Shoulder arthroscopy rotator cuff repair postoperative rehabilitation propaganda, education and tracking training system

Abstract

The invention relates to a shoulder arthroscopy rotator cuff repair postoperative rehabilitation propaganda, education and tracking training system, which is characterized in that a doctor records a training doctor's advice in an electronic medical record file of a patient, and a background calls postoperative training video data of a corresponding training type according to the training frequency in the doctor's advice and issues the data to a patient end APP; the patient is reminded to carry out system tracking learning regularly, training action videos of the patient are collected and uploaded to the background, and the background judges whether the training actions of the patient are standard or non-standard through the model. When it is found that the action is not standard in the background, a corresponding action training warning can be generated. Therefore, a doctor can log in the background to check the postoperative training actions of each patient and the judgment result of whether the actions are standard or not, and meanwhile, the exercise frequency of the patient can be determined, so that the patient can recover faster and better. The patient receives background information and tracks training through the APP, postoperative rehabilitation training can be systematically and regularly carried out, and the good rehabilitation purpose is achieved.

Description

Technical Field

[0001] The present disclosure relates to the field of medical rehabilitation technology, and in particular to a post-operative rehabilitation education and tracking training system for rotator cuff repair under shoulder arthroscopy, an application method thereof, and an electronic device. Background Art

[0002] Arthroscopic rotator cuff repair is a minimally invasive surgical technique for rotator cuff tears. This technique uses advanced shoulder arthroscopy equipment to enter the shoulder joint through a tiny incision, thus avoiding the large-area incision of traditional surgery and significantly reducing surgical trauma and postoperative recovery time.

[0003] During the operation, the doctor will first make several small incisions in the patient's shoulder, and then introduce the shoulder arthroscopy and other necessary surgical instruments into the shoulder joint cavity through these incisions. Under the direct vision of the shoulder arthroscopy, the doctor can clearly see the location and extent of the rotator cuff tear and the condition of the surrounding tissues.

[0004] Next, the doctor will use appropriate surgical techniques to repair the rotator cuff based on the specific circumstances of the tear and the individual differences of the patient. This may include cleaning the inflamed synovium, bursa and adhesions around the rotator cuff in the shoulder joint, implanting medial and lateral anchors to fix the torn rotator cuff, and using sutures to sew the torn rotator cuff back to its original anatomical position.

[0005] The entire surgical process requires doctors to have superb surgical skills and rich clinical experience to ensure the success of the operation and the safety of the patient. Through arthroscopic rotator cuff repair, patients can regain the stability and function of the shoulder joint in a shorter recovery time and improve their quality of life.

[0006] Postoperative rehabilitation training after arthroscopic rotator cuff repair is an important part of restoring shoulder joint function, reducing complications and accelerating the recovery process. The following is an overview of the necessity and content of postoperative rehabilitation training:

[0007] 1. Necessity of postoperative rehabilitation training

[0008] 1. Promote wound healing: Rehabilitation training helps improve local blood circulation and accelerate the wound healing process.

[0009] 2. Prevent adhesion: Through appropriate activities, you can prevent adhesion of the soft tissue around the shoulder joint and avoid affecting the joint function.

[0010] 3. Restore joint range of motion: Gradually increase the range of joint motion to ensure that the shoulder joint can return to normal or near-normal levels of activity.

[0011] 4. Enhance muscle strength: Through targeted strength training, you can enhance the strength of the muscles around the shoulders and improve joint stability.

[0012] 5. Relieve pain: Rehabilitation training helps to relieve postoperative pain and improve the patient's quality of life.

[0013] (II) Contents of postoperative rehabilitation training

[0014] 1. Early rehabilitation training (1-6 weeks after surgery)

[0015] Wearing a brace: Wear a shoulder joint protection brace according to the doctor's instructions to limit joint movement and protect the repaired area.

[0016] Passive activities: With the help of the healthy limb or others, perform passive activities of the shoulder joint, such as grasping training, elbow flexion and extension training, shoulder joint rotation training, etc.

[0017] Pain management: Use anti-inflammatory analgesics to control pain and ensure that rehabilitation training can proceed smoothly.

[0018] 2. Mid-term rehabilitation training (6 weeks to 3 months after surgery)

[0019] Active activities: gradually transition from passive activities to active activities, such as pendulum training, wall climbing training, head holding training, etc., to gradually increase the range of joint movement.

[0020] Strength training: Start with light strength training, such as isometric muscle contraction training of the shoulder muscles using dumbbells, being careful to avoid excessive force.

[0021] Daily living activities: Patients are encouraged to participate in daily living activities under protection, such as dressing, combing hair, washing face, etc., to improve their ability to take care of themselves.

[0022] 3. Late rehabilitation training (3 to 6 months after surgery)

[0023] Comprehensive functional recovery: Strengthen the range of motion training of the shoulder joint in all directions, such as shoulder flexion, abduction, and extension, to ensure comprehensive recovery of joint function.

[0024] Resistance training: Use equipment such as elastic bands to perform resistance training to further enhance the strength and stability of the shoulder muscles.

[0025] Functional activities: According to the patient's specific situation, functional activities such as swimming, yoga, etc. are gradually introduced to improve the flexibility and coordination of the shoulder joint.

[0026] However, for some patients who have undergone arthroscopic rotator cuff repair, many patients need to go home and complete their postoperative rehabilitation exercises by themselves, but this will cause the following problems:

[0027] Doctors are not sure whether the shoulder rehabilitation exercises performed by patients at home are standard, and cannot obtain the patients' exercise frequency in real time;

[0028] Regarding the systematic postoperative rehabilitation training for arthroscopic rotator cuff repair, some elderly patients often forget the training plan told by the doctor, and therefore cannot effectively complete the systematic and regular postoperative rehabilitation training. Summary of the invention

[0029] In order to solve the above problems, the present application proposes a postoperative rehabilitation education and tracking training system for shoulder arthroscopic rotator cuff repair and its application method and electronic equipment.

[0030] On the one hand, the present application proposes a postoperative rehabilitation education and tracking training system for shoulder arthroscopic rotator cuff repair, comprising:

[0031] A doctor-side PC is used to generate corresponding training instructions for a patient who has undergone arthroscopic rotator cuff repair and write the training instructions of the patient into an electronic medical record system, wherein the training instructions include the training frequency and training type of the patient;

[0032] The backend server is used to execute the training doctor's orders and regularly send the postoperative education data of the corresponding training type to the patient's patient APP;

[0033] The patient-side APP is used to regularly display the postoperative education data and upload the training action video of the patient when training according to the postoperative education data to the background server;

[0034] The doctor-side PC and the patient-side APP are respectively connected to the backend server for communication.

[0035] As an optional implementation scheme of the present application, optionally, the background server is also used for:

[0036] Storing the training action video uploaded by the patient;

[0037] Perform frame image processing on the training action video to obtain a frame image set consisting of a number of postoperative training action frame images:

[0038] S={S(1), S(1)...S(n)},

[0039] S(1), S(1)...S(n) are postoperative training action frame images of different frames, and S is a frame image set;

[0040] Real-time and orderly identification of each of the postoperative training action frame images in the frame image set, and training action analysis to determine whether the patient's training action is standardized:

[0041] If it is normal, traverse and identify the next postoperative training action frame image;

[0042] If it is not standardized, a corresponding training warning notification will be generated and sent to the patient-side APP.

[0043] As an optional implementation scheme of the present application, optionally, the background server is deployed with:

[0044] An electronic medical record system for recording the training instructions of the patient after arthroscopic rotator cuff repair surgery;

[0045] A postoperative education system, used for configuring and storing the postoperative education data corresponding to different training types;

[0046] A patient management system is used to manage patients who have undergone arthroscopic rotator cuff repair. According to the patient's training instructions, the postoperative education data of the corresponding training type is retrieved from the postoperative education system at regular intervals and sent to the patient-side APP through an API interface;

[0047] An image processing module is used to perform frame image processing on the training action video to obtain a frame image set consisting of a plurality of frames of postoperative training action frame images;

[0048] The intelligent model of shoulder arthroscopic rotator cuff repair is used to identify each postoperative training action frame image in the frame image set in real time and in an orderly manner, and to perform training action analysis to determine whether the patient's training action is standardized and output the corresponding judgment result.

[0049] As an optional implementation scheme of the present application, optionally, the background server is also deployed with:

[0050] Backend database, used for background data storage.

[0051] As an optional implementation scheme of the present application, optionally, the background server is also deployed with:

[0052] API interface, used for patient-side APP communication.

[0053] As an optional implementation scheme of the present application, optionally, the background server is also deployed with:

[0054] The model API is used to call the arthroscopic rotator cuff repair intelligent model.

[0055] On the other hand, the present application proposes an application method of a postoperative rehabilitation education and tracking training system for shoulder arthroscopic rotator cuff repair, comprising the following steps:

[0056] The doctor logs into the backend server through the doctor's PC and writes the patient's training instructions into the electronic medical record system;

[0057] The patient management system reads and retrieves the postoperative education data of the corresponding training type from the postoperative education system regularly according to the patient's training instructions, and sends it to the patient's patient APP through the API interface;

[0058] The patient displays the postoperative education data regularly through the patient-side APP, and uploads the training action video trained according to the postoperative education data to the background server;

[0059] The backend server imports the training action video into the image processing module, performs frame image processing on the training action video, obtains a frame image set corresponding to a plurality of postoperative training action frame images, and imports the frame image set into the intelligent model of shoulder arthroscopic rotator cuff repair;

[0060] Through the intelligent model of shoulder arthroscopic rotator cuff repair, each of the postoperative training action frame images in the frame image set is identified in real time and orderly, and the training action analysis is performed to determine whether the patient's training action is standardized, and the corresponding judgment result is output:

[0061] If it is normal, traverse and identify the next postoperative training action frame image;

[0062] If it is not standardized, a corresponding training warning notification will be generated and sent to the patient-side APP.

[0063] In another aspect, the present application further provides an electronic device, comprising:

[0064] processor;

[0065] a memory for storing processor-executable instructions;

[0066] Wherein, the processor is configured to implement the application method when executing the executable instructions.

[0067] Technical effects of the present invention:

[0068] This application records the training instructions in the patient's electronic medical record file by the doctor, and reads the training instructions of each patient through the patient management system in the background, and retrieves the postoperative training video data of the corresponding training type according to the training frequency and sends it to the patient-side APP, reminding the patient to conduct system tracking and learning regularly, and at the same time collects the patient's training action video and uploads it to the background, which uses the model to determine whether its training action is standardized or not. When the background finds that its action is not standardized, it can generate a corresponding action training warning.

[0069] Therefore, through the tracking training system of the present invention, doctors can log in to the background to view the postoperative training movements of each patient and the results of the determination of whether the movements are standardized, so that doctors can clearly know whether the rehabilitation training movements done by the patient at home are standardized, and at the same time can determine the patient's exercise frequency to make them recover faster and better. Patients receive background information and track training through the APP, and can carry out postoperative rehabilitation training systematically and regularly to achieve good rehabilitation goals.

[0070] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.

[0072] Figure 1 Shown is a schematic diagram of the system composition structure of the present invention;

[0073] Figure 2 It is a schematic diagram of the postoperative rehabilitation training action of the shoulder arthroscopic rotator cuff repair according to the present invention;

[0074] Figure 3 Shown is a schematic diagram of model training of the present invention;

[0075] Figure 4 Shown is the accuracy curve of the model verification of the present invention;

[0076] Figure 5 It is a schematic diagram showing the application of the electronic device of the present invention. DETAILED DESCRIPTION

[0077] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0078] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0079] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the main purpose of the present disclosure.

[0080] Example 1

[0081] like Figure 1 As shown, on one hand, the present application proposes a postoperative rehabilitation education and tracking training system for shoulder arthroscopic rotator cuff repair, comprising:

[0082] A doctor-side PC is used to generate corresponding training instructions for a patient who has undergone arthroscopic rotator cuff repair and write the training instructions of the patient into an electronic medical record system, wherein the training instructions include the training frequency and training type of the patient;

[0083] The backend server is used to execute the training doctor's orders and regularly send the postoperative education data of the corresponding training type to the patient's patient APP;

[0084] The patient-side APP is used to regularly display the postoperative education data and upload the training action video of the patient when training according to the postoperative education data to the background server;

[0085] The doctor-side PC and the patient-side APP are respectively connected to the backend server for communication.

[0086] The present invention proposes a medical rehabilitation system for postoperative rehabilitation education and follow-up training of shoulder joint repair surgery, which involves a background server, a doctor side and a patient side. The doctor can log in to the background server through the doctor side PC to issue corresponding medical orders and manage patients, and the patient can log in to the background server through an APP to carry out corresponding education and follow-up training learning.

[0087] The communication and login operations between the PC and the APP and the background can refer to the existing terminal communication.

[0088] Doctors can formulate corresponding training instructions for each postoperative patient based on their condition, write them into the electronic medical record system, and record the training instructions in the patient's electronic medical record file. The training instructions include the frequency of postoperative exercise for the patient, such as training once a week after surgery or 2 to 4 weeks after surgery, as well as the corresponding training types at different training stages. The specific training frequency and training type can be formulated by the doctor based on the patient's condition after surgery and recorded in the instructions. The training types can be different training movements, such as shrugs, shoulder circles, pendulum movements, etc., which can be referred to Figure 2 Types of postoperative joint exercises shown.

[0089] Postoperative education data of different training types can be stored on the background server. For example, postoperative education training action videos for different training types, etc., can also be divided according to the intensity of training (when the doctor issues medical orders, the training intensity corresponding to each stage (frequency) is different, for example, the training intensity of the first week is: 5 groups). The background reads the training orders of each patient through the patient management system, and retrieves the postoperative training video data of the corresponding training type according to the training frequency and sends it to the patient-side APP, reminding patients to conduct system tracking learning regularly, and follow the postoperative education training videos issued by the system for exercise.

[0090] At the same time, the patient-side APP activates the terminal camera system to collect the patient's training action video and upload it to the background. Let the patient follow the educational video to perform postoperative exercises, and the terminal device where the patient is located collects the training action video and uploads it to the background, which will further determine whether the training action is standardized or not. When the background finds that the action is not standardized, a corresponding action training warning can be generated and sent to the APP terminal. The type of warning sent can be an image or text, or a color, or an indicator light, etc., which can be set specifically.

[0091] Therefore, rapid recovery is now advocated, and the hospital stay will be shorter. For patients who have undergone hip replacement or knee replacement, rehabilitation training is crucial to their recovery. Through a scientific and systematic rehabilitation training program, it can effectively promote wound healing, prevent adhesions, restore joint mobility and enhance muscle strength, and ultimately help patients return to normal life and work.

[0092] As an optional implementation scheme of the present application, optionally, the background server is also used for:

[0093] Storing the training action video uploaded by the patient;

[0094] Perform frame image processing on the training action video to obtain a frame image set consisting of a number of postoperative training action frame images:

[0095] S={S(1), S(1)...S(n)},

[0096] S(1), S(1)...S(n) are postoperative training action frame images of different frames, and S is a frame image set;

[0097] Real-time and orderly identification of each of the postoperative training action frame images in the frame image set, and training action analysis to determine whether the patient's training action is standardized:

[0098] If it is normal, traverse and identify the next postoperative training action frame image;

[0099] If it is not standardized, a corresponding training warning notification will be generated and sent to the patient-side APP.

[0100] After the backend receives the training action video uploaded by the patient through the APP, it can process the action video into a frame image set. The steps for processing the video into the corresponding frame image are as follows:

[0101] Method 1: Use professional software (such as Adobe Premiere Pro, Final Cut Pro, etc.)

[0102] 1. Import video: Open the video editing software and import the video file to be processed into the project.

[0103] 2. Locate the frame: Find the frame you want to export as an image on the timeline. You can locate it by playing the video and pausing it at a specific frame.

[0104] 3. Export frame: Most video editing software provides the function of exporting the current frame as an image. Usually, this can be achieved by right-clicking the frame on the timeline or using the export / save as image option provided by the software.

[0105] 4. Set export parameters: Set the format, resolution, quality and other parameters of the exported image as needed.

[0106] 5. Save the image: Select a save location and name the image file, then click Save.

[0107] Method 2: Use command line tools (such as FFmpeg)

[0108] 1. Install FFmpeg: If you haven’t installed FFmpeg yet, download it from the official website and install it.

[0109] 2. Open a command line interface: You can use Command Prompt or PowerShell on Windows, or Terminal on macOS or Linux.

[0110] 3. Run FFmpeg command: Use a command similar to the following to export specific frames or all frames in the video as images. For example, to export the first frame per second, you can use the "ffmpeg -iinput.mp4 -vffps=1output%d.png" command. Here "fps=1" means exporting 1 frame per second, and "output%d.png" is the file name template of the output image, where "%d" will be replaced by the frame number.

[0111] 4. View results: View the exported image files in the specified output directory.

[0112] Method 3: Use online tools

[0113] 1. Visit an online video to image converter: Search and visit a reliable online video to image converter in your browser.

[0114] 2. Upload video: Follow the tool's instructions to upload the video file that needs to be processed.

[0115] 3. Select settings: Select the frame rate, format and other settings for the exported image as needed.

[0116] 4. Start conversion: Click the Convert button to start processing the video.

[0117] 5. Download image: After the conversion is complete, download the generated image file to your local computer.

[0118] Therefore, the present invention can perform frame image processing on the training action video to obtain a frame image set consisting of several frames of postoperative training action frame images, and call the corresponding arthroscopic rotator cuff repair intelligent model through the model API, traverse (real-time and orderly identify) each of the postoperative training action frame images in the frame image set, and perform training action analysis to determine whether the patient's training movements are standardized.

[0119] The training and generation of the intelligent model of shoulder arthroscopy rotator cuff repair is described in the following. It can be deployed in the background in advance, or deployed on a third-party cloud server, and called through the model API (cloud API) (shoulder arthroscopy rotator cuff repair intelligent model).

[0120] As an optional implementation scheme of the present application, optionally, the background server is deployed with:

[0121] An electronic medical record system for recording the training instructions of the patient after arthroscopic rotator cuff repair surgery;

[0122] A postoperative education system, used for configuring and storing the postoperative education data corresponding to different training types;

[0123] A patient management system is used to manage patients who have undergone arthroscopic rotator cuff repair. According to the patient's training instructions, the postoperative education data of the corresponding training type is retrieved from the postoperative education system at regular intervals and sent to the patient-side APP through an API interface;

[0124] An image processing module is used to perform frame image processing on the training action video to obtain a frame image set consisting of a plurality of frames of postoperative training action frame images;

[0125] The intelligent model of shoulder arthroscopic rotator cuff repair is used to identify each postoperative training action frame image in the frame image set in real time and in an orderly manner, and to perform training action analysis to determine whether the patient's training action is standardized and output the corresponding judgment result.

[0126] Electronic medical record systems, such as HIS systems, can record patients' relevant medical information and build electronic medical record files for patients. Therefore, during the medical consultation process, doctors can log in to the backend and generate and record corresponding training instructions for patients in the electronic medical record system.

[0127] The postoperative education system pre-configures and stores postoperative education data of different training types. The postoperative education data mainly includes videos of postoperative rehabilitation exercises of different training intensities. Therefore, the subsequent patient management system can send training videos regularly according to the training orders of each patient. It can retrieve the postoperative rehabilitation training action videos regularly according to the training frequency in the training orders, and select the video data of the corresponding intensity to send to the patient APP where the patient is located.

[0128] As for the binding management of each patient-side APP and each patient, the administrator can bind them in the patient management system in advance.

[0129] Action videos require frame image processing. Referring to the above frame image operations, the corresponding frame image processing can be directly implemented in the image processing module. Subsequently, the image processing module sends the frame images of the patient's training action video to the model one by one. The model sequentially identifies whether the training actions in each frame image are standardized. If not, the model outputs the corresponding judgment results. The background generates corresponding alarm information based on the judgment results and sends it to the patient-side APP through the APP to remind the patient to pay attention to the current training actions.

[0130] When reminding the patient of the training process, the corresponding standardized movement features can be synchronously sent to the APP so that the patient can make corrections.

[0131] As attached Figure 3 Shown is the training flow chart of the intelligent model of rotator cuff repair under shoulder surgery.

[0132] Building a CNN (convolutional neural network) action recognition model to identify whether the user's action is standard can be done through the following steps:

[0133] 1. Data collection and preprocessing

[0134] 1) Collect data: First, you need to collect video or image datasets containing various user actions. These datasets should contain samples of standard and irregular actions with corresponding labels. For details, please refer to the attached Figure 2 The individual action image sets shown.

[0135] 2) Data preprocessing:

[0136] Perform frame extraction on video data and decompose the video into a series of image frames.

[0137] The image frames are normalized, cropped, scaled, etc. to ensure the consistency of the input data.

[0138] Data augmentation, such as rotation, flipping, scaling, etc., may also be required to increase the generalization ability of the model.

[0139] 3). Action characteristics

[0140] The action features on each action image can be extracted using feature engineering.

[0141] The methods for identifying and extracting action features from action images mainly involve computer vision and image processing technology. The following are some of the main methods and steps:

[0142] Image preprocessing

[0143] Grayscale conversion: Convert color images into grayscale images to reduce the amount of computation and eliminate the interference of color information on motion feature extraction.

[0144] Denoising: Use filters and other methods to remove noise from the image, improve image quality, and facilitate subsequent feature extraction.

[0145] Feature extraction

[0146] Appearance-based feature extraction

[0147] Color features: Although color features are not the main ones in action recognition, they still play a certain role in some scenarios (such as distinguishing sportswear of different colors).

[0148] Texture features: The texture features of the image are extracted through methods such as LBP (local binary pattern). These features are very helpful in describing the surface characteristics of the object.

[0149] Shape features: Shape features in images are obtained using edge detection, contour extraction, and other methods. These features are important for identifying actions with specific shapes (such as jumping and boxing).

[0150] Based on motion feature extraction

[0151] Optical flow method: extract motion features by analyzing the motion trajectory of pixels or feature points in the image. The optical flow method can capture the changes of pixels or feature points in the image over time, thereby reflecting the motion state of the object.

[0152] Background difference method: extract the motion area by comparing the difference between the current frame and the background frame. This method is simple and effective, but it is easily affected by factors such as lighting changes and dynamic background.

[0153] Inter-frame difference method: Extract motion information by comparing the differences between adjacent frames. Inter-frame difference method is effective for detecting fast-moving targets, but it is easy to produce "hole" phenomenon.

[0154] Feature Description and Coding

[0155] Feature descriptors: Use algorithms such as SIFT (Scale Invariant Feature Transform) and SURF (Speeded Up Robust Features) to describe the extracted features and generate feature descriptors. These descriptors have characteristics such as scale invariance and rotation invariance, and can better represent the features in the image.

[0156] Feature encoding: Encode the extracted feature descriptors for subsequent classification and recognition. Common encoding methods include Bag of Words (BoW) and Fisher Vector (FV).

[0157] Classification and identification

[0158] Classifier training: Use the extracted features to train a classifier (such as SVM, decision tree, random forest, etc.) to learn the difference between different actions.

[0159] Action recognition: Input the action image to be recognized into the trained classifier for classification and recognition. The classifier will output the corresponding action category as the recognition result.

[0160] Deep Learning Methods

[0161] With the development of deep learning, deep learning models such as convolutional neural networks (CNN) and recurrent neural networks (RNN) have been widely used in action recognition. These methods can achieve higher-level image analysis and understanding by automatically learning hierarchical features in images. For example, when using CNN for action recognition, the local features of the image can be extracted through the convolution layer, the calculation amount and overfitting problem can be reduced through the pooling layer, and finally the action classification can be achieved through the fully connected layer.

[0162] The specific feature engineering method can be selected by the administrator.

[0163] 2. Design CNN model

[0164] 1). Choose a network architecture: Choose a suitable CNN architecture, such as AlexNet, VGG, ResNet, etc. These architectures have achieved remarkable results in the field of image recognition.

[0165] 2). Define the network hierarchy:

[0166] Convolutional layer: used to extract features from images.

[0167] Pooling layer: used to reduce the spatial size of the feature map and reduce the amount of calculation.

[0168] Activation function: such as ReLU, used to increase the nonlinearity of the network.

[0169] Fully connected layer: Converts the feature map into the output vector required for classification.

[0170] 3). Determine hyperparameters: such as convolution kernel size, step size, padding, learning rate, etc.

[0171] 3. Model training

[0172] 1). Data division: Divide the dataset into training set, validation set and test set.

[0173] 2). Define the loss function: Choose a loss function suitable for the classification task, such as cross entropy loss.

[0174] 3). Select an optimizer: such as SGD, Adam, etc., to optimize model parameters.

[0175] 4). Training model: Use the training set data to train the model and update the network weights through the back-propagation algorithm.

[0176] 5). Verification and tuning: Use the validation set to evaluate model performance and perform model tuning as needed. You can use f1, AUC, accuracy curves, etc. to verify the model. Figure 4 As shown, it is the accuracy verification curve of the model of the present invention. In the verification of different times, through the optimization iterative training, the recognition accuracy of the model for action features is gradually improved.

[0177] 4. Model evaluation and testing

[0178] 1). Performance evaluation: Use the test set to evaluate the model’s accuracy, recall, F1 score and other performance indicators.

[0179] 2). Visual analysis: Analyze the model’s prediction results and error types through visualization tools for further improvement.

[0180] 5. Deployment and Application

[0181] 1). Model deployment: Deploy the trained model to actual application scenarios, such as smart monitoring systems, fitness applications, etc.

[0182] 2). Real-time recognition: extract and preprocess frames of real-time video streams, and then input them into the model for action recognition.

[0183] 3). Result feedback: Feedback the recognition results to the user or system for further processing or response.

[0184] Through the above steps, a CNN-based action recognition model can be built to identify whether the user's action is standard. It should be noted that the performance of the model is affected by many factors, including the quality of the data set, the choice of network architecture, the adjustment of hyperparameters, etc. Therefore, continuous optimization and improvement are required in practical applications. The specific iteration conditions are set by the administrator.

[0185] As an optional implementation scheme of the present application, optionally, the background server is also deployed with:

[0186] Backend database is used for backend data storage. You can use MYSQL database for backend data storage.

[0187] As an optional implementation scheme of the present application, optionally, the background server is also deployed with:

[0188] API interface, used for patient-side APP communication.

[0189] As an optional implementation scheme of the present application, optionally, the background server is also deployed with:

[0190] The model API is used to call the arthroscopic rotator cuff repair intelligent model.

[0191] The deployment of each API interface can be configured accordingly according to the application situation of the present invention.

[0192] Obviously, those skilled in the art should understand that the implementation of all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. Those skilled in the art can understand that the implementation of all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated as: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0193] Example 2

[0194] Based on the implementation principle of Example 1, the present application, on the other hand, proposes an application method of a postoperative rehabilitation education and tracking training system for shoulder arthroscopic rotator cuff repair, comprising the following steps:

[0195] The doctor logs into the backend server through the doctor's PC and writes the patient's training instructions into the electronic medical record system;

[0196] The patient management system reads and retrieves the postoperative education data of the corresponding training type from the postoperative education system regularly according to the patient's training instructions, and sends it to the patient's patient APP through the API interface;

[0197] The patient displays the postoperative education data regularly through the patient-side APP, and uploads the training action video trained according to the postoperative education data to the background server;

[0198] The backend server imports the training action video into the image processing module, performs frame image processing on the training action video, obtains a frame image set corresponding to a plurality of postoperative training action frame images, and imports the frame image set into the intelligent model of shoulder arthroscopic rotator cuff repair;

[0199] Through the intelligent model of shoulder arthroscopic rotator cuff repair, each of the postoperative training action frame images in the frame image set is identified in real time and orderly, and the training action analysis is performed to determine whether the patient's training action is standardized, and the corresponding judgment result is output:

[0200] If it is normal, traverse and identify the next postoperative training action frame image;

[0201] If it is not standardized, a corresponding training warning notification will be generated and sent to the patient-side APP.

[0202] Please understand it in conjunction with Example 1 for details.

[0203] The modules or steps of the present invention described above can be implemented by a general-purpose computing system, they can be concentrated on a single computing system, or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by a program code executable by a computing system, so that they can be stored in a storage system and executed by the computing system, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0204] Example 3

[0205] As attached Figure 5 As shown, further, in another aspect, the present application also proposes an electronic device, including:

[0206] processor;

[0207] a memory for storing processor-executable instructions;

[0208] Wherein, the processor is configured to implement the application method described in Example 2 when executing the executable instructions.

[0209] The electronic device of the embodiment of the present disclosure includes a processor and a memory for storing instructions executable by the processor, wherein the processor is configured to implement the application method described in the above embodiment 2 when executing the executable instructions.

[0210] Here, it should be noted that the number of processors can be one or more. At the same time, the electronic device of the embodiment of the present disclosure may also include an input system and an output system. Among them, the processor, memory, input system and output system may be connected through a bus or in other ways, which are not specifically limited here.

[0211] The memory is a computer-readable storage medium that can be used to store software programs, computer executable programs, and various modules, such as the program or module corresponding to the application method of the embodiment of the present disclosure. The processor executes various functional applications and data processing of the electronic device by running the software programs or modules stored in the memory.

[0212] The input system can be used to receive input numbers or signals. The signal can be a key signal related to user settings and function control of the device / terminal / server. The output system can include display devices such as display screens.

[0213] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A postoperative rehabilitation education and tracking training system for shoulder arthroscopic rotator cuff repair, characterized in that: include: A doctor-side PC is used to generate corresponding training instructions for a patient who has undergone arthroscopic rotator cuff repair and write the training instructions of the patient into an electronic medical record system, wherein the training instructions include the training frequency and training type of the patient; The backend server is used to execute the training doctor's orders and regularly send the postoperative education data of the corresponding training type to the patient's patient APP; The patient-side APP is used to regularly display the postoperative education data and upload the training action video of the patient when training according to the postoperative education data to the background server; The doctor-side PC and the patient-side APP are respectively connected to the backend server for communication.

2. The postoperative rehabilitation education and follow-up training system for shoulder arthroscopic rotator cuff repair according to claim 1 is characterized in that: The backend server is also used for: Storing the training action video uploaded by the patient; Perform frame image processing on the training action video to obtain a frame image set consisting of a number of postoperative training action frame images: S={S(1), S(1)...S(n)}, S(1), S(1)...S(n) are postoperative training action frame images of different frames, and S is a frame image set; Real-time and orderly identification of each of the postoperative training action frame images in the frame image set, and training action analysis to determine whether the patient's training action is standardized: If it is normal, traverse and identify the next postoperative training action frame image; If it is not standardized, a corresponding training warning notification will be generated and sent to the patient-side APP.

3. The postoperative rehabilitation education and follow-up training system for shoulder arthroscopic rotator cuff repair according to claim 1 is characterized in that: The backend server is deployed with: An electronic medical record system for recording the training instructions of the patient after arthroscopic rotator cuff repair surgery; A postoperative education system, used for configuring and storing the postoperative education data corresponding to different training types; A patient management system is used to manage patients who have undergone arthroscopic rotator cuff repair. According to the patient's training instructions, the postoperative education data of the corresponding training type is retrieved from the postoperative education system at regular intervals and sent to the patient-side APP through an API interface; An image processing module is used to perform frame image processing on the training action video to obtain a frame image set consisting of a plurality of frames of postoperative training action frame images; The intelligent model of shoulder arthroscopic rotator cuff repair is used to identify each postoperative training action frame image in the frame image set in real time and in an orderly manner, and to perform training action analysis to determine whether the patient's training action is standardized and output the corresponding judgment result.

4. The postoperative rehabilitation education and follow-up training system for shoulder arthroscopic rotator cuff repair according to claim 3 is characterized in that: The backend server is also equipped with: Backend database, used for background data storage.

5. The postoperative rehabilitation education and follow-up training system for shoulder arthroscopic rotator cuff repair according to claim 3 is characterized in that: The backend server is also equipped with: API interface, used for patient-side APP communication.

6. The postoperative rehabilitation education and follow-up training system for shoulder arthroscopic rotator cuff repair according to claim 3, characterized in that: The backend server is also equipped with: The model API is used to call the arthroscopic rotator cuff repair intelligent model.

7. An application method of the postoperative rehabilitation education and follow-up training system for shoulder arthroscopic rotator cuff repair according to any one of claims 1 to 6, characterized in that: The steps include: The doctor logs into the backend server through the doctor's PC and writes the patient's training instructions into the electronic medical record system; The patient management system reads and retrieves the postoperative education data of the corresponding training type from the postoperative education system regularly according to the patient's training instructions, and sends it to the patient's patient APP through the API interface; The patient displays the postoperative education data regularly through the patient-side APP, and uploads the training action video trained according to the postoperative education data to the background server; The backend server imports the training action video into the image processing module, performs frame image processing on the training action video, obtains a frame image set corresponding to a plurality of postoperative training action frame images, and imports the frame image set into the intelligent model of shoulder arthroscopic rotator cuff repair; Through the intelligent model of shoulder arthroscopic rotator cuff repair, each of the postoperative training action frame images in the frame image set is identified in real time and orderly, and the training action analysis is performed to determine whether the patient's training action is standardized, and the corresponding judgment result is output: If it is normal, traverse and identify the next postoperative training action frame image; If it is not standardized, a corresponding training warning notification will be generated and sent to the patient-side APP.

8. An electronic device, characterized in that include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the application method described in any one of claims 1 to 6 when executing the executable instructions.

Images