Elbow joint fracture postoperative full-cycle quantitative rehabilitation method for rehabilitation robot

Abstract

The invention discloses a rehabilitation robot-oriented elbow joint fracture postoperative full-cycle quantitative rehabilitation method, which comprises the following steps of: analyzing to obtain a rehabilitation stage according to the clinical manifestation of a patient after an elbow joint fracture operation; if in the early rehabilitation stage, a reciprocating passive training method is adopted; if in the middle stage of rehabilitation, an active activity range training method is adopted; if in the later period of rehabilitation, an on-load muscle strength training method is adopted. The invention provides a complete cycle elbow joint fracture postoperative rehabilitation method, and provides a reciprocating passive training method, an active motion range training method and an on-load muscle force training method corresponding to early, middle and later rehabilitation stages respectively. The invention provides a rehabilitation robot-oriented elbow joint fracture postoperative full-cycle quantitative rehabilitation method, and solves the problems of blindness and experience of elbow joint fracture postoperative full-cycle rehabilitation training.

Description

Technical Field

[0001] The present invention belongs to the cross - technical field of robotics and rehabilitation medicine engineering, and specifically relates to a full - cycle quantitative rehabilitation method for elbow fractures after surgery for rehabilitation robots. Background Art

[0002] The elbow joint is a very important joint in the human body, assisting in more than 60% of hand movements, and involving 15 skeletal muscles in the upper limb of the human body. Relevant research shows that more than 178 million new fracture patients are added globally every year, of which more than 12 million are elbow fracture patients. Fracture patients need surgical reduction. If the postoperative rehabilitation is improper, complications such as muscle atrophy, adhesion of surrounding tissues, and joint stiffness are likely to occur. Therefore, the rehabilitation effect of elbow fractures directly affects the normal life and work of patients.

[0003] The treatment process of elbow fractures can be divided into three basic links: fracture reduction surgery, postoperative fixation, and rehabilitation training. Generally speaking, the postoperative rehabilitation training methods for elbow fractures are divided into two modes: passive training and active training. Among them, active training can be divided into unloaded active training and loaded active training. It has been proven that passive training methods can effectively reduce complications such as adhesion of surrounding tissues and joint stiffness around the elbow joint, and active training methods can guide patients to move actively, effectively reduce muscle atrophy, and improve muscle strength. However, the most commonly used current rehabilitation method for elbow joint motor function is for hospital physicians to drive the affected limb of the patient for rehabilitation exercise. This method lacks quantitative data, resulting in the lack of an accurate quantitative method for elbow joint rehabilitation training.

[0004] Investigating and summarizing the existing elbow joint rehabilitation methods based on rehabilitation robots, the vast majority of which are aimed at neurological rehabilitation and cannot achieve postoperative fracture rehabilitation. The main defects are as follows: The rehabilitation movement of the elbow joint is a single - degree - of - freedom rotation, lacking traction movement, and the rehabilitation effect on soft tissues such as muscles and ligaments near the elbow joint is very limited. For example, the elbow joint rehabilitation robot control method described in patent CN 119367174 A analyzes the motion intention based on IMU data and adjusts the spatial position and joint angle of the elbow joint. Another example is patent CN 118787531 A, which predicts the motion trend based on surface electromyogram signals, but neither has the traction effect on the soft tissues of the elbow joint, easily leading to problems such as joint stiffness, and is not suitable for postoperative fracture rehabilitation. Summary of the Invention

[0005] The present invention is proposed to solve the problems existing in the prior art, and its purpose is to provide a full - cycle quantitative rehabilitation method for elbow fractures after surgery for rehabilitation robots.

[0006] The technical solution of the present invention is: A full - cycle quantitative rehabilitation method for elbow fractures after surgery for rehabilitation robots, including the following steps:

[0007] A. Analyze the rehabilitation stage based on the clinical manifestations after the patient's elbow joint fracture surgery;

[0008] B. If in the early stage of rehabilitation, adopt the reciprocating passive training method;

[0009] C. If in the middle stage of rehabilitation, adopt the active range of motion training method;

[0010] D. If in the late stage of rehabilitation, adopt the loaded muscle force training method.

[0011] Furthermore, the rehabilitation exercises of the elbow joint in the reciprocating passive training method, active range of motion training method, and loaded muscle force training method include joint rotation movement and joint traction movement.

[0012] Furthermore, the reciprocating passive training method has the following specific process:

[0013] First, determine the maximum flexion and maximum hyperextension positions in the elbow joint rotation movement based on a goniometer;

[0014] Then, use the method of robot teaching to obtain the maximum joint traction displacement in the elbow joint traction movement;

[0015] Finally, perform reciprocating passive training based on the maximum flexion, maximum hyperextension positions, and maximum joint traction displacement.

[0016] Furthermore, the active range of motion training method has the following specific process:

[0017] First, based on the surface electromyogram signals of the biceps brachii and triceps brachii, offline train a neural network model that can predict joint angles and traction forces;

[0018] Then, use the neural network model to online predict the movement angles and traction forces of the patient's elbow joint;

[0019] Finally, perform active range of motion training based on the movement angles and traction forces.

[0020] Furthermore, the loaded muscle force training method has the following specific process:

[0021] First, simulate the muscle load of joint rotation based on the admittance control model;

[0022] Finally, perform loaded muscle force training based on the muscle load.

[0023] Furthermore, in the active range of motion training method, a method for continuously estimating joint angles and traction forces based on surface electromyogram signals is adopted.

[0024] Furthermore, the data acquisition and preprocessing algorithm in the continuous estimation method has the following specific process:

[0025] First, the collected sEMG signals are divided into short-time sliding windows;

[0026] Then, each channel of the sEMG signal is segmented;

[0027] After that, notch filters and band-pass filters are used to solve interference;

[0028] After that, normalization is performed using the maximum value of MVC;

[0029] Finally, the time-domain features of the surface electromyogram signals of two muscles are extracted, and normalization of the eigenvalue is performed.

[0030] Furthermore, the time-domain features include absolute average value, standard deviation, root mean square, and wavelength.

[0031] Furthermore, the process of segmenting each channel of the sEMG signal is as follows:

[0032] Each channel of the sEMG signal is segmented using a sliding window with a fixed sample size, and the interval of the sliding window is set at the same time.

[0033] The beneficial effects of the present invention are as follows:

[0034] The present invention provides a complete full-cycle rehabilitation method for elbow fractures after surgery, and proposes a reciprocating passive training method, an active range of motion training method, and a load-bearing muscle force training method corresponding to the early, middle, and late rehabilitation stages respectively.

[0035] The present invention proposes a full-cycle quantitative rehabilitation method for elbow fractures after surgery for rehabilitation robots, which solves the problems of blindness and empiricism in the full-cycle rehabilitation training for elbow fractures after surgery. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is a flowchart of the method of the present invention;

[0037] Figure 2 is a schematic diagram of the stage division after elbow fracture surgery in the present invention;

[0038] Figure 3 is a schematic curve diagram of the reciprocating passive training trajectory in the present invention;

[0039] Figure 4 is a flowchart of the acquisition and processing of electromyogram signals in the present invention;

[0040] Figure 5 is a schematic structural diagram of the neural network in the present invention;

[0041] Figure 6 is a schematic curve diagram of the active range of motion training trajectory in the present invention;

[0042] Figure 7 It is a schematic diagram of the curve of the trajectory of muscle force training with load in the present invention; Specific embodiments

[0043] Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and embodiments:

[0044] As Figures 1 to 7 shown, a full-cycle quantitative rehabilitation method for postoperative elbow joint fractures of a rehabilitation robot includes the following steps:

[0045] A. Analyze the rehabilitation stage based on the clinical manifestations after the patient's elbow joint fracture surgery;

[0046] B. If in the early stage of rehabilitation, adopt the reciprocating passive training method;

[0047] C. If in the middle stage of rehabilitation, adopt the active range of motion training method;

[0048] D. If in the late stage of rehabilitation, adopt the muscle force training method with load.

[0049] The rehabilitation movements of the elbow joint in the reciprocating passive training method, the active range of motion training method, and the muscle force training method with load include joint rotation movement and joint traction movement.

[0050] The reciprocating passive training method has the following specific process:

[0051] First, determine the maximum flexion and maximum hyperextension positions in the elbow joint rotation movement based on a goniometer;

[0052] Then, obtain the maximum joint traction displacement in the elbow joint traction movement by using the robot teaching method;

[0053] Finally, perform reciprocating passive training based on the maximum flexion, maximum hyperextension positions, and maximum joint traction displacement.

[0054] The active range of motion training method has the following specific process:

[0055] First, based on the surface electromyogram signals of the biceps brachii and triceps brachii, an offline trained neural network model capable of predicting joint angles and traction forces is obtained;

[0056] Then, use the neural network model to online predict the movement angles and traction forces of the patient's elbow joint;

[0057] Finally, perform active range of motion training based on the movement angles and traction forces.

[0058] The muscle force training method with load has the following specific process:

[0059] First, based on the admittance control model, simulate the muscle load during joint rotation;

[0060] Finally, based on the muscle load, conduct muscle force training with load.

[0061] In the active range of motion training method, a continuous estimation method for joint angle and traction force based on surface electromyogram (sEMG) signals is adopted.

[0062] The data acquisition and preprocessing algorithm in the continuous estimation method is as follows:

[0063] First, the collected sEMG signals are divided into short-time sliding windows;

[0064] Then, each channel of the sEMG signal is segmented;

[0065] Next, a notch filter and a band-pass filter are used to solve interference;

[0066] Next, normalization is performed using the maximum value of MVC;

[0067] Finally, the time-domain features of the sEMG signals of two muscles are extracted, and normalization of the eigenvalues is performed.

[0068] The time-domain features include absolute mean value, standard deviation, root mean square, and wavelength.

[0069] The process of segmenting each channel of the sEMG signal is as follows:

[0070] Each channel of the sEMG signal is segmented using a sliding window with a fixed sample size, and the interval of the sliding window is set.

[0071] Step A analyzes the rehabilitation stage through the clinical manifestations after the patient's elbow joint fracture surgery, and the specific process is as follows:

[0072] First, in the rehabilitation stage, the early stage is from 0 to 2 weeks after surgery. The clinical features are mainly local swelling, pain, muscle spasm, and the fracture site is still unstable and prone to re-displacement.

[0073] Then, the middle stage is from 3 to 6 weeks after surgery. The clinical features are mainly that the local soft tissue trauma has basically healed, fibrous tissue at the fracture end has increased and callus has formed, and the fracture site is relatively stable.

[0074] Finally, the late stage is from 7 to 12 weeks after surgery. The clinical features are mainly callus formation and the local pain has basically disappeared.

[0075] Specifically, the reciprocating passive training method provides a personalized rehabilitation trajectory for simulating rehabilitation training movements, and the motion parameters of the joint motor and the traction cylinder are customized.

[0076] Specifically, the reciprocating passive training method is as follows:

[0077] b1. Use a goniometer to measure the maximum flexion angle and maximum hyperextension angle of the patient's elbow joint rotation.

[0078] b2. The speed loop and position loop controllers of the joint motor adopt PID regulators to drive the patient's elbow joint to swing in a trigonometric function trajectory between the maximum flexion angle and the maximum hyperextension angle, and use an inertial measurement unit to achieve closed-loop control of speed and position.

[0079] b3. Use the admittance mode to determine the maximum traction displacement of the patient. The traction electric cylinder adopts a servo controller and moves in a trigonometric function trajectory between 0 mm and the maximum traction displacement.

[0080] More specifically, the rehabilitation training trajectory of the reciprocating passive training method is as Figure 3 shown. Send a control command every once in a while using a timer. The admittance mode formulas of the motor and the electric cylinder are respectively as in (1) and (2)

[0081]

[0082] where T θ (t) is the torque on the motor shaft, calculated by the force sensor at the end of the electric cylinder, is the angular acceleration, is the angular velocity, M θ is the inertia coefficient, B θ is the damping coefficient.

[0083]

[0084] where F cy (t) is the force on the electric cylinder, obtained by the force sensor at the end of the electric cylinder, is the acceleration of the electric cylinder movement, is the speed of the electric cylinder movement, M d is the inertia coefficient, B d is the damping coefficient.

[0085] Specifically, the active range of motion training method is as follows:

[0086] c1. Use a deep learning algorithm to construct a model between surface electromyogram signals, joint angles, and traction forces.

[0087] c2. After the surface electromyogram signal is collected and processed by the lower computer, it is sent to the upper computer via Bluetooth, and a pre-trained neural network is called to predict the joint angle and traction displacement.

[0088] Compared with using a motor equipped with a torque sensor at the elbow joint, this method has higher precedence and better continuity in predicting the movement effect.

[0089] Specifically, the load-bearing muscle force training method is as follows:

[0090] After the patient's elbow joint flexes / extends to the extreme position, the robot will apply a small resistance and oppose the patient at the extreme position to improve the patient's muscle strength.

[0091] The rehabilitation training trajectory of the load-bearing muscle force training is as Figure 7 shown. The formulas of the motor and the electric cylinder under the load-bearing muscle force training are as in (3) and (4)

[0092]

[0093] In the formula, T0 is the applied joint rotation torque load.

[0094]

[0095] In the formula, F0 is the applied joint rotation torque load.

[0096] Specifically, the neural network model based on deep learning in the method for continuously estimating joint angle and traction force based on surface electromyogram signals is as follows:

[0097] First, two neural networks are used to predict the elbow joint angle and traction force respectively, and the input signals are the time-domain features of two muscles.

[0098] Then, a deep learning algorithm is used to solve the problem of continuously estimating the elbow joint rotation angle and traction force. The schematic of the network model structure is as Figure 5 shown. It mainly includes a convolutional layer, an LSTM layer, an average pooling layer, and a fully connected layer;

[0099] Finally, the sliding window image of muscle activation is used as the input, and the corresponding elbow joint angle and traction force are used as the output.

[0100] In the present invention, the rehabilitation training movement of the elbow joint is flexion / extension movement and traction movement. The reciprocating passive training method determines the maximum flexion and maximum extension positions of the elbow joint based on a goniometer, and uses the robot teaching method to obtain the maximum traction displacement of the joint. The active range of motion training method is based on the surface electromyogram signals of the biceps brachii and triceps brachii, offline trains a neural network model that can predict the joint angle and traction force, and uses this model to online predict the movement angle and traction force of the patient's elbow joint. The load-bearing muscle force training method simulates the joint rotation load based on the admittance control model to train the patient's muscle force.

[0101] The present invention provides a complete full-cycle postoperative rehabilitation method for elbow fractures, and proposes a reciprocating passive training method, an active range of motion training method, and a load-bearing muscle strength training method corresponding to different rehabilitation stages.

[0102] The present invention proposes a full-cycle quantitative rehabilitation method for elbow fractures after surgery for rehabilitation robots, solving the problems of blindness and empiricism in the full-cycle rehabilitation training for elbow fractures after surgery.

[0103] The present invention can be directly applied to the field of robot-assisted rehabilitation engineering, reducing the probability of postoperative complications, greatly reducing the time cost and economic cost, and improving the rehabilitation quality.

Claims

1. A full-cycle quantitative rehabilitation method for elbow fracture surgery based on rehabilitation robots, characterized by: The following steps are involved: A. Analyze the patient's clinical manifestations after elbow fracture surgery to determine the patient's recovery stage; B. If you are in the early stages of rehabilitation, use reciprocating passive training methods; C. If you are in the middle stage of recovery, use active range of motion training; D. If you are in the late stage of recovery, use loaded muscle strength training methods.

2. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 1, characterized in that: The rehabilitation exercises of the elbow joint in the reciprocating passive training method, active range of motion training method, and loaded muscle strength training method include joint rotation exercises and joint traction exercises.

3. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 1, characterized in that: Reciprocating passive training method, the specific process is as follows: First, the maximum flexion and maximum hyperextension positions in elbow rotational movement were determined based on the joint protractor; Then, the robot teaching method is used to obtain the maximum joint traction displacement in the elbow traction movement; Finally, reciprocating passive training was performed based on maximum flexion, maximum hyperextension position, and maximum joint traction displacement.

4. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 1, characterized in that: Active range of motion training method, the specific process is as follows: Firstly, based on the surface electromyographic signals of the biceps and triceps, a neural network model capable of predicting joint angles and traction forces was obtained through offline training. Then, a neural network model is used to predict the patient's elbow joint motion angle and traction force online; Finally, perform active range of motion training based on movement angle and traction.

5. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 1, characterized in that: The specific process of loaded muscle strength training is as follows: First, the muscle load of joint rotation is simulated based on the admittance control model; Finally, perform loaded muscle strength training based on muscle load.

6. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 1, characterized in that: The active range of motion training method uses a continuous estimation method of joint angle and traction force based on surface electromyographic signals.

7. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 6, characterized in that: The data collection and preprocessing algorithm in the continuous estimation method is as follows: First, the acquired sEMG signals are divided into short-time sliding windows; Then, each channel of the sEMG signal is segmented; Then, use notch filters and bandpass filters to resolve interference; Then, the maximum value of MVC is used for normalization; Finally, the time domain features of the surface EMG signals of the two muscles are extracted, and the normalization of the feature values ​​is performed.

8. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 7, characterized in that: The time domain characteristics include absolute mean value, standard deviation, root mean square, and wavelength.

9. The full-cycle quantitative rehabilitation method for elbow fracture surgery oriented to a rehabilitation robot according to claim 1, characterized in that: Each channel of the sEMG signal is segmented as follows: Each channel of the sEMG signal was segmented using a sliding window with a fixed sample size while setting the interval of the sliding window.

Images