Intelligent orthopedic traction method, system and device

Abstract

The application belongs to the technical field of orthopedic traction and discloses an intelligent orthopedic traction method, system and device; the method comprises the following steps: collecting clinical data and DICOM images, extracting anatomical parameters through three-dimensional reconstruction, obtaining an initial scheme set through a parameter recommendation model in combination with the clinical data, performing traction with the initial parameters, collecting and preprocessing original signals to obtain biomechanical characteristics, inputting the fear analysis model to obtain a fear level, statistically analyzing the original signals after traction to obtain muscle fatigue results, evaluating the curative effect based on the original signals and daily fatigue results, optimizing a digital twin in combination with the biomechanical characteristics and the evaluation results, and obtaining an optimized traction scheme in combination with reinforcement learning; the application realizes the synergistic optimization of orthopedic traction treatment in physiological recovery and psychological care, and significantly improves the level of personalized treatment and the effect of rehabilitation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of orthopedic traction, more particularly, the present application relates to an intelligent orthopedic traction method, system and device. BACKGROUND

[0002] Orthopedic traction is a common method for treating orthopedic diseases such as fractures and dislocations. Traditional orthopedic traction methods have many shortcomings. For example, most orthopedic traction beds are manually controlled to adjust the traction force, which is not convenient for precise control of the force during the stretching process, and can easily cause secondary injury to the stretched part. The existing various traction methods use fixed supports with traction weights, which have potential risks of misalignment of the traction force line angle, and continuous static traction can easily increase the incidence of complications such as thrombosis and soft tissue injury. In addition, when medical staff perform orthopedic traction work, there is a lack of good auxiliary methods to improve the success rate and recovery condition before and after traction, and the recovery speed of the patient's affected limb cannot be guaranteed. Moreover, different symptoms require different traction forces, and medical staff need to spend a lot of effort to control the traction equipment.

[0003] Chinese patent application No. CN117860374A discloses an orthopedic traction auxiliary method and system based on artificial intelligence: S1: data classification: access the current hospital database, process the past orthopedic traction data, classify them by the limbs and causes of traction for subsequent treatment; S2: data reference: upload the patient's affected limb data, access the database, search and process it according to its cause, generate a three-dimensional graph of the uploaded affected limb condition, extract and compare the generated three-dimensional graph features with similar case features, and reference the data of similar case treatment and recovery after comparison; S3: processing of auxiliary scheme: generate images of the traction method according to the data of similar case treatment and recovery for medical staff to assist in viewing, including displacement distance of the traction device, spacing adjustment of the traction device, displacement direction of the traction device, and pressure control during use of the traction device. When the traction device is in use, the pressure value needs to be extracted and adjusted through experiments to determine whether the pressure is qualified or not, and the traction work of the patient's affected limb is carried out accordingly; S4: auxiliary nursing: medical staff log in to the system to process and record the nursing of the patient, including the patient's blood condition, temperature condition, skin condition, traction equipment condition, placement condition of the affected limb position, and complication prevention treatment; S5: measuring and processing the patient's affected limb: measuring the length of the patient's affected limb and healthy limb and comparing them to process the traction equipment according to the measurement, including adjusting the weight of the traction equipment and controlling the reset timing.

[0004] The above method can meet most scenarios, but research and actual application of the above method and prior art find that the above method and prior art at least have the following defects:

[0005] Psychological fear can cause abnormal muscle resistance and affect coordination. The above method does not consider the psychological fear of patients caused by treatment, so that the traction strategy cannot adapt to the psychology of patients, and the patient compliance and rehabilitation effect are reduced.

[0006] Therefore, the present application provides an intelligent orthopedic traction method, system and device to solve the above problems. SUMMARY

[0007] In order to overcome the above defects of the prior art, in order to achieve the above purpose, the present application provides the following technical scheme: an intelligent orthopedic traction method, comprising:

[0008] Collecting clinical data and DICOM images, performing three-dimensional reconstruction on the DICOM images, and extracting anatomical parameters; inputting the clinical data and the anatomical parameters into a parameter recommendation model to obtain an initial scheme set;

[0009] Using the parameters in the initial scheme set as initial execution parameters for traction, collecting an original signal set in the traction process, pre-processing the original signal set to obtain a signal set, analyzing the signal set to obtain biomechanical characteristics, inputting the biomechanical characteristics into a fear analysis model to obtain a fear level;

[0010] After each group of traction ends, statistically analyzing each group of original signal sets to obtain muscle fatigue results;

[0011] Efficacy evaluation is performed on the original signal set and the daily muscle fatigue results, and the digital twin is dynamically optimized based on the biomechanical characteristics and the evaluation results to obtain a dynamic twin, and a reinforcement learning algorithm is used to optimize the scheme to obtain an optimized traction scheme.

[0012] Further, the signal set includes left and right hand grip force signals, traction force signals and muscle group surface electromyography signals.

[0013] The method for obtaining biomechanical characteristics comprises:

[0014] The left and right hand grip force signals are averaged to obtain a double-hand average grip force signal;

[0015] The average value and the standard deviation of the double-hand average grip force signal in a preset period are calculated, and the ratio of the standard deviation to the average value is calculated to obtain a grip force fluctuation rate;

[0016] The time offset corresponding to the maximum value of the cross-correlation function of the double-hand average grip force signal and the traction force signal is calculated to obtain a response phase difference;

[0017] calculating the covariance of the average grip force signal of both hands and the traction force signal, and the standard deviation of the traction force signal within a preset time period, calculating the Pearson correlation coefficient of the average grip force signal of both hands and the traction force signal to obtain force coordination;

[0018] identifying the peak value of the average grip force signal of both hands in the phase where the traction force signal remains constant, and calculating the time required for the peak value to decay to half of the peak value to obtain relaxation delay time;

[0019] calculating the marginal probability density function of the average grip force signal of both hands, the marginal probability density function of the traction force signal, and the marginal probability density function of the average grip force signal of both hands and the traction force signal to obtain mutual information entropy; the marginal probability density function is calculated by kernel density estimation method;

[0020] performing wavelet packet decomposition on the muscle group surface electromyography signal to obtain B frequency bands, each frequency band corresponding to a different muscle group state, calculating the energy of each frequency band, and taking the normalized energy of the fatigue frequency band as the frequency band energy;

[0021] concatenating the grip force volatility, response phase difference, force coordination, relaxation delay time, mutual information entropy, and frequency band energy to obtain biomechanical features.

[0022] Further, the method for obtaining muscle fatigue results comprises:

[0023] extracting the peak values of the left and right hand grip force signals in the Kth set of original signals, and determining muscle fatigue when the peak values of the left and right hand grip force signals decay to M% of the initial maximum value;

[0024] calculating the mean value of the action peak values of the flexor muscle group surface electromyography signal in the previous F times, and combining the flexor muscle group adjustment factor to calculate the flexor muscle group fatigue threshold;

[0025] calculating the mean value of the action peak values of the adductor muscle group surface electromyography signal in the previous F times, and combining the adductor muscle group adjustment factor to calculate the adductor muscle group fatigue threshold;

[0026] if the flexor muscle group surface electromyography signal is not higher than the flexor muscle group fatigue threshold, determining that the flexor muscle group is fatigued;

[0027] if the adductor muscle group surface electromyography signal is not higher than the adductor muscle group fatigue threshold, determining that the adductor muscle group is fatigued.

[0028] Further, the method for evaluating the efficacy of the original signal set and the daily muscle fatigue result comprises:

[0029] calculating the average fear rating of the previous day and the average fear rating of the current day to obtain the fear improvement rate;

[0030] The muscle growth rate, including the flexor muscle group muscle growth rate and the adductor muscle group muscle growth rate, is calculated by taking the average of the left and right hand grip peak values of each group on the previous day and the average of the left and right hand grip peak values on the day;

[0031] The difference between the flexor muscle group muscle growth rate and the adductor muscle group muscle growth rate is calculated to obtain the muscle group recovery difference index;

[0032] The change curve of the traction force with the traction displacement is obtained, and the Kelvin-Voigt viscoelastic model is used for fitting; the muscle and ligament stiffness is calculated according to the Kelvin-Voigt viscoelastic model, and the muscle and ligament stiffness change is calculated by taking the average of the muscle and ligament stiffness on the previous day and the muscle and ligament stiffness on the day;

[0033] The fear improvement rate, muscle growth rate, muscle group recovery difference index, and muscle and ligament stiffness change are calculated to generate a treatment effect report.

[0034] Further, the method for obtaining a dynamic twin body comprises:

[0035] The ligament stiffness is corrected using the Hookean elastic model to obtain a corrected ligament stiffness;

[0036] The muscle stiffness is corrected according to the muscle viscoelastic coefficient, muscle growth rate, and muscle group recovery difference index to obtain a corrected muscle stiffness;

[0037] The traction force signal is time-discretized, and the change rate of the traction displacement with time is calculated; the fitting equation of the traction force signal with respect to the muscle viscoelastic coefficient is established based on the Kelvin-Voigt viscoelastic model, and the muscle and ligament viscoelastic correction coefficient is obtained by solving using the least squares method;

[0038] The digital twin body is updated according to the corrected ligament stiffness, corrected muscle stiffness, and muscle and ligament viscoelastic correction coefficient to obtain a dynamic twin body.

[0039] Further, the method for obtaining an optimized traction scheme comprises:

[0040] The state space is defined: composed of a state vector, the state vector including biomechanical characteristics, fear level, muscle fatigue result, fear improvement rate, muscle growth rate, muscle group recovery difference index, and stiffness change;

[0041] The action space is defined: composed of an action vector, the action vector including traction force increment, traction angle increment, and traction mode;

[0042] The reward function is designed: the reward function is obtained by weighted calculation of the safety reward, treatment effect reward, and comfort reward;

[0043] Initialize the state vector, randomly select the action vector, perform the traction simulation in the dynamic twin, obtain the next round state vector and the reward function value; generate experience data containing the current state vector, the selected action vector, the reward function value and the next round state vector, and store it in the experience replay pool;

[0044] The DQN model randomly samples W pieces of experience data from the experience replay pool, calculates the target action value function value and the current action value function value, minimizes the loss function, and updates the current network parameters through back propagation; the target action value function value is calculated according to the reward function value and the state of the dynamic twin When all action vectors The action value function value corresponding to the action vector with the maximum action value function value is calculated and obtained;

[0045] Using the trained DQN model to collect the state vector of the patient after the actual traction, calculating the action value function value of all action vectors, selecting the action vector with the maximum action value function value, and calculating the parameters corresponding to the optimized traction scheme.

[0046] Further, in the reward function, the safety reward is calculated according to the maximum stress of the ligament in the dynamic twin and the safety threshold of the human spinal ligament;

[0047] The efficacy reward is calculated according to the difference between the next round muscle growth rate predicted by the dynamic twin and the next round muscle recovery difference index predicted by the dynamic twin;

[0048] The comfort reward is calculated according to the current average fear level and the muscle viscoelastic coefficient before and after correction.

[0049] Further, the clinical data includes diagnosis results, age, weight and medical history.

[0050] Further, the method for obtaining the anatomical parameters comprises:

[0051] According to the size and resolution of the DICOM image, a three-dimensional voxel grid is constructed;

[0052] The entire three-dimensional voxel grid is traversed voxel by voxel; for each voxel, the gray values of its eight vertices are read and compared with the pre-determined bone gray threshold to determine the vertex state;

[0053] According to the vertex state of the voxel, the corresponding isosurface topology is found in the topology table, and the corresponding triangular facet model is generated;

[0054] After generating the preliminary triangular facet model, optimization processing is performed to obtain a three-dimensional reconstruction image;

[0055] The three-dimensional reconstruction image is taken as an input of a convolutional neural network to obtain a feature map, and the feature map is taken as an input of a segmentation model to obtain anatomical parameters; the anatomical parameters include intervertebral space height, Cobb angle of the spine and interarticular process joint spacing, and anatomical attachment point coordinates of adductor muscle groups and flexor muscle groups.

[0056] Further, the initial scheme set includes an initial traction scheme and an initial upper limb exercise scheme, the initial traction scheme includes an initial traction force, a traction angle, a traction time length and a traction mode; and the initial upper limb exercise scheme includes an exercise mode, an initial resistance and an exercise parameter.

[0057] An intelligent orthopedic traction system implements the intelligent orthopedic traction method, and comprises:

[0058] The acquisition and analysis module acquires clinical data and DICOM images, performs three-dimensional reconstruction on the DICOM images, and extracts anatomical parameters; the clinical data and the anatomical parameters are input into a parameter recommendation model to obtain an initial scheme set.

[0059] The fear analysis module performs traction with the parameters in the initial scheme set as initial execution parameters, acquires an original signal set in a traction process, analyzes the original signal set after preprocessing, obtains biomechanical characteristics, and inputs the biomechanical characteristics into a fear analysis model to obtain a fear level.

[0060] The fatigue analysis module statistically analyzes each group of original signal sets after each group of traction ends to obtain muscle fatigue results.

[0061] The traction optimization module performs efficacy evaluation on the original signal set and daily muscle fatigue results, dynamically optimizes a digital twin based on biomechanical characteristics and evaluation results to obtain a dynamic twin, and performs scheme optimization in combination with a reinforcement learning algorithm to obtain an optimized traction scheme.

[0062] An intelligent orthopedic traction device, a hoist rope is taken as a bearing basic component and is used for suspending a traction rope, one end of the traction rope is connected with the hoist rope through a rotating mechanism, and the other end is fixedly connected with a lifting ring gripper; a pressure detection film is attached to an outer surface of the lifting ring gripper; and the device implements the intelligent orthopedic traction method.

[0063] The intelligent orthopedic traction method, system and device provided by the application have the following technical effects and advantages:

[0064] The present application extracts anatomical parameters by collecting clinical data and DICOM images, and inputs the parameters into a recommended model to generate an initial scheme set, laying a personalized foundation for adapting to the physiological and psychological state of the patient; during traction, the original signal set is collected and preprocessed, and the biomechanical characteristics are analyzed to obtain the fear analysis model to quantify the fear level, breaking the limitations of existing methods that do not consider psychological fear, and accurately identifying the fear-muscle abnormality confrontation correlation; after each traction, the muscle fatigue result is determined by the peak value decay of grip strength and the electromyographic signal threshold, avoiding the misjudgment of muscle tension caused by fear as normal fatigue; based on the original signal and daily fatigue result, the fear improvement rate, muscle growth rate and other indicators are calculated to carry out efficacy evaluation, and the correlation between psychological and physiological recovery is clearly quantified; the dynamic twin body integrating the influence of fear is obtained by modifying the ligament / muscle stiffness and viscoelastic coefficient based on the Hook elastic model and the Kelvin-Voigt model, providing a precise virtual environment for strategy simulation; finally, the traction scheme is optimized by the state space containing fear-related indicators, the reward function considering comfort reward and the DQN model, generating a strategy that can adapt to the psychological fear state of the patient; effectively relieving the muscle abnormality confrontation caused by fear, improving the patient's cooperation degree and compliance, solving the problem that the existing method ignores psychological fear, leading to mismatch of traction strategy and poor rehabilitation effect, realizing the synergistic optimization of orthopedic traction treatment in physiological recovery and psychological care, and improving the level of personalized treatment and rehabilitation effect. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 It is an intelligent orthopedic traction method process schematic diagram of the present application;

[0066] Figure 2 It is an intelligent orthopedic traction method data flow schematic diagram of the present application;

[0067] Figure 3 It is an intelligent orthopedic split traction device schematic diagram of the present application;

[0068] Figure 4 It is an intelligent orthopedic gantry traction device schematic diagram of the present application;

[0069] Figure 5 It is an intelligent orthopedic traction system schematic diagram of the present application;

[0070] The meaning of the mark in the figure: 1-rotating mechanism; 2-traction rope; 3-hanging ring gripper; 4-pressure detection film; 5-hanging arm rope. DETAILED DESCRIPTION

[0071] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the protection scope of the present application.

[0072] Embodiment 1

[0073] Please refer to Figure 1 、 Figure 2 The present embodiment provides an intelligent orthopedic traction method, comprising the following steps:

[0074] Clinical data and DICOM images are collected, three-dimensional reconstruction is performed on the DICOM images, and anatomical parameters are extracted; the clinical data and the anatomical parameters are input into a parameter recommendation model to obtain an initial scheme set;

[0075] The clinical data includes diagnosis results, age, weight and medical history;

[0076] The method for obtaining the anatomical parameters comprises:

[0077] Each DICOM image is traversed in the order of CT scanning, and a three-dimensional voxel grid is constructed according to the size and resolution of the DICOM image;

[0078] The entire three-dimensional voxel grid is traversed voxel by voxel; for each voxel, the gray values of its eight vertices are read and compared with a predetermined bone gray threshold to determine the vertex state;

[0079] According to the vertex state of the voxel, the corresponding isosurface topology in the topology table is searched to generate a corresponding triangular facet model;

[0080] After generating the preliminary triangular facet model, optimization processing is performed to obtain a three-dimensional reconstruction image;

[0081] The three-dimensional reconstruction image is taken as the input of a convolutional neural network to obtain a feature map, and the feature map is taken as the input of a segmentation model to obtain anatomical parameters; the anatomical parameters include intervertebral space height, Cobb angle of the spine and joint process joint spacing, and anatomical attachment point coordinates of the adductor muscle group and the flexor muscle group.

[0082] The initial scheme set includes an initial traction scheme and an initial upper limb exercise scheme, the initial traction scheme includes an initial traction force, a traction angle, a traction time length and a traction mode; the initial upper limb exercise scheme includes an exercise mode, an initial resistance and exercise parameters.

[0083] By collecting clinical data including diagnosis results, age, weight and medical history, combining with anatomical parameters such as intervertebral space height and spinal Cobb angle extracted from three-dimensional reconstruction of DICOM images, inputting the parameters into the parameter recommendation model to generate an initial traction scheme covering initial traction force and traction angle, and an initial upper limb exercise scheme including exercise mode and initial resistance, the foundation is laid for subsequent dynamic optimization strategy combined with psychological fear factors from the source of precise and personalized scheme, which relieves the psychological fear that may be aggravated due to unsuitable scheme, reduces the risk of muscle abnormal resistance and affects the cooperation degree, and helps to improve patient compliance and rehabilitation effect.

[0084] The parameters in the initial scheme set are used as initial execution parameters for traction, and an original signal set in the traction process is collected. The original signal set is preprocessed to obtain a signal set. The signal set is analyzed to obtain biomechanical characteristics. The biomechanical characteristics are input into a fear analysis model to obtain a fear level.

[0085] The method for obtaining the signal set comprises:

[0086] All original signals in the original signal set are filtered, amplified and analog-digital converted to obtain digital signals. The digital signals are then given a unified timestamp to obtain the signal set.

[0087] The signal set comprises left and right hand grip force signals, traction force signals and muscle group surface electromyography signals. Figure 3 、 Figure 4 wherein, Figure 3 and Figure 4 The hanging ring handles 3 of the traction device are attached with pressure detection films 4 to detect the left and right hand grip force signals in real time. Each traction rope 2 suspended on the arm rope 5 is provided with a tension sensor to detect the traction force signals in real time. The end of each traction rope 2 and the arm rope 5 are each provided with a rotating mechanism 1 to manually adjust the rope length to adapt to different patients. The muscle group surface electromyography signals are collected by sensors directly attached to the patient's skin surface, which are not shown in the figure.

[0088] The method for obtaining the biomechanical characteristics comprises:

[0089] The left and right hand grip force signals are averaged to obtain the average grip force signals of both hands.

[0090] The average value and the standard deviation of the average grip force signals of both hands within a preset period, such as 30s, are calculated. The ratio of the standard deviation to the average value is calculated to obtain the grip force fluctuation rate.

[0091] The time offset corresponding to the maximum value of the cross-correlation function of the average grip force signals of both hands and the traction force signals is calculated to obtain the response phase difference. For example, the response phase difference wherein, is the mathematical expectation; is the time offset; is the traction force signal at time t; is the two-handed average grip force signal at time t;

[0092] The covariance of the two-handed average grip force signal and the traction force signal, and the standard deviation of the traction force signal within a preset time period, are calculated, the Pearson correlation coefficient of the two-handed average grip force signal and the traction force signal is calculated, and the force coordination is obtained; such as force coordination wherein, is the traction force signal and the two-handed average grip force signal ; is the standard deviation of the traction force signal within a preset time period; is the standard deviation of the two-handed average grip force signal within a preset time period;

[0093] In the stage where the traction force signal remains constant, the peak value of the two-handed average grip force signal is identified, and the time required for the peak value to decay to half of the peak value is counted to obtain the relaxation delay time;

[0094] The marginal probability density function of the two-handed average grip force signal, the marginal probability density function of the traction force signal, and the marginal probability density function of the two-handed average grip force signal and the traction force signal are calculated to obtain the mutual information entropy; the marginal probability density function can be calculated by kernel density estimation method; such as wherein, is the marginal probability density function of the two-handed average grip force signal and the traction force signal, is the two-handed average grip force signal; is the traction force signal; is the joint probability density function of and ; and are the marginal probability density functions of and , is the value of the two-handed average grip force signal, is the value of the traction force signal;

[0095] The surface electromyography signal of the muscle group is wavelet packet decomposed to obtain B frequency bands, each frequency band corresponds to a different muscle group state, such as 1-4Hz for fatigue signal and 20-50Hz for active contraction signal, the energy of each frequency band is calculated, and the normalized energy of the fatigue frequency band is taken as the frequency band energy; such as calculating the energy of each frequency band wherein, is the wavelet packet coefficient of the i-th frequency band; N is the number of sampling points in the window, and n is the index value of the number of sampling points in the window; the energy of each frequency band is normalized to obtain normalized energy, such as normalized energy wherein, is the normalized energy of the i-th frequency band, is the frequency band energy of the i-th frequency band; the normalized energy corresponding to the fatigue frequency band is selected as the frequency band energy;

[0096] The grip force fluctuation rate, response phase difference, force coordination, relaxation delay time, mutual information entropy and frequency band energy are spliced to obtain the biomechanical characteristics.

[0097] When the initial set of parameters in the initial scheme is used as the initial execution parameter to carry out traction, by collecting the original signal set containing the left and right hand grip forces, traction force and muscle surface electromyogram during traction, all original signals are first filtered, amplified, analog-digital converted and stamped with a uniform time stamp to obtain a synchronous signal set, and then key information is extracted from the signal set to analyze the biomechanical characteristics; the left and right hand grip forces are averaged to obtain a double-hand average grip force signal, the grip force fluctuation rate of the signal in a preset period, the response phase difference and force coordination with the traction force signal are calculated, the relaxation delay time is counted in the constant traction force stage, the mutual information entropy of the double-hand average grip force and the traction force signal is calculated by kernel density estimation method, and the frequency band energy of the muscle surface electromyogram signal is calculated by wavelet packet decomposition and normalized energy of the fatigue frequency band. After the biomechanical characteristics formed by splicing these features are input into the fear analysis model, the fear level is obtained. This process can accurately quantify the muscle abnormal resistance caused by psychological fear of patients, break through the limitation of existing methods that do not consider the psychological fear of patients during treatment, adapt the psychological state of patients based on the clear fear level for subsequent traction strategy adjustment, reduce the problem of reduced cooperation degree caused by fear, and thus improve the patient compliance and rehabilitation effect.

[0098] After each group of traction ends, statistical analysis is performed on each group of original signal sets to obtain muscle fatigue results;

[0099] The method for obtaining muscle fatigue results comprises:

[0100] The peak values of the left and right hand grip force signals in the Kth group of original signal sets are extracted, and when the peak values of the left and right hand grip force signals are attenuated to M% of the initial maximum value, muscle fatigue is determined;

[0101] The mean value of the action peak values of the flexor muscle surface electromyogram signals in the first F times is calculated, and the flexor muscle fatigue threshold is calculated by combining the flexor muscle adjustment factor;

[0102] The mean value of the action peak values of the adductor muscle surface electromyogram signals in the first F times is calculated, and the adductor muscle fatigue threshold is calculated by combining the adductor muscle adjustment factor;

[0103] If the flexor muscle group surface electromyogram signal is not higher than the flexor muscle group fatigue threshold, it is determined that the flexor muscle group is fatigued.

[0104] If the adductor muscle group surface electromyogram signal is not higher than the adductor muscle group fatigue threshold, it is determined that the adductor muscle group is fatigued.

[0105] After each set of traction, muscle fatigue results are obtained by statistical analysis of each set of original signal sets; the peak value of left and right hand grip force in the Kth set of original signal sets is extracted, and when the peak value decays to M% of the initial maximum value, muscle fatigue is determined, M can be determined by combining large sample clinical experiments and statistical analysis with clinical expert consensus, such as recruiting patients with different conditions and physical conditions for traction experiments, continuously monitoring the decay of left and right hand grip force peak value with time in each traction set, and recording the decay ratio of grip force peak value relative to the initial maximum value when muscle fatigue is clinically determined; then, statistical analysis is performed on these experimental data; finally, the specific percentage value of M is determined by combining the consensus of experts in the field of orthopedic rehabilitation on the muscle fatigue determination standard, to ensure that the threshold value can accurately reflect the physiological characteristics of muscle fatigue and adapt to the actual needs of clinical traction therapy; at the same time, the mean value of the action peak value of the surface electromyogram signal of the flexor muscle group and the adductor muscle group in the first F times is calculated, and the fatigue thresholds of the flexor muscle group and the adductor muscle group are obtained by combining the corresponding muscle group adjustment factors, and the specific muscle fatigue condition is determined by judging whether the electromyogram signals of the two types of muscle groups are higher than the corresponding thresholds. This process can accurately distinguish between muscle abnormalities caused by psychological fear and normal muscle fatigue: avoid misjudging muscle tension caused by fear as fatigue and blindly adjusting the traction load, and prevent the continuous application of uncomfortable load due to the failure to identify true muscle fatigue, which can help to more accurately adapt the traction strategy based on the fear level in the future, reduce the problem of strategy mismatch caused by signal misjudgment, further alleviate the fear of patients due to discomfort or misunderstanding, improve treatment compliance and adherence, and ultimately help to improve rehabilitation results, making up for the shortcomings of existing methods that do not consider psychological fear and are difficult to accurately adapt to patients' psychology.

[0106] The original signal set and the daily muscle fatigue results are evaluated for efficacy, and the digital twin is dynamically optimized based on the biomechanical characteristics and evaluation results to obtain a dynamic twin, and the scheme is optimized using a reinforcement learning algorithm to obtain an optimized traction scheme.

[0107] The method for evaluating the efficacy of the original signal set and the daily muscle fatigue results includes:

[0108] The average fear level of the previous day and the average fear level of the current day are statistically analyzed to calculate the fear improvement rate; if the fear improvement rate wherein, is the average fear level of the previous day; is the average fear level of the current day;

[0109] The average of the peak value of the grip strength of the left and right hands of each group of the previous day, and the average of the peak value of the grip strength of the left and right hands of the day, are calculated to obtain the muscle growth rate, including the flexor muscle group muscle growth rate and the adductor muscle group muscle growth rate; such as the flexor muscle group muscle growth rate , the adductor muscle group muscle growth rate , wherein, is the average of the peak value of the flexor muscle group of the current training cycle; is the average of the peak value of the flexor muscle group of the previous training cycle; is the average of the peak value of the adductor muscle group of the current training cycle; is the average of the peak value of the adductor muscle group of the previous training cycle;

[0110] The difference between the flexor muscle group muscle growth rate and the adductor muscle group muscle growth rate is calculated to obtain the muscle group recovery difference index;

[0111] The traction force signal and the corresponding traction displacement are obtained, and are fitted through the Kelvin-Voigt viscoelastic model; such as , wherein, is the traction force signal; is the muscle and ligament stiffness; is the traction displacement; is the muscle and ligament viscoelastic coefficient; is the rate of change of the traction displacement with time; the muscle and ligament stiffness is calculated according to the Kelvin-Voigt viscoelastic model, the muscle and ligament stiffness of the previous day and the muscle and ligament stiffness of the day are counted, and the muscle and ligament stiffness change is calculated; such as the muscle and ligament stiffness change , wherein, is the muscle and ligament stiffness of the day; is the muscle and ligament stiffness of the previous day;

[0112] The fear improvement rate, the muscle growth rate, the muscle group recovery difference index, and the muscle and ligament stiffness change are counted to generate a therapeutic effect report.

[0113] The method comprises the following steps:

[0114] The method for obtaining the dynamic twin body comprises the following steps:

[0115] The ligament stiffness is corrected by using the Hookean elastic model to obtain the corrected ligament stiffness; for example , wherein is the corrected ligament stiffness; is the ligament stiffness of the previous round in the dynamic twin body; is the ligament stiffness change;

[0116] The muscle stiffness is corrected according to the muscle viscoelastic coefficient, the muscle growth rate and the muscle group recovery difference index to obtain the corrected muscle stiffness, for example the corrected muscle stiffness , wherein is the muscle stiffness; is the muscle viscoelastic coefficient; is the muscle growth rate; is the muscle group recovery difference index;

[0117] The traction force signal is subjected to time discretization processing, and the change rate of the traction displacement with time is calculated; the fitting equation of the traction force signal about the muscle viscoelastic coefficient is established based on the Kelvin-Voigt viscoelastic model, and the least square method is used to solve to obtain the muscle and ligament viscoelastic correction coefficient; for example the traction force signal , wherein is the muscle and ligament correction stiffness; is the traction displacement; is the muscle and ligament viscoelastic correction coefficient; is the change rate of the traction displacement with time.

[0118] According to the correction of ligament stiffness, the correction of muscle stiffness and the muscle and ligament viscoelasticity correction coefficient, the digital twin is updated to obtain a dynamic twin.

[0119] By adopting the Hookean elastic model to correct the ligament stiffness combined with the change of ligament stiffness, the muscle stiffness is corrected according to the muscle viscoelasticity coefficient, the muscle growth rate and the muscle group recovery difference index, and then the traction displacement change rate is calculated after the time discretization of the traction force signal, the fitting equation is established based on the Kelvin-Voigt model and the muscle and ligament viscoelasticity correction coefficient is solved by the least square method, and finally the digital twin is updated by combining these corrected biomechanical parameters to obtain the dynamic twin, which can convert the muscle abnormal confrontation caused by psychological fear into quantifiable muscle and ligament stiffness and viscoelasticity changes and integrate them into the twin simulation: the dynamic twin can accurately reproduce the influence of muscle tension caused by fear on the traction process, avoiding the limitations of existing methods that cannot simulate the real physiological state due to not considering fear, and when subsequent traction strategy simulation optimization is based on the twin, the adaptation effect of different strategies on the fear-muscle confrontation associated state can be predicted in advance, providing precise virtual environment support for developing traction strategies that adapt to the psychological state of patients, thereby reducing the problem of muscle abnormal confrontation and decreased cooperation caused by the mismatch between the strategy and the fear state, and helping to improve patient compliance and rehabilitation effect.

[0120] The method for obtaining an optimized traction scheme comprises:

[0121] Defining a state space composed of a state vector, the state vector including biomechanical characteristics, fear level, muscle fatigue result, fear improvement rate, muscle growth rate, muscle group recovery difference index and muscle and ligament stiffness change;

[0122] Defining an action space composed of an action vector, the action vector including traction force increment, traction angle increment and traction mode;

[0123] Designing a reward function: the reward function is obtained by weighted calculation of safety reward, therapeutic effect reward and comfort reward;

[0124] The safety reward is calculated according to the simulated maximum stress of the ligament in the dynamic twin and the safety threshold of the human spinal ligament; for example, the safety reward is , wherein, is the simulated maximum stress of the ligament in the dynamic twin; is the safety threshold of the human spinal ligament;

[0125] The therapeutic effect reward is calculated according to the next round of muscle group growth rate predicted by the dynamic twin and the next round of muscle group recovery difference index predicted by the dynamic twin; for example, the therapeutic effect reward is , wherein, and The next round of muscle group growth rate predicted by the dynamic twin; The next round of muscle group recovery difference index predicted by the dynamic twin;

[0126] The comfort reward is calculated according to the current average fear level and the muscle viscoelastic coefficient before and after correction; if the comfort reward , wherein, The current average fear level is; And The muscle viscoelastic coefficient before and after correction;

[0127] Initialize the state vector, randomly select an action, perform traction simulation in the dynamic twin, obtain the next round of state vector and reward function value; generate experience data containing the current state vector, selected action, reward function value and next round of state vector, and store it in the experience replay pool;

[0128] The DQN model randomly samples W pieces of experience data from the experience replay pool, calculates the target action value function value and the current action value function value, minimizes the loss function, and updates the current network parameters through back propagation; the target action value function value is calculated according to the reward function value and the state of the dynamic twin , traverse all possible actions , so that the action value function value of the action corresponding to the action with the maximum action value function value is calculated;

[0129] Using the trained DQN model to collect the state of the patient after actual traction, calculating the action value function value of all actions, selecting the action with the maximum action value function value, and calculating the parameters corresponding to the optimized traction scheme.

[0130] In the process of obtaining the optimized traction scheme, the biomechanical characteristics, the fear level, the muscle fatigue result, the fear improvement rate, the muscle growth rate, the muscle group recovery difference index and the muscle and ligament stiffness change are included in the state space to ensure that the optimization process fully considers the correlation between psychological fear and physiological state; the traction force increment, the traction angle increment and the traction mode are taken as the action space to provide specific direction for strategy adjustment; a weighted reward function including safety reward, treatment reward and comfort reward is designed, wherein the comfort reward is calculated according to the current average fear level and the muscle group viscoelasticity coefficient after correction, so that the psychological feeling of the patient is considered when the scheme is optimized, and the defect that the existing method ignores the psychological factor is avoided; then, the state vector is initialized, the traction simulation is performed in the dynamic twin body in which the influence of fear is integrated to generate experience data and store it in the experience replay pool, the DQN model is used to sample the experience data to calculate the target and current action value function value, the loss function is minimized to update the network parameters, the model learns the adaptation rule of psychological fear-physiological response-traction action in the training, and finally the trained DQN model is used to select the action with the maximum action value function value in combination with the actual traction state of the patient to obtain the optimized parameters, so that the optimized traction scheme generated can accurately adapt to the psychological fear state of the patient; both the muscle abnormal antagonism caused by the fear aggravated by improper strategy and the patient's cooperation degree improved by considering the comfort can be improved, thereby effectively solving the problem of strategy mismatch caused by not considering psychological fear in the existing method, and the rehabilitation effect is ensured.

[0131] Embodiment 2

[0132] The embodiment provides a traction correction method, which comprises the following steps:

[0133] The initial traction force in the initial traction scheme is taken as the target traction force, the traction force signals of the preprocessed synchronous signal set are monitored in real time, and when the following conditions are met: if the difference between the traction force signal of the preprocessed synchronous signal set and the target traction force is higher than V% of the target value, and the duration exceeds H seconds, and the fear level is not higher than one level;

[0134] the traction force is determined to deviate, and a telescopic adjustment instruction is output according to the deviation direction.

[0135] Embodiment 3

[0136] Referring to FIG. 1 Figure 5 The embodiment provides an intelligent orthopedic traction system, which comprises the following parts:

[0137] The acquisition and analysis module acquires clinical data and DICOM images, performs three-dimensional reconstruction on the DICOM images, extracts anatomical parameters, inputs the clinical data and the anatomical parameters into a parameter recommendation model, and obtains an initial scheme set;

[0138] Fear Analysis Module: The module uses the parameters in the initial scheme set as the initial execution parameters for traction, collects the original signal set during the traction process, preprocesses the original signal set and then analyzes it to obtain biomechanical characteristics, and inputs the biomechanical characteristics into the fear analysis model to obtain the fear level.

[0139] Fatigue analysis module: After each traction session ends, statistical analysis is performed on the original signal set of each session to obtain muscle fatigue results;

[0140] Traction optimization module: It evaluates the efficacy of the original signal set and daily muscle fatigue results, dynamically optimizes the digital twin based on biomechanical characteristics and evaluation results to obtain a dynamic twin, and combines reinforcement learning algorithms to optimize the scheme and obtain an optimized traction scheme.

[0141] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent orthopedic traction method, characterized in that, include: Collect clinical data and DICOM images, perform three-dimensional reconstruction of the DICOM images, and extract anatomical parameters; Clinical data and anatomical parameters are input into the parameter recommendation model to obtain an initial set of treatment options. The parameters in the initial scheme set are used as the initial execution parameters for traction. The original signal set is collected during the traction process, the original signal set is preprocessed to obtain a signal set, the signal set is analyzed to obtain biomechanical characteristics, and the biomechanical characteristics are input into the fear analysis model to obtain the fear level. After each traction session, statistical analysis was performed on the original signal set of each session to obtain muscle fatigue results. The efficacy of the treatment was evaluated using the original signal set and daily muscle fatigue results. Based on biomechanical characteristics and evaluation results, the digital twin was dynamically optimized to obtain a dynamic twin. Furthermore, a reinforcement learning algorithm was used to further optimize the treatment plan, resulting in an optimized traction scheme. Define the state space: it consists of state vectors, which include biomechanical characteristics, fear level, muscle fatigue outcome, fear improvement rate, muscle growth rate, muscle group recovery difference index, and stiffness change. Define the motion space: it consists of motion vectors, which include traction force increment, traction angle increment, and traction mode; Design the reward function: The reward function is obtained by weighted calculation of safety reward, therapeutic effect reward and comfort reward; Initialize the state vector, randomly select the action vector, perform traction simulation in the dynamic twin, and obtain the state vector and reward function value of the next round; generate empirical data containing the current state vector, the selected action vector, the reward function value and the state vector of the next round, and store it in the experience replay pool; The DQN model randomly samples W empirical data points from the experience replay pool, calculates the target action value function and the current action value function, minimizes the loss function, and updates the current network parameters through backpropagation; the target action value function is determined based on the reward function value and the state of the dynamic twin. At that time, traverse all action vectors The action value function value corresponding to the action vector that maximizes the action value function value is obtained by calculation. The trained DQN model is used to collect the patient's state vector after actual traction, calculate the action value function value of all action vectors, select the action vector with the largest action value function value, and calculate the parameters corresponding to the optimized traction scheme.

2. The intelligent orthopedic traction method according to claim 1, characterized in that, The signal set includes left and right hand grip force signals, traction force signals, and surface electromyography signals of muscle groups; Methods for obtaining biomechanical characteristics include: The average grip force signals of the left and right hands are averaged to obtain the average grip force signal of both hands. Calculate the average and standard deviation of the average grip force signals of both hands within a preset time period, and calculate the ratio of the standard deviation to the average to obtain the grip force fluctuation rate; Calculate the time offset corresponding to the maximum value of the cross-correlation function between the average grip force signal and the traction force signal of both hands to obtain the response phase difference; Calculate the covariance of the average grip force signal and the traction force signal of both hands, as well as the standard deviation of the traction force signal within a preset time period. Calculate the Pearson correlation coefficient between the average grip force signal and the traction force signal of both hands to obtain force coordination. During the period when the traction force signal remains constant, identify the peak value of the average grip force signal of both hands, and calculate the time required for the signal to decay from the peak value to half of the peak value to obtain the relaxation delay time. The marginal probability density functions of the average grip force signal, the traction force signal, and the average grip force signal and traction force signal are calculated to obtain the mutual information entropy. The marginal probability density function is obtained by kernel density estimation. Wavelet packet decomposition was performed on the electromyographic signals on the surface of the muscle group to obtain B frequency bands. Each frequency band corresponds to a different muscle group state. The energy of each frequency band was calculated, and the normalized energy of the fatigue frequency band was taken as the frequency band energy. Biomechanical characteristics are obtained by splicing together grip force fluctuation rate, response phase difference, force coordination, relaxation delay time, mutual information entropy, and frequency band energy.

3. The intelligent orthopedic traction method according to claim 2, characterized in that, Methods for obtaining muscle fatigue results include: Extract the peak values ​​of the left and right hand gripping force signals from the Kth original signal set. When the peak values ​​of the left and right hand gripping force signals decay to M% of the initial maximum value, it is determined to be muscle fatigue. The mean value of the surface electromyographic signal of the flexor muscle group in the first F repetitions was calculated, and the fatigue threshold of the flexor muscle group was obtained by combining the flexor muscle group regulation factor. The mean value of the peak values ​​of the surface electromyographic signals of the adductor muscle group in the first F repetitions was calculated, and the fatigue threshold of the adductor muscle group was obtained by combining the adductor muscle group regulation factor. If the electromyographic signal on the surface of the flexor muscle group is not higher than the fatigue threshold of the flexor muscle group, it is determined to be flexor muscle group fatigue. If the electromyographic signal on the surface of the adductor muscle group is not higher than the fatigue threshold of the adductor muscle group, it is determined to be adductor muscle group fatigue.

4. The intelligent orthopedic traction method according to claim 3, characterized in that, Methods for evaluating the efficacy of raw signal sets and daily muscle fatigue results include: Calculate the fear improvement rate by comparing the average fear level of the previous day with the average fear level of the current day. The average peak grip strength of each hand in the previous day and the average peak grip strength of the left and right hands on the current day were calculated to obtain the muscle growth rate, including the growth rate of flexor muscles and the growth rate of adductor muscles. Calculate the difference between the growth rate of flexor muscles and the growth rate of adductor muscles to obtain the muscle recovery difference index; The curve of traction force versus traction displacement was obtained and fitted using the Kelvin-Voigt viscoelastic model. The muscle and ligament stiffness was calculated based on the Kelvin-Voigt viscoelastic model. The muscle and ligament stiffness of the previous day and the current day were statistically analyzed to calculate the change in muscle and ligament stiffness. The study statistically analyzes fear improvement rate, muscle growth rate, muscle group recovery difference index, and changes in muscle and ligament stiffness to generate a treatment report.

5. The intelligent orthopedic traction method according to claim 4, characterized in that, Methods for obtaining dynamic twins include: The ligament stiffness was corrected using the Hooke elasticity model to obtain the corrected ligament stiffness; The muscle stiffness is corrected based on the muscle viscoelasticity coefficient, muscle growth rate, and muscle group recovery difference index to obtain the corrected muscle stiffness. The traction force signal is discretized over time to calculate the rate of change of traction displacement over time. Based on the Kelvin-Voigt viscoelastic model, a fitting equation for the traction force signal with respect to the muscle viscoelastic coefficient is established, and the least squares method is used to solve for the viscoelastic correction coefficients of the muscle and ligament. The digital twin is updated based on the correction coefficients for ligament stiffness, muscle stiffness, and viscoelasticity of muscles and ligaments to obtain a dynamic twin.

6. The intelligent orthopedic traction method according to claim 1, characterized in that, In the reward function, a safety reward is calculated based on the simulated maximum stress of the ligaments in the dynamic twin and the safety threshold of the human spinal ligaments. The therapeutic reward is calculated based on the difference index between the muscle group growth rate predicted by the dynamic twin and the muscle group recovery rate predicted by the dynamic twin in the next round. Comfort bonuses are calculated based on the current average fear level and the muscle viscoelasticity coefficients before and after correction.

7. The intelligent orthopedic traction method according to claim 6, characterized in that, The clinical data includes diagnosis, age, weight, and past medical history.

8. The intelligent orthopedic traction method according to claim 7, characterized in that, Methods for obtaining anatomical parameters include: Each DICOM image is traversed sequentially according to the CT scan order, and a three-dimensional voxel mesh is constructed based on the size and resolution of the DICOM image. Traverse the entire 3D voxel mesh voxel by voxel; for each voxel, read the gray values ​​of its eight vertices and compare them with a pre-determined bone gray value threshold to determine the vertex state; Based on the voxel vertex state, the corresponding isosurface topology is found in the topology table, and the corresponding triangular patch model is generated. After generating a preliminary triangular patch model, optimization processing is performed to obtain a three-dimensional reconstructed image; The 3D reconstructed image is used as input to a convolutional neural network to obtain a feature map. The feature map is then used as input to a segmentation model to obtain anatomical parameters. The anatomical parameters include the intervertebral disc height, the Cobb angle of the spine, the facet joint spacing, and the coordinates of the anatomical attachment points of the adductor and flexor muscle groups.

9. The intelligent orthopedic traction method according to claim 8, characterized in that, The initial program set includes an initial traction program and an initial upper limb exercise program. The initial traction program includes initial traction force, traction angle, traction duration, and traction mode; the initial upper limb exercise program includes exercise mode, initial resistance, and exercise parameters.

10. An intelligent orthopedic traction system, implementing the intelligent orthopedic traction method according to any one of claims 1-9, characterized in that, include: Acquisition and Analysis Module: Acquires clinical data and DICOM images, performs 3D reconstruction of DICOM images, and extracts anatomical parameters; Clinical data and anatomical parameters are input into the parameter recommendation model to obtain an initial set of treatment options. Fear Analysis Module: The module uses the parameters in the initial scheme set as the initial execution parameters for traction, collects the original signal set during the traction process, preprocesses the original signal set and then analyzes it to obtain biomechanical characteristics, and inputs the biomechanical characteristics into the fear analysis model to obtain the fear level. Fatigue analysis module: After each traction session ends, statistical analysis is performed on the original signal set of each session to obtain muscle fatigue results; Traction optimization module: It evaluates the efficacy of the original signal set and daily muscle fatigue results, dynamically optimizes the digital twin based on biomechanical characteristics and evaluation results to obtain a dynamic twin, and combines reinforcement learning algorithms to optimize the scheme and obtain an optimized traction scheme.

Images