Intelligent monitoring method and system for postoperative recovery of orthopedic patient

Abstract

The invention discloses an intelligent monitoring method and system for postoperative recovery of an orthopedic patient. The method comprises the following steps: acquiring personal information data and postoperative recovery record data of the orthopedic postoperative patient; performing first processing and second processing on the postoperative rehabilitation record data to generate multi-modal rehabilitation data; extracting the multi-mode rehabilitation data, inputting the multi-mode rehabilitation data into an LSTM orthopedic recovery network, and performing model training on the LSTM orthopedic recovery network; using the LSTM orthopedic recovery network to predict the extracted multi-modal rehabilitation data, and generating a prediction result; according to the method, postoperative multi-dimensional data of the orthopedic patient is collected, the richness of the data and the accuracy of subsequent prediction are improved, then the rehabilitation stage of the patient is predicted by adopting the model, and a phased diagnosis report can be generated for the patient.

Description

Technical Field

[0001] The present invention relates to the field of smart medical care, and in particular to a method for intelligently monitoring the postoperative recovery of orthopedic patients. Background Art

[0002] In recent years, with the increasing aging of the global population, orthopedic issues among the elderly have become increasingly prominent. The recovery process after orthopedic surgeries such as joint replacements and fracture fixation directly impacts patients' functional recovery and quality of life. Traditional postoperative monitoring methods rely solely on physician experience and regular outpatient checkups, but are unable to track recovery status in real time and provide patients with more accurate, efficient, and personalized diagnostic advice.

[0003] Currently, the field of intelligent monitoring after orthopedic surgery is undergoing a transformation, driven by the advent of smart healthcare and advances in the Internet of Things, artificial intelligence, and telemedicine. Chinese Invention Application No. 201910432100.4 proposes a method and system for orthopedic postoperative rehabilitation monitoring. This method uses Fourier transforms to analyze patient gait data. While suitable for feature extraction and noise reduction of steady-state periodic signals, it fails to consider patient data other than gait, resulting in a limited analysis. Chinese Invention Application No. 202211504128.2 discloses a method and system for remote intelligent rehabilitation after knee replacement surgery. This invention uses an IMU sensor and a monocular camera to identify knee joint training movements. Requiring only an IMU and a standard camera, it can collect real-time patient data, resulting in low cost and easy scalability. However, the accuracy of standard cameras decreases dramatically in low light or under obstruction. Real-time fusion of IMU and visual data requires significant computing power, and low-end devices can result in high latency. This method is only suitable for individual patient monitoring of body data, preventing hospitals from receiving and analyzing orthopedic patients' recovery status in real time and making predictions about their progress. Summary of the Invention

[0004] The technical problem solved by the present invention is that: other body data of the patient except gait data are not taken into account, and the patient's current recovery status cannot be predicted in real time and an authoritative and reliable diagnosis report cannot be generated.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, a method for intelligently monitoring the postoperative recovery of orthopedic patients is provided, characterized in that it comprises:

[0006] Step S100, collecting personal information data and postoperative rehabilitation record data of orthopedic surgery patients;

[0007] Step S200, performing a first process, a second process, and a third process on the postoperative rehabilitation record data to generate multimodal rehabilitation data;

[0008] Step S300, extracting the multimodal rehabilitation data and inputting it into an LSTM orthopedic restoration network and performing model training on the LSTM orthopedic restoration network;

[0009] Step S400: Using the LSTM orthopedic restoration network, predict the extracted multimodal rehabilitation data to generate a prediction result.

[0010] The personal information data includes: patient name, patient gender, patient number, surgical site and test time;

[0011] The postoperative rehabilitation record data includes: joint angle data, electromyographic data and human-computer interaction data;

[0012] The multimodal rehabilitation data includes: patient number, joint angle data, electromyographic data, human-computer interaction data and clinical evaluation data.

[0013] Preferably, step S100 specifically includes:

[0014] Step S101, recording the patient's personal information data after surgery;

[0015] Step S102, using optical motion capture to capture the patient's joint angle data based on the personal information data, wherein the joint angle data includes hip flexion angle, knee flexion angle, ankle flexion angle, and three-dimensional angular velocity;

[0016] Step S103, using a surface electromyograph to collect electromyographic data of the patient, wherein the electromyographic data includes the RMS value and median frequency of n muscles;

[0017] Step S104, extracting human-machine interaction data between the patient and the robot according to the rehabilitation robot, wherein the human-machine interaction data includes left leg torque, right leg torque, energy consumption, power assistance ratio, and joint force value;

[0018] Step S105 , saving the patient number and detection time of the personal information data as well as the joint angle data, electromyographic data and human-computer interaction data as postoperative rehabilitation record data.

[0019] Preferably, step S200 specifically includes:

[0020] Step S200 specifically includes: a first process, a second process, and a third process;

[0021] The first processing includes: using a time synchronization module to normalize the joint angle data and electromyographic data in the postoperative rehabilitation record data and scale them to between thresholds; performing multi-dimensional stage division on the postoperative rehabilitation record data, classifying the data according to the size of the joint angle data of the postoperative rehabilitation record data, the power ratio data of the human-computer interaction data, and the muscle RMS value of the electromyographic data, dividing the postoperative rehabilitation record data into n types, and using numbers 0 to n-1 as type labels to obtain clinical evaluation label data, and making a one-to-one correspondence between the clinical evaluation label data and the postoperative rehabilitation record data to obtain orthopedic rehabilitation data;

[0022] The second processing includes: forming a joint mechanics relationship value based on a one-to-one correspondence between the hip flexion angle, the knee flexion angle, and the ankle flexion angle of the joint angle data and the joint force value of the human-computer interaction data;

[0023] The third processing includes: performing joint motion evaluation on the joint mechanics relationship value according to the multi-dimensional evaluation standard table, determining whether the joint mechanics relationship value is in the multi-dimensional evaluation standard table, and if not, determining it as a dislocated joint mechanics relationship value; if so, comparing the joint mechanics relationship value with a standard score, and using the standard score as a base score for further calculation;

[0024] Performing a high-level difference calculation on the joint angle data of the joint mechanics relationship value and the highest threshold standard of the level at which the matching standard score is located, performing a low-level difference calculation on the joint angle data of the joint mechanics relationship value and the lowest threshold standard of the level at which the matching standard score is located, obtaining a low-standard joint angle difference, calculating a difference ratio between the high-standard joint angle difference and the low-standard joint angle difference, calculating a weighted sum of all joint angle differences based on the difference ratio, and using the weighted sum of the joint angle differences as the patient's current recovery state value;

[0025] The patient's current recovery state value is matched one-to-one with the orthopedic rehabilitation data to obtain multimodal rehabilitation data.

[0026] Preferably, step S300 specifically includes:

[0027] Step S301, building the architecture of the LSTM orthopedic restoration network and setting hyperparameters;

[0028] Step S302 , dividing the multimodal rehabilitation data into a training set, a validation set, and a test set according to the patient number at a ratio of 7:1.5:1.5;

[0029] Step S303: importing the training set into the architecture of the LSTM orthopedic restoration network, training the LSTM orthopedic restoration network, and verifying the LSTM orthopedic restoration network using the verification set and the test set.

[0030] Preferably, step S400 includes:

[0031] According to the patient number, the patient's postoperative rehabilitation record data and personal information data are extracted, and the LSTM orthopedic recovery network is used to predict the rehabilitation stage and generate a patient rehabilitation diagnosis report.

[0032] Preferably, the dimensionality reduction and normalization processing includes:

[0033] Normalization of joint angle data and EMG RMS;

[0034] The joint angle data normalization includes: scaling the joint angles in the postoperative rehabilitation record data to within a threshold using a Min-Max normalization calculation expression, wherein the Min-Max normalization calculation expression is:

[0035]

[0036] Among them, X norm,i is the normalized joint angle data of the i-th patient, X i is the joint angle data of the i-th patient, X min is the minimum value of the joint angle data in the postoperative rehabilitation record data, X max The maximum value of the joint angle data in the postoperative rehabilitation record data;

[0037] The myoelectric RMS normalization includes: scaling the myoelectric data in the postoperative rehabilitation record data to within a specified threshold according to Z-Score normalization. The Z-Score normalization calculation expression is:

[0038]

[0039] Among them, X z is the normalized value of the electromyographic data, X is the original value of the electromyographic data, μ is the mean of the electromyographic data, and σ is the standard deviation of the electromyographic data;

[0040] The Sigmoid function is used to normalize the electromyographic data to the value X z Compressed into the threshold range, the calculation expression of the Sigmoid function is:

[0041]

[0042] Among them, Sigmoid(X z ) is X z The compressed EMG data value, e is a mathematical constant;

[0043] Preferably, the LSTM orthopedic restoration network includes: setting the network structure hyperparameter value and training hyperparameter value of the LSTM orthopedic restoration network, inputting the multimodal rehabilitation data into the LSTM orthopedic restoration network, training, and setting the loss function to obtain the loss result.

[0044] Preferably, the patient rehabilitation diagnosis report: utilizes the LSTM orthopedic recovery network card to perform real-time prediction of the multimodal rehabilitation data, generates the patient's recovery stage, and saves the recovery stage, the patient's personal information data and the currently collected patient postoperative rehabilitation record data as a rehabilitation diagnosis report.

[0045] Secondly, an intelligent monitoring system for postoperative recovery of orthopedic patients is implemented, including a data acquisition module, a data processing module, a network training module, and an outcome prediction module;

[0046] The data acquisition module is used to collect the physical condition of orthopedic patients after surgery. It uses optical motion capture, surface electromyographs and rehabilitation robots to collect the patient's postoperative rehabilitation record data.

[0047] The data processing module is used to perform dimensionality reduction and normalization processing on the postoperative rehabilitation record data, and to divide the postoperative rehabilitation record data into multi-dimensional stages using manual annotation according to clinical evaluation standards to obtain multimodal rehabilitation data;

[0048] The network model training module is used to perform model training on the LSTM orthopedic restoration network, input multimodal rehabilitation data into the LSTM orthopedic restoration network, and train an LSTM orthopedic restoration network with strong prediction capabilities based on parameter adjustment operations;

[0049] The result prediction module is used to make real-time predictions on the patient's rehabilitation data, and uses the LSTM orthopedic recovery network to make real-time predictions on part of the extracted multi-module rehabilitation data to obtain a rehabilitation stage prediction result.

[0050] Beneficial effects of the present invention: The present invention uses optical motion capture, surface electromyographs and rehabilitation robots to perform postoperative rehabilitation testing on patients, obtain postoperative rehabilitation record data of patients, and perform first processing and second processing on the postoperative rehabilitation record data of patients, normalize and manually label the data to obtain multimodal rehabilitation data, thereby improving the richness of the data and the accuracy of subsequent predictions, and uses the LSTM orthopedic recovery network to predict the multimodal rehabilitation data of patients, obtain rehabilitation stage prediction results, and generate personalized diagnostic reports for patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of the basic flow of an intelligent monitoring method for postoperative recovery of orthopedic patients provided by one embodiment of the present invention.

[0052] Figure 2 A schematic diagram of the basic flow of an intelligent monitoring system for postoperative recovery of orthopedic patients provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0054] Example 1, with reference to Figure 1 , as one embodiment of the present invention, provides a method for intelligently monitoring the postoperative recovery of orthopedic patients, comprising:

[0055] Step S100, collecting personal information data and postoperative rehabilitation record data of orthopedic surgery patients;

[0056] Step S200, performing first processing, second processing, and third processing on the postoperative rehabilitation record data to generate multimodal rehabilitation data;

[0057] Step S300, extracting multimodal rehabilitation data and inputting it into an LSTM orthopedic restoration network and performing model training on the LSTM orthopedic restoration network;

[0058] Step S400: Using the LSTM orthopedic restoration network, the extracted multimodal rehabilitation data is predicted to generate a prediction result.

[0059] In this embodiment, the collected orthopedic patient data is normalized and labeled through the first and second processing to facilitate the subsequent substitution of the data into the network model for network model training. Afterwards, the LSTM orthopedic recovery network is used to predict the patient's postoperative recovery stage based on the current orthopedic patient's data, and generate a phased diagnosis report. This implementation plan effectively solves the timing and multimodal fusion difficulties of customer rehabilitation evaluation and has high clinical value.

[0060] Personal information data includes: patient name, patient gender, patient number, surgical site, and test time;

[0061] Postoperative rehabilitation record data includes: joint angle data, electromyography data and human-computer interaction data;

[0062] Multimodal rehabilitation data includes: patient number, joint angle data, electromyographic data, human-computer interaction data, and clinical assessment data;

[0063] In this embodiment, the collected personal information data and postoperative rehabilitation data are introduced in detail to provide a clearer data architecture for data processing.

[0064] Step S101, recording the patient's personal information data after surgery;

[0065] Step S102, using optical motion capture based on the personal information data, to capture the patient's joint angle data, where the joint angle data includes hip flexion angle, knee flexion angle, ankle flexion angle, and three-dimensional angular velocity;

[0066] Step S103, using a surface electromyograph to collect electromyographic data of the patient, the electromyographic data including the RMS value and median frequency of n muscles;

[0067] Step S104: extracting human-machine interaction data between the patient and the robot according to the rehabilitation robot, the human-machine interaction data including left leg torque, right leg torque, energy consumption, power assistance ratio, and joint force value;

[0068] Step S105 , saving the patient number and detection time of the personal information data as well as the joint angle data, electromyographic data and human-computer interaction data as postoperative rehabilitation record data.

[0069] In this embodiment, the RMS values of 8 muscles are collected, the joint angle data is used to evaluate the degree of motor function recovery, the electromyography data reflects the neuromuscular control ability, and the human-computer interaction data is used to quantify the intensity of exoskeleton intervention and the patient's active participation. After the collection is completed, the multi-source data is associated with a unified patient number to obtain postoperative rehabilitation record data, ensuring the temporal consistency of subsequent analysis, and also providing a reliable data basis for the objective and digital evaluation of orthopedic rehabilitation.

[0070] Step S200 specifically includes: a first process, a second process, and a third process;

[0071] The first processing includes: using a time synchronization module to normalize the joint angle data and electromyographic data in the postoperative rehabilitation record data and scale them to a threshold value; dividing the postoperative rehabilitation record data into multiple stages, classifying the data by size based on the joint angle data of the postoperative rehabilitation record data, the power ratio data of the human-computer interaction data, and the muscle RMS value of the electromyographic data, dividing the postoperative rehabilitation record data into n types, and using numbers 0 to n-1 as type labels to obtain clinical evaluation label data, and then making a one-to-one correspondence between the clinical evaluation label data and the postoperative rehabilitation record data to obtain orthopedic rehabilitation data;

[0072] The second processing includes: forming a joint mechanics relationship value based on the one-to-one correspondence between the hip flexion angle, knee flexion angle and ankle flexion angle of the joint angle data and the joint force value of the human-computer interaction data;

[0073] The third processing includes: performing joint motion evaluation on the joint mechanical relationship value according to the multi-dimensional evaluation standard table, determining whether the joint mechanical relationship value is in the multi-dimensional evaluation standard table, and if not, determining it as a dislocated joint mechanical relationship value; if so, comparing the joint mechanical relationship value with the standard score, and using the standard score as the base score for further calculation;

[0074] Performing a high-level difference calculation on the joint angle data of the joint mechanics relationship value and the highest threshold standard of the matching standard score level, performing a low-level difference calculation on the joint angle data of the joint mechanics relationship value and the lowest threshold standard of the matching standard score level, obtaining a low-standard joint angle difference, calculating the difference ratio of the high-standard joint angle difference and the low-standard joint angle difference, calculating a weighted sum of all joint angle differences based on the difference ratio, and using the weighted sum of the joint angle differences as the patient's current recovery status value;

[0075] The patient's current recovery status value is matched one-to-one with the orthopedic rehabilitation data to obtain multimodal rehabilitation data.

[0076] In this embodiment, the postoperative rehabilitation record data are divided into four types, with type labels: acute phase, subacute phase, functional phase, and regression phase, corresponding to 0, 1, 2, and 3, respectively;

[0077] The multi-dimensional evaluation standard table is:

[0078]

[0079] Step S300 specifically includes:

[0080] Step S301, building the architecture of the LSTM orthopedic restoration network and setting hyperparameters;

[0081] Step S302 , dividing the multimodal rehabilitation data into a training set, a validation set, and a test set according to the patient number at a ratio of 7:1.5:1.5;

[0082] Step S303: import the training set into the architecture of the LSTM orthopedic restoration network, train the LSTM orthopedic restoration network, and verify the LSTM orthopedic restoration network using the validation set and the test set.

[0083] In this embodiment, through the gating mechanism of the LSTM network architecture, the forget gate, input gate and output gate are used to effectively model the joint acquisition and temporal dynamic changes of the electromyographic signals in the rehabilitation process, and the correlation between multimodal data is learned by dictionary learning. By using a scientific data partitioning strategy, the accuracy and generalization ability of orthopedic rehabilitation prediction are significantly improved.

[0084] Step S400 includes:

[0085] According to the patient number, the patient's postoperative rehabilitation record data and personal information data are extracted, and the LSTM orthopedic recovery network is used to predict the rehabilitation stage and generate the patient's rehabilitation diagnosis report.

[0086] In this embodiment, the LSTM orthopedic recovery network is deployed to predict the recovery stage of postoperative patients, generate a certificate renewal rehabilitation diagnosis report, improve the consistency of test diagnosis, and optimize the hospital's diagnosis and treatment efficiency.

[0087] Dimensionality reduction and normalization processing includes: joint angle data normalization and electromyography RMS normalization;

[0088] Normalization of joint angle data includes: using the Min-Max normalization calculation expression to scale the joint angles in the postoperative rehabilitation record data to within the threshold. The Min-Max normalization calculation expression is:

[0089]

[0090] Among them, X norm,i is the normalized joint angle data of the i-th patient, X i is the joint angle data of the i-th patient, X min is the minimum value of the joint angle data in the postoperative rehabilitation record data, X max It is the maximum value of the joint angle data in the postoperative rehabilitation record data;

[0091] Myoelectric RMS normalization includes: scaling the myoelectric data in the postoperative rehabilitation recording data to within the specified threshold according to Z-Score normalization. The Z-Score normalization calculation expression is:

[0092]

[0093] Among them, X z is the normalized value of the EMG data, X is the original value of the EMG data, μ is the mean of the EMG data, and σ is the standard deviation of the EMG data;

[0094] Use the Sigmoid function to normalize the electromyographic data to the value X z Compressed into the threshold range, the calculation expression of the Sigmoid function is:

[0095]

[0096] Among them, Sigmoid(X z ) is X z The compressed electromyographic data value, e is a mathematical constant.

[0097] In this embodiment, all thresholds are set between [0, 1]. The raw data of postoperative knee flexion ranges from 10° to 90°. After normalization and scaling to the specified threshold [0, 1], the range becomes 0.12→1.0, which directly reflects the percentage of functional recovery. The RMS Z-Score decreases from 1.5 to -0.5, and after Sigmoid mapping, it decreases from 0.82→0.38, quantitatively displaying the muscle status. This embodiment establishes a standardized preprocessing paradigm for multimodal data analysis by setting a dedicated normalization method for the characteristics of joint angles and electromyographic signals.

[0098] The LSTM orthopedic restoration network includes: setting network structure hyperparameter values and training hyperparameter values of the LSTM orthopedic restoration network, inputting multimodal rehabilitation data into the LSTM orthopedic restoration network, performing training, and setting a loss function to obtain a loss result.

[0099] In this embodiment, the LSTM network structure hyperparameter values are set as follows: the number of layers is 3, the number of units is 64, the dropout rate is set to 0.2-0.5, an Attention layer is added to the last layer of LSTM, and the softmax regression function is used for rehabilitation stage classification; the LSTM network training hyperparameter values are set as follows: the learning rate is set to 0.001, the batch size is set to 32, and the loss function is set to the weighted cross entropy function to process the classification task.

[0100] Patient rehabilitation diagnosis report: Utilize the LSTM orthopedic recovery network card to perform real-time predictions on multimodal rehabilitation data, generate the patient's recovery stage, and save the recovery stage, the patient's personal information data, and the currently collected patient postoperative rehabilitation record data as a rehabilitation diagnosis report.

[0101] In this embodiment, optical motion capture and other instruments are used to collect patient postoperative recovery data, which is input into the LSTM network for prediction, determines the patient's recovery stage, and generates a rehabilitation diagnosis report. This embodiment achieves the transition of orthopedic rehabilitation from empirical medicine to precision medicine through data-driven decision-making and human-computer collaborative diagnosis. Its standardized report output provides a reliable basis for medical digitization.

[0102] Example 2, reference Figure 2 This is another embodiment of the present invention. Different from the first embodiment, this embodiment provides an intelligent monitoring system for postoperative recovery of orthopedic patients. In order to verify and illustrate the technical effects adopted in this method, this embodiment adopts traditional technical solutions and the method of the present invention for comparative testing, and compares the test results by means of scientific demonstration to verify the real effect of this method.

[0103] An intelligent monitoring system for postoperative recovery of orthopedic patients, comprising a data acquisition module, a data processing module, a network training module and an outcome prediction module;

[0104] The data acquisition module is used to collect the physical condition of orthopedic patients after surgery. It uses optical motion capture, surface electromyographs, and rehabilitation robots to collect postoperative rehabilitation records for individual patients.

[0105] The data processing module is used to perform dimensionality reduction and normalization processing on the postoperative rehabilitation record data, and to divide the postoperative rehabilitation record data into multi-dimensional stages using manual annotation according to clinical evaluation standards to obtain multimodal rehabilitation data;

[0106] The network model training module is used to train the LSTM orthopedic restoration network. It inputs multimodal rehabilitation data into the LSTM orthopedic restoration network and trains an LSTM orthopedic restoration network with strong predictive capabilities based on parameter adjustment operations.

[0107] The result prediction module is used to make real-time predictions on the patient's rehabilitation data. It uses the LSTM orthopedic recovery network to make real-time predictions on some of the extracted multi-module rehabilitation data to obtain the predicted results of the rehabilitation stage.

[0108] In this embodiment, the invention is divided into four modules. First, through the synchronous collection of optical motion capture (joint kinematics), surface electromyometry (muscle activation status), and rehabilitation robots (human-computer interaction mechanics), a multi-dimensional rehabilitation evaluation system is constructed, covering the entire chain of biomechanics-physiology-external intervention data; then, the normalization threshold is dynamically adjusted for different surgical forms to avoid errors caused by unified standardization; finally, the LSTM network model architecture is deployed to effectively predict the patient's rehabilitation stage based on the collected patient data. This module upgrades orthopedic rehabilitation from experience-driven to data-driven, and has significant clinical and economic value.

[0109] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium may be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0110] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent monitoring method for postoperative recovery of orthopedic patients, characterized in that: include: Step S100, collecting personal information data and postoperative rehabilitation record data of orthopedic surgery patients; Step S200, performing a first process, a second process, and a third process on the postoperative rehabilitation record data to generate multimodal rehabilitation data; Step S300, extracting the multimodal rehabilitation data and inputting it into an LSTM orthopedic restoration network and performing model training on the LSTM orthopedic restoration network; Step S400: Using the LSTM orthopedic restoration network, predict the extracted multimodal rehabilitation data to generate a prediction result.

2. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 1, characterized in that: The personal information data includes: patient name, patient gender, patient number, surgical site and test time; The postoperative rehabilitation record data includes: joint angle data, electromyographic data and human-computer interaction data; The multimodal rehabilitation data includes: patient number, joint angle data, electromyographic data, human-computer interaction data, clinical assessment data and joint contact force.

3. The intelligent monitoring method for postoperative recovery of orthopedic patients according to claim 1, characterized in that: Step S100 specifically includes: Step S101, recording the patient's personal information data after surgery; Step S102, using optical motion capture to capture the patient's joint angle data based on the personal information data, wherein the joint angle data includes hip flexion angle, knee flexion angle, ankle flexion angle, and three-dimensional angular velocity; Step S103, using a surface electromyograph to collect electromyographic data of the patient, wherein the electromyographic data includes the RMS value and median frequency of n muscles; Step S104, extracting human-machine interaction data between the patient and the robot according to the rehabilitation robot, wherein the human-machine interaction data includes left leg torque, right leg torque, energy consumption, power assistance ratio, and joint force value; Step S105 , saving the patient number and detection time of the personal information data as well as the joint angle data, electromyographic data and human-computer interaction data as postoperative rehabilitation record data.

4. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 3, characterized in that: Step S200 specifically includes: a first process, a second process, and a third process; The first processing includes: using a time synchronization module to normalize the joint angle data and electromyographic data in the postoperative rehabilitation record data and scale them to between thresholds; performing multi-dimensional stage division on the postoperative rehabilitation record data, classifying the data according to the size of the joint angle data of the postoperative rehabilitation record data, the power ratio data of the human-computer interaction data, and the muscle RMS value of the electromyographic data, dividing the postoperative rehabilitation record data into n types, and using numbers 0 to n-1 as type labels to obtain clinical evaluation label data, and making a one-to-one correspondence between the clinical evaluation label data and the postoperative rehabilitation record data to obtain orthopedic rehabilitation data; The second processing includes: forming a joint mechanics relationship value based on a one-to-one correspondence between the hip flexion angle, the knee flexion angle, and the ankle flexion angle of the joint angle data and the joint force value of the human-computer interaction data; The third processing includes: performing joint motion evaluation on the joint mechanics relationship value according to the multi-dimensional evaluation standard table, determining whether the joint mechanics relationship value is in the multi-dimensional evaluation standard table, and if not, determining it as a dislocated joint mechanics relationship value; if so, matching the joint mechanics relationship value with a standard score, and using the standard score as the standard score; Performing a high-level difference calculation on the joint angle data of the joint mechanics relationship value and the highest threshold standard of the level at which the matching standard score is located to obtain a high-standard joint angle difference; performing a low-level difference calculation on the joint angle data of the joint mechanics relationship value and the lowest threshold standard of the level at which the matching standard score is located to obtain a low-standard joint angle difference; calculating a difference ratio between the high-standard joint angle difference and the low-standard joint angle difference; calculating a weighted sum value of all joint angle differences based on the difference ratio; and using the weighted sum value of the joint angle differences as the patient's current recovery state value; The patient's current recovery state value is matched one-to-one with the orthopedic rehabilitation data to obtain multimodal rehabilitation data.

5. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 1, characterized in that: Step S300 specifically includes: Step S301, building the architecture of the LSTM orthopedic restoration network and setting hyperparameters; Step S302 , dividing the multimodal rehabilitation data into a training set, a validation set, and a test set according to the patient number at a ratio of 7:1.5:1.5; Step S303: importing the training set into the architecture of the LSTM orthopedic restoration network, training the LSTM orthopedic restoration network, and verifying the LSTM orthopedic restoration network using the verification set and the test set.

6. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 1, characterized in that: Step S400 includes: According to the patient number, the patient's postoperative rehabilitation record data and personal information data are extracted, and the LSTM orthopedic recovery network is used to predict the rehabilitation stage and generate a patient rehabilitation diagnosis report.

7. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 3, characterized in that: The normalization process includes: joint angle data normalization and myoelectric RMS normalization; The joint angle data normalization includes: scaling the joint angles in the postoperative rehabilitation record data to within a threshold using a Min-Max normalization calculation expression, wherein the Min-Max normalization calculation expression is: Among them, X norm,i is the normalized joint angle data of the i-th patient, X i is the joint angle data of the i-th patient, X min is the minimum value of the joint angle data in the postoperative rehabilitation record data, X max The maximum value of the joint angle data in the postoperative rehabilitation record data; The myoelectric RMS normalization includes: scaling the myoelectric data in the postoperative rehabilitation record data to within a specified threshold according to Z-Score normalization. The Z-Score normalization calculation expression is: Among them, X z is the normalized value of the electromyographic data, X is the original value of the electromyographic data, μ is the mean of the electromyographic data, and σ is the standard deviation of the electromyographic data; The Sigmoid function is used to normalize the electromyographic data to the value X z Compressed into the threshold range, the calculation expression of the Sigmoid function is: Among them, Sigmoid(X z ) is X z The compressed electromyographic data value, e is a mathematical constant.

8. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 5, characterized in that: The LSTM orthopedic restoration network includes: setting network structure hyperparameter values and training hyperparameter values of the LSTM orthopedic restoration network, inputting the multimodal rehabilitation data into the LSTM orthopedic restoration network, training, and setting a loss function to obtain a loss result.

9. The intelligent monitoring method for orthopedic patients' postoperative recovery according to claim 6, characterized in that: The patient rehabilitation diagnosis report: uses the LSTM orthopedic recovery network to perform real-time prediction on the multimodal rehabilitation data to generate the patient's recovery stage, and saves the recovery stage, the patient's personal information data and the currently collected patient postoperative rehabilitation record data as a rehabilitation diagnosis report.

10. An intelligent monitoring system for postoperative recovery of orthopedic patients, which is based on the intelligent monitoring method for postoperative recovery of orthopedic patients according to any one of claims 1 to 9, and is characterized in that: It includes data acquisition module, data processing module, network training module and result prediction module; The data acquisition module is used to collect the physical condition of orthopedic patients after surgery. It uses optical motion capture, surface electromyographs and rehabilitation robots to collect the patient's postoperative rehabilitation record data. The data processing module is used to perform dimensionality reduction and normalization processing on the postoperative rehabilitation record data, and to divide the postoperative rehabilitation record data into multi-dimensional stages using manual annotation according to clinical evaluation standards to obtain multimodal rehabilitation data; The network training module is used to perform model training on the LSTM orthopedic restoration network, input multimodal rehabilitation data into the LSTM orthopedic restoration network, and train an LSTM orthopedic restoration network with strong prediction capabilities based on parameter adjustment operations; The result prediction module is used to make real-time predictions on the patient's rehabilitation data, and uses the LSTM orthopedic recovery network to make real-time predictions on part of the extracted multimodal rehabilitation data to obtain rehabilitation stage prediction results.

Images