Intelligent medical rehabilitation physiotherapy instrument control system

Abstract

The invention discloses an intelligent medical rehabilitation physiotherapy instrument control system, which relates to the technical field of medical instruments, and comprises the following steps: acquiring physiological data of a patient, the physiological data comprising electromyographic signals, heart rate and body temperature, and performing real-time processing on an edge end to obtain physiological features; the edge calculation unit uploads the physiological features to a cloud end, a preliminary state decision model is constructed at the cloud end according to the physiological features to judge the preliminary state of a patient, actual physiotherapy records are grouped according to the preliminary state of the patient, effective physiotherapy records are counted, and an initial physiotherapy parameter set quantum optimization physiotherapy parameter is generated; and inputting the quantum optimization physiotherapy parameters into a digital twin model for secondary simulation, generating a muscle state quantum vector, calculating a Hilbert space distance, carrying out collaborative optimization on the quantum optimization physiotherapy parameters, obtaining optimal physiotherapy parameters, returning the optimal physiotherapy parameters to an edge end, and starting physiotherapy. Through centralized processing of the cloud, the model can be dynamically updated to adapt to demand changes of different patients.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and in particular to an intelligent medical rehabilitation therapy instrument control system. Background Art

[0002] In recent years, intelligent rehabilitation therapy systems have made significant progress in integrating edge computing, cloud-based modeling, and quantum optimization. Traditional systems process physiological data in real time at the edge (such as fractal denoising of electromyographic signals and extraction of heart rate and body temperature features), combine this with cloud-based decision-making models (such as lightweight decision tree algorithms) to generate initial therapy parameters, and utilize digital twin models (based on the OpenSim biomechanics framework) to simulate therapy effects, significantly improving parameter adaptability. The introduction of quantum computing further optimizes the efficiency of multi-objective parameter optimization. For example, by combining variational quantum circuits (VQEs) with the COBYLA algorithm, optimal therapy parameter sets and physiological characteristics are mapped into quantum state space to obtain the optimal solution.

[0003] However, existing technologies still have key bottlenecks: quantum optimization results are disconnected from the patient's real-time physiological state. Because the one-way optimization process lacks a dynamic verification mechanism, when the patient experiences sudden physiological changes (such as muscle spasms or abnormal heart rate), fixed parameters can easily lead to therapeutic deviations and even safety hazards. At the same time, the consistency between digital twin simulation results and actual physiological responses cannot be quantitatively guaranteed, and the system cannot autonomously trigger parameter corrections, which limits the accuracy and safety of personalized treatment. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent medical rehabilitation therapy instrument control system to solve the problem of dynamic matching of quantum optimized therapy parameters with the patient's real-time physiological state.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: The present invention provides an intelligent medical rehabilitation therapy instrument control system, which includes: A data processing module is used to obtain the patient's physiological data, including electromyographic signals, heart rate, and body temperature, and process the data in real time on the edge to obtain physiological characteristics; The initial parameter module is used by the edge computing unit to upload physiological characteristics to the cloud. In the cloud, a preliminary state decision model is constructed based on the physiological characteristics to determine the patient's initial state. The actual physical therapy records are grouped according to the patient's initial state, and the effective physical therapy records are counted to generate an initial physical therapy parameter set. The parameter optimization module is used to combine the patient's personal information and physiological characteristics to build a digital twin model of the patient, simulate and verify the physical therapy effect based on the initial physical therapy parameter set, and output the optimal physical therapy parameter set; The quantum optimization module is used to map the optimal therapy parameter set and physiological characteristics into quantum states, construct quantum circuits, and calculate the quantum optimized therapy parameters based on the multi-objective optimization principle; The collaborative optimization module is used to input the quantum optimized therapy parameters into the digital twin model for secondary simulation, generate the muscle state quantum vector, and collaboratively optimize the quantum optimized therapy parameters by calculating the Hilbert space distance. The optimal therapy parameters are obtained and transmitted back to the edge end to start therapy.

[0007] As a preferred solution of the intelligent medical rehabilitation therapy instrument control system of the present invention, physiological data is processed in real time at the edge, including denoising and feature value extraction.

[0008] As a preferred solution of the control system of the intelligent medical rehabilitation therapy instrument of the present invention, the denoising is specifically performed as follows: The electromyographic signals collected each time are evenly divided into several sub-segments; The fractal dimension calculation is performed on each sub-segment using the Higuchi algorithm; Based on the calculation results of fractal dimension, the fractal characteristics of the electromyographic signal are extracted and the denoised electromyographic signal is reconstructed.

[0009] As a preferred solution of the control system of the intelligent medical rehabilitation therapy instrument of the present invention, the specific steps of judging the initial status of the patient are as follows: Build and train a preliminary state decision model based on a lightweight decision tree algorithm; The physiological features uploaded by the edge are input into the trained preliminary state decision model to output the patient's preliminary state.

[0010] As a preferred solution of the control system of the intelligent medical rehabilitation therapy instrument of the present invention, the specific steps of generating the initial therapy parameter set are as follows: Build a rule base for physical therapy parameters; According to the patient's preliminary state output by the preliminary state decision model and combined with the physical therapy parameter rule library, the initial physical therapy parameter set is obtained.

[0011] As a preferred solution of the intelligent medical rehabilitation therapy instrument control system of the present invention, the specific steps of constructing the therapy parameter rule library are as follows: Integrate the actual physical therapy records of medical institutions and organize them into physical therapy record forms; These include the patient's physiological characteristics, patient diagnostic status, treatment parameters, and effect evaluation level during each treatment session; Use Python script to group physical therapy records by status label; Calculate the statistical value of each group of physical therapy parameters in the physical therapy record table, retain the physical therapy records of effective physical therapy, and generate a physical therapy parameter rule base.

[0012] As a preferred solution of the intelligent medical rehabilitation therapy instrument control system of the present invention, the specific steps of constructing the patient's digital twin model are as follows: Build a digital twin model of the patient based on the open source biomechanics framework OpenSim; Load the predefined human skeletal muscle template and adjust the digital twin model parameters based on the patient's basic information.

[0013] As a preferred solution of the control system of the intelligent medical rehabilitation therapy instrument of the present invention, wherein: the output of the preferred therapy parameter set is carried out in the following specific steps: Input the patient's physiological characteristics into the patient's digital twin model and simulate the muscle and nerve status of the treatment area through finite element analysis; Using the simulation engine MATLAB Simulink, the initial set of physical therapy parameters was used as input to simulate the physical therapy effect and calculate the comprehensive effect index; According to the comprehensive effect index of all physiotherapy parameter combinations, the physiotherapy parameter combinations with a comprehensive effect index greater than 0 are screened as effective physiotherapy parameter combinations; Then, the effective physiotherapy parameter combinations are compared with the real-time physiological characteristics to verify the applicability. All effective physiotherapy parameter combinations that meet the applicability requirements are collected to obtain the optimal physiotherapy parameter set.

[0014] As a preferred solution of the control system of the intelligent medical rehabilitation therapy instrument of the present invention, wherein: the quantum optimization therapy parameters are collaboratively optimized to construct a quantum circuit, the specific steps are as follows: The optimal physical therapy parameters are input into the digital twin model for secondary simulation to generate the muscle state quantum vector; Based on the physiological characteristics transmitted in real time, the quantum state encoding method in the optimal parameter module is reused to obtain the physiological characteristic quantum vector; Calculate the Hilbert space distance between the muscle state quantum vector and the physiological characteristic quantum vector; By comparing the preset distance threshold with the Hilbert space distance, a decision signal is output; Based on the decision signal, the quantum optimization therapy parameters are collaboratively optimized.

[0015] As a preferred solution of the control system of the intelligent medical rehabilitation therapy instrument of the present invention, wherein: the calculation of quantum optimization therapy parameters, the specific steps are as follows: Based on the principle of multi-objective optimization, a weighted objective function is constructed and calculated using quantum measurement methods; The COBYLA algorithm is combined with VQE to iteratively adjust the quantum circuit parameters to the highest value of the weighted objective function, thus obtaining the optimal quantum state physical therapy parameters. Using the quantum state decoding method, the optimal quantum state therapy parameters are mapped back to physical parameters to obtain quantum optimized therapy parameters.

[0016] The beneficial effects of the present invention are as follows: the present invention avoids the problem of incomplete information that may be caused by a single data source through physiological data collection and real-time processing, improves the reliability and accuracy of the data, and by introducing a lightweight decision tree algorithm, not only improves the prediction accuracy of the model, but also ensures the generalization ability and stability of the model through cross-validation and pruning optimization, establishes a "quantum optimization-biomechanical simulation" dynamic closed loop, uses quantum state space distance as a feedback signal, and realizes real-time adaptive adjustment of treatment parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 It is a module diagram of the control system of the intelligent medical rehabilitation therapy instrument in the embodiment.

[0019] Figure 2 Flowchart of the intelligent medical rehabilitation therapy device control method in the embodiment.

[0020] Figure 3 4 is a flowchart of physiological data denoising in an embodiment.

[0021] Figure 4 This is a flowchart of quantum optimization and collaborative verification in the embodiment. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0025] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides an intelligent medical rehabilitation therapy instrument control system, comprising the following steps: Data processing module: acquires the patient's physiological data, including electromyographic signals, heart rate, and body temperature, and processes them in real time on the edge to obtain physiological characteristics.

[0026] Through sensors (such as electromyography sensors, heart rate sensors, and body temperature sensors), the patient's physiological data is collected at a preset frequency, organized according to the collection timestamp format, and transmitted to the edge for real-time processing; Real-time processing on the edge: Each acquired EMG signal is evenly divided into several sub-segments. The Higuchi algorithm is used to calculate the fractal dimension of each sub-segment. Based on the fractal dimension calculation results, the fractal features of the signal are extracted and the denoised EMG signal is reconstructed. A noise separation method based on signal fractal reconstruction is used to remove noise from electromyographic signals. Specifically, each collected electromyographic signal is evenly divided into several sub-segments (decomposing long signals into short time segments to facilitate subsequent fractal dimension analysis while ensuring that the amount of edge computing is controllable). After the segmentation is completed, the Higuchi algorithm is used to perform fractal dimension calculation on each sub-segment. The calculation formula is as follows: ; in, is the fractal dimension, which indicates that the self-similarity value of the electromyographic signal usually ranges from , the EMG signal is usually close to , noise (such as white noise) is usually close to 1, is the time interval, which indicates the step length between signal sampling points. It is usually an integer and can be 2 here. Further, Indicates the electromyographic signal at time interval The length of the curve under is used to measure the complexity of the electromyographic signal at a specific time interval. The calculation formula is as follows: ; in, is the starting sampling point index, the value range is , which means calculating the length of the curve with different starting points. The index variable for internal summation, indicating the starting sampling point for calculation and time interval Under this condition, the counter of the difference accumulation process between signal sampling points is the total number of sub-segment sampling points, that is, the length of the sub-segment. Indicates rounding down, indicating the number of available intervals, is the amplitude value of the electromyographic signal at a certain sampling point, for example, Indicates that the electromyographic signal is at the index position The amplitude value, Similarly; Furthermore, based on the calculation results of the fractal dimension, the fractal characteristics of the signal are extracted and the denoised EMG signal is reconstructed; Specifically, a fast Fourier transform is performed on each collected electromyographic signal to convert the time domain signal into the frequency domain to obtain the electromyographic signal power spectrum, the electromyographic signal power spectrum is evenly divided into several frequency bands according to the number of sub-segments, the sub-electromyographic signal of each frequency band is extracted by bandpass filtering, the fractal dimension is calculated for each sub-electromyographic signal to obtain the fractal dimension range of the sub-electromyographic signal, the signal amplitude values ​​that are not within the fractal dimension range of the sub-electromyographic signal are set to zero to remove noise, and finally the electromyographic signal power spectrum with noise removed is reconstructed into the electromyographic signal by inverse Fourier transform; For example, the bandpass filter is designed using a 4th-order Butterworth filter. The cutoff frequency is set for the effective frequency range of the EMG signal (20 Hz to 150 Hz, based on clinical EMG signal spectrum analysis). The low-frequency cutoff is 20 Hz to remove baseline drift, and the high-frequency cutoff is 150 Hz to filter out power supply interference (50 Hz multiplication). The fractal dimension of the sub-EMG signal in each frequency band is set to 1.5 to 1.8 (based on the statistical mean of 1.65 and standard deviation of 0.1 from 1000 patient training data). The amplitude of frequency bands below 1.5 (such as white noise characteristics, close to 1) or above 1.8 (abnormal spikes) is set to zero. The inverse Fourier transform is performed by the DSP unit of the edge ARM Cortex-M7 (supporting FFT / IFFT libraries such as CMSIS-DSP). The denoised EMG signal power spectrum (256 points in the frequency domain) is input and the time domain signal (sampling rate 200 Hz, length 1 second, approximately 200 points) is output. The effectiveness of denoising was verified by calculating the signal-to-noise ratio (SNR) of the EMG signal before and after denoising; It should be noted that heart rate can be collected using infrared photoplethysmography (MAX30102 sensor). Its noise mainly comes from light interference, patient finger movements, and changes in sensor fit. However, this noise usually manifests as slow drift or slight jitter. For example, if the heart rate fluctuates from 80 bpm to 81 bpm, the measurement range of the heart rate signal is 30 (usually <1 bpm, accounting for <1%). The MAX30102 has a built-in low-pass filter (cutoff frequency approximately 5 Hz) and mean smoothing algorithm, which has initially suppressed noise at the hardware level. Body temperature is collected by a thermistor sensor (TMP36). The main noise comes from ambient temperature fluctuations (such as air flow) and changes in contact thermal resistance. However, these changes are extremely slow (<0.1°C / s), and high-frequency noise (such as power supply interference) is almost non-existent. Therefore, it is only necessary to perform denoising operations on the EMG signals that are significantly affected by noise; The root mean square (RMS) formula, standard deviation (HRV) formula and rate of change formula were used to calculate the characteristic values ​​of electromyographic signals, heart rate and body temperature to obtain physiological characteristics.

[0027] Initial parameter module: The edge computing unit uploads the physiological characteristics to the cloud, where it builds a preliminary state decision model to determine the patient's initial state based on the physiological characteristics. The actual physical therapy records are grouped according to the patient's initial state, and the effective physical therapy records are counted to generate an initial physical therapy parameter set.

[0028] Upload physiological characteristics to the cloud; Specifically, 24 bytes of data are transferred to the sending buffer through DMA (direct memory access), and the physiological characteristics are encapsulated into a data packet (the header is the device ID "RX001", the timestamp "2025-03-21 10:00:02.012", the data length, and the body is the characteristic data). The preset 256-bit key and initialization vector are loaded to encrypt the body of the data packet, and the header remains in plain text to obtain an encrypted data packet. Using the 5G network, a TCP / IP connection is established. The edge sends the encrypted data packet, the cloud replies with SYN / ACK, the edge confirms ACK, and the cloud returns a confirmation frame after receiving it. The encrypted data packet is decrypted using the same key and IV in the cloud, and the data integrity is verified to obtain the physiological characteristics. Organize patients’ historical physical therapy records into a training set for the preliminary state decision model; Each set of data in the training set includes physiological characteristics and corresponding state labels, and the state labels include normal, mild fatigue, moderate fatigue, and severe fatigue; For example, the training set is derived from the rehabilitation records of 1,000 patients. Each patient was collected 5 times a day (2 hours apart) for a total of 3 months, with a sample size of approximately 150,000 groups (1,000×5×90). Each group of data includes physiological characteristics (RMS, HRV, temperature change rate) and state labels (normal, mild fatigue, moderate fatigue, severe fatigue), which are annotated by rehabilitation physicians based on patient feedback and tests. The preprocessing steps include: (1) filling in missing values. If RMS is missing (accounting for about 0.5%, sensor failure), it is filled with the patient's average value of the day (such as 2.0mV), which takes 0.5 seconds; (2) removing outliers. RMS>5mV or HRV>50ms is considered invalid (accounting for 0.2%), and the corresponding records are deleted, which takes 0.3 seconds; (3) using the Z-score method for standardization. After preprocessing, the data is stored as an HDF5 file. A preliminary state decision model is constructed based on a lightweight decision tree algorithm (using XGBoost as an example). The number of decision trees in the preliminary state decision model and the bifurcation level of each decision tree are set according to the amount of data in the training set (for example, if there are 10,000 data points in the training set, the number of decision trees is set to 100, and the bifurcation level of each decision tree is 5). The probabilities of various preliminary states are obtained by minimizing the multi-classification logarithmic loss function. Specifically, the multi-classification logarithmic loss function is calculated as follows: ; ; ; in, is the multi-classification logarithmic loss function value, which represents the total loss of the initial state decision model. is the number of training data sets, is the index of the number of training data groups, is the prediction loss function value, which measures the The error between the true label and the predicted value of the training data set, and Respectively represent The true labels and predicted values ​​of the training data sets, is the number of preliminary status categories, is the preliminary status category index, is the number of decision trees, is the number index of decision trees, Indicates the The regularization term of the tree output, is the minimum splitting loss parameter, represents the number of leaf nodes of the tree, is the L2 regularization coefficient, is the weight of the leaf node (i.e. the predicted value given by the leaf node); During the training of the preliminary state decision model, 10-fold cross validation was used to ensure evaluation stability and generalization ability; Specifically, the training set is randomly divided into 10 parts using random seeds. In each iterative training, 9 parts are training sets and 1 part is validation set. Furthermore, a gradient boosting algorithm is performed on each training set. First, the first training set is input into the preliminary state decision model to generate the first decision tree, output the predicted initial probability (the initial probability is uniformly distributed, that is, the probability of each state label is 0.25), and calculate the probability residual (the actual label value minus the predicted label value, for example, the actual label value of "mild muscle fatigue" is 1, the predicted label value is 0.25, and the residual is 0.75); The residuals are fitted in the remaining decision trees in turn, and each decision tree selects the best split point through information gain, which is calculated as follows: ; in, is the information gain, which measures the reduction in loss after splitting, and the unit is dimensionless. and are the training data sets of the left leaf node and the right leaf node after splitting, and are the residual sums of the left and right leaf nodes respectively; After each fold is completed, the multi-classification accuracy and F1 score are calculated on the validation set. After training, the information gain is extracted, and decision tree nodes with information gain less than 0.01 are removed based on pruning optimization to reduce complexity. Input the physiological features uploaded by the edge end into the trained preliminary state decision model and output the patient's preliminary state; Based on the actual physical therapy records of the medical institution, including the patient's physiological characteristics, patient diagnostic status, physical therapy parameters and effect evaluation level (excellent, good, moderate and poor) during each physical therapy session; Among them, the patient's diagnostic status refers to the patient's diagnosis of normal, mild fatigue, moderate fatigue, or severe fatigue recorded in the actual physical therapy record before the physical therapy. The physical therapy parameters refer to the intensity, frequency, and duration of the physical therapy equipment set by the doctor based on the patient's diagnostic status. The effect evaluation level refers to the physical therapy effect evaluation conducted by the medical staff after the patient undergoes physical therapy and the effect evaluation level given; Use Excel or Python scripts (Pandas library) to organize actual physical therapy records and generate a physical therapy parameter rule library; Specifically, the patient's diagnostic status was classified according to the status label of the preliminary status decision model. A Python script was used to group the physical therapy record table by status label. The statistical values ​​of the physical therapy parameters were calculated for each group. The physical therapy records of effective physical therapy (with an effect evaluation grade of excellent or good) were retained to ensure the validity of the physical therapy parameters. A patient status and physical therapy parameter table was generated. A Python script (json library) was used to convert the patient status and physical therapy parameter table into JSON to generate a physical therapy parameter rule library. According to the patient's preliminary state output by the preliminary state decision model, combined with the physical therapy parameter rule library, the physical therapy parameters of the patient's preliminary state are collected to obtain the initial physical therapy parameter set (including the range of intensity, frequency and duration).

[0029] Parameter optimization module: Combines the patient's personal information and physiological characteristics to build a digital twin model of the patient, simulates and verifies the therapy effect based on the initial therapy parameter set, and outputs the optimal therapy parameter set.

[0030] Build a digital twin model of the patient based on the open source biomechanics framework OpenSim; Specifically, a predefined human skeletal muscle template is loaded, and the parameters of the digital twin model are adjusted according to the patient's basic information (including age, height, weight, and physical treatment area symptoms). For example, the age factor is set to 50 (muscle elasticity coefficient decreases by 10%), the gender is male (muscle volume increases by 15%), and the target is the right arm after stroke (the nerve conduction efficiency of the right arm is set to 60% of the normal value); The patient's physiological characteristics are then input, and the muscle and nerve states in the treatment area are simulated through Finite Element Analysis (FEA). FEA divides the treatment area into several grid cells, assigns attributes to each grid cell based on the physiological characteristics (such as muscle stiffness 1000Pa, nerve impedance 50Ω), and generates a three-dimensional digital twin model. Using the simulation engine MATLAB Simulink, the initial set of physical therapy parameters was used as input to simulate the physical therapy effect; For example, the frequency of electrical stimulation applied to the treatment area (target muscle in the right arm) is set to 20-30 Hz, the intensity to 2-4 mA, and the duration to 750-900 seconds. The simulation effect is calculated based on the muscle dynamics equation as follows: ; ; ; ; ; ; in, It is the muscle contraction force after physical therapy. Indicates the patient's maximum muscle contraction force, It is the frequency influence coefficient, which usually increases the power by 1% per Hz. is the electrical stimulation frequency, is the electrical stimulation intensity, is the baseline activation factor, dimensionless, with a value of 0.1 (initial muscle activity), is the RMS to power conversion factor, is the root mean square value of the electromyographic signal, For blood flow after treatment, is the baseline blood flow, is the flow gain coefficient, which increases by 0.1% per Hz×mA. is the standard deviation of the patient's heart rate, is the standard deviation of heart rate in healthy adults, usually 20ms. The standard blood flow rate for healthy adults is usually 250 mL / min. is the nerve conduction velocity after physical therapy, is the baseline nerve conduction velocity, is the frequency gain coefficient, which increases by 0.5% per Hz. is the standard nerve conduction velocity for healthy adults, is the patient's body temperature change rate, is the body temperature change rate of a healthy adult; The simulation engine runs all the combinations of therapy parameters in the initial therapy parameter set. After each simulation, the comprehensive effect index is calculated using the following formula: ; ; ; ; in, 、 and They are the percentage of improvement of muscle contraction force, blood flow and nerve conduction velocity, respectively, which is the comprehensive effect index; Furthermore, the physiotherapy parameter combinations are based on the initial physiotherapy parameter set (where intensity, frequency, and duration are all ranges), and a grid search method is used to exhaustively search all physiotherapy parameter combinations. According to the comprehensive effect index of all physiotherapy parameter combinations, the physiotherapy parameter combinations with a comprehensive effect index greater than 0 are screened as effective physiotherapy parameter combinations; Then, the effective physiotherapy parameter combination is compared with the real-time physiological characteristics to verify the applicability; Specifically, for example, when the root mean square value of the electromyographic signal is 1.8 mV, which supports the therapy intensity in the valid therapy parameter combination (e.g., RMS < 2.0 mV is suitable for moderate intensity), the heart rate standard deviation is stable (e.g., < 15 ms is safe), and the temperature change rate is normal (e.g., < 0.1°C / s is not overheated), it means that the current valid therapy parameter combination meets the applicability requirements and is determined to be the preferred therapy parameter combination. Otherwise, it is discarded. All effective physiotherapy parameter combinations that meet the applicability requirements are collected to obtain the optimal physiotherapy parameter set.

[0031] Quantum optimization module: maps the optimal therapy parameter set and physiological characteristics into quantum states, constructs quantum circuits, and calculates quantum optimized therapy parameters based on multi-objective optimization principles.

[0032] The data in the optimal physiotherapy parameter set and physiological characteristics are standardized, and a quantum bit state encoding method is used to assign a quantum bit to each standardized result in the optimal physiotherapy parameter set and physiological characteristics. Each standardized value in the optimal physiotherapy parameter set and physiological characteristics is encoded into a quantum state through an RY rotation gate; Based on parameterized quantum circuits, construct quantum circuits of several layers; Specifically, in the cloud quantum computing platform, the Qiskit library is loaded, and a circuit object is created through the QuantumCircuit function. The number of quantum bits is set, and six quantum bits respectively input physiological characteristics and preferred physical therapy parameters. A layered design method is adopted, and each layer includes a rotation gate and an entanglement gate. The RY rotation gate is applied to each quantum bit to adjust the quantum state angle, and the CNOT entanglement gate is applied to adjacent bits. Based on the principle of multi-objective optimization, a weighted objective function is constructed, which can be calculated using the quantum measurement method. The calculation formula is as follows: ; in, is the weighted objective function value, which represents the comprehensive score of the therapeutic effect. The larger the better. It is the muscle activity improvement index, dimensionless. For example, the root mean square value of the electromyographic signal increases from 1.8mV to 2.2mV. , is the penalty for heart rate fluctuation, dimensionless, for example, the standard deviation of heart rate increases from 10ms to 11ms, , is the temperature limit penalty, dimensionless, for example, the temperature is lower than the normal human body temperature (for example, more than 38℃ is considered as fever), that is, , otherwise, , To measure the patient's temperature; Use the COBYLA (Constrained Optimization BY Linear Approximation) classical optimization algorithm combined with VQE to iteratively adjust quantum circuit parameters; By iteratively calculating all optimal therapy parameter sets, the quantum circuit parameters are updated to maximize the weighted objective function value, and the optimal quantum state therapy parameters are obtained. Using the quantum state decoding method, the optimal quantum state therapy parameters are mapped back to physical parameters to obtain quantum optimized therapy parameters.

[0033] Collaborative optimization module: The quantum optimized therapy parameters are input into the digital twin model for secondary simulation to generate the muscle state quantum vector. By calculating the Hilbert space distance, the quantum optimized therapy parameters are collaboratively optimized to obtain the optimal therapy parameters and transmit them back to the edge end to start therapy.

[0034] Using quantum optimized therapy parameters as input conditions, the muscle and nerve states of the therapy area are simulated to obtain simulated muscle state data, including calculation of muscle contraction force, blood flow, and nerve conduction velocity. Through quantum state encoding and RY revolving gate operation, the simulated muscle state data is mapped to quantum states, and all quantum states are combined to form a muscle state quantum vector (a complex vector of length N). The physiological characteristics transmitted in real time are normalized, and each physiological characteristic value is mapped to a quantum state through the RY revolving gate. The physiological characteristic quantum vector is combined to obtain the physiological characteristic quantum vector. The Hilbert space distance between the muscle state quantum vector and the physiological characteristic quantum vector is calculated based on the inner product of the quantum states. The calculation formula is as follows: in, is the Hilbert space distance, with a value range of , is the muscle state quantum vector, is the physiological characteristic quantum vector; By analyzing the statistical consistency between the digital twin model simulation results and the quantum state of physiological characteristics in actual physical therapy, a distance threshold is set; Compare the distance threshold with the Hilbert space distance and output a decision signal, where the decision signal is a binary value, including re-optimization and maintenance of quantum optimized therapy parameters. The specific contents are as follows: When the Hilbert space distance is greater than or equal to the distance threshold, the quantum optimization therapy parameters need to be re-optimized. Specifically, the distance interpolation between the Hilbert space distance and the distance threshold is calculated, and the variational quantum circuit (VQE) and COBYLA algorithm of the quantum optimization module are reused to adjust the variational quantum circuit parameters to obtain the updated quantum optimization therapy parameters until the Hilbert space distance is less than the distance threshold. When the Hilbert space distance is less than the distance threshold, the current quantum optimized therapy parameters are returned and input to the edge as the optimal therapy parameters to start therapy. Specifically, the quantum optimized therapy parameters are encapsulated into byte packets, the text is encrypted using the OKD key, and the encrypted byte packets are transmitted to the edge via the 5G network. The encrypted byte packets are decrypted and verified at the edge to obtain the quantum optimized therapy parameters. The intelligent medical therapy device uses the quantum optimized therapy parameters to treat patients.

[0035] In summary, the present invention avoids the problem of incomplete information that may be caused by a single data source through physiological data collection and real-time processing, improves the reliability and accuracy of the data, and introduces a lightweight decision tree algorithm to not only improve the prediction accuracy of the model, but also ensure the generalization ability and stability of the model through cross-validation and pruning optimization, establishes a "quantum optimization-biomechanical simulation" dynamic closed loop, and uses quantum state space distance as a feedback signal to achieve real-time adaptive adjustment of treatment parameters.

[0036] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent medical rehabilitation therapy instrument control system, characterized by: include, A data processing module is used to obtain the patient's physiological data, including electromyographic signals, heart rate, and body temperature, and process the data in real time on the edge to obtain physiological characteristics; The initial parameter module is used by the edge computing unit to upload physiological characteristics to the cloud. In the cloud, a preliminary state decision model is constructed based on the physiological characteristics to determine the patient's initial state. The actual physical therapy records are grouped according to the patient's initial state, and the effective physical therapy records are counted to generate an initial physical therapy parameter set. The parameter optimization module is used to combine the patient's personal information and physiological characteristics to build a digital twin model of the patient, simulate and verify the physical therapy effect based on the initial physical therapy parameter set, and output the optimal physical therapy parameter set; The quantum optimization module is used to map the optimal therapy parameter set and physiological characteristics into quantum states, construct quantum circuits, and calculate the quantum optimized therapy parameters based on the multi-objective optimization principle; The collaborative optimization module is used to input the quantum optimized therapy parameters into the digital twin model for secondary simulation, generate the muscle state quantum vector, and collaboratively optimize the quantum optimized therapy parameters by calculating the Hilbert space distance. The optimal therapy parameters are obtained and transmitted back to the edge end to start therapy.

2. The intelligent medical rehabilitation therapy instrument control system according to claim 1, characterized in that: Process physiological data in real time at the edge, including denoising and feature extraction.

3. The intelligent medical rehabilitation therapy instrument control system according to claim 2, characterized in that: The denoising steps are as follows: The electromyographic signals collected each time are evenly divided into several sub-segments; The fractal dimension calculation is performed on each sub-segment using the Higuchi algorithm; Based on the calculation results of fractal dimension, the fractal features of EMG signals are extracted and the denoised EMG signals are reconstructed.

4. The intelligent medical rehabilitation therapy instrument control system according to claim 1, characterized in that: The specific steps for judging the patient's initial condition are as follows: Build and train a preliminary state decision model based on a lightweight decision tree algorithm; The physiological features uploaded by the edge are input into the trained preliminary state decision model to output the patient's preliminary state.

5. The intelligent medical rehabilitation therapy instrument control system according to claim 1, characterized in that: The specific steps of generating the initial therapy parameter set are as follows: Build a rule base for physical therapy parameters; According to the patient's preliminary state output by the preliminary state decision model and combined with the physical therapy parameter rule library, the initial physical therapy parameter set is obtained.

6. The intelligent medical rehabilitation therapy instrument control system according to claim 5, characterized in that: The specific steps of constructing the therapy parameter rule base are as follows: Integrate the actual physical therapy records of medical institutions and organize them into physical therapy record forms; These include the patient's physiological characteristics, patient diagnostic status, treatment parameters, and effect evaluation level during each treatment session; Use Python script to group physical therapy records by status label; Calculate the statistical value of each group of physical therapy parameters in the physical therapy record table, retain the physical therapy records of effective physical therapy, and generate a physical therapy parameter rule base.

7. The intelligent medical rehabilitation therapy instrument control system according to claim 2, characterized in that: The specific steps of building a digital twin model of a patient are as follows: Build a digital twin model of the patient based on the open source biomechanics framework OpenSim; Load the predefined human skeletal muscle template and adjust the digital twin model parameters based on the patient's basic information.

8. The energy storage efficiency improvement system combined with energy management according to claim 1, characterized in that: The specific steps of outputting the optimal therapy parameter set are as follows: Input the patient's physiological characteristics into the patient's digital twin model and simulate the muscle and nerve status of the treatment area through finite element analysis; Using the simulation engine MATLAB Simulink, the initial set of physical therapy parameters was used as input to simulate the physical therapy effect and calculate the comprehensive effect index; According to the comprehensive effect index of all physical therapy parameter combinations, effective physical therapy parameter combinations are screened out; The effective physiotherapy parameter combinations are compared with the real-time physiological characteristics to verify the applicability, and all effective physiotherapy parameter combinations that meet the applicability requirements are collected to obtain the optimal physiotherapy parameter set.

9. The energy storage efficiency improvement system combined with energy management according to claim 1, characterized in that: The specific steps of collaboratively optimizing the quantum optimization therapy parameters are as follows: The optimal physical therapy parameters are input into the digital twin model for secondary simulation to generate the muscle state quantum vector; Based on the physiological characteristics transmitted in real time, the quantum state encoding method in the optimal parameter module is reused to obtain the physiological characteristic quantum vector; Calculate the Hilbert space distance between the muscle state quantum vector and the physiological characteristic quantum vector; By comparing the preset distance threshold with the Hilbert space distance, a decision signal is output; Based on the decision signal, the quantum optimization therapy parameters are collaboratively optimized.

10. The energy storage efficiency improvement system combined with energy management according to claim 1, characterized in that: The specific steps of calculating quantum optimization physical therapy parameters are as follows: Based on the principle of multi-objective optimization, a weighted objective function is constructed and calculated using quantum measurement methods; The COBYLA algorithm is combined with VQE to iteratively adjust the quantum circuit parameters to the highest value of the weighted objective function, thus obtaining the optimal quantum state physical therapy parameters. Using the quantum state decoding method, the optimal quantum state therapy parameters are mapped back to physical parameters to obtain quantum optimized therapy parameters.

Images