Multi-mode electrical stimulation cognitive recovery system

Abstract

The invention relates to the technical field of cognitive function recovery, and discloses a multi-mode electrical stimulation cognitive recovery system. The patient data acquisition module is used for acquiring brain function related data of a patient to initialize a neural network state; the neurodynamics modeling module is used for constructing a dynamic neural network model based on the brain function network of the patient, and the model describes the spatial and temporal distribution of the neural network and the response of the neural network to the multi-mode electrical stimulation; and the optimal control calculation module is used for calculating an optimal parameter combination of multi-mode electrical stimulation in real time based on an optimal control theory. According to the technical scheme of multi-module cooperative work, real-time monitoring and dynamic adjustment of the brain function state of a patient are achieved through accurate connection of the patient data acquisition module, the neurodynamics modeling module and the closed-loop feedback module. Compared with a scheme in which the state of a patient depends on static preset parameters in the prior art, the technical effect of personalized electrical stimulation parameter optimization is achieved, and the problem that a static method is insufficient in adaptability in a complex dynamic brain function state is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of cognitive function recovery, in particular to a multi-mode electrical stimulation cognitive recovery system. Background Art

[0002] In the rehabilitation treatment of neurocognitive dysfunction, electrical stimulation technology is widely used as a non-invasive means, such as transcranial direct current stimulation and transcranial magnetic stimulation. These technologies act on the patient's brain through external electric or magnetic fields to regulate the activity level of the neural network. However, the existing electrical stimulation methods have many shortcomings in practical applications, which affects the personalized effect and adaptability of the treatment.

[0003] Traditional electrical stimulation therapy usually relies on preset static parameters. Parameters include stimulation intensity, frequency, and duration, which are often based on fixed standards or limited information about the patient's initial state. This static model ignores the dynamic characteristics of brain functional networks, especially the patient's need to adjust the state of brain activity in real time during treatment. Therefore, this type of method is inefficient when facing complex cognitive impairment rehabilitation problems and cannot adapt to individualized treatment needs.

[0004] Existing electrical stimulation technologies usually lack accurate modeling of the complex signal transmission laws of brain functional networks. Most studies only use simple linear models or empirical formulas to try to describe the relationship between neural activity and electrical stimulation. However, in reality, the spatiotemporal distribution characteristics and nonlinear dynamic characteristics of neural networks are extremely complex. It is difficult to fully reflect the true mechanism of signal propagation and multimodal stimulation response between brain regions by using simplified models alone. This technical limitation of lack of modeling accuracy makes it difficult to obtain a scientific basis for the optimization of stimulation parameters.

[0005] In addition, existing multi-modal electrical stimulation devices often work independently and lack the ability to coordinate control between systems. The time, frequency and spatial parameters of multiple stimulation modes are not matched and optimized, which easily causes mutual interference of stimulation signals or waste of resources. This phenomenon of isolated operation of devices seriously limits the actual application effect of multi-modal stimulation technology in the recovery of complex brain functions. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a multi-modal electrical stimulation cognitive recovery system, which solves the problems of lack of dynamic personalized parameter optimization, insufficient accuracy of neural network modeling, lack of real-time feedback regulation ability and poor effect of multi-modal stimulation collaborative control in the process of cognitive function recovery.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-mode electrical stimulation cognitive recovery system, comprising:

[0008] A patient data collection module, used to collect patient brain function related data to initialize the neural network state;

[0009] The neurodynamic modeling module is used to build a dynamic neural network model based on the patient's brain functional network. The model describes the spatiotemporal distribution of the neural network and its response to multimodal electrical stimulation;

[0010] An optimal control calculation module, used for calculating the optimal parameter combination of multi-mode electrical stimulation in real time based on optimal control theory;

[0011] A closed-loop feedback module is used to monitor the patient's brain function status in real time, modify the neural network model according to the feedback data, and dynamically adjust the input of the optimal control module;

[0012] The multi-mode electrical stimulation execution module is used to control the synchronous operation of multiple electrical stimulation devices according to the optimal parameter combination output by the optimal control module, and implement targeted electrical stimulation plans.

[0013] Preferably, the patient data acquisition module comprises:

[0014] An EEG data acquisition unit, used to collect the frequency characteristics of the patient's brain area electrical activity;

[0015] Functional imaging data acquisition unit, used to acquire the spatial distribution of functional connections in brain regions;

[0016] The noise filtering and preprocessing unit is used to clean up artifacts and noise in the acquired data to improve data quality.

[0017] Preferably, the neural dynamics modeling module includes:

[0018] A model building unit is used to build a dynamic model of a neural network and describe the spatiotemporal distribution and dynamic evolution of the neural network through partial differential equations;

[0019] A model parameter initialization unit, used to initialize diffusion coefficients and nonlinear neuron activation function parameters according to patient data;

[0020] The model verification and correction unit is used to verify the model output and correct the model parameters that are inconsistent with the patient data.

[0021] Preferably, the optimal control calculation module includes:

[0022] An optimization objective function construction unit is used to construct an objective function that balances the benefits of enhancing neural activity in brain regions and the energy consumption of electrical stimulation;

[0023] An optimal control algorithm unit, used for calculating an optimal combination of multi-mode electrical stimulation parameters based on optimal control theory;

[0024] The control parameter adjustment unit is used to dynamically adjust the weight coefficient in the objective function according to the optimization result.

[0025] Preferably, the closed-loop feedback module includes:

[0026] A real-time monitoring unit, used to collect the patient's brain function activities during electrical stimulation;

[0027] A state estimation unit, used for dynamically estimating the patient's neural activity state through Kalman filtering;

[0028] The model correction unit is used to correct the neural dynamics model parameters according to the feedback data to improve the modeling accuracy.

[0029] Preferably, the multi-mode electrical stimulation execution module includes:

[0030] An electrical stimulation parameter allocation unit, used to allocate the parameters calculated by the optimal control module to electrical stimulation devices of different modes;

[0031] A multi-mode collaborative control unit for synchronously controlling the temporal intensity and spatial position of multiple stimulation modes;

[0032] Real-time stimulation feedback unit, used to monitor the output signal of the electrical stimulation device and calibrate the stimulation accuracy.

[0033] Preferably, the EEG data acquisition unit is capable of analyzing specific frequency changes in the patient's EEG activity and generating time series signals to evaluate the activity level of brain functional areas.

[0034] Preferably, the model building unit can describe the signal propagation between brain regions through diffusion terms, describe the dynamic behavior of neuronal excitation and inhibition through nonlinear dynamic terms, and simulate the influence of external noise through random disturbance terms.

[0035] Preferably, the optimization objective function construction unit is capable of dynamically adjusting weight parameters in the objective function to adapt to patient-specific treatment goals, including activating specific brain regions or reducing stimulation side effects.

[0036] Preferably, the state estimation unit is capable of adjusting the gain parameters of the Kalman filter based on the patient's real-time neural activity feedback data to improve the accuracy of the patient's brain function state estimation.

[0037] The present invention provides a multi-mode electrical stimulation cognitive recovery system. It has the following beneficial effects:

[0038] 1. The present invention adopts a technical solution of multi-module collaborative work, and realizes real-time monitoring and dynamic adjustment of the patient's brain function state through the precise connection of the patient data acquisition module, the neurodynamic modeling module and the closed-loop feedback module. It achieves the technical effect of personalized electrical stimulation parameter optimization. Compared with the solution in the prior art that relies on static preset parameters for the patient's state, it solves the problem of insufficient adaptability of static methods under complex dynamic brain function states.

[0039] 2. The present invention achieves the technical effect of describing the spatiotemporal dynamic characteristics of brain networks with high precision by adopting a neurodynamic modeling method based on partial differential equations, combined with dynamic simulation of diffusion terms, nonlinear terms and multi-modal electrical stimulation input. Compared with the shortcomings of single linear modeling in the prior art, the problem of lack of accurate mathematical description between the complexity of brain region signal interaction and multi-modal stimulation response is solved.

[0040] 3. The present invention realizes efficient correction of the patient's brain function state and real-time response of the system through real-time monitoring of the closed-loop feedback module and dynamic estimation of the Kalman filter technology. Compared with the static stimulation scheme in the prior art that ignores real-time data feedback, it solves the problem that the robustness and accuracy of the system under dynamic brain activity are difficult to guarantee.

[0041] 4. The present invention achieves the technical effect of efficiently activating specific brain areas and balancing brain function networks by adopting the coordinated control of multi-mode electrical stimulation execution modules, matching the timing of multiple stimulation modes and optimizing harmonic frequencies. Compared with the limitations of the single electrical stimulation mode in the prior art, it solves the problem of limited effect in the complex brain function recovery process. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a system framework diagram of the present invention;

[0043] Figure 2 It is a schematic diagram of a patient data acquisition module of the present invention;

[0044] Figure 3 It is a schematic diagram of the neural dynamics modeling module of the present invention;

[0045] Figure 4 It is a schematic diagram of the optimal control calculation module of the present invention;

[0046] Figure 5 It is a schematic diagram of a closed-loop feedback module of the present invention;

[0047] Figure 6 It is a schematic diagram of a multi-mode electrical stimulation execution module of the present invention;

[0048] Figure 7 This is a logic diagram of the system operation of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the specification 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.

[0050] Example:

[0051] Please refer to the attached Figure 1 -Attached Figure 7 , an embodiment of the present invention provides a multi-modal electrical stimulation cognitive recovery system, including;

[0052] A patient data collection module, used to collect patient brain function related data to initialize the neural network state;

[0053] Specifically, in this embodiment, the patient data acquisition module is mainly responsible for acquiring various types of patient neural activity information such as EEG signals and functional imaging data, and preprocessing the acquired data to ensure the accuracy and validity of the data. The output of this module is directly connected to the neural dynamics modeling module to initialize the state of the dynamic neural network model. The patient data acquisition module also has a real-time function, and its continuously collected feedback data provides the closed-loop feedback module with the necessary real-time neural activity information;

[0054] In this embodiment, the EEG data acquisition unit is used to collect electrical activity information of the patient's brain area. The EEG signal reflects the synchronized discharge activity of neurons in a specific area, and its characteristic frequency band can reveal the patient's neurological function status;

[0055] The electrode array covers multiple key areas of the patient's scalp to collect electrical signals from the brain. Specifically, the EEG signal can be represented by a time series function:

[0056]

[0057] in:

[0058] EEG(t) is the EEG signal at time t;

[0059] A i is the signal amplitude of the i-th channel;

[0060] f i is the frequency of the signal;

[0061] φ i is the initial phase of the signal;

[0062] N is the total number of electrode channels;

[0063] The EEG data acquisition unit converts the time domain signal into the frequency domain signal through fast Fourier transform to extract the energy value of different frequency components. In the specific implementation process, the power spectrum of the following frequency bands can be analyzed in detail;

[0064] Alpha waves (8-12 Hz); associated with relaxation and basal cognitive states;

[0065] Beta waves (13-30Hz); related to higher cognitive functions such as attention and task execution;

[0066] Theta waves (4-7 Hz); related to memory and emotion regulation;

[0067] In this embodiment, the functional imaging data acquisition unit is used to collect functional connections and activity distribution between brain regions. Specifically, the functional imaging data acquisition unit evaluates neural activity by collecting changes in blood oxygen levels in brain regions. The BOLD signal can be represented by the following function:

[0068] BOLD(t)=S(t)·H(t)

[0069] in:

[0070] BOLD(t) is the functional imaging signal at time t;

[0071] S(t) is the neural activity signal;

[0072] H(t) is the hemodynamic response function;

[0073] The functional imaging data acquisition unit constructs a brain functional network based on the acquired signals. The nodes of the network represent brain regions, and the edges represent the functional connection strength between different brain regions. The functional connection can be calculated using coherence analysis. The formula is as follows;

[0074]

[0075] in:

[0076] C ij is the functional connection strength between brain regions i and j;

[0077] S i and S j is the neural activity signal of nucleus j in brain region i;

[0078] Cov is the covariance;

[0079] Var is the variance;

[0080] In this embodiment, the noise filtering and preprocessing unit is used to clean up artifacts and noise in the EEG and functional imaging data, improve the quality of the collected data, and use a bandpass filter to retain the signal within the target frequency range. Specifically, the filtered signal can be expressed as;

[0081] EEG filtered (t)=EEG(t)*h(t)

[0082] in:

[0083] EEG filtered (t) is the filtered signal;

[0084] * is the convolution operation;

[0085] h(t) is the impulse response function of the filter

[0086] The noise filtering and preprocessing unit can also apply the independent component analysis method. ICA effectively removes the artifact component by decomposing the mixed signal into independent components. The decomposed signal can be expressed as;

[0087] X=A·S

[0088] in:

[0089] X is the observation signal matrix;

[0090] A is the mixing matrix;

[0091] S is the independent component matrix;

[0092] The unit can also normalize the preprocessed signals for subsequent modeling and analysis; the patient data acquisition module can provide a high-quality data basis for subsequent neurodynamic modeling and optimal control calculations.

[0093] The neurodynamic modeling module is used to build a dynamic neural network model based on the patient's brain functional network. The model describes the spatiotemporal distribution of the neural network and its response to multimodal electrical stimulation;

[0094] Specifically, in this embodiment, the neural dynamics modeling module realizes the state prediction of the dynamic neural network through partial differential equation (PDE) modeling, and combines data verification and parameter correction to ensure the accuracy and individual characteristics of the model;

[0095] In this embodiment, the model building unit is responsible for establishing a dynamic model of the neural network, using partial differential equations to describe the spatiotemporal distribution and dynamic evolution behavior of the brain functional network;

[0096] There is a complex signal transmission relationship between the patient's brain regions, which can be described by diffusion terms. The core equation of the model is as follows:

[0097]

[0098] in:

[0099] x(r,t) represents the neural activity state at time t and spatial position r;

[0100] D is the diffusion coefficient, which is used to quantify the signal propagation strength between brain regions;

[0101] It is the Laplace operator of the signal, describing the diffusion of the signal in space;

[0102] f(x(r,t)) is the nonlinear neuron excitation and inhibition function, usually expressed as;

[0103] f(x(r,t))=-αx(r,t)+βx 2 (r,t)-γx 3 (r,t)

[0104] in;

[0105] α>0 is the natural attenuation coefficient of neural activity;

[0106] β>0 is a low-order excitation factor;

[0107] γ>0 is a high-order inhibition factor;

[0108] Represents the input effect of multimodal electrical stimulation on the neural network;

[0109] in;

[0110] u i (t) is the stimulus intensity of the ith mode at time t;

[0111] g i (r) is the spatial distribution function of the pattern;

[0112] ξ(r,t) represents the random disturbance term, simulating the external environmental noise, and its distribution usually obeys the Gaussian distribution and satisfies;

[0113]

[0114] Among them: 2 is the noise intensity;

[0115] The diffusion coefficient D can be designed as a brain region-dependent parameter D(r) to more accurately reflect the differences in signal propagation between different brain regions;

[0116] In this embodiment, the model parameter initialization unit is used to personalize the parameters of the neural dynamics model;

[0117] This unit calculates the diffusion coefficient D, the excitation inhibition function parameters α, β, and γ by fitting the EEG signals and functional imaging data provided by the patient data acquisition module;

[0118] The initialization process uses the least squares method to fit the data, and the objective function is as follows;

[0119]

[0120] in:

[0121] θ={D,α,β,γ} are model parameters;

[0122] x model (r k ,t k ; θ) is the model at position r k and time t k Output:

[0123] x data (r k ,t k ) is the actual data of the patient;

[0124] The parameter initialization unit can also use a Bayesian method to estimate parameters by combining prior distribution with patient data;

[0125] In this embodiment, the model verification and correction unit is used to evaluate the accuracy of the model and correct the model parameters according to the error, by comparing the model prediction output with the actual patient data, calculating the prediction error, and if the error exceeds a preset threshold, the parameter correction is performed;

[0126] Specifically, the error is defined as;

[0127] ∈(t)=||x model (t)-x data (t)||2

[0128] in:

[0129] ∈(t) is the error at time t;

[0130] ||·||2 represents the Euclidean norm

[0131] The correction algorithm uses the gradient descent method to update the model parameters by minimizing the error. The update formula is:

[0132]

[0133] in:

[0134] θ (n) is the parameter value of the nth iteration;

[0135] η is the learning rate;

[0136] Through the collaborative work of the above units, the neural dynamics modeling module can establish an accurate and personalized neural network dynamic model.

[0137] An optimal control calculation module, used for calculating the optimal parameter combination of multi-mode electrical stimulation in real time based on optimal control theory;

[0138] Specifically, the optimal control calculation module in this embodiment dynamically adjusts the intensity, frequency and spatial distribution of electrical stimulation by constructing an optimization objective function and combining the optimal control theory, thereby achieving precise control of the neural network;

[0139] In this embodiment, the optimization objective function construction unit is used to define the objective of electrical stimulation parameter optimization, and the objective function needs to comprehensively consider the benefits of brain area neural activity, electrical stimulation energy consumption, and system constraints;

[0140] Specifically, the optimization objective function is in the following form;

[0141]

[0142] in:

[0143] J(u) represents the objective function;

[0144] t0 and t f are the initial time and the end time of optimization respectively;

[0145] Ω is the spatial definition domain of the brain functional area;

[0146] R(x(r,t)) represents the enhancement effect of neural activity in the brain region, which can be defined as the square value of the neural activity level;

[0147] R(x(r,t))=x 2 (r,t)

[0148] C(u(t)) represents the energy consumption of electrical stimulation, which is generally defined as the sum of the squares of the stimulus intensities;

[0149]

[0150] in;

[0151] N is the number of electrical stimulation patterns;

[0152] w1 and w2 are weight parameters used to balance brain area benefits and energy consumption;

[0153] The unit can also add safety constraints to the objective function to limit the maximum value of the neural activity level x(r,t) in the brain area to not exceed the preset threshold xmax ;

[0154]

[0155] In this embodiment, the optimal control algorithm unit is used to solve the control input u(t) in the optimization objective function. The unit uses the Pontryagin maximum principle in the optimal control theory for calculation. Specifically, the Hamiltonian function H is first constructed, and its expression is as follows:

[0156]

[0157]

[0158] in:

[0159] H represents the Hamiltonian function;

[0160] λ(r,t) is a state variable, which is used to describe the sensitivity of the objective function to the state variable;

[0161] and ξ(r,t) are the diffusion term, nonlinear term, electrical stimulation input term and random term in the neural dynamics model respectively;

[0162] According to the Pontryagin maximum principle, the optimal control input u * (t) The following conditions are met:

[0163]

[0164] This unit solves the above optimal control problem by numerical method, and uses gradient descent method or Lagrange multiplier method to iteratively optimize the control input. The update formula of gradient descent method is:

[0165]

[0166] in:

[0167] is the control input for the kth iteration;

[0168] η is the learning rate;

[0169] In this embodiment, the control parameter adjustment unit is used to dynamically adjust the weight parameters w1 and w2 in the objective function. The setting of the weight parameters depends on the patient's treatment goal. For patients whose main goal is to restore cognitive function, the weight w1 should be appropriately increased, while w2 can be relatively reduced. On the contrary, for patients who are sensitive to energy consumption, the value of w2 needs to be appropriately increased. In a possible implementation, the unit calculates the dynamic adjustment value of the weight parameter by real-time monitoring of the brain region neural activity level x(r,t) and the electrical stimulation energy consumption C(u(t)). The adjustment formula is as follows;

[0170]

[0171]

[0172] in:

[0173] R(x real (t)) and C real (u(t)) are the actual brain area benefits and energy consumption, respectively;

[0174] R(x target (t)) and C target (u(t)) is the preset target value;

[0175] α and β are adjustment coefficients;

[0176] The optimal control calculation module can also combine the patient's real-time feedback data to dynamically modify the optimization objective function. For patients with fast state changes, a real-time feedback term F(t) can be added, and its expression is:

[0177]

[0178] in:

[0179] F(t) is the feedback error term;

[0180] x real (t′) and x pred (t′) are the actual state and the predicted state respectively;

[0181] Through the above method, the module can efficiently calculate the optimal combination of electrical stimulation parameters that adapt to individual differences and provide reliable input for the multi-mode electrical stimulation execution module.

[0182] A closed-loop feedback module is used to monitor the patient's brain function status in real time, modify the neural network model according to the feedback data, and dynamically adjust the input of the optimal control module;

[0183] Specifically, in this embodiment, the closed-loop feedback module is closely connected with the patient data acquisition module to obtain the patient's neural activity data in real time; at the same time, it interacts with the neural dynamics modeling module and the optimal control calculation module to complete the closed-loop feedback optimization process;

[0184] The module ensures that the system can adapt to the dynamic changes of the patient's status through continuous iteration of state estimation, model correction and feedback closed-loop signal;

[0185] In this embodiment, the real-time monitoring unit is used to obtain the patient's neural activity feedback data during the electrical stimulation process. The real-time monitoring unit processes the signals provided by the patient data acquisition module, including multiple sources such as EEG signals and functional imaging data;

[0186] Specifically, during the monitoring of EEG signals, the unit evaluates the patient's neural activity response by extracting the power changes in a specific frequency band. The real-time changes in the power of alpha waves (8-12Hz) can be expressed as;

[0187]

[0188] in:

[0189] P α (t) is the power of the α wave at time t;

[0190] Δt is the length of the sliding window;

[0191] EEG α (τ) is the filtered α wave signal;

[0192] The real-time monitoring unit can also obtain the blood oxygen changes between brain regions through functional imaging data to evaluate the strength of neural activity. The blood oxygen change signal can be expressed as;

[0193] ΔHb(t)=HbO(t)-HbR(t)

[0194] in:

[0195] HbO(t) is the oxygenated hemoglobin concentration at time t;

[0196] HbR(t) is the concentration of deoxyhemoglobin at time t;

[0197] The real-time monitoring unit can also fuse multimodal data, for example, combining the power changes of EEG signals with the blood oxygen signals of functional imaging to generate a more comprehensive map of the patient's neural activity;

[0198] In this embodiment, the state estimation unit is used to dynamically estimate the patient's brain function state in a random noise environment. The unit uses Kalman filtering technology to correct the state of the system by observing the difference between the data and the model prediction value;

[0199] Specifically, the Kalman filter process includes two steps: state prediction and state update. The state prediction formula is:

[0200] x pred (t+1)=Ax est (t)+Bu(t)

[0201] The state update formula is:

[0202] x est (t+1)=x pred (t+1)+K(t+1)(z(t+1)-Hx pred (t+1))

[0203] in:

[0204] x pred (t+1) is the state prediction value at time t+1;

[0205] x est (t+1) is the estimated value of the state at time t+1;

[0206] A is the system state transfer matrix;

[0207] B is the control input matrix;

[0208] u(t) is the control input;

[0209] z(t+1) is the observed value at time t+1;

[0210] H is the observation matrix;

[0211] K(t+1) is the Kalman gain, and the calculation formula is as follows;

[0212]

[0213] Where P pred (t+1) is the prediction error covariance matrix, R is the observation noise covariance matrix;

[0214] The unit can enhance the dynamic response capability to the patient's state by adjusting the Kalman gain K(t+1). When the noise of the observed data is large, the Kalman gain is reduced to reduce error propagation; conversely, when the noise is small, the Kalman gain is increased to track state changes faster.

[0215] In this embodiment, the model correction unit is used to adjust the parameters of the neurodynamic model according to the real-time monitoring and state estimation results, and the unit updates the model parameters by recursive least squares method;

[0216] Specifically, the parameter update formula of the recursive least squares method is as follows:

[0217]

[0218]

[0219] in:

[0220] θ(t) is the model parameter vector at time t;

[0221] P(t) is the parameter error covariance matrix at time t;

[0222] X(t) is the input data vector at time t;

[0223] y(t) is the target output at time t;

[0224] The model correction unit can make local adjustments to specific parameters of the neurodynamic model and adjust the diffusion coefficient D in real time to more accurately reflect the signal propagation characteristics between brain regions;

[0225] The unit can also dynamically adapt to individual differences among patients through online learning algorithms, using neural networks or other machine learning models to further improve the flexibility and accuracy of parameter correction;

[0226] The closed-loop feedback module plays an important dynamic regulatory role in the entire multimodal electrical stimulation cognitive recovery system. The design of the module ensures the real-time adaptability of the system to the patient's condition, while providing the necessary real-time support for optimal control calculation and neural dynamics modeling.

[0227] A multi-mode electrical stimulation execution module is used to control multiple electrical stimulation devices to work synchronously according to the optimal parameter combination output by the optimal control module, and implement a targeted electrical stimulation plan;

[0228] Specifically, in this embodiment, the multi-mode electrical stimulation execution module needs to work closely with the closed-loop feedback module to correct the stimulation output in real time. At the same time, the operation of the module also depends on the parameter input provided by the patient data acquisition module and the neural dynamics modeling module to ensure the dynamic adaptation of the stimulation strategy to the actual state of the patient;

[0229] In this embodiment, the electrical stimulation parameter allocation unit is responsible for allocating the stimulation parameters output by the optimal control calculation module to devices with different stimulation modes; the parameters received by the unit include the stimulation intensity u i (t), stimulation frequency f i and the spatial distribution function g i (r), the action mechanism of the parameter can be described by the following formula;

[0230] I i (r,t)=u i (t) g i (r)·sin(2πf i t+φ i )

[0231] in:

[0232] I i (r, t) is the output signal of the ith stimulation pattern at time t and spatial position r;

[0233] u i (t) is the stimulus intensity at time t;

[0234] g i (r) is the spatial distribution function, defining the action area of ​​the stimulation signal;

[0235] f i is the stimulation frequency, which is usually selected based on the frequency of neural oscillations in the target brain area;

[0236] φ i is the initial phase of the stimulus signal;

[0237] This unit can be adjusted by g i The function form of (r) can realize the precise positioning of a specific brain area. If the stimulation signal needs to be concentrated in a certain brain area, the Gaussian distribution function can be used;

[0238]

[0239] in:

[0240] r c is the central location of the target brain area;

[0241] σ is the distribution width parameter, which is used to control the signal diffusion range;

[0242] In this embodiment, the multi-mode collaborative control unit is responsible for synchronously coordinating the operation of multiple electrical stimulation devices. Generally, the unit needs to control devices of different modes at the same time. In order to achieve synergy between different modes, the unit optimizes the stimulation signal through dynamic matching of time and frequency. If a superposition effect is required between two modes, the stimulation frequency can be set to a harmonic relationship;

[0243] f j =n·f i

[0244] in:

[0245] f j and f i are the stimulation frequencies of the j-th and i-th modes, respectively;

[0246] n is a positive integer, indicating the harmonic multiple;

[0247] The multi-mode cooperative control unit can also achieve timing matching through delay control. In scenarios where alternating effects are required, the stimulation signals of the modes are set with different phase delays;

[0248] φ j =φ i +Δφ

[0249] in:

[0250] φ j and φ i are the initial phases of the j-th and i-th modes, respectively;

[0251] Δφ is the phase delay;

[0252] The multi-modal collaborative control unit can also adjust the ratio of stimulation intensity according to the specific needs of the brain area. For scenarios where the signals of two brain areas need to be balanced, the intensity ratio can be calculated by the following formula;

[0253]

[0254] in:

[0255] Target j and Target i are the activation requirements of the j-th and i-th target brain regions, respectively;

[0256] In this embodiment, the real-time stimulation feedback unit is used to monitor the output signal of the electrical stimulation device and dynamically calibrate the stimulation parameters. The unit collects the real-time output data of the electrical stimulation device, compares it with the expected signal, and calculates the output error. The error is defined as;

[0257] ∈(t)=||I real (r,t)-I target (r,t)||2

[0258] in:

[0259] I real (r, t) is the actual output signal of the device;

[0260] I target (r, t) is the target output signal;

[0261] ||·||2 is the Euclidean norm;

[0262] The real-time stimulation feedback unit corrects the output error through proportional integral derivative control, and the control formula is as follows;

[0263]

[0264] in:

[0265] Δu(t) is the adjustment amount of the control input;

[0266] K p , K i and K d are proportional, integral and derivative gains respectively;

[0267] The unit can also dynamically adjust the operating frequency of the electrical stimulation device in combination with feedback data. When the actual signal frequency deviates from the target frequency, the frequency correction amount is calculated by the following formula;

[0268] Δf=f target -f real

[0269] Through the implementation of the above functions, the multi-mode electrical stimulation execution module can flexibly adapt to the individual needs of patients while maintaining a high degree of coordination with other modules of the entire system.

[0270] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multimodal electrical stimulation cognitive recovery system, characterized in that: include; A patient data collection module, used to collect patient brain function related data to initialize the neural network state; The neurodynamic modeling module is used to build a dynamic neural network model based on the patient's brain functional network. The model describes the spatiotemporal distribution of the neural network and its response to multimodal electrical stimulation; An optimal control calculation module, used for calculating the optimal parameter combination of multi-mode electrical stimulation in real time based on optimal control theory; A closed-loop feedback module is used to monitor the patient's brain function status in real time, modify the neural network model according to the feedback data, and dynamically adjust the input of the optimal control module; The multi-mode electrical stimulation execution module is used to control the synchronous operation of multiple electrical stimulation devices according to the optimal parameter combination output by the optimal control module, and implement targeted electrical stimulation plans.

2. The multimodal electrical stimulation cognitive recovery system according to claim 1, characterized in that: The patient data acquisition module comprises: An EEG data acquisition unit, used to collect the frequency characteristics of the patient's brain area electrical activity; Functional imaging data acquisition unit, used to acquire the spatial distribution of functional connections in brain regions; The noise filtering and preprocessing unit is used to clean up artifacts and noise in the acquired data to improve data quality.

3. The multimodal electrical stimulation cognitive recovery system according to claim 1, characterized in that: The neural dynamics modeling module includes: A model building unit is used to build a dynamic model of a neural network and describe the spatiotemporal distribution and dynamic evolution of the neural network through partial differential equations; A model parameter initialization unit, used to initialize diffusion coefficients and nonlinear neuron activation function parameters according to patient data; The model verification and correction unit is used to verify the model output and correct the model parameters that are inconsistent with the patient data.

4. The multimodal electrical stimulation cognitive recovery system according to claim 1, characterized in that: The optimal control calculation module includes: An optimization objective function construction unit is used to construct an objective function that balances the benefits of enhancing neural activity in brain regions and the energy consumption of electrical stimulation; An optimal control algorithm unit, used for calculating an optimal combination of multi-mode electrical stimulation parameters based on optimal control theory; The control parameter adjustment unit is used to dynamically adjust the weight coefficient in the objective function according to the optimization result.

5. The multi-modal electrical stimulation cognitive recovery system according to claim 1, characterized in that: The closed-loop feedback module comprises: A real-time monitoring unit, used to collect the patient's brain function activities during electrical stimulation; A state estimation unit, used for dynamically estimating the patient's neural activity state through Kalman filtering; The model correction unit is used to correct the neural dynamics model parameters according to the feedback data to improve the modeling accuracy.

6. The multi-modal electrical stimulation cognitive recovery system according to claim 1, characterized in that: The multi-mode electrical stimulation execution module comprises: An electrical stimulation parameter allocation unit, used to allocate the parameters calculated by the optimal control module to electrical stimulation devices of different modes; A multi-mode collaborative control unit for synchronously controlling the temporal intensity and spatial position of multiple stimulation modes; Real-time stimulation feedback unit, used to monitor the output signal of the electrical stimulation device and calibrate the stimulation accuracy.

7. The multi-modal electrical stimulation cognitive recovery system according to claim 2, characterized in that: The EEG data acquisition unit is capable of analyzing specific frequency changes in the patient's EEG activity and generating time series signals to evaluate the activity level of brain functional areas.

8. The multi-modal electrical stimulation cognitive recovery system according to claim 3, characterized in that: The model building unit can describe the signal propagation between brain regions through diffusion terms, describe the dynamic behavior of neuron excitation and inhibition through nonlinear dynamic terms, and simulate the influence of external noise through random disturbance terms.

9. The multi-modal electrical stimulation cognitive recovery system according to claim 4, characterized in that: The optimization objective function construction unit can dynamically adjust the weight parameters in the objective function to adapt to the patient's specific treatment goals, including activating specific brain regions or reducing stimulation side effects.

10. The multi-modal electrical stimulation cognitive recovery system according to claim 5, characterized in that: The state estimation unit can adjust the gain parameters of the Kalman filter based on the patient's real-time neural activity feedback data to improve the accuracy of the patient's brain function state estimation.

Images