basketball and running at the second stage of test. ( ${}^{*}P<$ 0.05, ${}^{**}P<$ 0.01, ${}^{***}P<$ 0.001).

4. Results

After 14 days of the first stage testing, none of the participants reported discomfort or pain in their knee joint. Figure 3a shows the healthy time for different genders during the four exercises without a break in the first stage of test which could keep the fatigue degree of knee less than 80 present. The male gender’s healthy time for all these four sports was significantly different from the female gender’s one. Figure 3b shows the healthy time for participants with different BMI scores. The recommended time for participants whose BMI $>$ 24 and BMI $\leqslant$ 24 was significantly different when playing badminton and table tennis but the difference was not significant when running and playing basketball. As shown in Fig. 3c, a short break during sports could significantly extend the healthy time for these four exercises and long-term high-intensity continuous exercise might increase the risk of knee fatigue.

At the second stage of the experiment, the testing group was equipped with the sensoring device for warning the fatigue of knee joint while the control group depended on their natural judgement. As shown in Fig. 4a and b, the Lysholm scores and IKDC scores of the control group significantly decreased after running and playing basketball while the ones of the testing group did not change obviously. Because playing basketball is more intense than running for the knee joint, the reduction in scales is more significant, and the changes in Fig. 4a and b conformed to this tendency. Figure 4c and d show the Lysholm sores and IKDC scores of the control group for different genders before and after playing basketball and running. The mean Lysholm scale for men decreased by 6 points after running, while the mean scale for women decreased by 8.9 points. For basketball playing, the mean corresponding reduced Lysholm scores were 7.7 and 9.5 respectively. Furthermore, the mean IKDC scale for men decreased by 10 points after running, while the mean scale for women decreased by 12.1 points. For basketball playing, the mean corresponding reduced IKDC scores were 11.3 and 13.6 respectively. Figure 4e and f show that the BMI scores could influence the fatigue degree of the knee joint during sports in the second stage of test. The mean Lysholm scale for participant with BMI $\leqslant$ 24 decreased by 7.6 points after running, while the mean scale for participant with BMI $>$ 24 decreased by 7.2 points. For basketball playing, the mean corresponding reduced Lysholm scores were 8.0 and 9.4 respectively. Moreover, the mean IKDC scale for participant with BMI $\leqslant$ 24 decreased by 10.8 points after running, while the mean scale for participant with BMI $>$ 24 decreased by 11.3 points. For basketball playing, the mean corresponding reduced IKDC scores were 12.1 and 12.9 respectively.

5. Discussion

According to the results of the experiment in Figs 3 and 4, only depending on human perception to prevent the fatigue of knee joints and avoid injuries may fail; and the designed estimation and predicting algorithm can successfully warn against knee joint injuries during exercise. The economical electronic prototype of the sensoring system is developed to acquire the data successfully in real time to make a correct decision. Moreover, a short time break was found to prolong the healthy time for knees during exercise, and this may be because a short rest could give the soft tissues and muscles of the knee joint timely nutritious supplements. On the contrary, continuous high-intensity exercise without a break will most likely cause excessive wear and injury to the knee joint. Therefore, people should avoid non-stop high-intensity exercise, even with warning systems.

To the best of the authors’ knowledge, no studies have yet taken the knee motion angle and load during exercise into consideration while estimating knee fatigue and developing algorithm to prevent knee joint injuries. Thus, this research may greatly benefit sports enthusiasts and athletes. One limitation of the current study is the limited number of volunteers recruited in the experiments due to strict exclusion criteria. However, it does not prevent us from exploring interesting tendencies in experiments based on previous literature. In the future research, experiments with a larger sample size will be conducted to further confirm these tendencies. As shown in Fig. 3a, it was obvious that women’s knees were more likely to be injured during these exercises and the tendency was more significant in sports like basketball and running in which the movements were more intense in the lower limb. Although we have not found literature on knee fatigue injuries related to gender, related literature show that women have higher peak pressure on their hallux, toes, forefoot and medial of the foot compared to men while standing and walking [25]. Furthermore, women have a higher pelvic tilt and a center of gravity anterior to men [26] and weaker gluteus medius strength compared to men [27]. Based on the principle of force interaction and the connection between the foot and the knee joint through the tibia, women’s knee joints may bear heavier loads than men during exercise, which may be one of the reasons why their knee joints were more vulnerable to fatigue in our experiments. It can also be found in Fig. 4c and d that female suffered more decrement in scale sores than male while playing basketball and running, which also implied that women may be more vulnerable to knee injuries than men in these sports. This tendency is consistent with the observation in the first stage of test.

As shown in Fig. 3b, when playing badminton and table tennis, the participants with BMI scores less than 24 were able to have more time to stay healthy for the knee joint. The differences were not significant in playing basketball and running although their mean values had the same tendency. This might be due to the large time variance caused by the deviation of people’s personal tolerance to playing basketball and running. Thus, the people with higher BMI scores may have higher risk of knee injuries when playing badminton and table tennis. This can be due to the fact that excessive body weight increases the mechanical load on knee joints while doing sports or other physical activity [18]. This tendency and reason have also been reported in previous literature [28, 29, 30]. Thus, it is really important for the overweight to protect their knee joints during daily exercises and the equipment developed in this research will be a good choice. Therefore, female gender and a high BMI score may be two factors to increase the risk of knee injuries, and they could be further verified by subsequent large sample experiments. In addition, the tendencies found from the data measured by the designed wearable system were consistent with the previous literature, which can further confirm the effectiveness and scientificity of the system.

6. Conclusion

In order to reduce the risk of knee injuries during exercise, an economical wearable system based on the STM32F103C8T6 microcontroller and multiple sensors was developed to acquire the parameters of the knee joint. Then an acceleration-weighed curve fitting algorithm was proposed to estimate the fatigue state. According to the results of the experiment, the system is helpful to prevent knee joint from injuries induced by excessive exercise, and lead the users to maintain a suitable exercise rhythm, intensity and duration.

Author contributions

JX, TZ: Conceptualization, Investigation, Supervision, Methodology, Formal analysis, Funding acquisition, Resources, Writing – original draft, Writing – review & editing, Visualization, Software. HH, JC: Investigation, Data curation. XH: Investigation, Formal analysis. PW: Investigation. All authors read and approved the final manuscript.

Availability of data and materials

Not applicable.

Consent to participate

Not applicable.

Funding

This study was supported by the National Key Technologies Research and Development Program of China (2019YFC0118203), Outstanding Youth Fund of Fujian Agriculture and Forestry University (XJQ201820).

Footnotes

Acknowledgments

The authors would like to thank the doctors at the Department of Orthopedics, Fujian Medical University Union Hospital, for their assistance in scale testing and experimental data collection.

Conflict of interest

None to report.

References

Zago

Esposito

Bertozzi

Tritto

Rampichini

Galvani

, et al. Kinematic effects of repeated turns while running. European Journal of Sport Science. 2019; 19(8): 1072-81.

Almaawi

Awwad

Bamugaddam

Alasheikh

Muaddi

Almutair

, et al. Prevalence of knee injuries among male college students in Riyadh, Kingdom of Saudi Arabia. Journal of Orthopaedic Surgery and Research. 2020; 15(1): 126.

Notarnicola

Maccagnano

Barletta

Ascatigno

Astuto

Panella

, et al. Returning to sport after anterior cruciate ligament reconstruction in amateur sports men: a retrospective study. Muscles Ligaments Tendons J. 2016; 6(4): 486-91.

Chan

LLY

Wong

AYL

Wang

. Associations between sport participation and knee symptoms: a cross-sectional study involving 3053 undergraduate students. BMC Sports Science. Medicine and Rehabilitation. 2020; 12(1): 20.

Drew

Finch

. The Relationship Between Training Load and Injury, Illness and Soreness: A Systematic and Literature Review. Sports Medicine. 2016; 46(6): 861-83.

Bercovitz

Herman

Solomonow-Avnon

Wolf

Kodesh

. Plantar pressure modifications in experienced runners following an exhaustive run. Sports Biomech. 2020; 1-11.

Dannenmaier

Kaltenbach

Kolle

Krischak

. Application of functional data analysis to explore movements: walking, running and jumping – a systematic review. Gait & Posture. 2020; 77: 182-9.

Konrath

Karatsidis

Schepers

Bellusci

De Zee

Andersen

. Estimation of the knee adduction moment and joint contact force during daily living activities using inertial motion capture. Sensors. 2019; 19(7): 1681.

Dai

Dong

Liu

. A textile sensor for long durations of human motion capture. Sensors. 2019; 19(10): 2369.

10.

Bessone

Petrat

Schwirtz

. Ground reaction forces and kinematics of ski jump landing using wearable sensors. Sensors. 2019; 19(9): 2011.

11.

Wang

Jones

Chapman

Siddle

Russell

Alazmani

, et al. A review of wearable sensor systems to monitor plantar loading in the assessment of diabetic foot ulcers. IEEE Trans Biomed Eng. 2020; 67(7): 1989-2004.

12.

Escamilla-Martínez

Gómez-Martín

Fernández-Seguín

Martínez-Nova

Pedrera-Zamorano

Sánchez-Rodríguez

. Longitudinal Analysis of Plantar Pressures with Wear of a Running Shoe. Int J Environ Res Public Health. 2020; 17(5).

13.

Jaysrichai

Suputtitada

Khovidhungij

. Mobile sensor application for kinematic detection of the knees. Annals of Rehabilitation Medicine. 2015; 39(4): 599-608.

14.

Tognetti

Lorussi

Carbonaro

De Rossi

. Wearable goniometer and accelerometer sensory fusion for knee joint angle measurement in daily life. Sensors. 2015; 15(11): 28435-55.

15.

Tian

Zheng

Huang

Zhu

, et al. The effects of marathon running on three-dimensional knee kinematics during walking and running in recreational runners. Gait & Posture. 2020; 75: 72-7.

16.

Woo

SLC

Debski

Withrow

Janaushek

. Biomechanics of knee ligaments. American Journal of Sports Medicine. 1999; 27(4): 533-43.

17.

Woo

SLC

Abramowitch

Kilger

Liang

. Biomechanics of knee ligaments: Injury, healing, and repair. Journal of Biomechanics. 2006; 39(1): 1-20.

18.

Ward

Bodiwala

Thomas

. The importance of lower limb injuries in car crashes when cost and disability are considered. Accident Analysis & Prevention. 1992; 24(6): 613-20.

19.

Chen

Hong

Wang

. Plantar stress-related injuries in male basketball players: variations on plantar loads during different maximum-effort maneuvers. BioMed Research International. 2018; 2018: 4523849.

20.

Khan

Sengottuvel

Patel

Gireesan

Baskaran

Mani

. Design and implementation of a discrete-time proportional integral (PI) controller for the temperature control of a heating pad. SLAS Technol. 2018; 23(6): 614-23.

21.

Randsborg

Tajet

Negård

Røtterud

. Manipulation under anesthesia for stiffness of the knee joint after total knee replacement. Arthroplast Today. 2020; 6(3): 470-4.

22.

Niethammer

Altmann

Holzgruber

Goller

Fischer

Müller

. Third generation autologous chondrocyte implantation is a good treatment option for athletic persons. Knee Surg Sports Traumatol Arthrosc, 2020.

23.

Roger

Bertani

Vigouroux

Mottier

Gaillard

Have

, et al. ACL reconstruction using a quadruple semitendinosus graft with cortical fixations gives suitable isokinetic and clinical outcomes after 2 years. Knee Surg Sports Traumatol Arthrosc, 2020.

24.

Ciloglu

Karaali

Gorgulu

Ekiz

. Ultrasonographic evaluation of the patellar tendon length and elasticity after open-wedge high tibial osteotomy: A comparison with radiological and clinical parameters. Knee. 2020; 27(4): 1128-34.

25.

Yamamoto

Hoshino

Kanzaki

Nukuto

Yamashita

Ibaraki

, et al. Plantar pressure sensors indicate women to have a significantly higher peak pressure on the hallux, toes, forefoot, and medial of the foot compared to men. Journal of Foot and Ankle Research. 2020; 13(1): 40.

26.

Medina McKeon

Hertel

. Sex differences and representative values for 6 lower extremity alignment measures. J Athl Train. 2009; 44(3): 249-55.

27.

Nakagawa

Moriya

Maciel

Serrão

. Trunk, pelvis, hip, and knee kinematics, hip strength, and gluteal muscle activation during a single-leg squat in males and females with and without patellofemoral pain syndrome. J Orthop Sports Phys Ther. 2012; 42(6): 491-501.

28.

Richmond

Nettel-Aguirre

Doyle-Baker

Macpherson

Emery

. Examining measures of weight as risk factors for sport-related injury in adolescents. Journal of Sports Medicine. 2016; 2016: 7316947.

29.

Kaplan

Digel

Scavo

Arellana

. Effect of obesity on injury risk in high school football players. Clin J Sport Med. 1995; 5(1): 43-7.

30.

Pihlajamäki

Parviainen

Kautiainen

Kiviranta

. Incidence and risk factors of exercise-related knee disorders in young adult men. BMC Musculoskeletal Disorders. 2017; 18(1): 340.