Brain function recovery device and system based on photobiomodulation and transcranial magnetic combination

Abstract

The present invention discloses a brain function recovery device and system based on the combination of photobiomodulation and transcranial magnetism. The device includes: a data acquisition module for acquiring initial physiological state data, including EEG data, heart rate data, blood oxygen data and skin conductance data; a parameter optimization module for optimizing the initial parameters of the model and initializing the model according to the initial physiological state data to obtain an initial control model; a model training module for training according to historical physiological state data and historical control data as input data to obtain a standard control model; a control module for inputting the physiological state data into the standard control model when the control time is less than or equal to the preset maximum physical therapy time, adjusting the control strategy until the control time is greater than the preset maximum physical therapy time to obtain an optimal control strategy; and an execution module for generating a pulsed physical therapy magnetic field according to the optimal control strategy. The device can adjust the physical therapy strategy in real time and simplify the physical therapy process.

Description

Technical Field

[0001] The present invention relates to the technical field of brain function therapy devices, and in particular to a brain function recovery device and system based on the combination of photobiomodulation and transcranial magnetism. Background Art

[0002] Transcranial magnetic stimulation is a non-invasive neuromodulation technology that stimulates specific areas of the brain through magnetic fields, which can regulate neural activity and thus relieve depression, anxiety, neuralgia and other brain diseases. With the deepening of research on brain function, the demand for transcranial magnetic devices that can be precisely controlled and personalized is growing.

[0003] In the existing technology, transcranial magnetic stimulation devices are usually equipped with an adjustable magnetic field generator and a control unit, allowing medical personnel to set stimulation parameters such as frequency, intensity, and duration according to the patient's specific situation. In addition, some advanced devices also integrate sensors and feedback systems to monitor the patient's physiological response and magnetic field distribution, thereby achieving personalized and precise control of physical therapy to a certain extent.

[0004] Existing technologies require manual adjustment of parameters in actual applications, which is a cumbersome and time-consuming process. There is a lack of efficient real-time decision-making mechanism, and it is impossible to adjust the therapy strategy in time according to subtle changes during the therapy process. Summary of the invention

[0005] The present invention provides a brain function recovery device and system based on the combination of photobiomodulation and transcranial magnetism to achieve real-time and automatic adjustment of physical therapy strategies, simplify the physical therapy process, and achieve flexible and efficient physical therapy.

[0006] In the first aspect, in order to solve the above technical problems, the present invention provides a brain function recovery device based on photobiomodulation and transcranial magnetic combination, comprising:

[0007] A data acquisition module, used to acquire initial physiological state data, wherein the initial physiological state data includes EEG data, heart rate data, blood oxygen data and skin conductance data;

[0008] A parameter optimization module, used to optimize the initial parameters of the model according to the initial physiological state data and use the initial parameters to initialize the model to obtain an initial control model;

[0009] A model training module, used to perform experience playback training on the initial control model based on historical physiological state data and historical control data as input data to obtain a standard control model;

[0010] A control module, for inputting the current physiological state data into the standard control model when the control time is less than or equal to the preset maximum physical therapy time, and cyclically adjusting the control strategy until the control time is greater than the preset maximum physical therapy time, thereby obtaining an optimal control strategy;

[0011] The execution module is used to generate a corresponding pulse therapy magnetic field according to the optimal control strategy.

[0012] As an optional implementation, the parameter optimization module includes:

[0013] A genetic module, used for initializing individuals in the population and performing fitness evaluation on the individuals in the population according to the initial physiological state data, so as to obtain the fitness of the individuals in the population;

[0014] A crossover module, used for performing a crossover operation on individuals in the population to obtain crossover individuals;

[0015] A mutation module, used for performing a mutation operation according to the crossover individuals to obtain mutant individuals;

[0016] A selection module is used to select individuals with high fitness from the initial population individuals and the variant individuals according to the fitness of the initial population individuals and the variant individuals in the population to enter the next generation, so as to gradually optimize the initial parameters of the model;

[0017] An initialization module, used to initialize the model using the initial parameters to obtain an initial control model;

[0018] Among them, the individuals in the population represent a model network parameter, including the weight value and bias term connecting the input layer and the hidden layer, and the weight value and bias term connecting the hidden layer and the output layer.

[0019] As an optional implementation, the genetic module is used to:

[0020] Randomly initialize the model network structure;

[0021] Initialize the population according to the model network structure;

[0022] According to the initial physiological state data and the model network structure, the fitness of the individual population is evaluated, and the calculation formula is as follows:

[0023]

[0024] in, Indicates The fitness value of each individual; Indicates the total number of initial physiological state data; Represents the instant reward of the jth sample data; represents the discount factor; represents the target Q-value function; represents the action Q-value function; Indicates the next state, including the EEG data, the heart rate data, the blood oxygen data and the skin conductance data; Indicates that according to the next state The control strategies adopted; Indicates Network parameters of each individual; Represents the physiological state of the jth sample data; represents the control strategy adopted according to the j-th sample data and the model network; Represents the maximum value function.

[0025] As an optional implementation manner, the cross module is used to:

[0026] Based on the individuals in the population, a candidate solution is initialized for each individual;

[0027] Based on the individuals in the population, a random number is generated for each dimension of each individual;

[0028] According to the candidate solution and the random number, a cross calculation is performed, and the calculation formula is as follows:

[0029]

[0030] in, Indicates at time At that time, The cross-individual of the j-th dimension of individuals; represents the j-th dimension of the mutation vector generated in the mutation module; represents the jth dimension of the current individual; Represents a random number generation function that generates a random number in the interval [0,1]; represents the crossover probability; Represents a randomly chosen dimension index, ensuring that at least one dimension is inherited from the mutation vector.

[0031] As an optional implementation, the mutation module is used to:

[0032] Based on the crossover individuals in the population, randomly select a crossover individual as a basis vector;

[0033] Based on the crossover individuals in the population, randomly select another two different crossover individuals, and calculate the difference between the two different crossover individuals to obtain a difference vector;

[0034] Based on the basis vector and the difference vector, a mutation calculation is performed to obtain a mutant individual. The calculation formula is as follows:

[0035]

[0036] in, represents a variant individual; represents basis vectors; Represents the mutation factor, which is used to control the degree of mutation; represents the difference vector.

[0037] As an optional implementation manner, the selection module is used to:

[0038] Based on the initial population individuals and the mutant individuals in the population, the mutant individual fitness is calculated to obtain the mutant individual fitness;

[0039] When the fitness of the variant individual is greater than the fitness of the initial population individual, the initial population individual is updated to the variant individual;

[0040] When the fitness of the candidate solution is less than or equal to the fitness of the individual in the initial population, the individual in the initial population is kept unchanged;

[0041] According to the individuals in the updated population, the individuals with the highest fitness are selected as the next generation of population individuals to gradually optimize the model parameters.

[0042] As an optional implementation, the model training module is used to:

[0043] Collect and store historical physiological status data and historical control data;

[0044] Using the historical physiological state data and the historical control data as input data, an action Q value and a target Q value are obtained by calculation through an initial control model;

[0045] Performing a timing difference error calculation according to the action Q value and the target Q value to obtain a timing difference error value;

[0046] According to the time series difference error value, parameter optimization is performed to obtain model optimization parameters;

[0047] Based on the model optimization parameters, the control model is initialized to carry out the next round of training;

[0048] When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed and a standard control model is obtained.

[0049] As an optional implementation manner, the parameter optimization is performed according to the timing difference error value to obtain the model optimization parameters, and the parameter optimization formula is as follows:

[0050] ,

[0051] in, Represents model parameters The loss function of represents the time series difference error value function; Indicates immediate reward; represents the discount factor; and Represents the parameters of the current and target networks; Indicates the next state, including the EEG data, the heart rate data, the blood oxygen data and the skin conductance data; Indicates that according to the next state The control strategies adopted; represents the learning rate; represents the loss value; represents the expected function; Indicates the current state; Indicates that according to the current state Control strategies adopted.

[0052] As an optional implementation manner, the execution module further includes:

[0053] An execution time module, used for, according to the optimal control strategy, executing the strategy according to the preset time so that the pulse magnetic field output module generates the corresponding pulse therapy magnetic field;

[0054] The strategy update module is used to re-collect the current physiological state data, generate the next round of control strategy and execute it when the execution time is greater than or equal to the preset strategy execution time.

[0055] In a second aspect, the present invention provides a brain function recovery system based on the combination of photobiomodulation and transcranial magnetic resonance imaging, the system comprising a head-mounted module, a touch control module, a pulsed magnetic field output module, a coil module, and a brain function recovery device based on the combination of photobiomodulation and transcranial magnetic resonance imaging as described in any of the above items.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention provides a brain function recovery device based on the combination of photobiomodulation and transcranial magnetism, comprising: a data acquisition module, used to acquire initial physiological state data, wherein the initial physiological state data includes electroencephalogram data, heart rate data, blood oxygen data and skin conductance data; a parameter optimization module, used to optimize the initial parameters of the model according to the initial physiological state data and use the initial parameters to initialize the model to obtain an initial control model; a model training module, used to perform experience playback training on the initial control model according to historical physiological state data and historical control data as input data to obtain a standard control model; a control module, used to input the current physiological state data into the standard control model when the control time is less than or equal to the preset maximum physical therapy time, cyclically adjust the control strategy until the control time is greater than the preset maximum physical therapy time, and obtain the optimal control strategy; an execution module, used to generate a corresponding pulsed physical therapy magnetic field according to the optimal control strategy. The present invention trains a standard control model by integrating data acquisition, parameter optimization, model training, self-control and execution modules, generates and adjusts the control strategy according to the current physiological state, thereby realizing real-time and automatic adjustment of the control strategy, simplifying the physical therapy process, and improving the flexibility and efficiency of the physical therapy strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a structural schematic diagram of a brain function recovery device based on photobiomodulation and transcranial magnetism provided by an embodiment of the present invention;

[0059] Figure 2 It is a schematic diagram of the structure of a brain function recovery system based on the combination of photobiomodulation and transcranial magnetism provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] Transcranial magnetic stimulation is a non-invasive neuromodulation technology that stimulates specific areas of the brain through magnetic fields, which can regulate neural activity and thus treat a variety of brain diseases such as depression, anxiety, and neuralgia. With the deepening of research on brain function, the demand for transcranial magnetic devices that can be precisely controlled and personalized is growing.

[0062] In the existing technology, transcranial magnetic stimulation devices are usually equipped with an adjustable magnetic field generator and a control unit, allowing medical personnel to set stimulation parameters such as frequency, intensity, and duration according to the patient's specific situation. In addition, some advanced devices also integrate sensors and feedback systems to monitor the patient's physiological response and magnetic field distribution, thereby achieving personalized and precise control of physical therapy to a certain extent.

[0063] In summary, the existing technology requires manual adjustment of parameters in actual application, which is a cumbersome and time-consuming process. It lacks an efficient real-time decision-making mechanism and cannot adjust the physical therapy strategy in time according to subtle changes in the physical therapy process.

[0064] In order to solve the above problems, a brain function recovery device based on the combination of photobiomodulation and transcranial magnetism provided in an embodiment of the present application will be introduced and explained in detail through the following specific embodiments.

[0065] This embodiment provides a brain function recovery device based on photobiomodulation and transcranial magnetism, wherein: Figure 1 It is a specific component module of a brain function recovery device based on photobiomodulation and transcranial magnetism in an embodiment of the present invention, such as Figure 1 As shown, the component modules include a data acquisition module 10, a parameter optimization module 20, a model training module 30, a control module 40 and an execution module 50.

[0066] The data acquisition module 10 is used to acquire initial physiological state data, wherein the initial physiological state data includes EEG data, heart rate data, blood oxygen data and skin conductance data.

[0067] The parameter optimization module 20 is used to optimize the initial parameters of the model according to the initial physiological state data and use the initial parameters to initialize the model to obtain an initial control model.

[0068] The model training module 30 is used to perform experience playback training on the initial control model based on historical physiological state data and historical control data as input data to obtain a standard control model.

[0069] The control module 40 is used to input the current physiological state data into the standard control model when the control time is less than or equal to the preset maximum therapy time, and cyclically adjust the control strategy until the control time is greater than the preset maximum therapy time to obtain the optimal control strategy.

[0070] The execution module 50 is used to generate a corresponding pulse therapy magnetic field according to the optimal control strategy.

[0071] In the data acquisition module, initial physiological state data is obtained, including EEG data, heart rate data, blood oxygen data and skin conductance data.

[0072] The preset maximum therapy time refers to the upper limit of the duration of therapy in the transcranial magnetic brain functional therapy system, which is preset to ensure patient safety and therapy effect. This time limit is based on clinical experience and research data, and is intended to avoid the risks and side effects that may be caused by excessive therapy. In this embodiment, the preset maximum therapy time is set to 1 hour. Depending on the specific therapy situation, the preset maximum therapy time can also be set to 30 minutes, 2 hours, etc., which is not limited by the present invention.

[0073] It should be noted that in the data acquisition module, the process of obtaining initial physiological state data is crucial, because these data provide the basis for subsequent parameter optimization, model training, and control strategy adjustment.

[0074] In specific applications, EEG data refers to brain electrical activity signals recorded by an EEG device, which are expressed as brain waves of different frequencies and amplitudes. Multiple electrodes are placed on the patient's scalp to capture weak electrical signals generated by the brain, which are then amplified and filtered to evaluate the functional state of the brain, monitor changes in EEG activity, guide the parameter setting of transcranial magnetic therapy, and evaluate the effect of therapy. Heart rate data refers to the heart rate measured by an electrocardiogram or pulse sensor. In this embodiment, it is expressed in heart beats per minute. Electrodes or sensors attached to the patient's chest or wrist detect heart electrical activity or pulse waveforms, and calculate heart rate. It is used to monitor the patient's heart function and physiological stress state, ensure the safety of transcranial magnetic therapy, and evaluate the patient's overall physiological response. Blood oxygen data is obtained through a pulse oximeter, which reflects the oxygen content in the blood. Changes in blood oxygen levels can indicate the patient's respiratory state, circulatory system function, or physiological changes during therapy. Skin conductance data is acquired through a skin conductance sensor and reflects the electrical conductivity of the skin. Changes in skin conductance can indicate the patient's emotional state, physiological stress, or activity of the autonomic nervous system.

[0075] In the parameter optimization module, according to the initial physiological state data, the initial parameters of the model are optimized and the model is initialized to obtain an initial control model, including:

[0076] A genetic module, used for initializing individuals in the population and performing fitness evaluation on the individuals in the population according to the initial physiological state data, so as to obtain the fitness of the individuals in the population;

[0077] A crossover module, used for performing a crossover operation on individuals in the population to obtain crossover individuals;

[0078] A mutation module, used for performing a mutation operation according to the crossover individuals to obtain mutant individuals;

[0079] A selection module is used to select individuals with high fitness from the initial population individuals and the variant individuals according to the fitness of the initial population individuals and the variant individuals in the population to enter the next generation, so as to gradually optimize the initial parameters of the model;

[0080] An initialization module, used to initialize the model using the initial parameters to obtain an initial control model;

[0081] Among them, the individuals in the population represent a model network parameter, including the weight value and bias term connecting the input layer and the hidden layer, and the weight value and bias term connecting the hidden layer and the output layer.

[0082] It should be noted that the parameter optimization module adopts the genetic algorithm framework to simulate the process of natural selection and genetic evolution through operations such as inheritance, crossover, mutation and selection to optimize the model parameters. Among them, crossover and mutation operations increase the diversity of the population, improve the exploration ability of the algorithm, and help to find the global optimal solution; the selection operation gradually optimizes the model parameters by retaining individuals with high fitness, and promotes the algorithm to converge to a better solution. The fitness of an individual is an indicator to measure the performance of the weight matrix and bias vector represented by the individual in the current iteration. The smaller the fitness, the smaller the error of the model corresponding to the individual, and the better the performance of the model. In the genetic optimization algorithm, fitness is used to guide the movement of individuals in the search space to find the optimal weight matrix and the optimal bias vector.

[0083] In this embodiment, the genetic module is used to perform the following steps:

[0084] Randomly initialize the model network structure;

[0085] Initialize the population according to the model network structure;

[0086] According to the initial physiological state data and the model network structure, the fitness of the individual population is evaluated, and the calculation formula is as follows:

[0087]

[0088] in, Indicates The fitness value of each individual; Indicates the total number of initial physiological state data; Represents the instant reward of the jth sample data; represents the discount factor; represents the target Q-value function; represents the action Q-value function; Indicates the next state, including the EEG data, the heart rate data, the blood oxygen data and the skin conductance data; Indicates that according to the next state The control strategies adopted; Indicates Network parameters of each individual; Represents the physiological state of the jth sample data; represents the control strategy adopted according to the j-th sample data and the model network; Represents the maximum value function.

[0089] It is worth noting that the model network structure includes two Q value networks, an action Q value network and a target Q value network. The Q value network mainly has three layers, including an input layer, a hidden layer, and an output layer. Among them, the input layer receives the patient's physiological state data as input; the hidden layer is composed of multiple neurons, which are used to extract and process the features of the input data, and usually adopts nonlinear activation functions such as ReLU, sigmoid and other functions to increase the expression ability of the model; the output layer generates the Q value of all possible actions in the current state, indicating the expected effect of taking a certain action in the current physiological state. The action Q value network is used to evaluate the value of the current state and action in real time, and the target Q value network is used to provide a stable target Q value to reduce instability during training. The parameters of the target Q network are regularly updated from the main Q network, but the update frequency is low to maintain the stability of the target. The fitness calculation formula is a key component in the genetic algorithm, which is used to evaluate the performance of each individual in the population, thereby guiding the evolutionary direction of the population. In this embodiment, the fitness is calculated based on the average square error of the individual's time series difference error on all sample data.

[0090] Among them, the control strategy refers to the method or rule used to adjust the transcranial magnetic therapy parameters under a specific physiological state. The therapy parameters include magnetic field intensity, frequency and duration, etc. The control strategy is used to adjust the therapy parameters in real time during transcranial magnetic therapy to adapt to the individual differences of patients and changes in the therapy process, which helps to improve the personalization and accuracy of therapy and ensure the maximum effect of therapy.

[0091] In a specific application, the control module receives real-time physiological state data from the head-mounted module, and inputs these data as input data into the standard control model for real-time analysis to identify the patient's current physiological state and physical therapy needs, and generate corresponding control strategies. The control module adjusts the transcranial magnetic therapy parameters according to the control strategy, including magnetic field strength, frequency, and duration. The control module can dynamically adjust according to the real-time data during the physical therapy process to adapt to the patient's physiological changes and physical therapy responses. In other words, the optimal control strategy in the current state is generated based on the current data and historical data, and the strategy duration is preset; after the duration is reached, the optimal strategy is adjusted again based on the current data and the historical data within this duration to obtain the optimal strategy within the next duration. Exemplarily, when the patient's EEG data shows an increase in abnormal activity, the control module may increase the magnetic field strength and frequency to better control the abnormal activity; when the patient's skin conductance data shows signs of tension or anxiety, the control module may adjust the magnetic field strength and duration to alleviate these symptoms.

[0092] In this embodiment, the cross-connect module is used to perform the following steps:

[0093] Based on the individuals in the population, a candidate solution is initialized for each individual;

[0094] Based on the individuals in the population, a random number is generated for each dimension of each individual;

[0095] According to the candidate solution and the random number, a cross calculation is performed, and the calculation formula is as follows:

[0096]

[0097] in, Indicates at time At that time, The cross-individual of the j-th dimension of individuals; represents the j-th dimension of the mutation vector generated in the mutation module; represents the jth dimension of the current individual; Represents a random number generation function that generates a random number in the interval [0,1]; represents the crossover probability; Represents a randomly chosen dimension index, ensuring that at least one dimension is inherited from the mutation vector.

[0098] It should be noted that the crossover module is a core component in the genetic algorithm, which is responsible for exchanging information between the genes of individuals in the population to generate new individuals. It generates offspring individuals by mixing the characteristics of the parent individuals, thereby introducing new genetic variations. The candidate solution represents a dimension of the newly generated individual. The crossover probability represents a preset parameter that determines the probability of inheriting a dimension from the mutation vector and is a key control parameter. In this embodiment, the crossover probability is set to 0.25. Depending on the application scenario and user needs, the crossover probability can also be set to 0.1, 0.3, etc., which is not limited by the present invention.

[0099] Wherein, based on the individuals in the population, a random number is generated for each dimension of each individual, and the random number is used to determine whether to inherit the value of the dimension from the mutation vector or keep the value of the dimension of the current individual unchanged.

[0100] In this embodiment, the mutation module is used to perform the following steps:

[0101] Based on the crossover individuals in the population, randomly select a crossover individual as a basis vector;

[0102] Based on the crossover individuals in the population, randomly select another two different crossover individuals, and calculate the difference between the two different crossover individuals to obtain a difference vector;

[0103] Based on the basis vector and the difference vector, a mutation calculation is performed to obtain a mutant individual. The calculation formula is as follows:

[0104]

[0105] in, represents a variant individual; represents basis vectors; Represents the mutation factor, which is used to control the degree of mutation; represents the difference vector.

[0106] It should be noted that the mutation module is a key component in the genetic algorithm, responsible for introducing new genetic mutations in the population. It generates new individuals by adding random changes to existing individuals. The basis vector is the starting point of the mutation operation and represents the basic characteristics of the mutant individual. It is an individual randomly selected from the population and used as the starting point of the mutation operation to provide basic characteristics for generating new mutant individuals. The difference vector represents the genetic difference between two different individuals. It is obtained by calculating the difference between two different individuals randomly selected from the population. It is used to introduce new genetic information and increase the diversity of the population by combining with the basis vector.

[0107] Among them, the mutation factor is a preset parameter used to adjust the degree of mutation, ensuring that the mutation operation can introduce sufficient diversity without destroying the excellent characteristics of existing individuals. It determines the degree of influence of the difference vector on the basis vector. In this embodiment, the mutation factor is set to 0.4. Depending on the application scenario and user needs, the crossover probability can also be set to 0.1, 0.5, etc., which is not limited by the present invention.

[0108] In this embodiment, the selection module is used to perform the following steps:

[0109] Based on the initial population individuals and the mutant individuals in the population, the mutant individual fitness is calculated to obtain the mutant individual fitness;

[0110] When the fitness of the variant individual is greater than the fitness of the initial population individual, the initial population individual is updated to the variant individual;

[0111] When the fitness of the candidate solution is less than or equal to the fitness of the individual in the initial population, the individual in the initial population is kept unchanged;

[0112] According to the individuals in the updated population, the individuals with the highest fitness are selected as the next generation of population individuals to gradually optimize the model parameters.

[0113] It is worth noting that the formula for the above selection process is as follows:

[0114] ,

[0115] in, Indicates that at time At that time, The current state of the initial population individuals; Indicates at time At that time, The variant individuals of the initial population; Represents a variant individual The fitness value of Indicates Initial population individuals The fitness value of Indicates that at time At that time, The final choice of an individual.

[0116] It is worth mentioning that the selection operation is a core component in the genetic algorithm, which is responsible for deciding which individuals will be retained to the next generation based on their fitness values. Through these steps, the selection operation can effectively retain individuals with better performance in the genetic algorithm, eliminate individuals with poor performance, and promote the evolution of the population towards a better solution. The selection process is based on fitness, and only candidate solutions that are better than the current solution are accepted, which helps the algorithm converge quickly to the local optimal solution. If the candidate solution is not better than the current solution, the current solution is retained, which helps maintain the diversity of the population and avoid premature convergence.

[0117] In this embodiment, the training module is used to perform the following steps:

[0118] Collect and store historical physiological status data and historical control data;

[0119] Using the historical physiological state data and the historical control data as input data, an action Q value and a target Q value are obtained by calculation through an initial control model;

[0120] Performing a timing difference error calculation according to the action Q value and the target Q value to obtain a timing difference error value;

[0121] According to the time series difference error value, parameter optimization is performed to obtain model optimization parameters;

[0122] Based on the model optimization parameters, the control model is initialized to carry out the next round of training;

[0123] When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed and a standard control model is obtained.

[0124] Among them, the maximum number of training times is a preset termination condition used to control the duration of the training process. When the maximum number of training times is reached, it is considered that the model has been fully trained and can be put into practical use as a standard control model. In this embodiment, the preset maximum number of training times is set to 200 times, which is a result of experimental verification and ensures that a calibration control model with acceptable performance is obtained within a reasonable time. However, the maximum number of training times is not fixed and may need to be adjusted accordingly according to the characteristics of the specific problem and the needs of the actual application. For example, the maximum number of training times can also be set to 100, 300, etc., which is not limited by the present invention.

[0125] It should be noted that historical physiological state data and historical control data are the basis for training models. These data include the patient's physiological indicators and corresponding physical therapy parameters at different time points. These data are used to train the model so that it can learn the relationship between physiological state and physical therapy parameters, so as to make more reasonable decisions in practical applications. Action Q value and target Q value are core concepts in reinforcement learning, representing the expected long-term reward and target long-term reward for taking an action in a specific state, respectively. By calculating these values, the model can evaluate the potential value of different actions and provide a basis for selecting the optimal action.

[0126] In this embodiment, according to the time series difference error value, parameter optimization is performed to obtain model optimization parameters, and the parameter optimization formula is as follows:

[0127] ,

[0128] in, Represents model parameters The loss function of represents the time series difference error value function; Indicates immediate reward; represents the discount factor; and Represents the parameters of the current and target networks; Indicates the next state, including the EEG data, the heart rate data, the blood oxygen data and the skin conductance data; Indicates that according to the next state The control strategies adopted; represents the learning rate; represents the loss value; represents the expected function; Indicates the current state; Indicates that according to the current state Control strategies adopted.

[0129] It should be noted that the temporal difference (TD) error is an indicator of the accuracy of model prediction, which reflects the difference between the current prediction and the actual reward. The temporal difference error is used to guide the parameter update of the model, and improve the prediction ability of the model by reducing the error. Parameter optimization is the process of reducing the prediction error by adjusting the model parameters. Through parameter optimization, the model can better adapt to the data and improve its performance in practical applications. In this embodiment, the stochastic gradient descent algorithm is used for parameter optimization. Of course, the parameter optimization process can also use adaptive learning rate method and momentum method, etc., which are not limited by the present invention.

[0130] In this embodiment, the execution module also includes the following modules:

[0131] An execution time module, used for, according to the optimal control strategy, executing the strategy according to the preset time so that the pulse magnetic field output module generates the corresponding pulse therapy magnetic field;

[0132] The strategy update module is used to re-collect the current physiological state data, generate the next round of control strategy and execute it when the execution time is greater than or equal to the preset strategy execution time.

[0133] It should be noted that the pulse therapy magnetic field is generated according to the optimal control strategy, which includes specific parameters such as magnetic field strength, frequency and duration. The strategy execution time refers to the length of time that this optimal control strategy will continue to be executed. In this embodiment, the preset strategy execution time is set to 5 minutes. Of course, depending on the application scenario and user needs, the strategy execution time can also be set to 3 minutes, 10 minutes, etc., and the present invention is not limited to this. After the strategy execution time ends, the device will re-collect the patient's current physiological state data and generate a new control strategy based on these new data. By dynamically adjusting the control strategy, the device can adapt to changes in the patient's physiological state and ensure that the adjustment of the therapy parameters is based on the patient's real-time physiological state, thereby improving the personalization and effectiveness of the therapy.

[0134] In order to facilitate the understanding of the present invention, some preferred embodiments of the present invention are further described below.

[0135] In this embodiment, the transcranial magnetic brain function therapy control device is installed in the neurotherapy area of ​​a professional medical institution, which involves the application of transcranial magnetic technology in the treatment of depression. The control device is a transcranial magnetic brain function therapy system, and the data acquisition device is the patient's physiological monitoring equipment.

[0136] The working process is as follows:

[0137] In step 1, the patient enters the physical therapy room, and the doctor uses a control device to connect the patient's physiological monitoring equipment, including electroencephalogram (EEG), electrocardiogram (ECG), pulse oximeter, and skin conductance sensor;

[0138] Step 2: The physiological monitoring device starts to collect the patient's initial physiological state data and sends it to the control device;

[0139] Step 3: The control device obtains the current physiological state data and inputs it into a pre-trained standard control model to obtain the optimal control strategy;

[0140] Step 4: According to the optimal control strategy, the control device generates a corresponding pulse therapy magnetic field and adjusts the magnetic field strength, frequency and duration of the transcranial magnetic device;

[0141] Step 5: The transcranial magnetic device starts to perform physical therapy on the patient for a preset strategy execution time;

[0142] Step 6: When the execution time is greater than or equal to the preset strategy execution time, the control device re-collects the current physiological state data, generates the next round of control strategy, and continues to perform physical therapy until the predetermined physical therapy cycle is completed;

[0143] Step 7. After the physical therapy is completed, the doctor evaluates the patient's response and the physical therapy effect, and adjusts the subsequent physical therapy plan as needed.

[0144] Through this process, the control device is able to achieve precise control of transcranial magnetic therapy, ensuring that the adjustment of therapy parameters is based on the patient's real-time physiological state, thereby improving the personalization and effectiveness of therapy.

[0145] In one embodiment, the pre-trained standard control model in step 3 includes:

[0146] By collecting the patient's historical physiological state data and historical control data as input, the action Q value and target Q value are calculated through the initial control model; according to the timing difference error calculation, parameter optimization is performed to obtain the model optimization parameters; based on the model optimization parameters, the control model is initialized and the next round of training is carried out until the number of training times reaches the preset maximum number of training times, and a pre-trained standard control model is obtained.

[0147] In this embodiment, the pre-trained standard control model is trained using the patient's historical data, and parameters are optimized. The number of training times reaches a preset maximum number of training times to obtain a standard control model.

[0148] In another embodiment, the pre-trained standard control model in step 3 includes:

[0149] By pre-training the basic model with large-scale historical physical therapy data, the pre-trained model parameters are transferred to the standard control model, and the patient's specific data is used to fine-tune it to adapt to the physiological state of the specific patient, thus obtaining a pre-trained standard control model.

[0150] In this embodiment, the pre-trained standard control model is trained by large-scale historical physical therapy data, and parameters are optimized, and the trained basic model is fine-tuned to obtain the pre-trained standard control model.

[0151] Reference Figure 2 The second embodiment of the present invention provides a brain function recovery system based on photobiomodulation and transcranial magnetic combination, comprising:

[0152] The head-mounted module is used to collect the patient's physiological status data, including EEG data, ECG data, blood oxygen data, and skin conductance data;

[0153] The control module is used to receive physiological data from the head-mounted module and generate a physical therapy strategy based on a pre-trained standard control model. It is responsible for communicating with the touch control module and the pulsed magnetic field output module to ensure the accurate execution of the physical therapy strategy.

[0154] The touch module is used to provide a user-friendly interface, allowing doctors or patients to manually input physical therapy parameters or adjust physical therapy progress, and display real-time physical therapy data and patient physiological status so that doctors can monitor and adjust in real time;

[0155] A pulse magnetic field output module, used for generating and outputting the pulse magnetic field required for physical therapy;

[0156] The coil module is used to accurately transmit the magnetic field generated by the pulsed magnetic field output module to a specific brain area of ​​the patient.

[0157] It should be noted that a brain function recovery system based on photobiomodulation and transcranial magnetic combination provided by an embodiment of the present invention includes a brain function recovery device based on photobiomodulation and transcranial magnetic combination of the above-mentioned embodiment, and the working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0158] In summary, the present invention provides a brain function recovery device based on the combination of photobiomodulation and transcranial magnetism, including: a data acquisition module, used to acquire initial physiological state data, the initial physiological state data including EEG data, heart rate data, blood oxygen data and skin conductance data; a parameter optimization module, used to optimize the initial parameters of the model according to the initial physiological state data and use the initial parameters to initialize the model to obtain an initial control model; a model training module, used to perform experience playback training on the initial control model based on historical physiological state data and historical control data as input data to obtain a standard control model; a control module, used to input the current physiological state data into the standard control model when the control time is less than or equal to the preset maximum therapy time, and cyclically adjust the control strategy until the control time is greater than the preset maximum therapy time to obtain an optimal control strategy; an execution module, used to generate a corresponding pulse therapy magnetic field according to the optimal control strategy. The present invention trains a standard control model by integrating data acquisition, parameter optimization, model training, self-control and execution modules, generates and adjusts the control strategy according to the current physiological state, thereby realizing real-time and automatic adjustment of the control strategy, simplifying the physical therapy process, and improving the flexibility and efficiency of the physical therapy strategy.

[0159] An embodiment of the present invention also provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a brain function recovery program based on the combination of photobiomodulation and transcranial magnetism. When the processor executes the computer program, the steps in the above-mentioned brain function recovery device embodiments based on the combination of photobiomodulation and transcranial magnetism are implemented. Alternatively, when the processor executes the computer program, the functions of each module / unit in the above-mentioned system embodiments are implemented.

[0160] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0161] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or may combine certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0162] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0163] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0164] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or system capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0165] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0166] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A brain function recovery device based on photobiomodulation and transcranial magnetic resonance imaging, characterized in that: include: A data acquisition module, used to acquire initial physiological state data, wherein the initial physiological state data includes EEG data, heart rate data, blood oxygen data and skin conductance data; A parameter optimization module, used to optimize the initial parameters of the model according to the initial physiological state data and use the initial parameters to initialize the model to obtain an initial control model; A model training module, used to perform experience playback training on the initial control model based on historical physiological state data and historical control data as input data to obtain a standard control model; A control module, for inputting the current physiological state data into the standard control model when the control time is less than or equal to the preset maximum physical therapy time, and cyclically adjusting the control strategy until the control time is greater than the preset maximum physical therapy time, thereby obtaining an optimal control strategy; An execution module, used for generating a corresponding pulse therapy magnetic field according to the optimal control strategy; Wherein, the parameter optimization module includes: A genetic module, used for initializing individuals in the population and performing fitness evaluation on the individuals in the population according to the initial physiological state data, so as to obtain the fitness of the individuals in the population; A crossover module, used for performing a crossover operation on individuals in the population to obtain crossover individuals; A mutation module, used for performing a mutation operation according to the crossover individuals to obtain mutant individuals; A selection module is used to select individuals with high fitness from the initial population individuals and the variant individuals according to the fitness of the initial population individuals and the variant individuals in the population to enter the next generation, so as to gradually optimize the initial parameters of the model; An initialization module, used to initialize the model using the initial parameters to obtain an initial control model; Among them, the individuals in the population represent a model network parameter, including the weight value and bias term of the connection between the input layer and the hidden layer, and the weight value and bias term of the connection between the hidden layer and the output layer; Wherein, the genetic module is used for: Randomly initialize the model network structure; Initialize the population according to the model network structure; According to the initial physiological state data and the model network structure, the fitness of the individual population is evaluated, and the calculation formula is as follows: in, Indicates The fitness value of each individual; Indicates the total number of initial physiological state data; Represents the instant reward of the jth sample data; represents the discount factor; represents the target Q-value function; represents the action Q-value function; Indicates the next state, including the EEG data, the heart rate data, the blood oxygen data and the skin conductance data; Indicates that according to the next state The control strategies adopted; Indicates Network parameters of each individual; Represents the physiological state of the jth sample data; represents the control strategy adopted according to the j-th sample data and the model network; represents the maximum value function; Wherein, the cross module is used for: Based on the individuals in the population, a candidate solution is initialized for each individual; Based on the individuals in the population, a random number is generated for each dimension of each individual; According to the candidate solution and the random number, a cross calculation is performed, and the calculation formula is as follows: in, Indicates at time At that time, The cross-individual of the j-th dimension of individuals; represents the j-th dimension of the mutation vector generated in the mutation module; represents the jth dimension of the current individual; Represents a random number generation function that generates a random number in the interval [0,1]; represents the crossover probability; represents a randomly selected dimension index, ensuring that at least one dimension is inherited from the mutation vector; Wherein, the mutation module is used to: Based on the crossover individuals in the population, randomly select a crossover individual as a basis vector; Based on the crossover individuals in the population, randomly select another two different crossover individuals, and calculate the difference between the two different crossover individuals to obtain a difference vector; Based on the basis vector and the difference vector, a mutation calculation is performed to obtain a mutant individual. The calculation formula is as follows: in, represents a variant individual; represents basis vectors; Represents the mutation factor, which is used to control the degree of mutation; represents the difference vector; Wherein, the selection module is used to: Based on the initial population individuals and the mutant individuals in the population, the mutant individual fitness is calculated to obtain the mutant individual fitness; When the fitness of the variant individual is greater than the fitness of the initial population individual, the initial population individual is updated to the variant individual; When the fitness of the candidate solution is less than or equal to the fitness of the individual in the initial population, the individual in the initial population is kept unchanged; According to the individuals in the updated population, the individuals with the highest fitness are selected as the next generation of population individuals to gradually optimize the model parameters; Wherein, the model training module is used to: Collect and store historical physiological status data and historical control data; Using the historical physiological state data and the historical control data as input data, the action Q value and the target Q value are obtained by calculation through the initial control model; Performing a timing difference error calculation according to the action Q value and the target Q value to obtain a timing difference error value; According to the time series difference error value, parameter optimization is performed to obtain model optimization parameters; Based on the model optimization parameters, the control model is initialized and the next round of training is performed; When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed and a standard control model is obtained; Among them, the parameter optimization is performed according to the time series difference error value to obtain the model optimization parameters, and the parameter optimization formula is as follows: ， in, Represents model parameters The loss function of represents the time series difference error value function; Indicates immediate reward; represents the discount factor; and Represents the parameters of the current and target networks; Indicates the next state, including the EEG data, the heart rate data, the blood oxygen data and the skin conductance data; Indicates that according to the next state The control strategies adopted; represents the learning rate; represents the loss value; represents the expected function; Indicates the current state; Indicates that according to the current state Control strategies adopted.

2. The brain function recovery device based on photobiomodulation and transcranial magnetic combination according to claim 1, characterized in that: The execution module further includes: An execution time module, used for, according to the optimal control strategy, executing the strategy according to the preset time so that the pulse magnetic field output module generates the corresponding pulse therapy magnetic field; The strategy update module is used to re-collect the current physiological state data, generate the next round of control strategy and execute it when the execution time is greater than or equal to the preset strategy execution time.

3. A brain function recovery system based on photobiomodulation and transcranial magnetic resonance imaging, characterized in that: The system includes a head-mounted module, a touch control module, a pulsed magnetic field output module, a coil module, and a brain function recovery device based on photobiomodulation and transcranial magnetism as described in any one of claims 1-2.

Images