Knee joint rehabilitation auxiliary training system

Abstract

The application relates to the field of healthcare information, and discloses a knee joint rehabilitation auxiliary training system, which comprises a data acquisition unit, a central processing unit, a display unit and the like. The central processing unit is used for acquiring multi-modal data, including plantar pressure, a flexion angle of a knee joint and activation data of at least one muscle, at least part of the data being from the data acquisition unit; calculating an activation level of the muscle based on the activation data of the muscle and corresponding personal maximum voluntary contraction data, comparing the activation level of the muscle with a corresponding target threshold to obtain a first comparison result, and generating first feedback information corresponding to the muscle based on the first comparison result; calculating a dynamic load index of the knee joint by using a dynamic load index calculation model based on at least the flexion angle and the plantar pressure, comparing the dynamic load index with a safe load interval to obtain a second comparison result, and generating second feedback information based on the second comparison result; and the display unit is used for displaying the first feedback information and the second feedback information to assist a patient in achieving a training target. The safety is improved.

Description

Technical Field

[0001] This application relates to the field of healthcare information, and more particularly to a knee joint rehabilitation assistive training system. Background Technology

[0002] During knee osteoarthritis rehabilitation training, some patients reduce their activity due to pain and fear, leading to quadriceps atrophy and creating a vicious cycle. Knee joint rehabilitation training systems are intelligent rehabilitation medical systems that can assist patients in knee joint rehabilitation training, helping them to restore knee joint function to a certain extent.

[0003] However, knee rehabilitation training systems in related technologies focus on gait correction or load reduction through orthotics, which cannot effectively train patients' active neuromuscular control and dynamic stability, resulting in poor safety of knee rehabilitation training. Summary of the Invention

[0004] The purpose of this application is to provide at least one knee joint rehabilitation auxiliary training system that can at least solve the problem of poor safety of knee joint rehabilitation training systems and at least improve the safety of knee joint rehabilitation auxiliary training systems.

[0005] To address the aforementioned technical problems, this application provides a knee joint rehabilitation assistive training system, comprising:

[0006] Data acquisition unit;

[0007] A central processing unit is used to acquire multimodal data of a patient during knee joint rehabilitation training. The multimodal data includes plantar pressure, knee flexion angle, and activation data of at least one muscle associated with the knee joint. At least a portion of the multimodal data comes from the data acquisition unit. For each type of muscle activation data, the unit calculates the muscle activation level based on the muscle activation data and the corresponding individual maximum voluntary contraction data. The activation level of the muscle is compared with a corresponding target threshold to obtain a first comparison result. Based on the first comparison result, a first feedback information corresponding to the muscle is generated. At least based on the flexion angle and plantar pressure, a dynamic load index of the knee joint is calculated using a dynamic load index calculation model. The dynamic load index is compared with a safe load range to obtain a second comparison result. Based on the second comparison result, a second feedback information is generated.

[0008] The display unit is used to display the first feedback information and the second feedback information to assist the patient in achieving the training goal, which includes maintaining the dynamic load index within the safe load range and the activation level of the muscle reaching the corresponding target threshold.

[0009] Optionally, the central processing unit is specifically used to perform the following calculation steps using the dynamic load index calculation model:

[0010] Obtain the pre-defined and stored correspondence between the knee flexion angle and the lever arm, which is personalized for the patient.

[0011] Based on the correspondence, determine the lever arm corresponding to the flexion angle of the patient during the knee joint rehabilitation training;

[0012] Based on the plantar pressure, the ground reaction force experienced by the patient during the knee joint rehabilitation training is determined;

[0013] Muscle strength is estimated based on the activation level of the at least one muscle; wherein the at least one muscle includes the quadriceps femoris, and the muscle strength is the activation level of the quadriceps femoris; or the at least one muscle includes the quadriceps femoris and the hamstrings, and the muscle strength is the weighted sum of the activation levels of the quadriceps femoris and the hamstrings.

[0014] Calculate the product of the ground reaction force and the lever arm, and calculate the ratio of the product to the muscle strength to obtain the dynamic load index.

[0015] Optionally, the data acquisition unit includes: a first sub-acquisition unit and a second sub-acquisition unit; or, the data acquisition unit includes: a first sub-acquisition unit, a second sub-acquisition unit and a third sub-acquisition unit;

[0016] The first sub-acquisition unit is used to acquire activation data of the at least one muscle.

[0017] The second sub-acquisition unit is used to acquire the buckling angle, and the second sub-acquisition unit includes a wireless inertial measurement unit or an optical motion capture system;

[0018] The third sub-acquisition unit is used to collect the plantar pressure, and the third sub-acquisition unit includes a pressure insole or a force measuring platform.

[0019] Optionally, the first sub-acquisition unit includes a surface electromyography (EMG) sensor, and the activation data includes EMG signals; the central processing unit is specifically used to calculate the root mean square (RMS) value of the EMG signals of the muscle acquired in real time for a preset duration to obtain a real-time RMS value; and to calculate the ratio of the real-time RMS value to the corresponding individual maximum voluntary contraction data to obtain the activation level of the muscle.

[0020] Alternatively, the first sub-acquisition unit includes an ultrasound imaging device, and the activation data includes ultrasound images; the central processing unit is specifically used to identify the target morphological parameters of the muscle from the real-time acquired ultrasound images of the muscle, calculate the ratio of the target morphological parameters to the corresponding individual maximum voluntary contraction data, and obtain the activation level of the muscle.

[0021] Optionally, the central processing unit is further configured to detect whether the quadriceps muscle is delayed in activation, and when the quadriceps muscle is delayed in activation, generate third feedback information.

[0022] The first sub-acquisition unit is also used to display the third feedback information to the patient to indicate that the quadriceps activation is delayed.

[0023] In the case where the data acquisition unit includes the first sub-acquisition unit and the second sub-acquisition unit, the central processing unit is specifically used to calculate the change in the flexion angle; when the change in the flexion angle exceeds a first preset value, it is determined that the patient has stood up; the time difference between the start time of guiding the patient to perform rehabilitation training movements and the time when the patient is detected to stand up is calculated to obtain the delay time; whether the delay time exceeds a second preset value is detected, and if so, the quadriceps activation delay is detected;

[0024] In the case where the data acquisition unit includes the first sub-acquisition unit, the second sub-acquisition unit, and the third sub-acquisition unit, the central processing unit is specifically used to calculate the change in the flexion angle; calculate the change in the plantar pressure; determine that the patient has stood up when the change in the flexion angle exceeds a first preset value and / or the change in the plantar pressure exceeds a third preset value; calculate the time difference between the start time of guiding the patient to perform rehabilitation training movements and the time when the patient is detected to stand up, to obtain the delay time; detect whether the delay time exceeds a second preset value, and if so, detect the quadriceps activation delay.

[0025] Optionally, the central processing unit is specifically used for:

[0026] The flexion angle, plantar pressure, and activation data of at least one muscle are input into the dynamic load index calculation model to obtain the dynamic load index output by the dynamic load index calculation model; wherein, the dynamic load index calculation model is obtained by model training based on flexion angle samples, plantar pressure samples, activation data samples of at least one muscle, and dynamic load index samples.

[0027] Alternatively, the flexion angle and the plantar pressure can be input into the dynamic load index calculation model to obtain the dynamic load index output by the dynamic load index calculation model; wherein, the dynamic load index calculation model is obtained by model training based on flexion angle samples, plantar pressure samples and dynamic load index samples.

[0028] Optionally, the central processing unit is further configured to:

[0029] Acquire patient flexion angle samples, plantar pressure samples, and activation data samples of at least one muscle;

[0030] Obtain the correspondence between the knee flexion angle sample and the lever arm sample, which are pre-calibrated and stored for the patient.

[0031] Based on the corresponding relationship sample, determine the lever arm sample corresponding to the flexion angle sample of the patient when performing knee joint rehabilitation training;

[0032] Based on the plantar pressure samples, the ground reaction force samples experienced by the patient during knee joint rehabilitation training were determined.

[0033] For each type of muscle activation data sample, an activation level sample of the muscle is calculated based on the activation data sample of the muscle and the corresponding individual maximum voluntary contraction data sample.

[0034] Based on the activation level samples of at least one of the muscles, estimate muscle strength samples;

[0035] Calculate the product of the ground reaction force sample and the lever arm sample, and use it as the product sample. Calculate the ratio of the product sample to the muscle force sample to obtain the dynamic load index sample.

[0036] Optionally, the display unit includes a display unit for displaying the first feedback information and the second feedback information;

[0037] The first feedback information includes a safety load bar, which is used to display the dynamic load index by color. When the dynamic load index is within the safety load range, the safety load bar is a first color; when the dynamic load index exceeds the safety load range, the safety load bar is a second color, which is different from the first color.

[0038] The second feedback information includes a muscle activation bar, which is used to display the activation level of the muscle through color; when the activation level of the muscle is less than the corresponding target threshold, the muscle activation bar is a third color; when the activation level of the muscle is greater than or equal to the corresponding target threshold, the muscle activation bar is a fourth color, which is different from the third color.

[0039] Optionally, the central processing unit is further configured to select a task template from multiple task templates in a task template library via a virtual training task generator, and generate a virtual training task for the knee joint rehabilitation training based on the selected task template, wherein the task template is used to simulate the daily life movements of the knee joint; display the action guidance information of the virtual training task through a two-dimensional display interface or an immersive environment to guide the patient to perform the rehabilitation training movements of the knee joint; when the patient reaches the training goal, determine that the virtual training task has been successfully completed, and allocate electronic resources to the patient;

[0040] The action guidance information includes at least one of the following: text, icons, and animations.

[0041] Optionally, the central processing unit is also used to create a personal profile for the patient, the personal profile including the initial safe load range and the target threshold, as well as the recorded number of successful virtual training tasks;

[0042] When the ratio of the number of successful attempts to N in N consecutive virtual training tasks is greater than or equal to a first preset ratio, the difficulty of the virtual training task is increased.

[0043] When the ratio of the number of successful attempts to N in N consecutive virtual training tasks is less than or equal to a second preset ratio, the difficulty of the virtual training task is reduced.

[0044] Where N is a positive integer greater than 1;

[0045] The difficulty of the virtual training task can be increased or decreased by at least one of the following methods: adjusting the upper limit of the safe load range, adjusting the target threshold, and adjusting the parameters of the task template.

[0046] The advantages of this application compared to the prior art are:

[0047] The knee joint rehabilitation assistive training system of this application includes a data acquisition unit, a central processing unit, and a display unit. The central processing unit acquires multimodal data of the patient during knee joint rehabilitation training. This multimodal data includes plantar pressure, knee flexion angle, and activation data of at least one muscle associated with the knee joint. At least a portion of the multimodal data originates from the data acquisition unit. For each type of muscle activation data, the unit calculates the muscle activation level based on the activation data and corresponding individual maximum voluntary contraction data. The activation level is compared with a corresponding target threshold to obtain a first comparison result, and first feedback information is generated based on the first comparison result. Furthermore, based at least on the flexion angle and plantar pressure, a dynamic load index is calculated for the knee joint using a dynamic load index calculation model. The dynamic load index is compared with a safe load range to obtain a second comparison result, and second feedback information is generated based on the second comparison result. The display unit displays the first and second feedback information to the patient to assist the patient in achieving training goals, ensuring the dynamic load index remains within the safe load range and the muscle activation level reaches the corresponding target threshold. In this way, by integrating multimodal data, a real-time dual-channel feedback mechanism for muscle activation level and dynamic load index is achieved, which helps guide patients to control knee joint load and muscle force on their own, thus ensuring the safety of training. Attached Figure Description

[0048] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0049] Figure 1 This is a schematic diagram of a knee joint rehabilitation assistive training system provided in one embodiment of this application;

[0050] Figure 2 This is a schematic diagram of a display interface provided in another embodiment of this application. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0052] This application proposes a knee joint rehabilitation assistive training system. The implementation details of the knee joint rehabilitation assistive training in this embodiment are described in detail below. The following content is only for the convenience of understanding and is not necessary for implementing this solution.

[0053] This embodiment provides a knee joint rehabilitation assistive training system, such as Figure 1 As shown, it includes:

[0054] Data acquisition unit 101;

[0055] The central processing unit 102 is used to acquire multimodal data of the patient during knee joint rehabilitation training. The multimodal data includes plantar pressure, knee flexion angle, and activation data of at least one muscle associated with the knee joint. At least a portion of the multimodal data comes from the data acquisition unit. For each muscle activation data, the activation level of the muscle is calculated based on the muscle activation data and the corresponding individual maximum voluntary contraction data. The activation level of the muscle is compared with the corresponding target threshold to obtain a first comparison result. Based on the first comparison result, a first feedback information corresponding to the muscle is generated. At least based on the flexion angle and plantar pressure, the dynamic load index of the knee joint is calculated using a dynamic load index calculation model. The dynamic load index is compared with a safe load range to obtain a second comparison result. Based on the second comparison result, a second feedback information is generated.

[0056] The display unit 103 is used to display first feedback information and second feedback information to help the patient achieve training goals, including keeping the dynamic load index within a safe load range and the activation level of muscles reaching the corresponding target threshold.

[0057] The central processing unit includes tablet computers or minicomputers, etc.

[0058] At least one muscle associated with the knee joint includes the quadriceps femoris, and may also include the hamstrings, etc.

[0059] Part of the multimodal data originates from the data acquisition unit. For example, the data acquisition unit includes a first sub-acquisition unit and a second sub-acquisition unit. The first sub-acquisition unit acquires activation data of at least one muscle. The second sub-acquisition unit acquires flexion angles. Therefore, the muscle activation data and knee flexion angles in the multimodal data originate from the data acquisition unit. The central processing unit can acquire pre-stored patient weight and determine plantar pressure based on the patient's weight.

[0060] Alternatively, all multimodal data may originate from a data acquisition unit. For example, the data acquisition unit may include a first sub-acquisition unit, a second sub-acquisition unit, and a third sub-acquisition unit. The first sub-acquisition unit is used to acquire activation data of at least one muscle. The second sub-acquisition unit is used to acquire flexion angles. The third sub-acquisition unit is used to acquire plantar pressure. Based on this, plantar pressure, muscle activation data, and knee flexion angles in the multimodal data originate from the data acquisition unit.

[0061] For example, the second sub-acquisition unit includes a wireless inertial measurement unit or an optical motion capture system. Specific implementations of the wireless inertial measurement unit or optical motion capture system can be found in relevant technologies and will not be elaborated upon here.

[0062] The third data acquisition unit includes a pressure insole or a force-measuring platform. A pressure insole is installed in the shoe to sense plantar pressure, and this pressure is collected through the insole. Alternatively, a force-measuring platform can be directly installed on training equipment (such as steps or pedals) to collect plantar pressure.

[0063] The first sub-acquisition unit includes a surface electromyography (EMG) sensor, and correspondingly, the activation data includes EMG signals. Based on this, the central processing unit specifically calculates the root mean square (RMS) value of the EMG signals of the muscles acquired in real-time for a preset duration, obtaining the real-time RMS value; it then calculates the ratio of the real-time RMS value to the corresponding individual's maximum voluntary contraction data to obtain the muscle activation level. The preset duration can be set according to actual needs. The RMS value effectively reflects the muscle activation level (i.e., discharge intensity) over a period of time and is highly correlated with the degree of muscle exertion.

[0064] Alternatively, the first sub-acquisition unit includes an ultrasound imaging device, and correspondingly, the activation data includes ultrasound images. Based on this, the central processing unit is specifically used to identify target morphological parameters of the muscle from the real-time acquired ultrasound images of the muscle, calculate the ratio of the target morphological parameters to the corresponding individual maximum voluntary contraction data, and obtain the muscle activation level. Here, the target morphological parameters are parameters that can characterize the muscle activation level, such as muscle thickness, etc.

[0065] In practical applications, during the initial setup or evaluation phase of a knee joint rehabilitation assistive system, patients are asked to perform multiple specific knee joint movements with maximum effort (e.g., fully extending the knee against resistance), and muscle activation data is recorded. Based on this recorded muscle activation data, the individual's maximum voluntary contraction data is obtained. For electromyography (EMG) signals, the root mean square (RMS) value is calculated from the recorded EMG signals, which is then used as the individual's maximum voluntary contraction EMG RMS value, i.e., the individual's maximum voluntary contraction data. For ultrasound images, the target morphological parameters of the muscle are identified based on the recorded ultrasound images of the muscle, and these parameters are used as the individual's maximum voluntary contraction data.

[0066] By normalizing muscle activation data by incorporating patients' individual maximum voluntary contraction data, differences between patients can be eliminated, resulting in personalized activation levels.

[0067] In this scheme, the activation level of a muscle is compared with a corresponding target threshold to obtain a first comparison result. If the first comparison result shows that the activation level of the muscle is less than the corresponding target threshold, the muscle has not reached an effective activation level. If the first comparison result shows that the activation level of the muscle is greater than or equal to the corresponding target threshold, the muscle has reached an effective activation level. Based on the first comparison result, first feedback information corresponding to the muscle is generated, which is used to characterize the first comparison result. At least based on the flexion angle and plantar pressure, the dynamic load index of the knee joint is calculated using a dynamic load index calculation model. The dynamic load index is compared with a safe load range to obtain a second comparison result. If the second comparison result shows that the dynamic load index is within the safe load range, the current rehabilitation training movement is safe. If the second comparison result shows that the dynamic load index exceeds the safe load range, the current rehabilitation training movement is unsafe. Based on the second comparison result, second feedback information is generated, which is used to characterize the second comparison result. The first and second feedback information are displayed through a display unit, realizing a dual-channel feedback mechanism. This allows patients to actively adjust their rehabilitation training movements, keeping the dynamic load index within the safe load range and ensuring that the muscle activation level reaches the corresponding target threshold, thereby gradually and safely achieving the training goal.

[0068] The target thresholds for the safe load range and the muscle activation level can be set according to the patient's actual situation.

[0069] The dual-channel feedback mechanism provides patients with real-time, parallel biofeedback control logic and implementation mechanisms for two dimensions—dynamic load index and muscle activation level—during dynamic functional training. It is a real-time, demonstrable dual-channel feedback synchronized with the movement, enabling patients to make immediate adjustments and guiding them to control knee joint load and muscle exertion based on their own abilities, thus ensuring the safety of training.

[0070] Thus, this solution provides an intelligent knee joint rehabilitation assistive training system that can quantify and control the internal load of the knee joint in real time during dynamic functional training, while ensuring effective muscle activation, to help patients safely complete high-intensity, function-oriented training, thereby breaking the vicious cycle of pain-atrophy-avoidance.

[0071] The knee joint rehabilitation assistive training system of this embodiment includes a data acquisition unit, a central processing unit, and a display unit. The central processing unit acquires multimodal data of the patient during knee joint rehabilitation training. This multimodal data includes plantar pressure, knee flexion angle, and activation data of at least one muscle associated with the knee joint. At least a portion of the multimodal data originates from the data acquisition unit. For each muscle activation data point, the unit calculates the muscle activation level based on the activation data and the corresponding individual maximum voluntary contraction data. The activation level is compared with a corresponding target threshold to obtain a first comparison result, and first feedback information is generated based on the first comparison result. Furthermore, based at least on the flexion angle and plantar pressure, the system calculates the dynamic load index of the knee joint using a dynamic load index calculation model. The dynamic load index is compared with a safe load range to obtain a second comparison result, and second feedback information is generated based on the second comparison result. The display unit presents the first and second feedback information to the patient to assist in achieving the training goals, ensuring the dynamic load index remains within the safe load range and the muscle activation level reaches the corresponding target threshold. In this way, by integrating multimodal data, a real-time dual-channel feedback mechanism for muscle activation level and dynamic load index is achieved, which helps guide patients to control knee joint load and muscle force on their own, thus ensuring the safety of training.

[0072] In some embodiments, the central processing unit is specifically configured to perform the following calculation steps using a dynamic load index calculation model:

[0073] Obtain the pre-defined and stored correspondence between the knee flexion angle and the lever arm, which is personalized for each patient.

[0074] Based on the correspondence, determine the lever arm corresponding to the flexion angle of the patient during knee joint rehabilitation training;

[0075] Based on plantar pressure, determine the ground reaction force experienced by the patient during knee joint rehabilitation training;

[0076] Muscle strength is estimated based on the activation level of at least one muscle; wherein at least one muscle includes the quadriceps femoris and the muscle strength is the activation level of the quadriceps femoris; or at least one muscle includes the quadriceps femoris and the hamstrings and the muscle strength is the weighted sum of the activation levels of the quadriceps femoris and the hamstrings.

[0077] Calculate the product of the ground reaction force and the lever arm, and then calculate the ratio of this product to the muscle force to obtain the dynamic load index.

[0078] For example, based on the knee flexion angle measured by the second sub-acquisition unit, the correspondence between the flexion angle and the lever arm is individually calibrated and stored for each patient. The lever arm reflects the vertical distance of the ground reaction force vector relative to the center of rotation of the knee joint at different knee flexion angles (i.e., the torque arm of the external load on the knee joint). The personalized calibration process is as follows: During the initialization phase of the knee joint rehabilitation assistive training system, the patient is guided to complete several standardized static postures (such as standing fully, knee flexion of approximately 30°, 60°, etc.). Combining the collected center position of the plantar pressure with the center position of the knee joint obtained by the second sub-acquisition unit, a unique correspondence curve between the knee flexion angle and the lever arm is calculated and fitted using a geometric model. Based on this curve, a patient-specific lookup table is generated, resulting in a tabular representation of the correspondence between the flexion angle and the lever arm. This significantly improves the individualized accuracy of load estimation.

[0079] Specifically, the dynamic load index = (ground reaction force × lever arm) / muscle strength.

[0080] In this embodiment, instead of directly measuring the absolute force generated by the muscle (which would require complex force measurement equipment), the relative force level is estimated indirectly and in real time by measuring muscle activation data, such as electrophysiological activity (electromyographic signals) or ultrasound images, and combining this with the patient's individual maximum voluntary contraction data to normalize the muscle activation level, thus obtaining muscle strength.

[0081] The dynamic load index calculation model considers not only external loads but also the active protective capacity of muscles. Even with high loads, the knee joint may remain in a relatively safe state if muscles contract in a coordinated manner.

[0082] The following example illustrates the calculation steps of the dynamic load index using the fusion of plantar pressure, knee flexion angle, and electromyographic (EMG) signals from the quadriceps and hamstring muscles. Surface EMG sensors are attached to the skin surfaces of the patient's quadriceps and hamstring muscles, respectively. These sensors acquire the minute electrical signals generated during muscle contraction (i.e., the raw EMG signals) in real time. The acquired raw EMG signals are very weak and contain noise, requiring amplification and filtering to obtain clean EMG signals. Then, feature analysis is performed on the processed EMG signals; for example, the root mean square (RMS) value of the EMG signals for a preset duration is calculated. The RMS value effectively reflects the muscle activation level over a period of time and is highly correlated with the degree of muscle exertion. Next, individual ability normalization is performed: this is the main step in achieving personalization. In order to eliminate the differences in absolute muscle strength between different patients (for example, the electromyographic signal intensity of a strong person and a weak person is very different), a baseline value for each patient is needed. This baseline value is the individual maximum voluntary contraction data. In the initial setup or assessment phase of the knee joint rehabilitation assistive training system, patients are asked to perform multiple specific movements with maximum effort (for example, fully straightening the knee against resistance), and the root mean square values ​​of electromyography (RMS) of the quadriceps and hamstrings are obtained at this time. This serves as a marker of the patient's 100% effort level, that is, the individual maximum voluntary contraction data for the quadriceps and hamstrings are obtained respectively. In subsequent training, the real-time RMS values ​​of the quadriceps and hamstrings are divided by their respective individual maximum voluntary contraction data. Based on this, the muscle activation level = (real-time RMS value of electromyography / individual maximum voluntary contraction data) × 100%, thus obtaining the real-time, normalized activation level of the quadriceps and hamstrings. Muscle strength reflects the activation level of muscles. Since the quadriceps femoris is the primary muscle group involved in knee extension and support, its activation level can be used as the main representative, and thus, muscle strength can be represented by its activation level. The synergistic contraction of the quadriceps femoris and hamstrings, these antagonistic muscles, can also be considered. Their combined action better stabilizes the knee joint, and the weighted sum of their activation levels can be used as muscle strength. The weighting coefficients for the quadriceps femoris and hamstring activation levels can be pre-set based on biomechanical studies. A dynamic load index is obtained through ground reaction force, lever arm, and muscle strength.

[0083] The dynamic load index calculation model in this embodiment is a biomechanical model. By fusion of multimodal data to calculate the dynamic load index, the quantification and active control of the internal load of the knee joint are realized, which helps to provide a more comprehensive view of knee joint safety.

[0084] In addition to using biomechanical models to calculate dynamic load indices, machine learning models (such as neural network models or support vector machines) can also be used to achieve accurate dynamic load calculations.

[0085] In some embodiments, the central processing unit is specifically used for:

[0086] The dynamic load index calculation model is obtained by inputting flexion angle, plantar pressure and activation data of at least one muscle into the dynamic load index calculation model. The dynamic load index calculation model is obtained by training the model based on flexion angle samples, plantar pressure samples, activation data samples of at least one muscle and dynamic load index samples.

[0087] Alternatively, the flexion angle and plantar pressure can be input into the dynamic load index calculation model to obtain the dynamic load index output by the dynamic load index calculation model; wherein, the dynamic load index calculation model is obtained by training the model based on flexion angle samples, plantar pressure samples and dynamic load index samples.

[0088] Specifically, an initial neural network model is trained using plantar pressure samples, flexion angle samples, and activation data samples of at least one muscle as input, and dynamic load index samples as output, to establish an end-to-end dynamic load index calculation model. In the simplified version, the dynamic load index can be calculated primarily based on the two most direct mechanical signals: plantar pressure samples and flexion angle samples, while still achieving basic load management functions.

[0089] Among these, the dynamic load index sample can be obtained in advance. Based on this, the central processing unit is also used for:

[0090] Acquire patient flexion angle samples, plantar pressure samples, and activation data samples of at least one muscle;

[0091] Obtain the correspondence between the knee flexion angle sample and the lever arm sample, which are pre-calibrated and stored for the patient.

[0092] Based on the corresponding relationship samples, determine the lever arm samples corresponding to the flexion angle samples of patients during knee joint rehabilitation training;

[0093] Based on plantar pressure samples, samples of ground reaction forces experienced by patients during knee joint rehabilitation training were determined.

[0094] For each type of muscle activation data sample, the muscle activation level sample is calculated based on the muscle activation data sample and the corresponding individual maximum voluntary contraction data sample.

[0095] Estimate muscle strength samples based on activation level samples of at least one muscle.

[0096] Calculate the product of the ground reaction force sample and the lever arm sample, and use it as the product sample. Calculate the ratio of the product sample to the muscle force sample to obtain the dynamic load index sample.

[0097] In this case, at least one muscle includes the quadriceps femoris, and the muscle strength sample is an activation level sample of the quadriceps femoris; or at least one muscle includes the quadriceps femoris and the hamstrings, and the muscle strength sample is a weighted sum of the activation level samples of the quadriceps femoris and the activation level samples of the hamstrings.

[0098] In this embodiment, dynamic load index samples are obtained by using a biomechanical model, providing an accurate data foundation for training machine learning models.

[0099] In some embodiments, the display unit includes a display unit for displaying first feedback information and second feedback information.

[0100] The first feedback information includes a safety load bar, which is used to display the dynamic load index through color. When the dynamic load index is within the safety load range, the safety load bar is the first color; when the dynamic load index exceeds the safety load range, the safety load bar is the second color, which is different from the first color.

[0101] The second feedback information includes a muscle activation bar, which is used to display the activation level of the muscle through color. When the activation level of the muscle is less than the corresponding target threshold, the muscle activation bar is the third color; when the activation level of the muscle is greater than or equal to the corresponding target threshold, the muscle activation bar is the fourth color, which is different from the third color.

[0102] In the safe load bar, if the dynamic load index does not exceed the upper limit of the safe load range and the difference between the upper limit of the safe load range and the dynamic load index is less than or equal to a preset difference, the first color is the first sub-color used for warning, such as yellow. If the dynamic load index does not exceed the upper limit of the safe load range and the difference between the upper limit of the safe load range and the dynamic load index is greater than a preset difference, the first color is the second sub-color used to indicate safety, such as green. See also Figure 2 The display device shows a vertical safety load bar, illustrating the upper and lower limits of the safety load range. The safety load bar uses color to indicate the dynamic load index in real time: green indicates safety, yellow indicates warning, and red indicates danger. When the dynamic load index exceeds the upper limit of the safety load range, it can flash red and play an alarm sound.

[0103] See Figure 2 The display device shows a horizontal muscle activation bar, with blue indicating that the quadriceps activation level has reached the target threshold.

[0104] For example, a patient's training goal includes keeping the safe load bar green and the quadriceps muscle activation bar blue while performing knee rehabilitation exercises to ensure that the muscles effectively share the load.

[0105] Furthermore, a muscle activation bar corresponding to the hamstrings can be added, using color to indicate the activation level of the hamstrings. The display device can be a wearable device, such as a virtual reality (VR) or augmented reality (AR) headset. The display device shows the first and second feedback information through an immersive environment, thus providing a more immersive training environment.

[0106] Alternatively, the display device can be a flat-panel display device, such as a mobile phone or television. The display device displays the first and second feedback information through a two-dimensional display interface.

[0107] In this embodiment, the first feedback information and the second feedback information are displayed in a visual manner through the display unit, so that the patient can see his / her training status intuitively and adjust the rehabilitation training movements in a timely manner.

[0108] In addition to visual feedback, the display unit can also enhance auditory feedback, such as different frequencies of tone representing the dynamic load index, or it can provide tactile feedback, such as the intensity of vibration from a wearable vibration motor to indicate the magnitude of the dynamic load index, which is especially helpful for elderly patients with poor vision.

[0109] In some embodiments, the central processing unit is further configured to select a task template from a plurality of task templates in a task template library via a virtual training task generator, generate a virtual training task for knee joint rehabilitation training based on the selected task template, wherein the task template is used to simulate daily life movements of the knee joint; display the action guidance information of the virtual training task through a two-dimensional display interface or an immersive environment to guide the patient to perform knee joint rehabilitation training movements; and determine that the virtual training task has been successfully completed when the patient reaches the training goal, and allocate electronic resources to the patient.

[0110] The action guidance information includes at least one of the following: text, icons, and animations.

[0111] For example, the task template library includes multiple task templates, which are used to simulate daily life movements of the knee joint, such as sitting-to-standing transfers, going up and down stairs, walking, etc.

[0112] The text and icons in the motion guidance information can display simple instructions, such as "Prepare to stand up" or "Step onto the platform." Animations can demonstrate rehabilitation training movements (such as a virtual character completing a sit-to-stand transfer), helping patients to imitate them. During virtual training tasks, the motion guidance information can provide real-time directional arrows, progress circles, or counters to guide the timing and trajectory of movements. Furthermore, the motion guidance information is displayed on a single interface, allowing patients to adjust their movements simply by focusing on this interface, ensuring intuitive and efficient training.

[0113] For example, electronic resources can be virtual coupons or points, etc.

[0114] For example, the central processing unit integrates a dynamic load index calculation model, a virtual training task generator, and a dual-channel feedback interface. The virtual training task generator generates step-up and step-down training tasks; see [link to documentation]. Figure 2 Patients follow the on-screen instructions to take steps. A safety load bar on one side displays the dynamic load index in real time, while a muscle activation bar on the other side shows the quadriceps activation level. Patients need to simultaneously maintain the dynamic load index within the safety load range and achieve the target quadriceps activation level to successfully complete the stair climbing training task and obtain electronic resources, thereby enhancing training enjoyment and adherence.

[0115] In this embodiment, through the design of virtual training tasks, various simple and complex daily life actions can be used as rehabilitation training actions, directly simulating these functional activities. This allows the rehabilitation effects to be better translated into improved quality of life. Furthermore, the immersive environment provides visual guidance and feedback, and electronic resources are offered during training, enhancing engagement and improving the patient experience.

[0116] In an exemplary embodiment, the central processing unit is also configured to create a personal profile for the patient, the personal profile including an initial safe load range and a target threshold, as well as the number of successful virtual training tasks recorded.

[0117] When the ratio of the number of successful attempts to N in N consecutive virtual training tasks is greater than or equal to a first preset ratio, the difficulty of the virtual training task is increased.

[0118] When the ratio of the number of successful attempts to N in N consecutive virtual training tasks is less than or equal to a second preset ratio, the difficulty of the virtual training task is reduced.

[0119] Where N is a positive integer greater than 1;

[0120] The difficulty of virtual training tasks can be increased or decreased in at least one of the following ways: adjusting the upper limit of the safe load range, adjusting the target threshold, and adjusting the parameters of the task template.

[0121] N, the first preset ratio and the second preset ratio can be set according to the actual situation.

[0122] The parameters of a task template include at least one of the following: action range, action speed, number of repetitions, and task complexity, etc.

[0123] Taking climbing stairs as an example, the height of the virtual stairs can be adjusted to regulate the range of motion. The height of the virtual stairs gradually increases from low to high (e.g., 10cm to 30cm), with the difficulty gradually increasing. This allows patients to be guided to climb virtual stairs of different heights. Movement speed controls the rhythm of the virtual training task (e.g., slow, medium, fast). This allows adjustment of the speed at which patients climb stairs. Task complexity can be adjusted by changing the number of obstacles or the direction of movement. For example, multi-level stairs or irregular ground can increase task complexity.

[0124] This embodiment provides a personalized and adaptive progression mechanism that can recommend or generate virtual training tasks suitable for the current difficulty based on the patient's historical performance data, such as the number of recorded successes. For example, if the patient succeeds consecutively, the difficulty of the virtual training task is automatically increased. This data-driven progression mechanism ensures the continuity and effectiveness of rehabilitation training. If the failure rate of the virtual training task is high, the virtual training task is simplified, thereby automatically triggering a reduction in difficulty. This ensures personalized and safe training and avoids excessive workload.

[0125] For example, when the ratio of the number of successful attempts to 3 in three consecutive virtual training tasks is less than or equal to a second preset ratio, the difficulty of the virtual training task is reduced. Specific methods include:

[0126] Lower the upper limit of the safe load range, for example, by reducing the upper limit of the safe load range by 5%, to reduce the load requirements on the knee joint.

[0127] Lowering the target threshold for muscle activation levels, for example, reducing the target threshold for quadriceps activation levels from 100% to 80%, makes it easier for patients to achieve the activation requirements.

[0128] The virtual training task can be simplified by adjusting the parameters of the task template, such as reducing the height of the virtual steps, slowing down the movement speed, or reducing the number of repetitions.

[0129] If the ratio of the number of successful attempts to 3 is greater than or equal to the first preset ratio, the difficulty of the virtual training task is increased. Specific methods include increasing the upper limit of the safe load range by 5%, or increasing the height of the virtual step, etc.

[0130] In some embodiments, the central processing unit is further configured to detect whether the quadriceps muscle activation is delayed, and when the quadriceps muscle activation is delayed, generate third feedback information; the third feedback information is used to characterize the quadriceps muscle activation delay.

[0131] The first sub-acquisition unit is also used to display third feedback information to the patient to indicate that the quadriceps activation is delayed.

[0132] In the case where the data acquisition unit includes a first sub-acquisition unit and a second sub-acquisition unit, the central processing unit is specifically used to calculate the change in flexion angle; when the change in flexion angle exceeds a first preset value, it is determined that the patient has stood up; the time difference between the start time of guiding the patient to perform rehabilitation training movements and the time when the patient is detected to stand up is calculated to obtain the delay time; whether the delay time exceeds a second preset value is detected, and if so, quadriceps activation delay is detected.

[0133] In the case where the data acquisition unit includes a first sub-acquisition unit, a second sub-acquisition unit, and a third sub-acquisition unit, the central processing unit is specifically used to calculate the change in flexion angle; calculate the change in plantar pressure; determine that the patient has stood up when the change in flexion angle exceeds a first preset value and / or the change in plantar pressure exceeds a third preset value; calculate the time difference between the start time of guiding the patient to perform rehabilitation training movements and the time when the patient is detected to stand up, and obtain the delay time; detect whether the delay time exceeds a second preset value, and if so, detect quadriceps activation delay.

[0134] The first, second, and third preset values ​​can be set according to the actual situation.

[0135] For example, a patient with moderate knee osteoarthritis used this knee rehabilitation assistive training system for sit-to-stand transfer training. The pressure insole and wireless inertial measurement unit detected that the dynamic load index rapidly entered the red high-risk zone the moment the patient stood up, while the electromyography sensor showed a delayed activation of the quadriceps. The system's display unit immediately alerted the patient with visual (e.g., a flashing red light on the display screen) and auditory cues. Based on the feedback, the patient learned to pre-activate the quadriceps before standing up and to stand up more slowly and in a controlled manner. After several attempts, the patient successfully maintained the dynamic load index within the green safe zone throughout the entire process, and the timing of quadriceps activation significantly improved.

[0136] If the electromyographic signal appears only after the moment of standing up (e.g., a delay of more than 100-200 milliseconds), the system identifies it as activation delay. This typically manifests as the quadriceps failing to activate beforehand to protect the knee joint.

[0137] In this embodiment, by detecting and providing feedback on the activation delay of the quadriceps, the activation timing of the quadriceps can be effectively improved, thereby enhancing the training effect.

[0138] Compared with existing technologies, the knee joint rehabilitation assistive training system of this application has the following significant advantages:

[0139] This technology enables precise quantification and active management of knee joint load: For the first time, it integrates multimodal data such as plantar pressure, knee flexion angle, and muscle activation data, and calculates a dynamic load index using a biomechanical model, achieving real-time, quantitative assessment of the internal load on the knee joint. This elevates rehabilitation training from a traditional, vague sensory level to a precise, data-driven level, resolving the core contradiction of balancing muscle strengthening and joint load reduction.

[0140] It provides an intuitive and efficient dual-channel biofeedback mechanism: through parallel real-time feedback of dynamic load index and muscle activation level, it guides patients to control knee joint load and ensure effective muscle exertion. This dual-channel design not only ensures the safety of training and effectively prevents secondary injuries during training, but also ensures the effectiveness of training, helping patients establish correct neuromuscular control patterns and breaking the vicious cycle of pain-atrophy-avoidance at its root.

[0141] The system enhances training safety and patient compliance: it sets dynamic safe load ranges, and if the dynamic load index exceeds these ranges, the system immediately issues warnings through visual, auditory, and other means, guiding the patient to adjust their movements. This immediate safety protection greatly reduces patients' (especially the elderly) fear of pain and injury, strengthens their confidence in training, and thus improves long-term training compliance.

[0142] This achieves truly personalized and adaptive rehabilitation: the system is not fixed but automatically adjusts the training difficulty (such as the upper limit of the safe load range and the target threshold for muscle activation levels) based on the patient's real-time performance and historical data. This allows the training program to closely match the dynamic changes in the patient's abilities, providing a tailored rehabilitation process for each patient, avoiding undertraining or overtraining, and maximizing rehabilitation efficiency.

[0143] It has a strong functional orientation: the training movements of this system (such as sitting-standing, going up and down stairs) highly simulate daily life, and its training effect can be directly converted into improvement in daily life ability (such as going up and down stairs, walking).

[0144] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A knee joint rehabilitation assistive training system, characterized in that, include: Data acquisition unit; A central processing unit is used to acquire multimodal data of a patient during knee joint rehabilitation training. The multimodal data includes plantar pressure, knee flexion angle, and activation data of at least one muscle associated with the knee joint. At least a portion of the multimodal data comes from the data acquisition unit. For each type of muscle activation data, the activation level of the muscle is calculated based on the muscle activation data and the corresponding individual maximum voluntary contraction data. The activation level of the muscle is compared with a corresponding target threshold to obtain a first comparison result. Based on the first comparison result, first feedback information corresponding to the muscle is generated. Based at least on the flexion angle and the plantar pressure, a dynamic load index of the knee joint is calculated using a dynamic load index calculation model. The dynamic load index is compared with a safe load range to obtain a second comparison result. Second feedback information is generated based on the second comparison result. Specifically, by combining the patient's individual maximum voluntary contraction data to normalize the muscle activation data, differences between different patients can be eliminated, resulting in personalized activation levels. The individual maximum voluntary contraction data is obtained based on muscle activation data recorded when the patient performs multiple specific knee joint movements with maximum effort. The display unit is used to display the first feedback information and the second feedback information to assist the patient in achieving the training goal, the training goal including the dynamic load index being maintained within the safe load range and the activation level of the muscle reaching the corresponding target threshold; The central processing unit is specifically used to perform the following calculation steps using the dynamic load index calculation model: Obtain the pre-defined and stored correspondence between the knee flexion angle and the lever arm, which is personalized for the patient. Based on the correspondence, determine the lever arm corresponding to the flexion angle of the patient during the knee joint rehabilitation training; Based on the plantar pressure, the ground reaction force experienced by the patient during the knee joint rehabilitation training is determined; Muscle strength is estimated based on the activation level of the at least one muscle; wherein the at least one muscle includes the quadriceps femoris, and the muscle strength is the activation level of the quadriceps femoris; or the at least one muscle includes the quadriceps femoris and the hamstrings, and the muscle strength is the weighted sum of the activation levels of the quadriceps femoris and the hamstrings. Calculate the product of the ground reaction force and the lever arm, and calculate the ratio of the product to the muscle strength to obtain the dynamic load index.

2. The knee joint rehabilitation assistive training system according to claim 1, characterized in that, The data acquisition unit includes: a first sub-acquisition unit and a second sub-acquisition unit; or, the data acquisition unit includes: a first sub-acquisition unit, a second sub-acquisition unit and a third sub-acquisition unit; The first sub-acquisition unit is used to acquire activation data of the at least one muscle. The second sub-acquisition unit is used to acquire the buckling angle, and the second sub-acquisition unit includes a wireless inertial measurement unit or an optical motion capture system; The third sub-acquisition unit is used to collect the plantar pressure, and the third sub-acquisition unit includes a pressure insole or a force measuring platform.

3. The knee joint rehabilitation assistive training system according to claim 2, characterized in that, The first sub-acquisition unit includes a surface electromyography (EMG) sensor, and the activation data includes EMG signals; the central processing unit is specifically used to calculate the root mean square (RMS) value of the EMG signals of the muscle acquired in real time for a preset duration, and obtain the real-time EMG RMS value. The activation level of the muscle is obtained by calculating the ratio of the real-time root mean square value of electromyography to the corresponding individual maximum voluntary contraction data. Alternatively, the first sub-acquisition unit includes an ultrasound imaging device, and the activation data includes ultrasound images; the central processing unit is specifically used to identify the target morphological parameters of the muscle from the real-time acquired ultrasound images of the muscle, calculate the ratio of the target morphological parameters to the corresponding individual maximum voluntary contraction data, and obtain the activation level of the muscle.

4. The knee joint rehabilitation assistive training system according to claim 2, characterized in that, The central processing unit is also used to detect whether the quadriceps muscle is activated late, and when the quadriceps muscle is activated late, it generates third feedback information. The first sub-acquisition unit is also used to display the third feedback information to the patient to indicate that the quadriceps activation is delayed. In the case where the data acquisition unit includes the first sub-acquisition unit and the second sub-acquisition unit, the central processing unit is specifically used to calculate the change in the flexion angle; when the change in the flexion angle exceeds a first preset value, it is determined that the patient has stood up; the time difference between the start time of guiding the patient to perform rehabilitation training movements and the time when the patient is detected to stand up is calculated to obtain the delay time; whether the delay time exceeds a second preset value is detected, and if so, the quadriceps activation delay is detected; In the case where the data acquisition unit includes the first sub-acquisition unit, the second sub-acquisition unit, and the third sub-acquisition unit, the central processing unit is specifically used to calculate the change in the flexion angle; calculate the change in the plantar pressure; and determine that the patient has stood up when the change in the flexion angle exceeds a first preset value and / or the change in the plantar pressure exceeds a third preset value. The time difference between the start time of guiding the patient to perform rehabilitation training movements and the time when the patient is detected standing up is calculated to obtain the delay time; If the delay time exceeds a second preset value, then the quadriceps activation delay is detected.

5. The knee joint rehabilitation assistive training system according to claim 1, characterized in that, The central processing unit is specifically used for: The flexion angle, plantar pressure, and activation data of at least one muscle are input into the dynamic load index calculation model to obtain the dynamic load index output by the dynamic load index calculation model; wherein, the dynamic load index calculation model is obtained by model training based on flexion angle samples, plantar pressure samples, activation data samples of at least one muscle, and dynamic load index samples. Alternatively, the flexion angle and the plantar pressure can be input into the dynamic load index calculation model to obtain the dynamic load index output by the dynamic load index calculation model; wherein, the dynamic load index calculation model is obtained by model training based on flexion angle samples, plantar pressure samples and dynamic load index samples.

6. The knee joint rehabilitation assistive training system according to claim 5, characterized in that, The central processing unit is also used for: Acquire patient flexion angle samples, plantar pressure samples, and activation data samples of at least one muscle; Obtain the correspondence between the knee flexion angle sample and the lever arm sample, which are pre-calibrated and stored for the patient. Based on the corresponding relationship sample, determine the lever arm sample corresponding to the flexion angle sample of the patient when performing knee joint rehabilitation training; Based on the plantar pressure samples, the ground reaction force samples experienced by the patient during knee joint rehabilitation training were determined. For each type of muscle activation data sample, an activation level sample of the muscle is calculated based on the activation data sample of the muscle and the corresponding individual maximum voluntary contraction data sample. Based on the activation level samples of at least one of the muscles, estimate muscle strength samples; Calculate the product of the ground reaction force sample and the lever arm sample, and use it as the product sample. Calculate the ratio of the product sample to the muscle force sample to obtain the dynamic load index sample.

7. The knee joint rehabilitation assistive training system according to claim 1, characterized in that, The display unit includes a display unit, which is used to display the first feedback information and the second feedback information; The first feedback information includes a safety load bar, which is used to display the dynamic load index by color. When the dynamic load index is within the safety load range, the safety load bar is a first color; when the dynamic load index exceeds the safety load range, the safety load bar is a second color, which is different from the first color. The second feedback information includes a muscle activation bar, which is used to display the activation level of the muscle through color; when the activation level of the muscle is less than the corresponding target threshold, the muscle activation bar is a third color; when the activation level of the muscle is greater than or equal to the corresponding target threshold, the muscle activation bar is a fourth color, which is different from the third color.

8. The knee joint rehabilitation assistive training system according to claim 7, characterized in that, The central processing unit is further configured to select a task template from multiple task templates in a task template library via a virtual training task generator, and generate a virtual training task for the knee joint rehabilitation training based on the selected task template, wherein the task template is used to simulate the daily life movements of the knee joint; display the action guidance information of the virtual training task through a two-dimensional display interface or an immersive environment to guide the patient to perform the rehabilitation training movements of the knee joint; when the patient reaches the training goal, determine that the virtual training task has been successfully completed, and allocate electronic resources to the patient; The action guidance information includes at least one of the following: text, icons, and animations.

9. The knee joint rehabilitation assistive training system according to claim 8, characterized in that, The central processing unit is also used to create a personal profile for the patient, the personal profile including the initial safe load range and the target threshold, as well as the recorded number of successful virtual training tasks; When the ratio of the number of successful attempts to N in N consecutive virtual training tasks is greater than or equal to a first preset ratio, the difficulty of the virtual training task is increased. When the ratio of the number of successful attempts to N in N consecutive virtual training tasks is less than or equal to a second preset ratio, the difficulty of the virtual training task is reduced. Where N is a positive integer greater than 1; The difficulty of the virtual training task can be increased or decreased by at least one of the following methods: adjusting the upper limit of the safe load range, adjusting the target threshold, and adjusting the parameters of the task template.

Images