Electromagnetic interference-free ultrashort wave intelligent physiotherapy self-adaptive regulation and control system based on biofeedback

Abstract

The invention discloses an electromagnetic interference-free ultrashort wave intelligent physiotherapy self-adaptive regulation and control system based on biofeedback, and belongs to the technical field of medical physiotherapy. The system solves the problems that an existing system cannot carry out real-time adjustment according to the physiological reaction of a patient individual, and treatment parameters are difficult to accurately regulate and control, and auxiliary features are generated by obtaining the individual features and real-time physiological data of the patient and combining context features; predicting probability distribution of tissue reaction by using a neural network model, calculating a confidence interval of the probability distribution, and automatically generating an initial treatment mode by comparing the confidence interval with a threshold value; the treatment effect is simulated, and possible complications and adverse reactions are predicted; evaluating curative effects and risks under different treatment parameter combinations by combining a Pareto frontier map, and automatically generating an optimal treatment parameter combination; treatment parameters are adjusted in time according to the real-time response of the patient, and the dynamic adaptability of the treatment process is ensured, so that personalized treatment is realized, and the treatment effect and the patient satisfaction are improved.

Description

Technical Field

[0001] This invention relates to the field of medical physiotherapy technology, specifically to an adaptive control system for intelligent physiotherapy based on biofeedback and electromagnetic interference-free ultra-shortwave therapy. Background Technology

[0002] Shortwave diathermy is a physical therapy method that uses the high-frequency oscillation of shortwave electromagnetic fields to generate heat in human tissues, thereby achieving therapeutic effects such as improving blood circulation, reducing inflammation, and relieving pain.

[0003] However, traditional shortwave diathermy equipment has some limitations, such as: its output power and frequency are usually fixed and cannot be adjusted in real time according to the individual patient's physiological response, resulting in large individual differences in treatment effects; in addition, it lacks an effective biofeedback mechanism, cannot monitor the patient's physiological state in real time during treatment, and is difficult to accurately control treatment parameters.

[0004] Therefore, to meet current needs, a biofeedback-based, electromagnetically interference-free ultra-shortwave intelligent physiotherapy adaptive control system is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a biofeedback-based, electromagnetically interference-free, ultra-shortwave intelligent physiotherapy adaptive control system. This system acquires individual patient characteristics and real-time physiological data, combines these with contextual features to generate auxiliary features; utilizes a neural network model to predict the probability distribution of tissue responses and calculates their confidence intervals, automatically generating an initial treatment mode by comparing it with a threshold; simulates treatment effects and predicts potential complications and adverse reactions; combines Pareto front plots to evaluate the efficacy and risks of different treatment parameter combinations, automatically generating the optimal treatment parameter combination; and adjusts treatment parameters promptly based on the patient's real-time responses to ensure dynamic adaptability of the treatment process, thereby achieving personalized treatment, improving treatment effectiveness and patient satisfaction, and solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A biofeedback-based, electromagnetically interference-free, ultra-shortwave intelligent physiotherapy adaptive control system includes: The biofeedback monitoring unit is configured to acquire, in real time, blood flow and velocity distribution maps, tissue elasticity, stiffness distribution maps, temperature distribution maps, and pH value distribution maps of the treatment target area through non-contact methods, forming a biofield map; The non-electromagnetic interference unit is configured to use a multi-modal programmable focused ultrasound phased array to generate ultra-short wave electromagnetic fields based on specific frequencies and pulse modes, in order to promote cell membrane permeability, improve microcirculation, and stimulate tissue repair. The intelligent analysis unit is configured to build a neural network model, taking the biofield map and treatment stage as input. The neural network model learns the complex mapping relationship between the biofield map and the healthy tissue pattern. Based on the current parameters and the biofield map, it predicts the tissue response trend in the future, judges its tolerance and response to ultra-shortwave therapy, and generates and outputs the optimal energy field regulation parameter instructions. The adaptive control unit is configured to automatically adjust the output power, frequency, and pulse mode parameters of the ultra-shortwave transmission module based on the optimal energy field control parameter command output, combined with patient feedback and changes in the biofield diagram, to achieve the best therapeutic effect.

[0007] Furthermore, the intelligent analysis unit includes: The auxiliary feature introduction module is configured to acquire the patient’s individual characteristics and real-time physiological data as static baseline features. Contextual features are introduced, including: whether the patient is currently in the acute, recovery, or rehabilitation phase; whether the goal of this treatment is to reduce inflammation, relieve pain, or promote tissue remodeling; and whether the patient is using anticoagulants or NSAIDs concurrently. These features are then fused with static baseline features as auxiliary features.

[0008] The treatment effect simulation module is configured to take into account the current energy field parameters and biofield diagram, and predict the probability distribution of the patient's tissue response within a specified time period through a neural network model. By introducing auxiliary features, calculating the confidence interval of the prediction results, and comparing it with a preset threshold, an initial treatment pattern is automatically generated. By utilizing patients' auxiliary characteristics, a personalized digital twin model is constructed to simulate the treatment effect of the initial treatment mode and predict whether complications and adverse reactions will occur during the treatment process; and a Pareto front plot of treatment parameters and expected efficacy is generated to show the trade-off between efficacy and risk.

[0009] Furthermore, the treatment effect simulation module introduces auxiliary features to calculate the confidence interval of the predicted results and compares it with a preset threshold to automatically generate an initial treatment mode, including: Obtain the predicted value vector of tissue response and its probability distribution parameters from the neural network model, and at the same time obtain the statistical features of the model training set; Based on the predicted value vector of tissue response, probability distribution parameters, and statistical characteristics of the model training set, output a prediction data package; Receive the prediction data packet and calculate the initial confidence interval, then output the initial confidence interval data packet; The initial confidence interval is corrected by fusing auxiliary features to obtain the corrected confidence interval data packet; Perform a comparison judgment based on the corrected confidence interval data and the preset threshold, and output the comparison result data packet; The initial treatment pattern is generated by combining rules based on the comparison results data packets.

[0010] Furthermore, the treatment effect simulation module utilizes the patient's auxiliary characteristics to construct a personalized digital twin model to simulate the treatment effect of the initial treatment mode, including: The scale-transformation neural network maps auxiliary features to a multi-scale physiological parameter set, which includes molecular scale parameters, cellular scale parameters, and tissue scale parameters. Based on the differences in blood flow distribution, elastic modulus, and pH value in different regions of the multi-scale physiological parameter set and biofield map, a non-uniform grid partitioning strategy is constructed. Based on a non-uniform meshing strategy, the tissue is divided into heterogeneous units with different electromagnetic susceptibility and thermal conductivity. A multi-source data fusion method based on parameter assignment outputs tissue heterogeneity model parameters corresponding to different heterogeneous units; A set of coupled equations is established using the finite element method. The parameters of the tissue heterogeneity model and the parameters of the initial treatment mode are used as inputs to the set of coupled equations, and the multiphysics spatiotemporal evolution data matrix output by the set of coupled equations is obtained. The spatiotemporal evolution data matrix of multiphysics fields is input into a time series convolutional neural network to obtain dynamic response prediction results; Based on the dynamic response prediction results and a pre-set clinical risk threshold library, a treatment outcome distribution is generated using the Monte Carlo simulation method; the treatment outcome distribution includes the risk probability of tissue burns, aggravated inflammation, and nerve damage; Based on the distribution of treatment results and real-time biofeedback data, the parameters of the digital twin model are dynamically corrected using a Bayesian update algorithm, and the corrected digital twin model and treatment effect prediction report are output.

[0011] Furthermore, the adaptive control unit includes: The strategy dynamic adjustment module is configured to evaluate the confidence interval of the prediction results and determine their reliability; based on the confidence interval of the prediction results, an initial treatment pattern is generated. By combining the Pareto frontier plot of treatment parameters and expected efficacy, the efficacy and risks of different combinations of treatment parameters are evaluated. Based on the treatment simulation results, the optimal combination of treatment parameters is generated; if the simulation results show that there is an imbalance between the efficacy and risk of the current treatment mode, the treatment parameters are automatically adjusted to optimize the treatment mode. Establish a real-time feedback mechanism to dynamically adjust personalized treatment models based on patients' real-time responses during treatment; Treatment parameters are adjusted in real time based on real-time feedback data and preset treatment goals; Establish an intelligent early warning mechanism that immediately issues an early warning signal and automatically initiates intervention strategies when an abnormal situation is detected.

[0012] Furthermore, the adaptive control unit further includes: The multi-dimensional threshold adjustment module is configured to dynamically adjust the preset threshold of the confidence interval based on the patient's individual characteristics and real-time physiological state. The multi-dimensional threshold comparison module is configured to perform multi-dimensional comparisons of the type and extent of tissue response when comparing the confidence intervals of the prediction results.

[0013] Furthermore, the electromagnetic interference-free unit includes: The parameter adjustment module is configured to acquire multi-dimensional characteristics of the patient and flexibly adjust parameters such as ultrasound frequency, pulse width, and pulse repetition frequency according to different treatment needs and patient conditions to achieve personalized treatment control. The electromagnetic interference suppression module is configured to use advanced electromagnetic shielding materials and technologies to ensure that broadband electromagnetic radiation is not generated while generating ultra-shortwave electromagnetic fields; and through electromagnetic compatibility design, it ensures that the equipment operates safely and stably in complex electromagnetic environments without adversely affecting surrounding medical equipment and personnel.

[0014] Furthermore, the biofeedback monitoring unit includes: The multimodal data acquisition module is configured to measure blood flow velocity and blood flow distribution in microvessels in real time through the laser Doppler effect; acquire high-resolution tomographic images of tissues in real time using the principle of optical coherence, including blood flow and velocity distribution maps; and acquire tissue temperature distribution maps in real time using infrared thermal imaging technology. The feature extraction module is configured to preprocess the collected multimodal data, including filtering, denoising, and normalization operations. Data analysis algorithms are used to extract characteristic parameters from the biofield diagram that reflect the physiological state and treatment response of tissues.

[0015] Furthermore, it also includes: The user interaction unit is configured to provide an operation interface for medical staff to input patient information, set treatment parameters, and view physiological data and equipment status during treatment. The treatment record unit is configured to automatically record the detailed process of each treatment, automatically generate a treatment report, and include, but are not limited to, comparisons of physiological data before and after treatment, evaluation of treatment effects, and export them in a standard format for easy archiving and sharing.

[0016] Furthermore, the user interaction unit includes: The data encryption module is configured to encrypt stored and transmitted patient information and treatment data to prevent unauthorized access. The data sharing module is configured to provide a standard data interface, supporting data sharing with other medical devices and hospital information systems.

[0017] Compared with the prior art, the beneficial effects of the present invention are: In this invention, by acquiring the patient's individual characteristics and real-time physiological data, and combining them with contextual features to generate auxiliary features, the treatment plan can be precisely tailored to the patient's specific condition. A neural network model is used to predict the probability distribution of tissue responses and calculate their confidence intervals, automatically generating an initial treatment mode by comparing it with a threshold. The treatment effect is simulated to predict possible complications and adverse reactions. Pareto front plots are used to evaluate the efficacy and risks under different combinations of treatment parameters, automatically generating the optimal combination of treatment parameters. Treatment parameters are adjusted promptly based on the patient's real-time response to ensure the dynamic adaptability of the treatment process, thereby achieving personalized treatment and improving treatment effectiveness and patient satisfaction. Attached Figure Description

[0018] Figure 1 This is a flowchart of the biofeedback-based, electromagnetic interference-free, ultra-shortwave intelligent physiotherapy adaptive control system of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] To address the limitations of existing shortwave diathermy equipment, such as its typically fixed output power and frequency, which cannot be adjusted in real-time according to individual patient physiological responses, leading to significant individual differences in treatment effectiveness; and the lack of an effective biofeedback mechanism to monitor the patient's physiological state during treatment and precisely control treatment parameters, please refer to [link to relevant documentation]. Figure 1 This embodiment provides the following technical solution: A biofeedback-based, electromagnetically interference-free, ultra-shortwave intelligent physiotherapy adaptive control system includes: The biofeedback monitoring unit is configured to acquire, in real-time, non-contact data on blood flow and velocity distribution, tissue elasticity, stiffness, temperature, and pH distribution in the treatment target area, forming a biofield map. Blood flow reflects metabolic and inflammatory states, tissue elasticity reflects the degree of edema and fibrosis, and temperature and pH further provide information on tissue metabolism and acid-base balance. The fusion of multimodal data constitutes a biofield map characterizing the physiological state of the tissue. The biofeedback monitoring unit includes: The multimodal data acquisition module is configured to measure blood flow velocity and blood flow distribution in microvessels in real time through the laser Doppler effect; laser Doppler technology can acquire high-resolution blood flow images in a non-contact state, reflecting the metabolic and inflammatory state of tissues; it can acquire high-resolution tomographic images of tissues in real time using the principle of optical coherence, including blood flow and velocity distribution maps; the principle of optical coherence technology can provide micron-level resolution in a non-contact state, suitable for blood flow monitoring in superficial tissues; and it can acquire temperature distribution maps of tissues in real time through infrared thermal imaging technology; infrared thermal imaging technology can monitor temperature changes in tissues in a non-contact state, reflecting inflammatory responses and tissue repair processes.

[0021] The feature extraction module is configured to preprocess the collected multimodal data, including filtering, denoising, and normalization operations, to ensure the accuracy and reliability of the data. Through data analysis algorithms, it extracts feature parameters reflecting the physiological state and treatment response of tissues from the biofield map. For example, it assesses the metabolic and inflammatory state of tissues through blood flow and velocity distribution maps; it assesses the degree of edema and fibrosis through tissue elastic modulus; it monitors the inflammatory response and tissue repair process through temperature distribution maps; and it assesses the acid-base balance of tissues through pH distribution maps.

[0022] The electromagnetic interference-free unit is configured to use a multimodal programmable focused ultrasound phased array to generate ultra-shortwave electromagnetic fields based on specific frequencies and pulse modes, thereby promoting cell membrane permeability, improving microcirculation, and stimulating tissue repair. The electromagnetic interference-free unit includes: The parameter adjustment module is configured to acquire multi-dimensional characteristics of the patient and flexibly adjust parameters such as ultrasound frequency, pulse width, and pulse repetition frequency according to different treatment needs and patient conditions to achieve personalized treatment control. For example, for acute inflammatory reactions, a lower frequency and a wider pulse width are used to reduce inflammation; for chronic injury repair, a higher frequency and a narrower pulse width are used to stimulate tissue regeneration. During treatment, the output parameters of ultrasound and shortwave are monitored in real time, and the treatment parameters are dynamically adjusted according to the patient's real-time physiological and tissue responses to ensure the safety and effectiveness of the treatment.

[0023] The electromagnetic interference suppression module is configured with advanced electromagnetic shielding materials and technologies to ensure that it generates ultra-shortwave electromagnetic fields without producing broadband electromagnetic radiation. Through electromagnetic compatibility design, it ensures safe and stable operation of the device in complex electromagnetic environments without adversely affecting surrounding medical equipment and personnel. Specifically, by emitting low-frequency, long-pulse focused ultrasound waves (e.g., 1-3 MHz), steady-state acoustic radiation is generated at the cellular scale, applying continuous, micro-Newton-level mechanical traction to the cytoskeleton and extracellular matrix, directly regulating the cell's mechanically sensitive channels and promoting the conversion of gene expression to a repair phenotype. Simultaneously with subwavelength modulation, a high-frequency, short-pulse nonlinear ultrasound component (e.g., 5-10 MHz) is superimposed. This component, as it propagates through tissue, generates abundant harmonics and frequency components, creating micro-perturbations in the tissue fluid and cell membrane lipid bilayer, safely enhancing the effect without generating heat. Molecular diffusion rate and membrane fluidity are enhanced, thereby promoting cell membrane permeability and improving microcirculation. By calculating and controlling the phase and amplitude of the emitted beam of each transducer in the control array, the sound waves of the above two mechanisms undergo constructive interference at specific three-dimensional coordinate points in the body, thus synthesizing a high-precision treatment focus at the core of the lesion far from the body surface. This disperses energy along the path, achieving deep-targeted stimulation of tissue repair. Its effective control depth can reach more than 8 cm under the skin. Its core working fundamental frequency is 0.5-10 MHz, and the instantaneous spatial peak time-averaged sound intensity can be precisely adjusted within the range of 10 mW / cm² to 5 W / cm². Furthermore, through electronic design, it ensures that any electromagnetic harmonic components generated are below the exemption limit of the electromagnetic compatibility standard for medical devices, thus completely eliminating the generation of interference-causing broadband electromagnetic radiation in terms of physical principles and engineering implementation.

[0024] The intelligent analysis unit is configured to construct a neural network model, taking the biofield map and treatment stage as input. The model learns the complex mapping relationship between the biofield map and healthy tissue patterns. Based on current parameters and the biofield map, it predicts future tissue response trends, assesses the tissue's tolerance and response to ultrashortwave therapy, and generates and outputs optimal energy field regulation parameter instructions. The intelligent analysis unit includes: The auxiliary feature introduction module is configured to acquire the patient's individual characteristics and real-time physiological data as static baseline features. Individual characteristics include: age, gender, body mass index, and basal metabolic rate; initial tissue health score based on quantitative analysis of pre-treatment medical images, including: fat content, fibrosis ratio, and baseline vascular density; gene polymorphism information or serum biomarker levels related to tissue repair and inflammatory response; real-time physiological data includes: heart rate variability and skin conductance response synchronously acquired through wearable devices, used to objectively assess stress levels and pain tolerance during treatment; real-time acquisition of the patient's local soreness and burning sensation levels through a simple human-machine interface, and time-series processing; and collection of sensitivity to different types of energy parameters, onset time, and adverse reaction history during historical treatments to form an individual's treatment response fingerprint; and the introduction of contextual features, including: acquiring whether the patient is currently in the acute, recovery, or rehabilitation phase; acquiring whether the goal of this treatment is anti-inflammatory, analgesic, or tissue remodeling promotion; and acquiring whether the patient is simultaneously using anticoagulants or NSAIDs, so as to adjust energy intensity accordingly to avoid bleeding or compound damage risks, and fusing these features with the static baseline features as auxiliary features.

[0025] The treatment effect simulation module is configured to take into account the current energy field parameters and biofield map. Using a neural network model, it predicts the probability distribution of tissue response within a specified timeframe, such as 5-60 seconds. It introduces auxiliary features, calculates the confidence interval of the prediction results, and compares it with a preset threshold to automatically generate an initial treatment mode. For example, for elderly patients, considering their weaker tissue repair capabilities, when the confidence level of the prediction result is below the threshold, it automatically switches to a conservative treatment mode, adjusting treatment parameters to avoid overstimulation. For patients with low pain thresholds, it optimizes the energy field intensity and pulse pattern to reduce discomfort during treatment. It utilizes the patient's auxiliary features to construct a personalized digital twin model, simulating the treatment effect of the initial treatment mode and predicting potential complications and adverse reactions during treatment. It also generates a Pareto front plot of treatment parameters versus expected efficacy, demonstrating the trade-off between efficacy and risk.

[0026] The beneficial effects achieved by the above are as follows: By acquiring individual patient characteristics and real-time physiological data, combined with contextual features such as disease stage, treatment goals, and medication use, unique static baseline and auxiliary features are generated for each patient, thereby achieving personalized treatment and improving treatment efficacy and patient satisfaction. Furthermore, using a neural network model, the probability distribution of future tissue responses is predicted based on current energy field parameters and biofield diagrams, and the confidence interval of the prediction results is calculated. By comparing with preset thresholds, an initial treatment pattern is automatically generated, and a personalized digital twin model is constructed to simulate treatment effects and predict possible complications and adverse reactions. Simultaneously, a Pareto front plot of treatment parameters versus expected efficacy is generated, intuitively demonstrating the trade-off between efficacy and risk, providing a scientific basis for optimizing treatment patterns, and enabling treatment plans to pursue optimal efficacy while minimizing risk.

[0027] The adaptive control unit is configured to automatically adjust the output power, frequency, and pulse mode parameters of the ultra-shortwave transmission module based on the optimal energy field control parameter commands output, combined with patient feedback and changes in the biofield diagram, to achieve the best therapeutic effect. The adaptive control unit includes: The strategy dynamic adjustment module is configured to evaluate the confidence interval of the prediction results and determine their reliability. Based on the confidence interval of the prediction results, an initial treatment mode is generated. For example, if the confidence interval is narrow and the prediction results are relatively reliable, the treatment mode is adjusted according to the prediction results; if the confidence interval is wide and the prediction results have high uncertainty, a more conservative strategy is adopted. The module also uses a Pareto front plot of treatment parameters versus expected efficacy to evaluate the efficacy and risk under different combinations of treatment parameters. If the prediction results show that the current treatment mode has a high risk, such as tissue damage or overstimulation, it automatically adjusts to a safer treatment mode. For example, it reduces the energy field intensity, increases the cooling time, or adjusts the pulse mode to reduce tissue stimulation. Based on the treatment simulation results, the module generates the optimal combination of treatment parameters. If the simulation results show an imbalance between efficacy and risk in the current treatment mode, the treatment parameters are automatically adjusted to optimize the treatment mode. For example, if the efficacy is poor but the risk is low, the energy field intensity is appropriately increased; if the risk is high but the efficacy is good, the energy field intensity is appropriately decreased or the pulse mode is adjusted.

[0028] Establish a real-time feedback mechanism to dynamically adjust personalized treatment modes based on the patient's real-time responses during treatment, thereby improving the dynamic adaptability of treatment. For example, if the patient reports severe pain, the energy field intensity or pulse mode is automatically adjusted to alleviate discomfort. Treatment parameters are adjusted in real time based on real-time feedback data and preset treatment goals. For example, if a good tissue response and low risk are detected, the energy field intensity is appropriately increased to accelerate the treatment effect; if a poor tissue response or high risk is detected, the energy field intensity is automatically reduced or treatment is paused. Establish an intelligent early warning mechanism to immediately issue warning signals and automatically activate intervention strategies when abnormalities are detected. For example, when tissue damage or increased inflammation is present, the energy field intensity is reduced, the cooling time is increased, or the pulse mode is adjusted to ensure the safety of the treatment process.

[0029] The beneficial effects achieved by the above are as follows: the reliability of the prediction results is assessed based on the confidence interval, and an initial treatment pattern is generated accordingly; the efficacy and risk of different combinations of treatment parameters are evaluated by combining Pareto front plots, and the optimal combination of treatment parameters is automatically generated; during the treatment process, the personalized treatment pattern is dynamically adjusted through a real-time feedback mechanism, and the treatment parameters are adjusted in a timely manner according to the patient's real-time response to ensure the dynamic adaptability of the treatment process; in addition, the intelligent early warning mechanism can immediately issue an early warning signal and initiate an intervention strategy when abnormalities are detected to ensure the safety of treatment.

[0030] The multi-dimensional threshold adjustment module is configured to dynamically adjust the preset threshold of the confidence interval based on the patient's individual characteristics and real-time physiological state. For example, for elderly patients or patients with chronic diseases, the threshold can be appropriately lowered to improve the conservatism of treatment; for young patients or patients with acute injuries, the threshold can be appropriately increased to enhance the aggressiveness of treatment.

[0031] The multi-dimensional threshold comparison module is configured to perform multi-dimensional comparisons of the type and degree of tissue response when comparing the confidence intervals of the prediction results; for example, a specific inflammation threshold is set for the prediction results of the inflammatory response, and a pain threshold is set for the prediction results of the pain response.

[0032] The beneficial effects achieved by the above are as follows: by dynamically adjusting the preset threshold of the confidence interval based on the patient's individual characteristics and real-time physiological state, the threshold is made to better fit the patient's actual situation; when comparing the confidence interval of the predicted results, the type and degree of tissue response are compared in multiple dimensions, which improves the accuracy and reliability of treatment decisions, further optimizes the treatment process, and provides patients with a more efficient, safe, and comfortable treatment experience.

[0033] The user interaction unit is configured to provide an operating interface for medical staff to input patient information, set treatment parameters, and view physiological data and equipment status during treatment. The patient information includes, but is not limited to, name, gender, age, medical history, genetic information, previous treatment records, and allergy history. The treatment parameters include, but are not limited to, the frequency, pulse mode, power, and treatment duration of the shortwave diathermy.

[0034] The treatment recording unit is configured to automatically record the detailed process of each treatment, automatically generate a treatment report, and include, but is not limited to, comparisons of physiological data before and after treatment, evaluation of treatment effectiveness, and export the report in a standard format for easy archiving and sharing. The user interaction unit includes: The data encryption module is configured to encrypt stored and transmitted patient information and treatment data. It supports multiple security authentication methods, such as username / password, fingerprint recognition, and facial recognition, to prevent unauthorized access and ensure data security and privacy.

[0035] The data sharing module is configured to provide a standard data interface, supporting data sharing with other medical devices and hospital information systems. Through data sharing, medical staff can gain a more comprehensive understanding of patients' conditions and treatment history, improving the continuity and consistency of treatment.

[0036] Working principle: By acquiring the patient's individual characteristics and real-time physiological data, and combining them with contextual features to generate auxiliary features, the treatment plan can be precisely tailored to the patient's specific situation; a neural network model is used to predict the probability distribution of tissue response and calculate its confidence interval, and an initial treatment mode is automatically generated by comparing it with a threshold; the treatment effect is simulated to predict possible complications and adverse reactions; the efficacy and risk of different treatment parameter combinations are evaluated by combining Pareto front plots, and the optimal treatment parameter combination is automatically generated; and the treatment parameters are adjusted in a timely manner according to the patient's real-time response to improve treatment effect and patient satisfaction.

[0037] In one embodiment, the treatment effect simulation module introduces auxiliary features, calculates the confidence interval of the predicted result, compares it with a preset threshold, and automatically generates an initial treatment mode, including: Obtain the predicted value vector of tissue response from the neural network model. and its probability distribution parameters Simultaneously, obtain statistical features of the model training set. ; in, The predicted mean of the neural network output. The standard deviation of the predictions output by the neural network; The mean of the features in the training set. The standard deviation of the sample is 1. For sample size; Based on the predicted value vector of tissue response, probability distribution parameters, and statistical characteristics of the model training set, a prediction data package is output. ;in, ; Receive prediction data packets and calculate the initial confidence interval, then output the initial confidence interval data packets. The specific calculation formula is as follows:

[0038]

[0039]

[0040] in, This is the upper bound of the initial confidence interval; This is the lower bound of the initial confidence interval; This is the initial confidence interval width. ; for The critical value of the distribution, which ranges from Query the distribution table; The significance level is usually set to 0.05 (corresponding to a 95% confidence level) or 0.01 (corresponding to a 99% confidence level). This is the training sample size; For the current input features; By fusing auxiliary features to correct the initial confidence interval, the corrected confidence interval data packet is obtained. The specific calculation formula is as follows:

[0041]

[0042]

[0043] in, This is the upper bound of the corrected confidence interval; This is the lower bound of the corrected confidence interval; This is the corrected confidence interval width; This is a comprehensive correction factor calculated based on auxiliary features. ; , Patient age characteristics are an auxiliary feature; (These correspond to the acute phase, recovery phase, and rehabilitation phase, respectively). (For those who received medication and those who did not); Perform a comparison judgment based on the corrected confidence interval data and the preset threshold, and output the comparison result data packet; In this embodiment, the comparison and judgment process includes confidence assessment and security assessment. Based on the results of the confidence assessment and security assessment, a comparison result data packet is generated and output, for example: Confidence assessment parameters The confidence assessment results include: high reliability ( ), moderately reliable ) and low reliability ( ); Safety assessment includes: if the predicted values ​​of all dimensions (tissue tolerance, inflammatory response, and tissue repair) are within the safe range, it is marked as "safe"; if a single dimension is close to the threshold but does not exceed it, it is marked as "cautious"; if any dimension exceeds the threshold, it is marked as "risk". Output the comparison result data packet {security assessment flag, confidence assessment result, preset threshold}.

[0044] The initial treatment pattern is generated by combining rules based on the comparison results data packets.

[0045] In this embodiment, the combination of comparison result data packets and their corresponding treatment strategies are preset manually, for example: Rule 1: Furthermore, the safety assessment is marked as "safe," and the corresponding strategies are: increase the energy field intensity by 20%, increase the pulse frequency, and prolong the treatment time; Rule 2: Furthermore, the safety assessment is marked as "safe" or "cautious," and the corresponding strategy is: to use standard energy field parameters and pulse mode equalization. Rule 3: Alternatively, if the safety assessment is marked as "risk", the corresponding strategy is to reduce the energy field intensity by 30%, increase the pulse interval, and shorten the single treatment time.

[0046] The working principle and beneficial effects of the above technical solution are as follows: This invention first obtains predicted tissue response values ​​and their probability distribution parameters from a neural network model, and calculates an initial confidence interval by combining it with statistical features of the training set to quantify the uncertainty of the prediction. Subsequently, auxiliary features such as age, disease stage, and medication status are introduced, and the confidence interval is individually adjusted through a comprehensive correction factor. Finally, the corrected confidence interval is compared with a preset threshold in two dimensions (confidence assessment and safety assessment) to generate an initial treatment pattern.

[0047] For example, the patient is a 65-year-old male with a BMI of 28.3, diagnosed with acute osteoarthritis of the right knee, with treatment goals of anti-inflammatory and analgesic effects, no history of shortwave diathermy, and is using warfarin (an anticoagulant).

[0048] Tissue response prediction vector = [0.65, 0.87, 0.42], probability distribution parameters = {0.68, 0.12}, statistical features of model training set = {5.2, 1.8, 1250}; Current input features (Patient Characteristic Composite Score), at a 95% confidence level, Calculated , Initial confidence interval width ; Calculate the correction factor: age correction factor Stage correction factor (Acute phase), drug correction factor (Using anticoagulants), comprehensive correction factor Corrected confidence interval width Corrected confidence interval , .

[0049] Preset thresholds: tolerance threshold 0.8, inflammation response threshold 0.7, repair threshold 0.6, and confidence interval width threshold 0.15. Confidence assessment. (Low reliability). The safety assessment indicates an inflammatory response level of 0.87, which is greater than the inflammatory response threshold of 0.7, and the upper bound of the corrected confidence interval is greater than 0.7; therefore, it is marked as "risk". The output alignment result is {low reliability, risk}. Generate initial treatment pattern: If the value is less than the confidence interval width threshold of 0.15, but the safety assessment is marked as "risk", a conservative treatment mode is generated, the energy field intensity is reduced by 30% (e.g., 25 watts × 0.7 = 17.5 watts), the pulse interval is increased (pause for 30 seconds every 2 minutes), and the single treatment time is shortened (8 minutes (standard 10 minutes × 0.8)).

[0050] This invention solves the technical problem that traditional shortwave diathermy equipment cannot be precisely adjusted according to individual patient differences. It transforms subjective judgment into objective quantitative indicators, significantly reducing the risk of overtreatment. At the same time, it provides more efficient treatment parameters for low-risk patients, improving the accuracy, safety and comfort of treatment.

[0051] In one embodiment, the treatment effect simulation module utilizes the patient's auxiliary characteristics to construct a personalized digital twin model to simulate the treatment effect of the initial treatment mode, including: The scale-transformation neural network maps auxiliary features to a multi-scale physiological parameter set, which includes molecular-scale parameters, cellular-scale parameters, and tissue-scale parameters.

[0052] In this embodiment, the scale-transformation neural network is a deep learning model that converts macroscopic patient characteristics (such as age and weight) into microscopic physiological parameters (such as cell activity and molecular concentration). Molecular scale parameters include ATP concentration, inflammatory factor concentration, and oxygen partial pressure. Cellular scale parameters include cell membrane potential, cell membrane permeability coefficient, and cell proliferation rate. Tissue scale parameters include vascular density, tissue thermal conductivity, and tissue dielectric constant. The scale-transformation neural network is trained by collecting patient macroscopic characteristics and corresponding tissue biopsy data from historical clinical data and performing laboratory calibration.

[0053] Based on the differences in blood flow distribution, elastic modulus, and pH value in different regions of the multi-scale physiological parameter set and biofield map, a non-uniform grid partitioning strategy is constructed.

[0054] In this embodiment, the non-uniform mesh generation strategy is a strategy that adaptively adjusts the mesh density based on tissue characteristics. For example, high-gradient regions (such as blood vessel boundaries) have dense meshes with a size of 0.1-0.5 mm, while uniform regions (such as the center of healthy tissue) have sparse meshes with a size of 2-5 mm. During construction, the Delaunay adaptive triangulation algorithm is used, and physiological parameter gradient thresholds are set (such as blood flow velocity gradient > 0.1 cm / s / mm, elastic modulus change rate > 5 kPa / mm), and the local mesh density is automatically adjusted according to the thresholds.

[0055] Based on a non-uniform meshing strategy, the tissue is divided into heterogeneous units with different electromagnetic susceptibility and thermal conductivity characteristics.

[0056] In this embodiment, the heterogeneous unit is a tissue microregion with unique physical properties (electromagnetic sensitivity and thermal conductivity).

[0057] A multi-source data fusion method based on parameter assignment outputs tissue heterogeneity model parameters corresponding to different heterogeneous units.

[0058] The parameters of the tissue heterogeneity model are the electromagnetic parameters (relative permittivity, electrical conductivity, relative magnetic permeability), thermal parameters (thermal conductivity, specific heat capacity, density, blood perfusion rate), geometric parameters (unit volume, unit center coordinates), and physiological state parameters (baseline temperature, pH value, elastic modulus) of each heterogeneous unit. These parameters are obtained through multi-source data fusion methods for parameter assignment, including: a lookup table method based on tissue type, a continuous mapping method based on biofield maps, and an inference method based on multi-scale parameters.

[0059] A set of coupled equations is established using the finite element method. The parameters of the tissue heterogeneity model and the initial treatment mode are used as inputs to the set of coupled equations, and the multiphysics spatiotemporal evolution data matrix output by the set of coupled equations is obtained.

[0060] In this embodiment, the coupling equation set consists of partial differential equations describing the interaction of electromagnetic, thermal, and biochemical fields; the multiphysics spatiotemporal evolution data matrix is ​​a three-dimensional tensor that records the energy absorption, temperature changes, and cellular responses of each heterogeneous unit during treatment. The spatiotemporal evolution data matrix of multiphysics fields is input into a time series convolutional neural network to obtain dynamic response prediction results; the dynamic response prediction results are the future expression level of inflammatory factors in tissues, changes in microvascular permeability, and tissue repair speed.

[0061] In this embodiment, a time-series convolutional neural network (TCN) is used to predict future biochemical, vascular, and cellular responses based on a multiphysics spatiotemporal evolution data matrix. It employs a hybrid architecture of one-dimensional convolutional layers (1D-Conv) and LSTM, taking 10-30 minutes of multiphysics data as input and outputting 72-hour predictions. The network consists of three convolutional blocks (each with two convolutional layers, batch normalization, and ReLU activation) and two LSTM layers. During training, measured data from patient treatment processes collected based on historical clinical data (including real-time sensor data and pre- and post-treatment blood tests) are input into a pre-defined TCN for training, and 5-fold cross-validation is used for evaluation. Inflammatory factor expression levels are achieved through indirect mapping (e.g., mapping the rate of temperature change to IL-6 levels).

[0062] Based on the dynamic response prediction results and a pre-set clinical risk threshold library, a treatment outcome distribution is generated using the Monte Carlo simulation method; the treatment outcome distribution includes the risk probability of tissue burns, aggravated inflammation, and nerve damage.

[0063] In this embodiment, the preset clinical risk threshold library is a safety standard database established based on clinical experience and literature, containing a database of critical values ​​for various tissue injuries, such as tissue burn temperature thresholds and nerve injury field strength thresholds. A treatment outcome distribution is generated using a Monte Carlo simulation method: an adaptive Monte Carlo algorithm is used to apply normal distribution random perturbations to key parameters (such as tissue thermal conductivity ±15% and blood perfusion rate ±20%). Parallel computing is used to accelerate each simulation, which is repeated more than 1000 times. The frequency of adverse reactions is statistically analyzed to generate a risk probability distribution.

[0064] Based on the distribution of treatment results and real-time biofeedback data, the parameters of the digital twin model are dynamically corrected using a Bayesian update algorithm, and the corrected digital twin model and treatment effect prediction report are output.

[0065] In this embodiment, the variational Bayesian inference algorithm is used to define the prior distribution, and the likelihood function is used to calculate the posterior distribution based on real-time blood flow / temperature change measurements. When the difference between the model prediction and the actual measurement exceeds the difference threshold, a full parameter update is triggered.

[0066] The working principle and beneficial effects of the above technical solution are as follows: This invention first maps patient auxiliary features to microscopic physiological parameters at three scales—molecular, cellular, and tissue—using a scale-transformation neural network, thus constructing a multi-scale physiological basis. Then, based on the differences between these multi-scale physiological parameters and blood flow distribution, elastic modulus, and pH value in the biofield diagram, a Delaunay adaptive triangulation algorithm is used to construct a non-uniform mesh, dividing the tissue into heterogeneous units with different electromagnetic susceptibility and thermal conductivity characteristics. Electromagnetic, thermal, geometric, and physiological parameters are assigned to each unit using a multi-source data fusion method, forming a precise tissue heterogeneity model. Next, a three-field coupled equation system of electromagnetic, thermal, and biochemical fields is established using the finite element method to simulate the physical changes in tissue under ultra-shortwave radiation. The calculation results are input into a time-series convolutional neural network to predict future tissue inflammatory factor expression, microvascular permeability changes, and repair speed. Furthermore, combined with a clinical risk threshold library, over 1000 Monte Carlo simulations are performed to quantify the risk probability of tissue burns, inflammation exacerbation, and nerve damage. Finally, a Bayesian update algorithm is used to dynamically adjust model parameters based on real-time biofeedback data, forming a closed-loop optimization. This invention achieves precise mapping from macroscopic features to microscopic responses, enabling early identification of potential risks, significantly improving treatment safety, and ensuring the personalization and dynamic adaptability of the treatment process through a continuous model correction mechanism.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or high-voltage switchgear that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or high-voltage switchgear.

[0068] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A biofeedback-based, electromagnetic interference-free, ultra-shortwave intelligent physiotherapy adaptive control system, characterized in that: include: The biofeedback monitoring unit is configured to acquire, in real time, blood flow and velocity distribution maps, tissue elasticity, stiffness distribution maps, temperature distribution maps, and pH value distribution maps of the treatment target area through non-contact methods, forming a biofield map; The non-electromagnetic interference unit is configured to use a multi-modal programmable focused ultrasound phased array to generate ultra-short wave electromagnetic fields based on specific frequencies and pulse modes, in order to promote cell membrane permeability, improve microcirculation, and stimulate tissue repair. The intelligent analysis unit is configured to build a neural network model, taking the biofield map and treatment stage as input. The neural network model learns the complex mapping relationship between the biofield map and the healthy tissue pattern. Based on the current parameters and the biofield map, it predicts the tissue response trend in the future, judges its tolerance and response to ultra-shortwave therapy, and generates and outputs the optimal energy field regulation parameter instructions. The adaptive control unit is configured to automatically adjust the output power, frequency, and pulse mode parameters of the ultra-shortwave transmission module based on the optimal energy field control parameter command output, combined with patient feedback and changes in the biofield diagram, to achieve the best therapeutic effect.

2. The adaptive control system for biofeedback-based electromagnetic interference-free ultra-shortwave intelligent physiotherapy according to claim 1, characterized in that, The intelligent analysis unit includes: The auxiliary feature introduction module is configured to acquire the patient’s individual characteristics and real-time physiological data as static baseline features. Contextual features are introduced, including: whether the patient is currently in the acute phase, recovery phase, or rehabilitation phase; whether the goal of this treatment is to reduce inflammation, relieve pain, or promote tissue remodeling; and whether the patient is using anticoagulants or NSAIDs. These features are then fused with static baseline features as auxiliary features. The treatment effect simulation module is configured to take into account the current energy field parameters and biofield diagram, and predict the probability distribution of the patient's tissue response within a specified time period through a neural network model. By introducing auxiliary features, calculating the confidence interval of the prediction results, and comparing it with a preset threshold, an initial treatment pattern is automatically generated. By utilizing patients' auxiliary characteristics, a personalized digital twin model is constructed to simulate the treatment effect of the initial treatment mode and predict whether complications and adverse reactions will occur during the treatment process; and a Pareto front plot of treatment parameters and expected efficacy is generated to show the trade-off between efficacy and risk.

3. The adaptive control system for biofeedback-based electromagnetic interference-free ultra-shortwave intelligent physiotherapy according to claim 2, characterized in that, The treatment effect simulation module introduces auxiliary features to calculate the confidence interval of the predicted results, compares it with a preset threshold, and automatically generates an initial treatment mode, including: Obtain the predicted value vector of tissue response and its probability distribution parameters from the neural network model, and at the same time obtain the statistical features of the model training set; Based on the predicted value vector of tissue response, probability distribution parameters, and statistical characteristics of the model training set, a prediction data package is output. Receive the prediction data packet and calculate the initial confidence interval, then output the initial confidence interval data packet; The initial confidence interval is corrected by fusing auxiliary features to obtain the corrected confidence interval data packet; Perform a comparison judgment based on the corrected confidence interval data and the preset threshold, and output the comparison result data packet; The initial treatment pattern is generated by combining rules based on the comparison results data packets.

4. The adaptive control system for biofeedback-based electromagnetic interference-free ultra-shortwave intelligent physiotherapy according to claim 2, characterized in that, The treatment effect simulation module utilizes the patient's auxiliary characteristics to construct a personalized digital twin model, simulating the treatment effect of the initial treatment mode, including: The scale-transformation neural network maps auxiliary features to a multi-scale physiological parameter set, which includes molecular scale parameters, cellular scale parameters, and tissue scale parameters. Based on the differences in blood flow distribution, elastic modulus, and pH value in different regions of the multi-scale physiological parameter set and biofield map, a non-uniform grid partitioning strategy is constructed. Based on a non-uniform meshing strategy, the tissue is divided into heterogeneous units with different electromagnetic susceptibility and thermal conductivity. A multi-source data fusion method based on parameter assignment outputs tissue heterogeneity model parameters corresponding to different heterogeneous units; A set of coupled equations is established using the finite element method. The parameters of the tissue heterogeneity model and the parameters of the initial treatment mode are used as inputs to the set of coupled equations, and the multiphysics spatiotemporal evolution data matrix output by the set of coupled equations is obtained. The spatiotemporal evolution data matrix of multiphysics fields is input into a time series convolutional neural network to obtain dynamic response prediction results; Based on the dynamic response prediction results and a pre-set clinical risk threshold library, a treatment outcome distribution is generated using the Monte Carlo simulation method; the treatment outcome distribution includes the risk probability of tissue burns, aggravated inflammation, and nerve damage; Based on the distribution of treatment results and real-time biofeedback data, the parameters of the digital twin model are dynamically corrected using a Bayesian update algorithm, and the corrected digital twin model and treatment effect prediction report are output.

5. The adaptive control system for biofeedback-based electromagnetic interference-free ultra-shortwave intelligent physiotherapy according to claim 2, characterized in that, The adaptive control unit includes: The strategy dynamic adjustment module is configured to evaluate the confidence interval of the prediction results and determine their reliability; based on the confidence interval of the prediction results, an initial treatment pattern is generated. By combining the Pareto frontier plot of treatment parameters and expected efficacy, the efficacy and risks of different combinations of treatment parameters are evaluated. Based on the treatment simulation results, the optimal combination of treatment parameters is generated; if the simulation results show that there is an imbalance between the efficacy and risk of the current treatment mode, the treatment parameters are automatically adjusted to optimize the treatment mode. Establish a real-time feedback mechanism to dynamically adjust personalized treatment models based on patients' real-time responses during treatment; Treatment parameters are adjusted in real time based on real-time feedback data and preset treatment goals; Establish an intelligent early warning mechanism that immediately issues an early warning signal and automatically initiates intervention strategies when an abnormal situation is detected.

6. The adaptive control system for intelligent physiotherapy based on biofeedback and electromagnetic interference-free ultra-shortwave therapy according to claim 5, characterized in that, The adaptive control unit further includes: The multi-dimensional threshold adjustment module is configured to dynamically adjust the preset threshold of the confidence interval based on the patient's individual characteristics and real-time physiological state. The multi-dimensional threshold comparison module is configured to perform multi-dimensional comparisons of the type and extent of tissue response when comparing the confidence intervals of the prediction results.

7. The adaptive control system for intelligent physiotherapy based on biofeedback and electromagnetic interference-free ultra-shortwave therapy according to claim 1, characterized in that, The electromagnetic interference-free unit includes: The parameter adjustment module is configured to acquire multi-dimensional characteristics of the patient and flexibly adjust parameters such as ultrasound frequency, pulse width, and pulse repetition frequency according to different treatment needs and patient conditions to achieve personalized treatment control. The electromagnetic interference suppression module is configured to use advanced electromagnetic shielding materials and technologies to ensure that broadband electromagnetic radiation is not generated while generating ultra-shortwave electromagnetic fields; and through electromagnetic compatibility design, it ensures that the equipment operates safely and stably in complex electromagnetic environments without adversely affecting surrounding medical equipment and personnel.

8. The adaptive control system for intelligent physiotherapy based on biofeedback and electromagnetic interference-free ultra-shortwave therapy according to claim 1, characterized in that, The biofeedback monitoring unit includes: The multimodal data acquisition module is configured to measure blood flow velocity and blood flow distribution in microvessels in real time through the laser Doppler effect; acquire high-resolution tomographic images of tissues in real time using the principle of optical coherence, including blood flow and velocity distribution maps; and acquire tissue temperature distribution maps in real time using infrared thermal imaging technology. The feature extraction module is configured to preprocess the collected multimodal data, including filtering, denoising, and normalization operations. Through data analysis algorithms, characteristic parameters reflecting the physiological state and treatment response of tissues are extracted from the biofield diagram.

9. The adaptive control system for intelligent physiotherapy based on biofeedback and electromagnetic interference-free ultra-shortwave therapy according to claim 1, characterized in that, Also includes: The user interaction unit is configured to provide an operation interface for medical staff to input patient information, set treatment parameters, and view physiological data and equipment status during treatment. The treatment record unit is configured to automatically record the detailed process of each treatment, automatically generate a treatment report, and include, but are not limited to, comparisons of physiological data before and after treatment, evaluation of treatment effects, and export them in a standard format for easy archiving and sharing.

10. The adaptive control system for intelligent physiotherapy based on biofeedback and electromagnetic interference-free ultra-shortwave therapy according to claim 9, characterized in that, The user interaction unit includes: The data encryption module is configured to encrypt stored and transmitted patient information and treatment data to prevent unauthorized access. The data sharing module is configured to provide a standard data interface, supporting data sharing with medical devices and hospital information systems.

Images