Rehabilitation exercise system and method based on intelligent interaction

Abstract

The invention discloses a rehabilitation exercise system and method based on intelligent interaction, and relates to the technical field of medical rehabilitation, intelligent management of basic information, medical history information, treatment information and medication information of patients is realized through a patient information management module; personalized exercise plans and real-time action guidance are provided for the patient through the rehabilitation exercise module; dynamically evaluating the rehabilitation state of the patient through a rehabilitation evaluation module; a friend communication platform is established through the community interaction module, and science popularization education is provided; and real-time monitoring and early warning of the doctor on the rehabilitation process of the patient are realized through the doctor supervision module. The artificial intelligence technology is adopted, the rehabilitation actions of the patient are monitored and corrected in real time through the camera, the rehabilitation state of the patient is evaluated through the electronic scale and the intelligent scoring system, the exercise plan is dynamically adjusted according to the rehabilitation condition of the patient, the rehabilitation exercise accuracy and the patient compliance are effectively improved, and the rehabilitation effect of the patient is improved. And the intelligent and electronic management of the whole rehabilitation exercise process can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of medical rehabilitation, and specifically to a rehabilitation exercise system and method based on intelligent interaction. Background Art

[0002] As an important part of the modern medical system, rehabilitation medicine plays a key role in clinical fields such as postoperative orthopedic recovery, comprehensive geriatric rehabilitation, pediatric rehabilitation, oncology rehabilitation, and sports injury rehabilitation. Through scientific and standardized rehabilitation exercises, the efficiency of patients' functional recovery can be improved, and at the same time, the incidence of complications can be reduced. With the acceleration of the global aging process and the continuous increase in the prevalence of chronic diseases, building an intelligent and precise rehabilitation service system has become an important direction for the development of medical technology.

[0003] In current clinical practice, the conventional rehabilitation training mode mainly adopts an independent training mechanism of "doctor-patient separation", which has significant technical limitations: (1) Lack of guidance: Patients lack real-time action correction by professional personnel in home or community training scenarios, and are prone to situations such as incorrect actions or inappropriate intensities, often making it difficult to achieve the expected exercise effect, and may even cause secondary injuries or delay the rehabilitation process. (2) Monitoring lag: Traditional evaluations rely on periodic outpatient follow-ups or medical staff phone follow-ups, lacking real-time monitoring and evaluation means, making it difficult for medical staff to timely grasp the patient's rehabilitation progress. (3) Adherence dilemma: The traditional one-way training mode has systematic deficiencies in dimensions such as psychological motivation (such as cultivating patients' self-efficacy), cognitive remodeling (structured transmission of disease knowledge), and behavior pattern formation (visual tracking of training data), resulting in a significant drop in adherence at key treatment nodes for a large number of patients due to "training isolation" and "fuzzy rehabilitation cognition", and this phenomenon is particularly prominent in the chronic disease rehabilitation group. (4) Rigidity of the plan: Existing training plans are mostly formulated based on static evaluations, lacking the ability to dynamically perceive and adaptively adjust to the patient's rehabilitation process, resulting in easy mismatch of phased training intensities, and thus the accuracy and effect of rehabilitation exercises are not fully guaranteed. Summary of the Invention

[0004] To solve the deficiencies mentioned in the above background art, the purpose of the present invention is to provide a rehabilitation exercise system and method based on intelligent interaction, realizing intelligent and electronic management of the entire process of rehabilitation exercise.

[0005] In the first aspect, the purpose of the present invention is achieved through the following technical solutions: A rehabilitation exercise system based on intelligent interaction, comprising:

[0006] A patient information management module, used to obtain and intelligently manage the patient's personal basic information, medical history information, treatment information, and medication information;

[0007] The rehabilitation exercise module is used to obtain the patient's initial rehabilitation status and provide an exercise plan and real-time action guidance according to the patient's rehabilitation status;

[0008] The rehabilitation assessment module is used to obtain the patient's real-time rehabilitation status and scientifically evaluate the patient's real-time rehabilitation status;

[0009] The community interaction module is used to pre-establish a communication platform for fellow patients and provide popular science education;

[0010] The doctor supervision module is used to collect the progress of the patient's rehabilitation process and conduct real-time monitoring and early warning on the progress of the patient's rehabilitation process.

[0011] Combined with the first aspect, in some implementation manners of the first aspect, the system further includes: the patient information management module includes a basic information management unit, a medical history information management unit, a treatment information management unit, and a medication information management unit.

[0012] Among them, the basic information management unit is used to manage the patient's personal basic information and the basic information of the emergency contact;

[0013] The medical history information management unit is used to manage the patient's medical history information and provide a reference when formulating a personalized exercise plan;

[0014] The treatment information management unit is used to manage the whole process information of the patient's medical intervention and completely record the clinical key data in the treatment preparation stage, implementation stage, and effect tracking stage;

[0015] The medication information management unit is used to manage the patient's medication information, provide corresponding drug instructions, and recommend suitable exercises according to the medication situation.

[0016] Combined with the first aspect, in some implementation manners of the first aspect, the system further includes: the rehabilitation exercise module includes an exercise plan generation unit, an action guidance unit, and an action detection unit;

[0017] Among them, the exercise plan generation unit is used to formulate and update a personalized exercise plan according to the patient's initial rehabilitation status and rehabilitation goals;

[0018] The action guidance unit is configured with a variety of rehabilitation exercise guidance videos and is used to provide real-time action guidance in the form of videos when the patient is performing rehabilitation exercises;

[0019] The action detection unit is used to real-time monitor the patient's rehabilitation actions through a camera, analyze the actions, compare the similarity with the standard actions, and give corrective suggestions using voice.

[0020] In combination with the first aspect, in certain implementations of the first aspect, the system further includes: The rehabilitation evaluation module includes an electronic scale unit and an intelligent scoring unit.

[0021] The electronic scale unit is used to provide diverse electronic evaluation scales for patients with different disease types and rehabilitation stages, and provide a filling interface to achieve scientific measurement and evaluation of the patient's rehabilitation status;

[0022] The intelligent scoring unit is used to perform intelligent analysis on the scoring items in the scale and give scores based on the scoring criteria of the scale.

[0023] In combination with the first aspect, in certain implementations of the first aspect, the system further includes: The community interaction module includes a health education unit, a patient communication unit, and a rehabilitation diary unit;

[0024] The health education unit is used to provide rehabilitation popular science knowledge for patients in the form of articles, animations, and videos. The rehabilitation popular science knowledge includes disease knowledge, rehabilitation knowledge, and exercise knowledge;

[0025] The patient communication unit is used to establish a patient communication platform to achieve communication and mutual assistance among patients, and use an exercise ranking and incentive mechanism to improve patient compliance;

[0026] The rehabilitation diary unit is used for patients to record and share their rehabilitation feelings and physical conditions every day.

[0027] In combination with the first aspect, in certain implementations of the first aspect, the system further includes: The doctor supervision module includes a rehabilitation progress tracking unit and a rehabilitation progress warning unit;

[0028] The rehabilitation progress tracking unit is used to monitor the patient's rehabilitation progress in real time, including the number of exercises, exercise time, exercise intensity, and exercise effect, generate an exercise report and summarize it to the doctor's end;

[0029] The rehabilitation progress warning unit is used to give warnings about the patient's rehabilitation progress, including abnormal behavior warnings and progress lack warnings. The abnormal behavior warnings and progress lack warnings will be fed back to the doctor's end in real time, enabling the doctor to communicate with the patient in a timely manner.

[0030] In combination with the first aspect, in certain implementations of the first aspect, the system further includes: The usage process of the motion detection unit:

[0031] The camera detects the patient's actions in real time; the recognition of human skeleton key points; the calculation of the human center line; the detection of the joint range of motion of each rehabilitation action; the analysis of motion characteristics; the comparison of similarity with standard actions; giving corrective suggestions based on the comparison results;

[0032] The recognition of the key points of the human skeleton adopts various methods, including but not limited to the Mediapipe framework, OpenPose, and HRNet;

[0033] The implementation steps of the Mediapipe framework are as follows:

[0034] Initialize Mediapipe Pose: Use the API provided by Mediapipe to initialize the MediaPipe Holistic model, configure the camera input source and output format, and set parameters, including the detection model and confidence threshold;

[0035] Capture the patient's action video: Capture the patient's video through the camera, and after processing with opencv, input each frame image into the Mediapipe Pose and Hands modules in sequence;

[0036] Identify and label the key points of the human body: The Mediapipe module processes each frame image, extracts the key points of the human body and the key points of the hands, labels the identified key points on the image, and records the coordinate information of each key point.

[0037] Combined with the first aspect, in some implementation manners of the first aspect, the system further includes: The joint angle calculation formula through the bone joint points is:

[0038]

[0039] where θ deg is the angle value between two vectors is the conversion relationship between radians and degrees, arccos is the inverse cosine function, x1; y1; x2; y2; x3; y3; x4; y4 are the coordinates of four different points P1, P2, P3, and P4 respectively;

[0040] The distance formula between joint points is:

[0041]

[0042] where d is the distance between two points, and x1; y1; x2; y2 are the coordinates of two points P1 and P2 respectively.

[0043] Combined with the first aspect, in some implementation manners of the first aspect, the system further includes: The community interaction module analyzes and scores the text in the discussions, replies, or chat exchanges initiated by users through a pre-established scoring model, and the process is as follows:

[0044] Select a text analysis model trained on a wide range of medical texts, such as Bio_ClinicalBERT, as the basic model;

[0045] Widely collect the text data of blogs, discussions, and replies actively posted by patients in the rehabilitation scenario, and perform data cleaning and desensitization;

[0046] Enhance the text data. On the one hand, expand the data set by using synonym replacement and syntactic structure transformation. On the other hand, add some open-source data in general scenarios;

[0047] Set the scoring range. The higher the score, the more positive it indicates, and vice versa. Label the data, and check the labeling quality based on the consistency of mutual evaluation;

[0048] On the basis of the basic model, by freezing some low-level parameters and only fine-tuning the high-level network, train a regression model for scoring, and use the cross-entropy loss function as the loss function. The calculation formula is:

[0049]

[0050] where c is the number of categories, y i is the true label, p i is the probability predicted by the model;

[0051] When the patient's content is posted in the community, use the model to analyze the content and calculate the original score I; at the same time, collect the interaction data of users on the content, such as the number of likes, comments, and forwards, and obtain the interaction index S through weighted average. Combine the original score I and the interaction index S, and use the reinforcement learning algorithm to update the model. The reward function formula is:

[0052] R = α * S + β * I

[0053] where both α and β are weight coefficients.

[0054] On the second aspect, in order to achieve the above purpose, the present invention discloses a rehabilitation exercise method based on intelligent interaction. The method includes the following steps:

[0055] Obtain the patient's personal basic information, medical history information, treatment information, and medication information;

[0056] Formulate a personalized exercise plan according to the patient's rehabilitation status and goals, and generate corresponding exercise guidance videos;

[0057] Guide the user to perform rehabilitation exercises, and use the camera to monitor the user's rehabilitation movements in real time, analyze the movements, compare the similarity with the standard movements, and give corrective suggestions;

[0058] Scientifically measure and evaluate the patient's rehabilitation status through an electronic scale;

[0059] Adjust the exercise plan according to the patient's rehabilitation status and goals, and generate new exercise guidance videos;

[0060] Through the patient community, publicize rehabilitation popular science knowledge and record rehabilitation feelings and physical conditions;

[0061] Monitor the patient's rehabilitation progress in real time, review the patient's exercise reports, receive early warnings of abnormal behavior and progress deficiencies of the patient, and communicate with the patient.

[0062] Advantages of the present invention:

[0063] By integrating modules of patient information management, intelligent rehabilitation exercise, rehabilitation assessment, community interaction and doctor supervision, the present invention realizes the whole-process intelligent management from patient data collection to exercise plan formulation, and then to real-time motion detection and feedback. Combining artificial intelligence technology, the system not only monitors the patient's movements in real time through a camera and compares them with standard movements, and gives voice correction suggestions in a timely manner, but also can dynamically adjust the exercise plan according to the patient's rehabilitation status and goals to ensure the scientific nature and personalization of the exercise. At the same time, through an electronic scale and an intelligent scoring mechanism, the rehabilitation effect is quantitatively evaluated, so that the rehabilitation progress can be accurately grasped, further improving the accuracy and safety of rehabilitation exercise. In addition, through the community interaction platform and the doctor supervision module, the present invention promotes the communication and mutual assistance among patients, and enables medical staff to monitor the patient's rehabilitation progress in real time and warn of abnormal situations, so as to provide intervention and guidance in a timely manner. With multi-level and all-round intelligent management means, the system not only effectively improves the patient's rehabilitation compliance and exercise effect, but also greatly reduces the risk of secondary injury, bringing a more convenient and efficient rehabilitation management solution to the medical rehabilitation field. Brief description of the drawings

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts;

[0065] Figure 1 It is a schematic block diagram of the structure of a rehabilitation exercise system based on intelligent interaction according to an embodiment of the present invention;

[0066] Figure 2 It is a schematic block diagram of the structure of a patient information management module in a rehabilitation exercise system based on intelligent interaction according to an embodiment of the present invention;

[0067] Figure 3 It is a flowchart of a rehabilitation exercise based on intelligent interaction according to an embodiment of the present invention;

[0068] Figure 4 It is a motion detection interface of a rehabilitation exercise based on intelligent interaction according to an embodiment of the present invention;

[0069] Figure 5 System measurement and manual measurement comparison chart for the embodiments of the present invention. Detailed implementation manners

[0070] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] Embodiment 1:

[0072] As Figure 1 shown, a rehabilitation exercise system based on intelligent interaction includes:

[0073] A patient information management module, configured to obtain and intelligently manage the patient's personal basic information, medical history information, treatment information, and medication information;

[0074] A rehabilitation exercise module, configured to obtain the patient's initial rehabilitation status and provide an exercise plan and real-time action guidance according to the patient's rehabilitation status;

[0075] A rehabilitation evaluation module, configured to obtain the patient's real-time rehabilitation status and scientifically evaluate the patient's real-time rehabilitation status;

[0076] A community interaction module, configured to pre-establish a communication platform for fellow patients and provide science popularization education;

[0077] A doctor supervision module, configured to collect the progress of the patient's rehabilitation process and perform real-time monitoring and early warning on the progress of the patient's rehabilitation process.

[0078] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the patient information management module is configured to intelligently manage the patient's personal basic information, medical history information, treatment information, and medication information. The patient information management module is mainly responsible for collecting, sorting, storing, and updating the patient's relevant basic information. The digital and automated management of this information not only makes data retrieval faster and updates more timely, but also greatly improves the security of information storage. Compared with the traditional paper management method, this module is more convenient and efficient, can provide perfect and accurate data support for doctors, and thus helps to make more accurate clinical decisions and formulate personalized rehabilitation plans. Figure 2 Schematic block diagram of the patient information management module in the rehabilitation exercise system based on intelligent interaction according to the embodiments of the present invention. As Figure 2 shown, the patient information management module includes a basic information management unit, a medical history information management unit, a treatment information management unit, and a medication information management unit.

[0079] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the basic information management unit is used to manage the personal basic information of the patient and the basic information of the emergency contacts. Among them, the personal basic information includes the patient's name, gender, date of birth, ID number, contact information, etc.; the basic information of the contacts includes the name and contact information of the emergency contacts, which are used to contact in time when an emergency occurs to the patient.

[0080] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the medical history information management unit is used to manage the medical history information of the patient, including current medical history, past medical history, family history, etc. These contents will be used as important reference information to provide reference when formulating a personalized exercise plan.

[0081] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the treatment information management unit is mainly used to centrally manage the whole-process clinical key data of the patient's previous treatments, including treatment plans, implementation processes, operation details, effect evaluations, and relevant imaging materials. It clearly shows the treatment process through a timeline and supports the comparison of effects at different stages. This unit integrates historical treatment data and imaging files, analyzes the correlation between treatment methods and rehabilitation effects, assists in formulating personalized rehabilitation plans, and at the same time ensures the integrity and traceability of treatment records, providing a reliable basis for dynamically adjusting the rehabilitation plan.

[0082] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the medication information management unit can store the patient's medication plan, including information such as medication type, medication time, dosage, medication method, and precautions. The medication plan can be automatically generated and updated according to the doctor's prescription, and the patient is reminded to take medicine on time regularly. The reminder methods include but are not limited to text messages, alarms, and push notifications from mobile applications, etc., to ensure that the patient can take the medicine correctly according to the doctor's advice. The patient can also query the detailed drug instructions through this module, including information such as the indications, contraindications, adverse reactions, and precautions of the drug, to ensure that the patient can have a full understanding of the drug. Particularly, the medication information management module will also record the patient's medication records, including medication time, dosage, and medication effects, and at the same time recommend appropriate exercises according to the medication situation to enhance the drug efficacy and accelerate the recovery of the body function.

[0083] In a specific embodiment of the present invention, assuming that a patient has taken a certain dose of central muscle relaxants orally as planned, 30 minutes after the patient takes the medicine, the medication management unit will push appropriate joint activity exercises and appropriately limit the intensity of the exercise items. In addition, this unit enables the intercommunication between the patient's drug use records and the relevant data of the rehabilitation assessment, providing a basis for the dynamic adjustment of the subsequent treatment plan.

[0084] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the rehabilitation exercise module is used to provide personalized exercise plans and real-time action guidance according to the patient's rehabilitation status. The rehabilitation exercise module is mainly responsible for formulating and updating personalized exercise plans according to the patient's rehabilitation status and rehabilitation goals, configuring a variety of rehabilitation exercise videos to provide real-time action guidance for the patient during the rehabilitation exercise process, and detecting, analyzing, and comparing the patient's rehabilitation actions in real time through a camera and giving corrective suggestions. This module is crucial for improving the patient's rehabilitation quality and optimizing the rehabilitation treatment process. The functions of the rehabilitation exercise module include an exercise plan generation unit, an action guidance unit, and an action detection unit.

[0085] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the exercise plan generation unit is used to design and timely update personalized exercise plans that meet the actual needs of patients according to the specific conditions of each patient (including but not limited to the patient's medical history information, treatment information, medication information, rehabilitation status, rehabilitation goals). In this way, for patients, the rehabilitation process can be made more targeted and scientific, improving the rehabilitation effect and enhancing the rehabilitation confidence. For medical staff, it can assist them in formulating precise rehabilitation exercise plans according to different patients, saving time and energy, and optimizing the work process.

[0086] In a specific embodiment of the present invention, assume that a patient undergoes hip replacement surgery, and the exercise plan generation unit plays the following roles: In the early postoperative rehabilitation stage (0 - 2 weeks), the goals of this stage are to relieve pain and swelling, promote wound healing, initially restore joint range of motion, and prevent complications. Ankle pump exercise, gluteal muscle contraction, and passive joint movement plans are formulated. For ankle pump exercise, when the patient lies flat, the lower limbs should be straightened, and the toes should be slowly hooked up, trying to point the toes towards oneself. When reaching the maximum limit, hold for 3 - 5 seconds, then slowly press down the toes. When reaching the maximum limit, hold for 3 - 5 seconds, and repeat this process. Each group consists of 20 - 30 times, and 3 - 4 groups are performed every day. For gluteal muscle contraction, when the patient is in the supine position, the buttocks muscles should be clamped and held for 5 - 10 seconds, then relaxed. Each group consists of 20 - 30 times, and 3 - 4 groups are performed every day. For passive joint movement, the patient should perform passive flexion and extension movements of the hip joint with the assistance of medical staff or family members. The range of motion should be gradually increased from small to large, and it is performed 2 - 3 times a day. In the functional recovery stage (2 - 6 weeks), the goals of this stage are to increase joint range of motion, enhance muscle strength, and improve balance ability. A joint movement exercise plan is formulated, including but not limited to hip joint flexion, extension, adduction, and abduction in the sitting or standing position. For strength training, isometric and isotonic contractions of the quadriceps femoris are performed. For isometric contraction of the quadriceps femoris, when the patient is in the supine position, the lower limbs should be straightened, the thigh muscles should be tightened, and the knee joint should be straightened as much as possible. Hold for 5 - 10 seconds, then relax. Each group consists of 20 - 30 times, and 3 - 4 groups are performed every day. In the strengthening training stage (6 - 12 weeks), the goals of this stage are to enhance muscle strength, improve joint stability, restore the ability to perform daily activities, and improve functional movement ability. Resistance training, walking training, and balance training plans are formulated. For resistance training, the patient uses an elastic band for resistance training, including but not limited to hip joint abduction and extension movements. Fix the elastic band on a stable object, put one end around the ankle, and perform resistance exercises in the corresponding directions. Each group consists of 10 - 15 times, and 3 - 4 groups are performed every day. For walking training, the patient should gradually increase the walking distance and speed, starting from a flat road surface and transitioning to a road surface with a certain slope. When walking, the correct posture should be maintained, with the head held high, the chest out, and the steps even. Each walk lasts for 20 - 30 minutes, and it is performed 1 - 2 times a day. For balance training, the patient performs balance training such as single-leg standing and closed-eye standing. Each time, it is held for 30 - 60 seconds, and each group consists of 3 - 5 times. It is performed 2 - 3 times a day. In the comprehensive recovery stage (12 weeks and later), the goals of this stage are to restore the ability to live and work normally and participate in daily sports activities. Daily recovery plans such as jogging, cycling, and swimming are formulated.

[0087] In a specific embodiment of the present invention, assume that a patient undergoes breast cancer surgery, and the exercise plan generation unit functions as follows: In the first stage of postoperative rehabilitation (1 - 7 days), the goal of this stage is to perform simple joint activities to prevent limb swelling, pain, and numbness while the surgical wound has not healed. For the exercise of clenching and extending fingers, the patient is required to lie flat and rest. The five fingers on the surgical side should be kept in a fully extended position, and then the fingers are forcefully clenched into a fist. The patient can also hold a rubber ball in the palm and hold the force for 2 seconds each time and then relax appropriately. In coordination with the wrist flexion movement, when the patient is in the fist - clenching state, the patient should perform forceful flexion and extension movements on the wrist, then clench the five fingers into a fist and keep the wrist rotating for one week. When performing the flexion and extension exercise of the elbow joint, keep the action of clenching the five fingers into a fist until the elbow is flexed to the shoulder and then keep it in the extended state. When sitting, perform deep breathing exercises, slowly inhale deeply and then exhale. Place both hands on the chest, and you can feel the movements of the chest rising and relaxing. After completing 1 - 2 sets of functional exercises, receive neck movement exercises. Keep the shoulder joint adducted and fixed, and practice the forward flexion, backward extension, etc. of the head. In the second stage (1 - 2 weeks), the goal of this stage is to start practicing the initial activities of the shoulder joint after the surgical incision skin flap has healed. Develop an exercise plan for shoulder joint adduction, forward flexion, and backward extension. The patient is required to slowly practice the inward and outward rotation movements of the shoulder joint on the horizontal plane. And practice the arm - rotating movement while standing. Let the upper limb hang down and move forward and backward with the shoulder as the axis for exercise. When practicing the arm - swinging movement, stand facing the wall and perform the finger - climbing - the - wall activity on the front and side. Control the elevation of the shoulder joint within 90 degrees to avoid excessive stretching. In the third stage (after 15 days), the goal of this stage is to appropriately increase the range of motion and further strengthen the functional exercise of the shoulder joint. Develop an exercise plan including actions such as swinging the arms, expanding the chest, rowing the arms, rotating the upper limb, and backward extension. In the fourth stage (chemotherapy intermission period), according to the patient's own recovery situation, when the limb function is weak, perform simple training such as rotating, backward extending, swinging the arms, and expanding the chest on the affected limb, 5 minutes each time, 3 - 5 times a day. When the limb function is strong, perform training such as washing the face and sweeping the floor with the affected limb, 2 times a day.

[0088] In the above - mentioned rehabilitation exercise system based on intelligent interaction, the action guidance unit is configured with a variety of rehabilitation exercise guidance videos, including but not limited to basic stretching videos, advanced strength training videos, and special joint range - of - motion practice videos designed for different diseases, different body parts, and different rehabilitation stages, which are used to provide real - time action guidance in the form of videos when patients are performing rehabilitation exercises. The aim is to enable patients to obtain professional and accurate rehabilitation action demonstrations and solve various action problems during the rehabilitation process. The clear action breakdown in the videos can enable patients to more intuitively and deeply understand the essentials of rehabilitation actions, greatly reducing the incidence of incorrect actions, so as to avoid poor rehabilitation effects or even secondary injuries to patients due to improper rehabilitation actions.

[0089] In a specific embodiment of the present invention, the hip joint mobility exercise video after hip replacement surgery includes, but is not limited to, flexion and extension, adduction, and abduction. The following is the specific content of the video. Hip joint flexion and extension: Stand steadily on both feet, adjust the distance between the feet to the same width as the shoulders, place the hands naturally on both sides of the waist, keep the chest and back straight and stretched, slowly lift the leg upward and forward to the maximum height that can be reached, pause slightly at the maximum height, and then slowly return to the original position. Repeat twice. Hip joint adduction: Stand steadily on both feet, adjust the distance between the feet to the same width as the shoulders, place the hands naturally on both sides of the waist, keep the chest and back straight and stretched, slowly move the leg to the opposite side of the body to the maximum height that can be reached, pause slightly at the maximum height, and then slowly return to the original position. Repeat twice. Hip joint abduction: Stand steadily on both feet, adjust the distance between the feet to the same width as the shoulders, place the hands naturally on both sides of the waist, keep the chest and back straight and stretched, slowly lift the leg outward and upward to the maximum height that can be reached, pause slightly at the maximum height, and then slowly return to the original position. Repeat twice.

[0090] In a specific embodiment of the present invention, the shoulder joint mobility exercise video after breast cancer surgery includes, but is not limited to, forward flexion, backward extension, and adduction. The following is the specific content of the video. Shoulder joint forward flexion: Stand steadily on both feet, adjust the distance between the feet to the same width as the shoulders, keep the body straight, let the arms hang naturally on both sides of the body with the palms facing inwards, slowly lift the arms upward and forward to the maximum height that can be reached, pause slightly at the maximum height, do not shrug the shoulders, and slowly return to the original position. Repeat twice. Shoulder joint backward extension: Stand steadily on both feet, adjust the distance between the feet to the same width as the shoulders, keep the body straight, let the arms hang naturally on both sides of the body with the palms facing inwards, slowly lift the arms upward and backward to the maximum height that can be reached, do not twist the body, and slowly return to the original position. Repeat twice. Shoulder joint adduction: Stand steadily on both feet, straighten the arms and let them hang naturally, then move them to the opposite side of the body, and slowly return to the original position. Repeat twice.

[0091] In the above-mentioned intelligent interaction-based rehabilitation exercise system, the motion detection unit includes: a camera to detect the patient's motion in real time; recognition of human body bone key points; calculation of the human body center line; detection of the joint mobility of each rehabilitation motion; motion feature analysis; comparison of similarity with standard motions; voice prompts to give corrective suggestions; and use of the detection system.

[0092] Among them, the camera detects the patient's motion in real time. Devices such as a mobile phone camera or a computer camera are used. The patient stands about 3 meters in front of the camera, and the whole body appears within the camera range. The patient performs rehabilitation exercises for the upper limbs, lower limbs, etc. The camera continuously captures and stores the patient's motions. The system calls the camera interface to capture the patient's motion video in real time. The captured video is converted into frame pictures through the opencv module, and the bone joint points are detected in sequence.

[0093] Furthermore, the recognition of human body bone key points adopts various methods, including but not limited to the Mediapipe framework, OpenPose, and HRNet.

[0094] Mediapipe is an open-source cross-platform machine learning framework that can efficiently process video and image data. Through the Pose module of Mediapipe, the real-world coordinate values (X, Y, Z) of 33 key points of the human body can be detected in real time, including the head, shoulders, elbows, wrists, hips, knees, and ankles, etc.; through the Hands module of Mediapipe, the real-world coordinate values (X, Y, Z) of 21 key points of both hands can be detected in real time, including the roots, joints, and ends of the index finger, middle finger, ring finger, little finger, and the roots, joints, and ends of the index finger and thumb, etc.

[0095] The Mediapipe framework is based on the Blaze-Pose algorithm, which combines a detector-tracker, a deep convolutional neural network, and a regression encoder network. It can efficiently and accurately identify human key points, thus providing real-time detection and guidance for the rehabilitation movements of patients. The detector-tracker is the preprocessing stage of the Mediapipe framework, mainly used to provide initial pose estimation and tracking. The detector is responsible for detecting the presence of the human body in the image. If a face is detected, it will use the position information of the face to calculate the vector formed by the midline of the hip bone and the face. This vector is used to correct the tilt angle of the human body and help determine the positions of other body parts, thus generating the initial pose estimation result. The tracker uses historical information and motion models to predict the position of the human body in the current frame. Through the tracker, the algorithm can continuously track the human pose in the video sequence, reducing the computational overhead required for re-detection in each frame. The deep convolutional neural network is the core part of the Mediapipe framework and consists of the following three parts: a heatmap prediction network, an offset network, and the combination of the heatmap and the offset map. The heatmap prediction network is used to output the heatmap of human key points. The heatmap is a probability distribution map, indicating the probability of each pixel point being a certain key point. The peak in the heatmap usually corresponds to the position of the key point. The offset network outputs an offset map, indicating the small offset of the key point relative to the predicted position in the heatmap. Due to the limited resolution of the heatmap, directly using the heatmap to locate the key points may have a certain error. The role of the offset map is to further refine the position of the key points to make it more accurate. By combining the predicted position in the heatmap with the corresponding offset map, the algorithm can more accurately locate the position of the key points. Specifically, the heatmap provides the rough position of the key points, while the offset map provides fine adjustment, thus obtaining more accurate key point coordinates. The regression encoder network is the last step of the Mediapipe framework, used to regress the key point positions to obtain the final key point coordinates. The regression encoder network receives the heatmap and offset map information from the deep convolutional neural network and further optimizes the position of the key points. Through the regression encoder, the algorithm can output more accurate key point coordinates, thus improving the accuracy of pose estimation.

[0096] The implementation steps of the Mediapipe framework are as follows:

[0097] Initialize the Mediapipe Pose module: Use the API provided by Mediapipe to initialize the MediaPipeHolistic model, configure the camera input source and output format. Set necessary parameters such as the detection model, confidence threshold, etc. to ensure the accuracy and real-time performance of key point recognition.

[0098] Capture the patient's action video: Capture the patient's video through the camera. After processing with opencv, each frame of the image is sequentially input into the Mediapipe Pose and Hands modules.

[0099] Identify and label the human body key points: The Mediapipe module processes each frame of the image, extracts 33 key points of the human body and 21 key points of the hand. Label the identified key points on the image and record the coordinate information of each key point.

[0100] Among them, to calculate the center line of the human body, according to the coordinates of the identified key points, calculate the center points of the shoulders and the center points of the hips of the human body respectively, and then comprehensively calculate the position of the center line of the human body.

[0101] Among them, the detection of the joint range of motion of each joint rehabilitation action is based on the skeletal key points obtained from the Mediapipe Pose and Hands modules. According to different rehabilitation actions, only select the minimum set of skeletal joint points related to the measurement of specific joint ranges of motion. Calculate the angles of the joints and the distances between the key points according to the actions for further analysis of the action characteristics.

[0102] When calculating the center point position of the human body, use the four joint points of the left shoulder, right shoulder, left hip, and right hip in the Pose module of Mediapipe to calculate the center points of the shoulders and the center points of the hips respectively, and draw the center line of the human body through the two points;

[0103] The skeletal joint point indices of the Mediapipe Pose and Hands modules used for detecting each part are shown in Table 1 below:

[0104] Table 1 Joint Index Table for Detection Parts

[0105] Part Meidapipe Pose Module Joint Index Meidapipe Hands Module Joint Index Hand 0,2,4,5,8,9,12,13,16,17,20 Wrist 13,14,15,16 20 Forearm 11,12,23,24 8,20 Shoulder 11,12,13,14,23,24 Elbow 11,12,13,14,15,16 Hip 11,12,23,24,25,26 Knee 23,24,25,26,27,28 Ankle 25,26,27,28,31,32

[0106] Furthermore, the formula for calculating the joint angle through the skeletal joint points is:

[0107]

[0108] Among them, θ deg is the angle value of the included angle between two vectors It is the conversion relationship between radians and degrees. For the arccos inverse cosine function, x1, y1, x2, y2, x3, y3, x4, and y4 are the coordinates of four different points P1, P2, P3, and P4.

[0109] Furthermore, the distance formula between joint points is as follows:

[0110]

[0111] Where d is the distance between two points, and x1, y1, x2, and y2 are the coordinates of two points P1 and P2 respectively.

[0112] Taking the detection of hand joint range of motion as an example, the opening and closing state of the palm is confirmed by selecting 11 joint points. The average value of the distances from the ends of the five fingers to the finger roots is calculated and compared with the average value of the distances from the metacarpophalangeal joints of the five fingers to the finger roots. If the average value of the distances from the ends of the fingers to the finger roots is greater than the average value of the distances from the metacarpophalangeal joints of the fingers to the finger roots, the palm is in an open state; if it is less, the palm is in a clenched state. Moreover, the greater the ratio, the greater the degree of palm opening, and the smaller the ratio, the greater the degree of palm clenching.

[0113] Among them, action feature analysis and similarity comparison include calculating the joint range of motion values of the patient and comparing them with the standard reference values to judge the patient's joint range of motion. The standard reference values are shown in Table 2 below:

[0114] Table 2 Standard Reference Values for Rehabilitation Actions

[0115]

[0116] Furthermore, the accuracy of the system is evaluated by comparing the detected joint range of motion values of the system with the standard values measured manually using an electronic goniometer. Algorithms including but not limited to dynamic time warping, Euclidean distance, and cosine similarity are used to compare the similarity with pre-recorded standard actions, calculate the similarity of the action sequence, and evaluate the accuracy of the patient's actions.

[0117] Among them, the voice prompt system generates corresponding voice prompt content according to the results of the similarity comparison. Through voice synthesis technology, the prompt content is converted into voice output and played to the patient in real time to guide them to improve their actions.

[0118] Among them, data storage and feedback store the data of each detection, including the joint angles, similarity scores, etc. of each detection action, in the database, and generate a detection report to be provided to the patient and medical staff to help them understand the rehabilitation progress and improvement direction.

[0119] Among them, the system uses but is not limited to PyQt6 to develop the user interface, implementing functions such as patient information entry, demonstration video playback, real-time video display, angle change curve, motion assessment, result display, and report storage. The detection interface is as follows Figure 4 : Select to detect the left hand or the right hand, click the start detection button, turn on the camera, start playing the demonstration video, click start recording and end recording, the camera captures the patient's video and detects the joint range of motion of this action through the detection function, and saves the detection results and detection pictures. Click the previous or next button to switch the detection action. When clicking the end detection button, the camera will be turned off, and all detection results will be output as a detection report.

[0120] In the hip joint range of motion exercise after hip replacement surgery, enter the patient information through the user interface, select the left hand or the right hand respectively, and then click start detection. The interface starts to play the demonstration video of the first action and the real-time captured picture of the patient's action by the camera at the same time. Complete the exercise action according to the action guidance and input the video and pictures into the detection module. In the detection module, select 4 joint points of the shoulder and hip and 1 joint point of the knee used for the center line drawn in the Pose module of Mediapipe. Through the shoulder center point, hip center point, hip joint point, and knee joint point, calculate the movement angles of hip flexion, extension, adduction, abduction and other actions through the joint angle calculation formula. After the detection is completed, the maximum joint range of motion will be calculated, the joint range of motion will be judged, and the corresponding pictures will be saved. According to the joint point coordinates of the detection action and the standard action, use the cosine similarity to calculate the similarity of the action and evaluate the accuracy of the patient's action. After the first action is detected, click the next button to start the second action exercise, and the process is the same as the first action. After all exercise actions are completed, click the end detection button to turn off the camera, and output the results of all exercise actions as a detection report for doctors' reference.

[0121] In the shoulder joint range of motion exercise after breast cancer surgery, the overall process is similar to that of the hip joint range of motion exercise, but different in the detection module. In the detection module, select 4 joint points of the shoulder and hip and 2 joint points of the shoulder and elbow used for the center line drawn in the Pose module of Mediapipe. Through the shoulder center point, hip center point, shoulder joint point, and elbow joint point, calculate the movement angles of shoulder abduction, adduction, forward flexion, backward extension and other actions through the joint angle calculation formula. Taking shoulder abduction as an example, through system measurement Figure 5 The shoulder joint range of motion is 154.94°. The standard value measured manually with an electronic goniometer is 159.05°. The two values are close and the system error is small. Through multiple measurements, the system error is about 5.80%.

[0122] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the rehabilitation assessment module is used to scientifically evaluate the patient's rehabilitation status. The basic principle is to digitize the commonly used scales in rehabilitation medicine, and use artificial intelligence technology to intelligently analyze the scoring items in the scale, and give a score according to the scoring criteria of the scale. Through this module, patients can complete rehabilitation assessments independently without relying on doctors, and the system can also dynamically adjust the exercise plan according to the patient's measured rehabilitation status. The rehabilitation assessment module includes an electronic scale unit and an intelligent scoring unit.

[0123] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the electronic scale unit has built-in multiple rehabilitation assessment scales, including but not limited to: Fugl-Meyer scale, WOMAC scale, Harris scale, Constant-Murley, etc. These scales are used to assess the patient's rehabilitation status, such as pain level, degree of functional impairment, quality of life, etc. The electronic scale module also provides a convenient filling interface, which is designed to receive the user's operating instructions, thereby realizing the automated management of the assessment process. This module can be adapted and optimized according to the specific needs of different question types to ensure the fluency and accuracy of the assessment process.

[0124] Specifically, this module uses an adaptive touch screen interface to realize the visual operation of the scale items, and all scoring items are equipped with voice-assisted prompts. For the self-assessment items of the patient's subjective symptom assessment and functional status in the rehabilitation assessment scale (such as the self-assessment items in the Barthel Index of Daily Living Ability), the patient directly selects the description that best suits his or her own situation by clicking on the touch screen, so that the patient can independently quantify the items such as his or her pain level, joint stiffness, and self-care ability. For items that require measurement of limb function (such as joint range of motion measurement), the patient is required to demonstrate the corresponding limb movements in front of the camera according to the voice prompts. Before officially recording the video, the system will play the corresponding sample video to help the patient or the patient's family clarify the measurement task requirements.

[0125] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the intelligent scoring unit is mainly responsible for processing the user's input and mapping it to the corresponding scoring items, and generating the scoring results according to the scoring criteria, which mainly includes the following steps:

[0126] Format, verify and clean the raw input data collected by the electronic scale unit to remove abnormal or missing data;

[0127] Map the preprocessed data into predefined scoring items;

[0128] Generate the scoring results according to the scoring criteria. Among them, for the measurement of joint range of motion, it is necessary to use the joint range of motion measurement function as described above based on the video recorded by the patient to obtain the movement angle of the target joint;

[0129] Transmit the generated scoring results to the rehabilitation evaluation module in real time to provide a basis for the subsequent adjustment of the rehabilitation plan of the system. And record the detailed process and log of each scoring, which is convenient for doctors to review and conduct subsequent data analysis.

[0130] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the community interaction module not only helps patients comprehensively understand and master rehabilitation theories and methods, improve their understanding and implementation ability of the rehabilitation plan, but also effectively relieves the psychological pressure of patients during the rehabilitation process and enhances their confidence and compliance by promoting the transmission of rehabilitation knowledge, communication among patients, and the active recording of the rehabilitation process by patients. These external methods have positive and profound significance in improving patients' mental health and promoting physiological recovery. The community interaction module includes a health education unit, a patient communication unit, and a rehabilitation diary unit.

[0131] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the health education unit stores a large amount of rehabilitation popular science knowledge collected and sorted from authoritative institutions such as the Chinese Rehabilitation Medicine Association, the World Health Organization, and the China Rehabilitation Research Center, including but not limited to forms such as articles, animations, and videos. This unit supports users to actively search and browse, and also supports personalized active recommendation by matching the disease type and rehabilitation stage of patients.

[0132] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the patient communication unit is used to provide a community for patient communication and interaction. Optionally, in an embodiment of the present invention, the patient communication unit includes but is not limited to the following functions: First, topic discussion and sharing: Patients can initiate discussions under specific topics and share their experiences, feelings, and experiences during the rehabilitation process. Other patients can comment and reply to the published discussions, and enhance the enthusiasm and sense of sharing of communication through mechanisms such as liking, rewarding, and collecting. Doctors can also actively invite patients with good replies to share their experiences actively. Second, chat groups: Provide instant messaging functions, support real-time interaction, and at the same time, groups can be created with specific diseases, rehabilitation stages, or interests as the theme, so that patients with the same needs can communicate more timely and deeply. Third, integral incentive and ranking list: Use the integral, badge, and ranking list system to encourage patients to continuously participate and share high-quality content by recording data such as the number of posts, patient activity, and rehabilitation check-ins, forming a positive incentive mechanism. Fourth, professional Q&A: Patients can post questions about rehabilitation, exercise, or health management and invite doctors, rehabilitation experts, or experienced patients to answer, forming the transmission and sharing of knowledge.

[0133] In particular, in a specific embodiment of the present invention, a scoring model is implemented to analyze and score the text in the discussions, replies or chat exchanges initiated by users. For the text analyzed as negative, the system will actively give users tips and psychological guidance to encourage them to face the rehabilitation process positively. If the negative degree is relatively serious, the text content will be marked as "not suitable for display" in the community, require manual review or even be blocked, and a doctor will be invited to intervene in the corresponding patients.

[0134] Specifically, on the open-source medical text analysis model, a medical text analysis dataset is used for fine-tuning to obtain a scoring model, and the interaction data of users is widely collected in the community to continuously improve the model using the reinforcement learning algorithm. The implementation of this model includes the following steps:

[0135] Select a text analysis model trained on a wide range of medical texts, such as Bio_ClinicalBERT, as the base model;

[0136] Widely collect text data such as blogs, discussions, and replies actively posted by patients in the rehabilitation scenario, and perform data cleaning and desensitization;

[0137] Enhance the text data. On the one hand, expand the dataset by using synonym replacement and syntactic structure transformation, and on the other hand, add some open-source data in general scenarios;

[0138] Set the scoring range to 0-10 points. The higher the score, the more positive it is, and vice versa. Label the data and check the labeling quality based on the consistency of mutual evaluation;

[0139] On the base model, by freezing some low-level parameters and only fine-tuning the high-level network, train a regression model for scoring, and use the cross-entropy loss function as the loss function. The calculation formula is:

[0140]

[0141] where C is the number of categories y i is the true label p i is the probability predicted by the model.

[0142] When the content of the patient is posted in the community, use the model to analyze the content and calculate the original score I; at the same time, collect the interaction data of users for this content, such as the number of likes, comments, and forwards, and obtain the interaction index S by weighted average. Combine the original score I and the interaction index S, and use the reinforcement learning algorithm to update the model. The reward function formula is:

[0143] R = α * S + β * I

[0144] where both α and β are weight coefficients.

[0145] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the rehabilitation diary unit provides a convenient interface for patients to record their subjective feelings and physical conditions during the rehabilitation process. Users can choose to share the diary content to the community or to share it with doctors so that doctors can better understand the patient's rehabilitation status.

[0146] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the doctor supervision module is mainly responsible for real-time monitoring and early warning of the patient's exercise progress. Doctors can use this module to learn about the patient's exercise progress in detail, so as to better understand the patient's rehabilitation status, communicate with the patient or the patient's family in time when abnormal conditions occur, and reduce the pressure of doctor follow-up. The doctor supervision module includes a rehabilitation progress tracking unit and a rehabilitation progress early warning unit.

[0147] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the rehabilitation progress tracking unit is mainly used to track the patient's rehabilitation status and rehabilitation process. In particular, the rehabilitation progress information will be updated in real time as the patient's rehabilitation process progresses. Among them, the rehabilitation status will be updated after the patient uses the rehabilitation assessment, and the rehabilitation process will be updated after the patient uses the rehabilitation exercise, including the number of exercises, exercise time, exercise intensity and exercise effect. Within a fixed period, the system will generate an exercise report based on the patient's rehabilitation status and rehabilitation process, and summarize it to the doctor's side.

[0148] In the above-mentioned rehabilitation exercise system based on intelligent interaction, the rehabilitation progress warning unit is used to warn the patient's rehabilitation progress. In the rehabilitation progress warning unit, the system can monitor the patient's various dynamic data during the rehabilitation exercise in real time, and judge whether abnormal behavior occurs by analyzing the patient's movement amplitude, balance state and posture changes, such as the patient falls, suddenly loses balance or suffers from severe dizziness, and other situations that may cause decreased movement coordination, body loss of control and joint sprains; at the same time, when the system detects that the patient has not exercised according to the scheduled plan for many consecutive days or the exercise frequency is significantly lower than the standard, it will automatically trigger a progress loss warning and notify the doctor in time so as to quickly intervene to adjust the rehabilitation plan to ensure the continuity and safety of the rehabilitation process.

[0149] Figure 3 FIG. 1 is a flowchart of a rehabilitation exercise based on intelligent interaction according to an embodiment of the present invention. Figure 3 As shown, the rehabilitation exercise method based on intelligent interaction includes the following steps:

[0150] Collect the patient's basic personal information, medical history information, treatment information and medication information;

[0151] Develop personalized exercise plans based on the patient's rehabilitation status and goals, and generate corresponding exercise guidance videos;

[0152] Guide the user to perform rehabilitation exercises, and use a camera to monitor the user's rehabilitation movements in real time, analyze the movements, compare the similarity with the standard movements, and give corrective suggestions using voice.

[0153] Scientifically measure and evaluate the patient's rehabilitation status through an electronic scale.

[0154] Adjust the exercise plan according to the patient's rehabilitation status and goals, and generate a new exercise guidance video.

[0155] Repeat the steps until the patient reaches the rehabilitation goal.

[0156] Optionally, in order to improve the patient's rehabilitation effect, the rehabilitation exercise method based on intelligent interaction in the embodiment of the present invention further includes: First, through the patient community, publicize rehabilitation popular science knowledge, record rehabilitation feelings and physical conditions, realize communication and mutual assistance among patients, and use an exercise ranking and incentive mechanism to improve the patient's compliance; Second, the doctor monitors the patient's rehabilitation progress in real time, consults the patient's exercise report, receives early warnings of abnormal behavior and progress lack of the patient, and communicates with the patient in a timely manner.

[0157] Through the patient community, publicize rehabilitation popular science knowledge, record rehabilitation feelings and physical conditions, realize communication and mutual assistance among patients, and use an exercise ranking and incentive mechanism to improve the patient's compliance.

[0158] The doctor monitors the patient's rehabilitation progress in real time, consults the patient's exercise report, receives early warnings of abnormal behavior and progress lack of the patient, and communicates with the patient in a timely manner.

[0159] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and is used to implement one or more instructions, specifically used to load and execute one or more instructions in the computer storage medium to implement the above method.

[0160] It should be further noted that, based on the same inventive concept, the present invention also provides a computer storage medium, on which a computer program is stored, and when the computer program is run by a processor, the above-mentioned method is executed. The storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0161] In the description of this specification, the description with reference to terms such as "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0162] The above shows and describes the basic principles, main features, and advantages of the present disclosure. Those skilled in the art of this industry should understand that the present disclosure is not limited by the above embodiments, and the above embodiments and the descriptions in the specification only illustrate the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure will have various changes and improvements, and these changes and improvements all fall within the scope of the present disclosure claimed.

Claims

1. A rehabilitation exercise system based on intelligent interaction, characterized in that: include: Patient information management module, used to obtain and intelligently manage the patient's basic personal information, medical history information, treatment information and medication information; The rehabilitation exercise module is used to obtain the patient's initial rehabilitation status and provide an exercise plan and real-time action guidance based on the patient's rehabilitation status; The rehabilitation assessment module is used to obtain the patient's real-time rehabilitation status and conduct a scientific assessment of the patient's real-time rehabilitation status; Community interaction module, used to pre-establish a patient communication platform and provide popular science education; The doctor supervision module is used to collect the progress of the patient's rehabilitation process and conduct real-time monitoring and early warning of the progress of the patient's rehabilitation process.

2. A rehabilitation exercise system based on intelligent interaction according to claim 1, characterized in that: The patient information management module includes a basic information management unit, a medical history information management unit, a treatment information management unit and a medication information management unit. Wherein, the basic information management unit is used to manage the patient's personal basic information and the basic information of the emergency contact; The medical history information management unit is used to manage the patient's medical history information and provide a reference when formulating a personalized exercise plan; The treatment information management unit is used to manage the whole process information of the patient's medical intervention, and fully record the key clinical data of the treatment preparation stage, implementation stage and effect tracking stage; The medication information management unit is used to manage the patient's medication information, provide corresponding medication instructions and recommend appropriate exercises based on medication conditions.

3. The intelligent interactive rehabilitation training system according to claim 1, characterized in that: The rehabilitation exercise module includes an exercise plan generating unit, an action guiding unit and an action detecting unit; Wherein, the exercise plan generating unit is used to formulate and update a personalized exercise plan according to the patient's initial rehabilitation status and rehabilitation goals; The action guidance unit is configured with a variety of rehabilitation exercise guidance videos, which are used to provide real-time action guidance in the form of videos when the patient is doing rehabilitation exercises; The motion detection unit is used to monitor the patient's rehabilitation movements in real time through a camera, analyze the movements, compare the similarity with standard movements, and give correction suggestions.

4. The intelligent interactive rehabilitation training system according to claim 1, characterized in that: The rehabilitation assessment module includes an electronic scale unit and an intelligent scoring unit; The electronic scale unit is used to provide a variety of electronic assessment scales for patients with different disease types and rehabilitation stages, and provide a filling interface to achieve scientific measurement and evaluation of the patient's rehabilitation status; The intelligent scoring unit is used to perform intelligent analysis on the scoring items in the scale and give a score based on the scoring criteria of the scale.

5. The intelligent interactive rehabilitation training system according to claim 1, characterized in that: The community interaction module includes a health education unit, a patient communication unit and a rehabilitation diary unit; The health education unit is used to provide patients with rehabilitation knowledge through articles, animations and videos, and the rehabilitation knowledge includes disease knowledge, rehabilitation knowledge and exercise knowledge; The patient communication unit is used to establish a patient communication platform to achieve communication and mutual assistance between patients, and use exercise rankings and incentive mechanisms to improve patient compliance; The rehabilitation diary unit is used for patients to record and share their rehabilitation feelings and physical conditions every day.

6. The intelligent interactive rehabilitation training system according to claim 1, characterized in that: The doctor supervision module includes a rehabilitation progress tracking unit and a rehabilitation progress early warning unit; The rehabilitation progress tracking unit is used to monitor the patient's rehabilitation progress in real time, including the number of exercises, exercise time, exercise intensity and exercise effect, generate an exercise report and summarize it to the doctor's end; The rehabilitation progress warning unit is used to warn the patient's rehabilitation progress, including abnormal behavior warning and progress loss warning. The abnormal behavior warning and progress loss warning will be fed back to the doctor in real time, so that the doctor can communicate with the patient in time.

7. The intelligent interactive rehabilitation training system according to claim 3, characterized in that: The use process of the motion detection unit: The camera detects the patient's movements in real time; identifies the key points of the human skeleton; calculates the center line of the human body; and detects the range of motion of each joint in rehabilitation movements; Movement feature analysis; Compare similarity with standard movements; give correction suggestions based on the comparison results; The recognition of the key points of the human skeleton is carried out in a variety of ways, including but not limited to the Mediapipe framework, OpenPose, and HRNet. The implementation steps of the Mediapipe framework are as follows: Initialize Mediapipe Pose: Use the API provided by Mediapipe to initialize the MediaPipe Holistic model, configure the camera input source and output format, and set parameters, including the detection model and confidence threshold; Capturing patient action videos: The patient videos captured by the camera are processed by opencv and each frame is input into the Mediapipe Pose and Hands modules in sequence; Identify and mark key points of the human body: The Mediapipe module processes each frame of the image, extracts the key points of the human body and the hand, marks the identified key points on the image, and records the coordinate information of each key point.

8. The intelligent interactive rehabilitation training system according to claim 7, characterized in that: The formula for calculating joint angles through bone joint points is: where θ deg is the angle between the two vectors, is the conversion relationship between radians and angles, arccos is the inverse cosine function, x1; y1; x2; y2; x3; y3; x4; y4 are the coordinates of four different points P1, P2, P3, P4 respectively; The distance formula between joint points is: Where d is the distance between two points, x1; y1; x2; y2 are the coordinates of the two points P1 and P2 respectively.

9. The intelligent interactive rehabilitation training system according to claim 5, characterized in that: The community interaction module analyzes and scores the text in the discussion, reply or chat initiated by the user through a pre-established scoring model. The process is as follows: Select a text analysis model trained on a wide range of medical texts, such as Bio_ClinicalBERT, as the base model; Extensively collect text data from blogs, discussions, and replies actively posted by patients in rehabilitation scenarios, and perform data cleaning and desensitization; Enhance text data by using synonym replacement and grammatical structure transformation to expand the data set, and add open source data for some common scenarios. Set a scoring range, where higher scores represent more positive and lower scores represent more negative. Label the data and check the quality of the labeling based on the consistency of peer reviews. On the basis of the basic model, by freezing some low-level parameters and only fine-tuning the high-level network, the regression model for scoring is trained, and the cross entropy loss function is used as the loss function. The calculation formula is: Among them, C is the number of categories, y i is the true label, p i The probability predicted by the model; When the patient's content is published in the community, the model is used to analyze the content and calculate the original score I. At the same time, the user's interaction data on the content, such as the number of likes, comments, and reposts, is collected, and the interaction index S is obtained by weighted average. The model is updated using the reinforcement learning algorithm based on the original score I and the interaction index S. The reward function formula is: R=α*S+β*I Among them, α, β and are weight coefficients.

10. A rehabilitation exercise method based on intelligent interaction, characterized in that: The method comprises the following steps: Obtain the patient's basic personal information, medical history information, treatment information, and medication information; Develop personalized exercise plans based on the patient's rehabilitation status and goals, and generate corresponding exercise guidance videos; Guide users to perform rehabilitation exercises, and use cameras to monitor users' rehabilitation movements in real time, analyze the movements, compare the similarity with standard movements, and give correction suggestions; Scientifically measure and evaluate the patient's recovery status through electronic scales; Adjust exercise plans based on the patient's recovery status and goals and generate new exercise instruction videos; Through the patient community, we spread knowledge about rehabilitation and recorded the feelings and physical condition of rehabilitation; Monitor the patient's rehabilitation progress in real time, review the patient's exercise report, receive early warnings of abnormal behavior and lack of progress, and communicate with the patient.

Images