Remote rehabilitation evaluation method and system after knee joint replacement

Abstract

The invention discloses a remote rehabilitation evaluation method and system after knee joint replacement, and the method comprises the steps: fusing the real-time objective motion data of a patient with the latest subjective feeling data, and obtaining the fusion input data; intelligently generating a plurality of personalized rehabilitation schemes according to the fused input data; and the personalized rehabilitation scheme is converted into rehabilitation suggestions based on natural languages for the user to select. According to the method, the targeted personalized rehabilitation scheme is intelligently generated according to the real-time data of the patient, the rehabilitation process of the patient can be accurately evaluated in real time, rehabilitation abnormity can be found in time, necessary adjustment suggestions are provided, on the basis, objective motion data and latest subjective feeling data are fused to generate the rehabilitation scheme, and the rehabilitation efficiency is improved. The rehabilitation scheme is highly matched with the actual demand of the patient, the invalid recovery time is shortened, and the postoperative rehabilitation quality and efficiency of the patient are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of remote rehabilitation technology, and more specifically, to a method and system for remote rehabilitation assessment after knee replacement surgery. Background Art

[0002] Knee replacement surgery, as an important means of treating knee arthritis and other degenerative knee diseases, has been widely used in clinical practice. The rehabilitation effect of patients after surgery largely depends on the timeliness, personalization, and scientific rehabilitation plan of rehabilitation training.

[0003] Traditional rehabilitation models mainly rely on regular hospital follow-ups and manual monitoring. With the rapid development of the Internet of Things (IoT) and artificial intelligence (AI), intelligent rehabilitation devices and remote medical systems have gradually been applied to the field of rehabilitation assessment. However, these two rehabilitation assessment methods have the following deficiencies:

[0004] 1. Lack of real-time performance: The traditional model relies on regular follow-up consultations and cannot capture the dynamic changes in the patient's rehabilitation status in real time, which may delay the intervention of abnormal situations. Most existing intelligent rehabilitation devices and remote medical systems rely on cloud computing, which is prone to data transmission delays and high power consumption, making it difficult to meet the requirements of remote real-time monitoring. Moreover, once a failure occurs in the cloud or transmission, the system cannot make a real-time response.

[0005] 2. Lack of personalization: Traditional rehabilitation models usually adopt a unified template and cannot fully consider the individual differences of patients in terms of age, weight, preoperative function, etc. Existing intelligent rehabilitation devices and remote medical systems lack intelligent decision support and personalized optimization capabilities. Therefore, both of these methods are difficult to meet the personalized rehabilitation needs.

[0006] 3. Single evaluation index: Traditional rehabilitation assessment methods mostly rely on a single index (such as gait analysis or pain management) and cannot comprehensively reflect the multi-dimensional rehabilitation effects such as knee joint flexibility, gait balance, and muscle strength recovery. Existing intelligent rehabilitation devices and remote medical systems also lack an effective balance mechanism for multiple objectives, which restricts the improvement of the overall rehabilitation effect. Summary of the Invention

[0007] This application provides a method and system for remote rehabilitation assessment after knee replacement surgery, which can intelligently generate a targeted personalized rehabilitation plan based on the patient's real-time data, can evaluate the patient's rehabilitation progress in real time and accurately, detect rehabilitation abnormalities in a timely manner and provide necessary adjustment suggestions. On this basis, the objective motion data and the latest subjective feeling data are fused to generate a rehabilitation plan, ensuring that the rehabilitation plan highly matches the actual needs of the patient, shortening the ineffective recovery time, and significantly improving the quality and efficiency of the patient's postoperative rehabilitation.

[0008] This application provides a method for remote rehabilitation assessment after knee replacement surgery, including:

[0009] Fuse the real-time objective motion data of the patient with the latest subjective feeling data to obtain fused input data;

[0010] Intelligently generate multiple personalized rehabilitation plans based on the fused input data;

[0011] Convert the personalized rehabilitation plans into rehabilitation suggestions based on natural language for the user to select.

[0012] Preferably, intelligently generating multiple personalized rehabilitation plans based on the fused input data specifically includes:

[0013] Generate multiple rehabilitation paths based on the fused input data;

[0014] Construct multiple rehabilitation scenarios;

[0015] For each rehabilitation path, simulate the implementation of the rehabilitation path in each rehabilitation scenario, and calculate the comprehensive rehabilitation result of the rehabilitation path based on the rehabilitation results of the same rehabilitation path under all rehabilitation scenarios;

[0016] Select multiple rehabilitation paths with the optimal comprehensive rehabilitation results as personalized rehabilitation plans.

[0017] Preferably, the rehabilitation result of each rehabilitation path in each rehabilitation scenario is obtained based on a multi-objective function and multiple single-objective functions. The multi-objective function is the weighted sum of multiple single-objective functions, and the input data of each single-objective function is one or more objective motion data.

[0018] Preferably, use an evolutionary algorithm to generate multiple rehabilitation paths. The rehabilitation path includes multiple rehabilitation training indicators, and use the non-dominated sorting method to select personalized rehabilitation plans.

[0019] Preferably, generate personalized rehabilitation plans and / or the conversion of personalized rehabilitation plans into rehabilitation suggestions through the patient's wearable device or the edge gateway around the patient.

[0020] Preferably, the method further includes:

[0021] Obtain the activity type of the patient corresponding to the real-time objective motion data based on the real-time objective motion data.

[0022] Preferably, determine the weights of each index in the rehabilitation path according to the activity type.

[0023] Preferably, the method further includes:

[0024] Evaluate the current rehabilitation effect according to the activity type;

[0025] Dynamically adjust the rehabilitation path according to the rehabilitation effect and the activity type.

[0026] The present application also provides a remote rehabilitation evaluation system after knee joint replacement, including a fusion module, a program generation module, and a conversion module;

[0027] The fusion module is used to fuse the real-time objective motion data of the patient with the latest subjective feeling data to obtain fusion input data;

[0028] The program generation module is used to intelligently generate multiple personalized rehabilitation programs based on the fusion input data;

[0029] The conversion module is used to convert the personalized rehabilitation program into a rehabilitation suggestion based on natural language for the user to select.

[0030] Preferably, the program generation module includes a rehabilitation path generation module, a scenario construction module, a simulation module, and a screening module;

[0031] The rehabilitation path generation module is used to generate multiple rehabilitation paths based on the fusion input data;

[0032] The scenario construction module is used to construct multiple rehabilitation scenarios;

[0033] The simulation module is used to, for each rehabilitation path, simulate the implementation of the rehabilitation path in each rehabilitation scenario, and calculate the comprehensive rehabilitation result of the rehabilitation path based on the rehabilitation results of the same rehabilitation path under all rehabilitation scenarios;

[0034] The screening module is used to screen out multiple rehabilitation paths with the optimal comprehensive rehabilitation results as the personalized rehabilitation programs.

[0035] Preferably, the rehabilitation result of each rehabilitation path in each rehabilitation scenario is obtained based on a multi-objective function and multiple single-objective functions. The multi-objective function is the weighted sum of the multiple single-objective functions, and the input data of each single-objective function is one or more objective motion data.

[0036] Preferably, the rehabilitation path generation module is used to generate multiple rehabilitation paths by using an evolutionary algorithm, and each rehabilitation path includes multiple rehabilitation training indicators; and

[0037] The screening module is used to screen out the personalized rehabilitation programs by using the non-dominated sorting method.

[0038] Preferably, the program generation module and / or the conversion module are embedded in the patient's wearable device or the edge gateway around the patient.

[0039] Preferably, the remote rehabilitation evaluation system after knee joint replacement further includes an activity type acquisition module, and the activity type acquisition module is used to obtain the activity type of the patient corresponding to the real-time objective motion data based on the real-time objective motion data.

[0040] Preferably, the rehabilitation path generation module is used to determine the weights of the various indicators in the rehabilitation path according to the activity type.

[0041] Preferably, the remote rehabilitation evaluation system after knee joint replacement further includes an effect evaluation module, which is used to evaluate the current rehabilitation effect according to the activity type;

[0042] The rehabilitation path generation module is used to dynamically adjust the rehabilitation path according to the rehabilitation effect and the activity type.

[0043] Other features and advantages of the present application will become clear through the following detailed description of the exemplary embodiments of the present application with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the present application and, together with the description, serve to explain the principles of the present application.

[0045] Figure 1 is a flowchart of the remote rehabilitation evaluation method after knee joint replacement provided by the present application;

[0046] Figure 2 is a flowchart of intelligently generating multiple personalized rehabilitation plans based on the fusion of input data provided by the present application;

[0047] Figure 3 is a structural diagram of the remote rehabilitation evaluation system after knee joint replacement provided by the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] Now, various exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings. It should be noted that: unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions and values set forth in these embodiments do not limit the scope of the present application.

[0049] The following description of at least one exemplary embodiment is merely illustrative in nature and in no way serves as a limitation to the present application or its application or use.

[0050] Techniques, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, the techniques, methods, and devices should be regarded as part of the specification.

[0051] In all the examples shown and discussed here, any specific value should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0052] This application provides a remote rehabilitation evaluation method and system for post-knee replacement. It can intelligently generate personalized rehabilitation plans based on patients' real-time data, accurately evaluate the rehabilitation progress of patients in real time, promptly detect rehabilitation abnormalities and provide necessary adjustment suggestions. On this basis, it combines objective motion data with the latest subjective feeling data to generate rehabilitation plans, ensuring a high degree of fit between the rehabilitation plans and the actual needs of patients, shortening the ineffective recovery time, and significantly improving the quality and efficiency of patients' postoperative rehabilitation. Moreover, when generating the rehabilitation path, a multi-objective combination and balance mechanism between objectives is adopted, making the rehabilitation plan more comprehensive and significantly enhancing the overall rehabilitation effect. In addition, the rehabilitation plan is obtained through edge computing by wearable devices or edge gateways around the patient, reducing data transmission power consumption, extending the device usage time, and ensuring rapid and stable system response.

[0053] As Figure 1 shown, the remote rehabilitation evaluation method for post-knee replacement provided by this application includes:

[0054] S110: Integrate the real-time objective motion data of the patient with the latest subjective feeling data to obtain integrated input data.

[0055] S120: Intelligently generate multiple personalized rehabilitation plans based on the integrated input data.

[0056] S130: Convert the personalized rehabilitation plan into rehabilitation suggestions based on natural language for the user to select.

[0057] Among them, in step S110, the real-time objective motion data is obtained by real-time collection through multi-modal wearable sensing devices worn by the patient, including objective physiological indicators such as knee flexion and extension angles, gait, stride, gait cycle, and load bearing. For example, an accelerometer is used to capture the acceleration changes during knee movement to assist in analyzing the motion pattern; a gyroscope is used to measure the rotation angle of the knee, especially rotation and flexion / extension movements; a force sensor is used to monitor the load borne by the knee. The wearable sensing devices also include intelligent knee pads, watches, etc.

[0058] The patient can report subjective feelings, including pain, fatigue level, range of motion, etc., regularly or irregularly through questionnaires, mobile APPs or remote medical platforms to form subjective feeling data . Common scales for subjective feeling data include:

[0059] VAS pain assessment scale: 0 points represents no pain, and 10 points represents severe pain);

[0060] Fatigue scale: Quantify the patient's fatigue state to assist clinicians or algorithms in adjusting the training intensity in a timely manner.

[0061] The latest subjective perception data reflects the patient's latest perception of rehabilitation training. By integrating it with real-time objective motion data, a personalized rehabilitation plan is obtained, which is more targeted.

[0062] Preferably, before the integration, the real-time objective motion data is preprocessed, and then the preprocessed real-time objective motion data is integrated with the latest subjective perception data.

[0063] As an example, the data preprocessing includes denoising, time series segmentation, and normalization.

[0064] Among them, high-frequency noise is filtered out through a low-pass filter to filter out the vibration interference generated during movement:

[0065] (1)

[0066] Among them, is the original sensor data of a certain wearable sensing device, is the data volume of the original sensor data, is the filter coefficient, is the filtered data.

[0067] As an example, the sliding window technique is used to divide the continuous sensor data after denoising into fixed-duration segments of 1 to 2 seconds, ensuring that each data segment can be an independent sample for model input. Subsequently, each sensor data is normalized to unify the dimension, ensuring that the deep learning model can efficiently and accurately process multi-source data.

[0068] As an example, a splicing method based on timestamps or record indexes is used to fuse objective and subjective data. First, the real-time objective motion data and subjective perception data are time-aligned according to timestamps, and then the two are spliced together to obtain the fused input data :

[0069] (2)

[0070] It can be understood that the real-time objective motion data and the latest subjective perception data can also be fused through the "text-numerical embedding" method or by using the attention mechanism.

[0071] Preferably, after obtaining the real-time objective motion data, the activity type of the patient corresponding to the real-time objective motion data is also obtained to provide data support for obtaining rehabilitation suggestions.

[0072] As an example, a deep learning model (DCNN) is used to identify the activity type of the patient. Specifically, the deep learning model includes a convolutional layer, a pooling layer, and a fully connected layer.

[0073] The convolutional layer is used to extract spatio-temporal features (such as the periodic change of flexion and extension angles) from the knee joint movement time series data:

[0074] (3)

[0075] Among them, represents the preprocessed data; represents the convolutional kernel (i.e., the weight); represents the convolutional layer bias; * represents the convolution operation; represents the feature map output by the convolutional layer.

[0076] The pooling layer is used to perform max-pooling or average-pooling operations within each local window to reduce the data dimension and mitigate the impact of noise:

[0077] (4)

[0078] Among them, represents the convolutional feature of the local window within the pooling window; is the pooling result.

[0079] After the pooled feature map is flattened, it is input into the fully connected layer, and the activity type is classified and recognized through activation functions (such as ReLU, Sigmoid):

[0080] (5)

[0081] Among them, is the flattened vector output by the pooling layer; and are the weight matrix and bias of the fully connected layer respectively; is the activation function; is the prediction output of the deep learning network for the activity type (such as walking, standing, climbing stairs, etc.).

[0082] Preferably, the deep learning model is embedded in the patient's wearable device or the edge gateway around the patient to achieve edge computing.

[0083] Preferably, to reduce the computational complexity and power consumption on the wearable sensing device or the edge node, the system adopts quantization technology to reduce the model computational complexity and energy consumption, and improve the device response speed and battery life:

[0084] (6)

[0085] Among them, represents the floating-point value of the model weight or sensor data; is the quantization step size; is the quantization result.

[0086] This quantization process is only used for hardware deployment and running efficiency optimization, and does not affect the calculation of subsequent metrics (such as gait symmetry, joint flexion and extension).

[0087] Continue Figure 1 , in step S120, as Figure 2 shown, as an embodiment, multiple personalized rehabilitation plans are intelligently generated based on the fused input data, specifically including:

[0088] S1201: Generate multiple rehabilitation paths based on the fused input data.

[0089] S1202: Construct multiple rehabilitation scenarios.

[0090] S1203: For each rehabilitation path, simulate the implementation of the rehabilitation path in each rehabilitation scenario, and calculate the comprehensive rehabilitation result of the rehabilitation path based on the rehabilitation results of the same rehabilitation path in all rehabilitation scenarios.

[0091] In each simulation process, the objective and subjective motion metrics of the patient are evaluated to obtain the performance distribution on each metric.

[0092] In each rehabilitation scenario, each rehabilitation path is simulated multiple times, and factors such as the patient's motion state, environmental interference, and subjective fatigue fluctuation are randomly sampled in each simulation, and thus a score is obtained for each simulation . The average value of the scores obtained from multiple simulations of the same rehabilitation path in the same rehabilitation scenario is used as the rehabilitation result of the rehabilitation path in that rehabilitation scenario :

[0093] (7)

[0094] Wherein, represents the rehabilitation result of the y-th rehabilitation path, represents the score of the j-th simulation of the y-th rehabilitation path, represents the total number of simulations.

[0095] S1204: Screen out multiple rehabilitation paths with the optimal comprehensive rehabilitation results as the personalized rehabilitation plan.

[0096] As an embodiment, the rehabilitation result of each rehabilitation path in each rehabilitation scenario is obtained based on a multi-objective function and multiple single-objective functions. The multi-objective function is the weighted sum of multiple single-objective functions, and the input data of each single-objective function is one or more objective motion data.

[0097] The multi-objective function is expressed as:

[0098] (8)

[0099] Among them, represents the weight of the i-th single-objective function, represents the i-th single-objective function.

[0100] As an example, the single-objective functions include a joint flexibility recovery function, a gait balance function, a pain management function, a muscle strength recovery function, etc.

[0101] Specifically, as an example, the joint flexibility recovery function is the knee joint flexion and extension angle recovery function:

[0102] (9)

[0103] Among them, represents the current knee joint flexion and extension angle (as the input data of this function), which is obtained through a wearable sensing device; represents the target flexion and extension angle.

[0104] The higher it is, the closer the knee joint mobility is to the target recovery level.

[0105] The gait balance function reflects the symmetry degree of the left and right foot steps:

[0106] (10)

[0107] Among them, the left foot step time and the right foot step time are the input data of this function, which are obtained by statistically calculating the sole contact timestamps detected by the acceleration and force sensors. Among them, the left foot step time is the difference between the contact times of the left foot for two consecutive times, and the right foot step time is the difference between the contact times of the right foot for two consecutive times.

[0108] The smaller it is, the more symmetrical and stable the gait is. If is greater than the threshold value, it indicates that the patient has a gait deviation.

[0109] If it is necessary to be homogenized with other objectives, conversion methods such as 1 - etc. can be used.

[0110] The pain management function and the muscle strength recovery function can be quantified accordingly according to the VAS pain assessment scale and the muscle strength ratio.

[0111] As an example, the muscle strength recovery function is:

[0112] (11)

[0113] Among them, the current muscle strength value is the input data of this function, and it is subjective data feedback by the patient.

[0114] Based on the above embodiments, the multi-objective function is expressed as:

[0115] (12)

[0116] Preferably, the weight of each single-objective function can be adjusted according to the patient's real-time feedback or objective rehabilitation effect evaluation to achieve the optimal balance of multiple objectives.

[0117] On this basis, as an embodiment, in step S1201, an evolutionary algorithm (such as a multi-objective genetic algorithm) is used to generate multiple rehabilitation paths, and the rehabilitation paths include multiple rehabilitation training indicators.

[0118] Specifically, first, multiple initial rehabilitation paths are generated according to the individual differences of the patient (such as age, weight, past medical history, etc.) and the fused input data to form an initial set of rehabilitation paths as the initial population . Among them, each initial rehabilitation path contains rehabilitation training indicators such as training actions, frequencies, intensities, rest durations, etc.:

[0119] (13)

[0120] Subsequently, the following steps are executed:

[0121] Selection: According to the multi-objective fitness, a higher survival probability is given to the solutions with excellent performance (less dominated).

[0122] Crossover: Randomly combine some parameters of two rehabilitation paths to generate new candidate paths.

[0123] Mutation: Make random perturbations to a small part of the parameters to avoid falling into local extrema and improve the population diversity.

[0124] After the above process, multiple rehabilitation paths are finally obtained.

[0125] On this basis, in this embodiment, in step S1203, when implementing the rehabilitation path in each rehabilitation scenario, the patient's movement is simulated according to all the above-mentioned rehabilitation training indicators, and objective movement data is collected. According to the objective movement data, the values of each single-objective function and the multi-objective function are calculated as the rehabilitation results.

[0126] In step S1204, as an embodiment, the preset number of rehabilitation paths with the maximum value of the multi-objective function is taken as the personalized rehabilitation plan.

[0127] In step S1204, preferably, by combining the multi-objective function and all single-objective functions, the non-dominated sorting method is used to screen out the personalized rehabilitation plan. Specifically, all rehabilitation paths are hierarchically classified according to the "non-dominated" relationship, and multiple rehabilitation paths (Pareto front solution set) in the optimal layer are the personalized rehabilitation plan.

[0128] The non-dominated sorting genetic algorithm (NSGA-II) is used to perform multi-objective optimization on multiple single-objectives involved in the patient's rehabilitation process (such as joint flexibility recovery, gait balance, and muscle strength recovery). By constructing a weighted combined objective function, the Pareto optimal rehabilitation plan is continuously screened out, and the large language model is used to explain each plan in detail, ultimately helping the patient achieve the best postoperative rehabilitation effect. This multi-objective optimization method can take into account multiple patient needs (such as pain reduction, speed increase, muscle strength enhancement, etc.) at the same time, making the patient's rehabilitation training of higher quality.

[0129] Preferably, in step S1204, the significantly inferior rehabilitation paths can be screened out first, and then the non-dominated sorting method is used to screen out the personalized rehabilitation plan.

[0130] Preferably, as an embodiment, on the basis of obtaining the activity type according to the real-time objective motion data, in step S1201, when generating the rehabilitation path, the weights of each index in the rehabilitation path are determined according to the activity type. For example, if it is detected that the patient can only perform simple activities (such as standing) currently, a rehabilitation path with lower intensity and mainly focusing on balance and stability training is generated; if it is recognized that the patient's activity ability has been significantly improved (such as from standing to climbing stairs), a more complex rehabilitation path with increased training difficulty or turning is generated.

[0131] Preferably, as another embodiment, on the basis of obtaining the activity type according to the real-time objective motion data, in step S1201, the current rehabilitation effect is first evaluated according to the activity type. For example, if it is recognized that the patient is in a walking state, the gait symmetry and stability are analyzed according to the real-time objective motion data, and then the patient's gait recovery situation is evaluated; if it is recognized that the patient is climbing stairs, the flexibility and load-bearing capacity of the knee joint are evaluated according to the real-time objective motion data. These evaluation results based on specific activity types are directly used to measure the rehabilitation effect and serve as important reference indicators for optimizing the personalized rehabilitation plan.

[0132] Then, the rehabilitation path is dynamically adjusted according to the rehabilitation effect and activity type. For example, when it is detected that the patient's pain rises sharply, fatigue is excessive, or the subjective pain level or objective movement data is far from the expected, it can be re-evaluated in time, and the training intensity, rest time interval, and weight of the single objective function can be automatically adjusted, or a new candidate solution can be directly inserted in the middle and several generations of iterative search can be restarted to quickly obtain a solution that is more suitable for the patient's current state. This real-time feedback mechanism ensures that the patient's training intensity matches their own state, reduces the risk of secondary injury or excessive fatigue, and continuously improves and customizes the rehabilitation plan that best suits the patient's personalized needs, thereby achieving the best recovery effect after surgery.

[0133] Preferably, as another embodiment, based on the activity type obtained according to the real-time objective motion data, in step S130, the effect of different rehabilitation paths on the patient's mobility improvement can be predicted according to the activity type, thereby generating personalized rehabilitation training recommendations (such as determining the training intensity, frequency and content).

[0134] Preferably, the rehabilitation pathway generation models can be embedded in the patient's wearable device or the edge gateway around the patient to achieve edge computing.

[0135] Continue Figure 1 ,In step S130, the personalized rehabilitation plan is converted into rehabilitation suggestions based on natural language through the vertical field large language model (LLM), including recommendations for appropriate sports, training intensity, frequency, etc.

[0136] Specifically, after reading the fused input data and multiple personalized rehabilitation plans, the vertical domain large language model combines its medical domain knowledge and reasoning ability to generate natural language rehabilitation suggestions for patients, such as: "It is recommended to perform 2 sets of flexion and extension exercises every day, 15 minutes each set" or "Appropriately extend the rest interval when the fatigue level reaches 7 points".

[0137] Preferably, on the basis of the above, the rehabilitation recommendations also include reasons for the recommendation and precautions (for example, prompting the performance differences of each personalized rehabilitation plan in dimensions such as joint flexibility, gait balance and muscle strength recovery). Users (patients and doctors) can quickly adjust the rehabilitation training plan accordingly, for example, selecting the final plan from multiple personalized rehabilitation plans according to the most concerned indicators (such as urgent need to relieve pain, or strengthen balance training).

[0138] Preferably, the vertical domain large language model can be embedded in the patient's wearable device or the edge gateway around the patient to achieve edge computing.

[0139] Based on the above, this application also provides a remote rehabilitation assessment system after knee replacement surgery. Figure 3As shown in the figure, the remote rehabilitation evaluation system after knee joint replacement includes a fusion module 310, a plan generation module 320, and a conversion module 330.

[0140] The fusion module 310 is used to fuse the real-time objective motion data of the patient with the latest subjective feeling data to obtain the fused input data.

[0141] The plan generation module 320 is used to intelligently generate multiple personalized rehabilitation plans based on the fused input data.

[0142] The conversion module 330 is used to convert the personalized rehabilitation plan into a rehabilitation suggestion based on natural language for the user to select.

[0143] Preferably, the plan generation module 320 includes a rehabilitation path generation module 3201, a scenario construction module 3202, a simulation module 3203, and a screening module 3204.

[0144] The rehabilitation path generation module 3201 is used to generate multiple rehabilitation paths based on the fused input data.

[0145] The scenario construction module 3202 is used to construct multiple rehabilitation scenarios.

[0146] For each rehabilitation path, the simulation module 3203 is used to simulate the implementation of the rehabilitation path in each rehabilitation scenario, and calculate the comprehensive rehabilitation result of the rehabilitation path based on the rehabilitation results of the same rehabilitation path in all rehabilitation scenarios.

[0147] The screening module 3204 is used to screen out multiple rehabilitation paths with the optimal comprehensive rehabilitation results as the personalized rehabilitation plan.

[0148] Preferably, the rehabilitation result of each rehabilitation path in each rehabilitation scenario is obtained based on a multi-objective function and multiple single-objective functions. The multi-objective function is the weighted sum of multiple single-objective functions, and the input data of each single-objective function is one or more objective motion data.

[0149] Preferably, the rehabilitation path generation module 3201 is used to generate multiple rehabilitation paths by using an evolutionary algorithm. Each rehabilitation path includes multiple rehabilitation training indicators. And the screening module 3204 is used to screen out the personalized rehabilitation plan by using the non-dominated sorting method.

[0150] Preferably, the remote rehabilitation evaluation system after knee joint replacement further includes an activity type acquisition module 340. The activity type acquisition module 340 is used to obtain the activity type of the patient corresponding to the real-time objective motion data based on the real-time objective motion data.

[0151] Preferably, the rehabilitation path generation module 3201 is used to determine the weight of each index in the rehabilitation path according to the activity type.

[0152] Preferably, the remote rehabilitation evaluation system after knee joint replacement further includes an effect evaluation module 350, which is used to evaluate the current rehabilitation effect according to the activity type.

[0153] The rehabilitation path generation module 3201 is used to dynamically adjust the rehabilitation path according to the rehabilitation effect and the activity type.

[0154] Although some specific embodiments of the present application have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for illustration and not for limiting the scope of the present application. Those skilled in the art should understand that the above embodiments can be modified without departing from the scope and spirit of the present application. The scope of the present application is defined by the appended claims.

Claims

1. A remote rehabilitation assessment method after knee replacement surgery, characterized in that: include: Fuse the patient's real-time objective motion data with the latest subjective feeling data to obtain fused input data; Intelligently generating multiple personalized rehabilitation plans based on the fused input data; The personalized rehabilitation plan is converted into rehabilitation suggestions based on natural language for user selection.

2. The remote rehabilitation assessment method after knee replacement surgery according to claim 1, characterized in that: Intelligently generate multiple personalized rehabilitation plans based on the fused input data, specifically including: generating a plurality of rehabilitation pathways based on the fused input data; Construct multiple rehabilitation scenarios; For each rehabilitation pathway, simulate the implementation of the rehabilitation pathway in each rehabilitation scenario, and calculate the comprehensive rehabilitation result of the rehabilitation pathway based on the rehabilitation results of the same rehabilitation pathway in all rehabilitation scenarios; A plurality of rehabilitation pathways with the best comprehensive rehabilitation results are screened out as the personalized rehabilitation program.

3. The remote rehabilitation assessment method after knee replacement surgery according to claim 2, characterized in that: The rehabilitation result of each rehabilitation path in each rehabilitation scenario is obtained based on a multi-objective function and multiple single-objective functions, where the multi-objective function is a weighted sum of multiple single-objective functions, and the input data of each single-objective function is one or more objective motion data.

4. The remote rehabilitation assessment method after knee replacement surgery according to claim 3, characterized in that: An evolutionary algorithm is used to generate multiple rehabilitation paths, which include multiple rehabilitation training indicators, and a non-dominated sorting method is used to screen out the personalized rehabilitation program.

5. The remote rehabilitation assessment method after knee replacement surgery according to claim 1, characterized in that: The personalized rehabilitation program and / or the conversion of the personalized rehabilitation program into rehabilitation recommendations are generated through the patient's wearable device or an edge gateway around the patient.

6. A remote rehabilitation assessment system after knee replacement surgery, characterized in that: It includes fusion module, solution generation module and transformation module; The fusion module is used to fuse the patient's real-time objective motion data with the latest subjective feeling data to obtain fused input data; The program generation module is used to intelligently generate multiple personalized rehabilitation programs based on the fused input data; The conversion module is used to convert the personalized rehabilitation plan into rehabilitation suggestions based on natural language for user selection.

7. The remote rehabilitation evaluation system after knee replacement surgery according to claim 6, characterized in that: The program generation module includes a rehabilitation path generation module, a scenario construction module, a simulation module and a screening module; The rehabilitation path generation module is used to generate a plurality of rehabilitation paths according to the fused input data; The scenario construction module is used to construct multiple rehabilitation scenarios; The simulation module is used to simulate the implementation of each rehabilitation pathway in each rehabilitation scenario, and calculate the comprehensive rehabilitation result of the rehabilitation pathway based on the rehabilitation results of the same rehabilitation pathway in all rehabilitation scenarios; The screening module is used to screen out multiple rehabilitation pathways with the best comprehensive rehabilitation results as the personalized rehabilitation program.

8. The remote rehabilitation evaluation system after knee replacement surgery according to claim 7, characterized in that: The rehabilitation result of each rehabilitation path in each rehabilitation scenario is obtained based on a multi-objective function and multiple single-objective functions, where the multi-objective function is a weighted sum of multiple single-objective functions, and the input data of each single-objective function is one or more objective motion data.

9. The remote rehabilitation evaluation system after knee replacement surgery according to claim 8, characterized in that: The rehabilitation path generation module is used to generate multiple rehabilitation paths using an evolutionary algorithm, each rehabilitation path including multiple rehabilitation training indicators; and The screening module is used to screen out the personalized rehabilitation plan by adopting a non-dominated sorting method.

10. The remote rehabilitation evaluation system after knee replacement surgery according to claim 6, characterized in that: The solution generation module and / or the conversion module are embedded in the patient's wearable device or an edge gateway around the patient.

Images