Neurological rehabilitation training system based on artificial intelligence and regulation method

Abstract

The application provides a neural internal medicine rehabilitation training system and regulation method based on artificial intelligence, extracts the rhythm power of a stimulation pulse of a rehabilitation training device to a patient and the blood oxygen concentration of a target nerve point of the patient, classifies the efficiency of the remodeling level of the neural function in the rehabilitation training through the blood oxygen concentration and the rhythm power, and obtains a graded remodeling index of the neural function regulation in the rehabilitation training; determines the cortex excitability index of the patient in the neural internal medicine rehabilitation training through the motor evoked potential of the target patient muscle cortex, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device; multi-modal synergistically integrates the cortex excitability index and the graded remodeling index of the neural function regulation, obtains the neural regulation target parameter of the remodeling index in the rehabilitation training device, and then dynamically regulates the stimulation parameter in the rehabilitation training device based on the neural regulation target parameter. Based on the above scheme, multi-modal coupling regulation of the stimulation parameter in the rehabilitation training process can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rehabilitation training, and more particularly to a neural internal medicine rehabilitation training system and a regulation method based on artificial intelligence. BACKGROUND

[0002] In recent years, the mature application of neural regulation technology marks the entry of neural rehabilitation into the era of targeted intervention. By adjusting the cortical excitability in a non-invasive manner, neural function reorganization is promoted. The integration of artificial intelligence and multi-modal signal analysis technology brings new breakthroughs to neural rehabilitation, enabling precise assessment and individualized regulation of neural function status, and promoting the transformation of neural rehabilitation from empirical treatment to intelligent and precise treatment, opening up new avenues for improving rehabilitation efficacy.

[0003] Traditional neural rehabilitation methods mainly rely on periodic functional scale assessment and imaging examination, which can only provide static snapshots at discrete time points and cannot capture key details of dynamic changes in neural plasticity. Although neural electrophysiological detection (e.g., evoked potentials) can reflect part of the neural conduction function, it lacks the ability to analyze microscopic mechanisms such as synaptic plasticity and neural network reorganization. At the same time, existing technologies have a detection delay of several hours to several days, completely missing the real-time window period of neural remodeling, resulting in a serious disconnection between rehabilitation intervention and neural remodeling process. Therefore, how to realize multi-modal coupling regulation of stimulation parameters during rehabilitation training has become a difficult problem in the industry. SUMMARY

[0004] The present application provides a neural internal medicine rehabilitation training system and a regulation method based on artificial intelligence, which can realize multi-modal coupling regulation of stimulation parameters during rehabilitation training.

[0005] In a first aspect, the present application provides a neural internal medicine rehabilitation training regulation method based on artificial intelligence, comprising:

[0006] Starting the rehabilitation training device to perform rehabilitation training on the patient with a nervous system disease, and collecting physiological signals of the patient's target neural target point and electromyography signals of the motor cortex under a specified stimulation pulse in the rehabilitation training device during the rehabilitation training;

[0007] Extracting the rhythm power of the stimulation pulse and the blood oxygen concentration of the patient's target neural target point from the physiological signals, classifying the remodeling level of neural function in the rehabilitation training according to the blood oxygen concentration and the rhythm power, and obtaining a classified remodeling index of neural function regulation in the rehabilitation training;

[0008] extracting a motor evoked potential of a target patient muscle cortex in the electromyography signal of the motor cortex, evaluating the excitability of the patient cortex through the motor evoked potential, a stimulation frequency and a stimulation intensity of a stimulation pulse in the rehabilitation training device, to obtain a cortex excitability index of the patient in the rehabilitation training;

[0009] integrating the cortex excitability index and the hierarchical remodeling indicator of the neural function regulation in multiple modes to obtain a neural regulation target parameter of the remodeling indicator in the rehabilitation training device, and then dynamically adjusting a stimulation parameter scheme of the rehabilitation training device based on the neural regulation target parameter, and outputting a visual rehabilitation treatment recommendation report to the patient.

[0010] In some embodiments, extracting the rhythm power of the stimulation pulse and the blood oxygen concentration of the target neural target point from the physiological signal specifically includes:

[0011] obtaining an electroencephalogram and a near-infrared spectrum of the patient in the rehabilitation training of the neurology department from the physiological signal;

[0012] extracting the rhythm power of the target frequency band in the stimulation pulse from the electroencephalogram;

[0013] extracting the change amount of the oxygenated hemoglobin concentration of the target area from the near-infrared spectrum as the blood oxygen concentration of the target neural target point of the patient.

[0014] In some embodiments, classifying the remodeling level of the neural function in the rehabilitation training through the blood oxygen concentration and the rhythm power to obtain the hierarchical remodeling indicator of the neural function regulation in the rehabilitation training specifically includes:

[0015] performing demand influence normalization on the blood oxygen concentration and the rhythm power to obtain a pulse influence parameter and a blood oxygen influence parameter;

[0016] performing parameter weighted fusion on the pulse influence parameter and the blood oxygen influence parameter to obtain the hierarchical remodeling indicator of the neural function regulation in the rehabilitation training.

[0017] In some embodiments, extracting the motor evoked potential of the target patient muscle cortex in the electromyography signal of the motor cortex specifically includes:

[0018] obtaining a signal peak amplitude and a peak latency interval of the target patient muscle cortex from the electromyography signal of the motor cortex;

[0019] filtering out the motor evoked potential of the target patient muscle cortex from the electromyography signal of the motor cortex through the signal peak amplitude and the peak latency interval.

[0020] In some embodiments, the excitability of the cortex of the patient is evaluated by the motor evoked potential, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device, and the cortex excitability index of the patient in the rehabilitation training specifically comprises:

[0021] The stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device are monitored in real time;

[0022] The basic cortex excitability of the cortex of the patient is determined by the signal peak amplitude of the motor evoked potential and the stimulation intensity of the stimulation pulse in the rehabilitation training device;

[0023] The basic cortex excitability is fitted and adjusted according to the stimulation frequency of the stimulation pulse in the rehabilitation training device, and the cortex excitability index of the patient in the rehabilitation training is obtained.

[0024] In some embodiments, the stimulation parameter scheme of the rehabilitation training device is dynamically adjusted based on the neuromodulation target parameter, and specifically comprises:

[0025] An online policy optimization model based on deep learning is initialized;

[0026] The neuromodulation target parameter is taken as a target value in the online policy optimization model;

[0027] The cortex excitability index is taken as a current state quantity in the online policy optimization model;

[0028] The online policy optimization model with the input target value and the current state quantity is used to perform gradient adjustment on the stimulation frequency and the stimulation intensity in the rehabilitation training device.

[0029] In some embodiments, the physiological signals include electroencephalogram and near-infrared spectrum.

[0030] In a second aspect, the present application provides a neurology rehabilitation training system based on artificial intelligence, comprising a rehabilitation training control unit, wherein the rehabilitation training control unit comprises:

[0031] The acquisition module is configured to start the rehabilitation training device to perform rehabilitation training on the patient with a nervous system disease, and acquire physiological signals of the target neural target of the patient and electromyography signals of the motor cortex of the patient under a specified stimulation pulse in the rehabilitation training device during the rehabilitation training;

[0032] The processing module is configured to extract the rhythm power of the stimulation pulse and the blood oxygen concentration of the target neural target of the patient from the physiological signals, perform performance grading on the remodeling level of the neural function in the rehabilitation training by the blood oxygen concentration and the rhythm power, and obtain a graded remodeling index of the neural function control in the rehabilitation training;

[0033] The processing module is further used to extract a motor evoked potential of a target patient muscle cortex in the motor cortex electromyogram signal, evaluate the excitability of the patient cortex through the motor evoked potential, a stimulation frequency and a stimulation intensity of a stimulation pulse in the rehabilitation training device, and obtain a cortex excitability index of the patient in the rehabilitation training.

[0034] The execution module is used to perform multi-modal collaborative integration on the cortex excitability index and the grading remodeling index of the neural function regulation, obtain a neural regulation target parameter of the remodeling index in the rehabilitation training device, dynamically adjust a stimulation parameter scheme of the rehabilitation training device based on the neural regulation target parameter, and output a visual rehabilitation treatment recommendation report to the patient.

[0035] In a third aspect, the present application provides a computer device, which comprises a memory and a processor, the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device executes the above-mentioned artificial intelligence-based rehabilitation training regulation method for neurology.

[0036] In a fourth aspect, the present application provides a computer readable storage medium, which stores instructions or codes, when the instructions or codes are run on a computer, the computer is caused to execute the above-mentioned artificial intelligence-based rehabilitation training regulation method for neurology.

[0037] The technical scheme provided by the embodiments of the present application has the following beneficial effects:

[0038] In the artificial intelligence-based rehabilitation training system and regulation method for neurology provided by the present application, after starting the rehabilitation training device to perform rehabilitation training on the patient with a nervous system disease, physiological signals of a target neural target point of the patient under a specified stimulation pulse and electromyogram signals of a motor cortex are collected during the rehabilitation training; the rhythm power of the stimulation pulse and the blood oxygen concentration of the target neural target point of the patient are extracted from the physiological signals, the remodeling level of the neural function in the rehabilitation training is graded in effectiveness through the blood oxygen concentration and the rhythm power, and a grading remodeling index of the neural function regulation in the rehabilitation training is obtained; a motor evoked potential of a target patient muscle cortex in the motor cortex electromyogram signal is extracted, the excitability of the patient cortex is evaluated through the motor evoked potential, a stimulation frequency and a stimulation intensity of a stimulation pulse in the rehabilitation training device, and a cortex excitability index of the patient in the rehabilitation training is obtained; the cortex excitability index and the grading remodeling index of the neural function regulation are subjected to multi-modal collaborative integration, a neural regulation target parameter of the remodeling index in the rehabilitation training device is obtained, the stimulation parameter scheme of the rehabilitation training device is dynamically adjusted based on the neural regulation target parameter, and a visual rehabilitation treatment recommendation report is output to the patient.

[0039] Therefore, in the present application, the cortical excitability index and the hierarchical remodeling index of neural function regulation are multi-modal synergistically integrated to obtain the neural regulation target parameter of the remodeling index in the rehabilitation training device, and then the stimulation parameter scheme of the rehabilitation training device is dynamically adjusted based on the neural regulation target parameter, and a visual rehabilitation treatment recommendation report is output to the patient. First, the hierarchical remodeling index of neural function regulation is determined to obtain the dynamic quantitative evaluation of neural remodeling efficiency, thereby realizing precise regulation based on metabolic-electrical activity coupling, establishing a dynamic hierarchical system of neural function remodeling level, and the blood oxygen concentration directly reflects the energy metabolism demand of neural activity, and the rhythm power characterizes the synchronization strength of specific frequency band neural oscillation, and the synergistic change of the two can accurately capture the metabolic-electrical activity coupling characteristics in the process of neural plasticity. When the blood oxygen concentration increases accompanied by the increase of gamma band rhythm power, it indicates that the neural compensation mechanism is effectively activated. If the blood oxygen is insufficient and the rhythm is disordered, it indicates that the neural remodeling is blocked. The limitations of traditional single index evaluation are overcome, and the key time window of neural remodeling can be automatically identified through the real-time efficiency grading system, and the stimulation parameters are dynamically adjusted. Then, the cortical excitability index is determined to obtain the objective evaluation of the excitability of the motor pathway, thereby realizing closed-loop regulation of neural-muscular coordination. The amplitude of the motor evoked potential directly reflects the recruitment degree of the motor neuron pool, and the latency reflects the nerve conduction velocity. When the amplitude of the motor evoked potential decreases but the stimulation intensity remains unchanged, it indicates that the cortical spinal cord bundle inhibition is enhanced, and the stimulation frequency needs to be lowered to avoid excessive inhibition. If the latency is prolonged accompanied by insufficient blood oxygen, it indicates that the white matter conduction is blocked, and the stimulation mode needs to be switched to promote myelin repair. By coupling the excitability index with the hierarchical remodeling index, the system can intelligently identify the type of neural-muscular decoding disorder and implement targeted intervention. In summary, based on the above scheme, multi-modal coupling regulation of stimulation parameters in the rehabilitation training process can be realized. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0041] Figure 1 is an exemplary flowchart of the neural internal medicine rehabilitation training regulation method based on artificial intelligence according to some embodiments of the present application;

[0042] Figure 2 is a logic diagram of neural function remodeling according to some embodiments of the present application;

[0043] Figure 3is a flowchart of a process for determining a cortical excitability index according to some embodiments of the present application;

[0044] Figure 4 is a structural diagram of a rehabilitation training regulation unit according to some embodiments of the present application;

[0045] Figure 5 is a structural diagram of a computer device for implementing an artificial intelligence-based neurology rehabilitation training regulation method according to some embodiments of the present application. DETAILED DESCRIPTION

[0046] In order to better understand the technical solutions of the present application, the technical solutions of the present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0047] Reference Figure 1 The figure is an exemplary flowchart of an artificial intelligence-based neurology rehabilitation training regulation method according to some embodiments of the present application, which mainly includes the following steps:

[0048] In step 101, start the rehabilitation training device to perform rehabilitation training on the patient with a nervous system disease, and collect the physiological signal and the electromyogram signal of the motor cortex of the patient's target nerve target point under the specified stimulation pulse in the rehabilitation training device during the rehabilitation training.

[0049] It should be noted that in the present application, the electromyogram signal represents the potential change of muscle electrical activity; the physiological signal represents the multi-modal bioelectricity characteristics of neural function activity, and the physiological signal includes electroencephalogram and near-infrared spectrum, wherein the electroencephalogram represents the electrical activity of cerebral cortex neurons, and the near-infrared spectrum represents the dynamic of brain tissue oxygen metabolism.

[0050] In a specific implementation, after starting the rehabilitation training device for the rehabilitation training of nervous system diseases, a non-invasive transcranial magnetic stimulation instrument is used to apply precise magnetic pulse stimulation to the motor cortex of the patient, and a high-precision multi-channel electroencephalogram device is used to synchronously collect the electroencephalogram of the patient, and the changes in the beta wave and gamma wave rhythms of the motor cortex area are monitored. In order to obtain the feedback signal of the motor nerve pathway, a medical-grade electromyography electrode is attached to the surface of the target muscle group to record the electromyography signal of the motor cortex induced by the transcranial magnetic stimulation in real time, and the electromyography signal includes potential amplitude and latency. Meanwhile, a multi-channel near-infrared optical probe is attached to the target area of the scalp of the patient, the emission end of the probe alternately irradiates the scalp tissue with near-infrared light of two wavelengths of 690 nm and 830 nm, and the multi-channel near-infrared optical probe of the receiving end synchronously detects the light intensity signal scattered by the tissue, and the sampling frequency is not less than 10 Hz. The light intensity signal is used as the near-infrared spectrum of the patient's head, and the combination of the electroencephalogram and the near-infrared spectrum is used as the physiological signal of the target neural target of the patient in the rehabilitation training device under the specified stimulation pulse, and the electromyography signal is used as the electromyography signal of the motor cortex of the patient.

[0051] In some embodiments, reference is made to Figure 2 The figure is a logical diagram of neural functional remodeling according to some embodiments of the present application, which shows the neural connections between the brain, spinal cord and hand and their structural changes, the brain part includes motor cortex remodeling, somatosensory cortex remodeling and brain connectivity network remodeling, emphasizing the corpus callosum signal of the contralateral hemisphere; the spinal cord part describes thalamic remodeling and dendritic spine remodeling, and the excitability change of the spinal cord anterior horn motor neuron; the hand part involves sensory afferent reduction, the reasons include trauma, amputation and immobilization, the figure also includes the influence of gene mutation (torsin-1a) and double hit (gene mutation plus peripheral injury) on neural remodeling, and points out the evidence of patients and animal models, and the figure shows that the volume of the nucleus of the bundle is reduced and the excitability of the neurons of the nucleus of the bundle is increased. The M1 and S1 regions in the brain are labeled.

[0052] In step 102, the rhythm power of the stimulation pulse and the blood oxygen concentration of the target neural target of the patient are extracted from the physiological signal, the remodeling level of the neural function in the rehabilitation training is graded in efficiency by the blood oxygen concentration and the rhythm power, and the grading remodeling index of the neural function regulation in the rehabilitation training is obtained.

[0053] In some embodiments, the extraction of the rhythm power of the stimulation pulse and the blood oxygen concentration of the target neural target of the patient from the physiological signal can be achieved by the following steps:

[0054] The electroencephalogram and near-infrared spectrum of the patient in the rehabilitation training of the department of neurology are obtained from the physiological signal;

[0055] The rhythm power of the target frequency band in the stimulation pulse is extracted from the electroencephalogram;

[0056] extracting a variation of oxygenated hemoglobin concentration of the target region from the near-infrared spectrum as the blood oxygen concentration of the target nerve point of the patient.

[0057] It should be noted that in the present application, the rhythm power represents the energy intensity of a specific brain electrical frequency band, and the blood oxygen concentration represents the oxygenated hemoglobin content variation of the local brain tissue; in specific implementation, first, the electroencephalogram and the near-infrared spectrum of the patient in the neurological rehabilitation training are obtained from the physiological signals; then, the rhythm power of the target frequency band in the stimulation pulse can be extracted from the electroencephalogram in the following manner, that is, the frequency domain signal of the stimulation pulse is obtained by performing fast Fourier transform on the electroencephalogram signal, so as to calculate the square modulus of the frequency domain signal as the power spectral density of each frequency band in the stimulation pulse, and then calculate the definite integral of all power spectral densities in the target frequency band (default γ band of 13-30 Hz) as the rhythm power of the target frequency band in the stimulation pulse; finally, the variation of oxygenated hemoglobin concentration of the target region can be extracted from the near-infrared spectrum as the blood oxygen concentration of the target nerve point of the patient in the following manner, that is, the near-infrared spectrum is analyzed using the near-infrared spectrum technology to obtain the variation of oxygenated hemoglobin concentration of the target region as the blood oxygen concentration of the target nerve point of the patient.

[0058] In addition, it should be noted that in the near-infrared spectrum technology, the concentration variations of oxygenated hemoglobin (HbO2) and deoxygenated hemoglobin (Hb) can be measured by analyzing the scattering and absorption characteristics of near-infrared light in tissues. This method utilizes the difference in absorption of near-infrared light by oxygenated hemoglobin and deoxygenated hemoglobin, and calculates blood oxygen parameters including oxygenated hemoglobin concentration, reduced hemoglobin concentration, total hemoglobin concentration, and tissue oxygen saturation, etc. through the modified Lambert-Beer law. The near-infrared spectrum technology can quantify the variation of oxygenated hemoglobin concentration of the target region, thereby serving as an indicator of the blood oxygen concentration of the target nerve point of the patient.

[0059] In some embodiments, the remodeling level of the neurological function in the rehabilitation training is graded in performance by the blood oxygen concentration and the rhythm power, and the graded remodeling indicator of the neurological function regulation in the rehabilitation training can be obtained by the following steps:

[0060] performing demand influence normalization on the blood oxygen concentration and the rhythm power to obtain a pulse influence parameter and a blood oxygen influence parameter;

[0061] performing parameter weighted fusion on the pulse influence parameter and the blood oxygen influence parameter to obtain the graded remodeling indicator of the neurological function regulation in the rehabilitation training.

[0062] It should be noted that in the present application, the hierarchical remodeling index represents the quantitative evaluation level of the neural function state; the pulse influence parameter represents the influence degree of the neural rhythm activity on the neural function remodeling level; and the blood oxygen influence parameter represents the influence degree of the brain tissue oxygen metabolism level on the neural function remodeling level.

[0063] In a specific implementation, first, the blood oxygen concentration and the rhythm power are subjected to demand influence normalization to obtain the pulse influence parameter and the blood oxygen influence parameter. The following method can be used: the baseline standardization method is used to divide the real-time monitored blood oxygen concentration by the maximum concentration change value of the patient in a resting state to obtain the blood oxygen influence parameter in the range of 0 to 1; for the rhythm power data, the ratio of the power value of the target frequency band to the total power of the full frequency band is calculated, and then divided by the reference power value of the target frequency band under the maximum stimulation intensity to obtain the pulse influence parameter in the range of 0 to 1, wherein the target frequency band under the maximum stimulation intensity can be obtained by statistical historical data; then, the pulse influence parameter and the blood oxygen influence parameter are subjected to parameter weighted fusion to obtain the hierarchical remodeling index of the neural function regulation in rehabilitation training. The following method can be used: based on the weight distribution scheme verified by clinical verification (usually the blood oxygen parameter weight is 0.6-0.7, and the rhythm power weight is 0.3-0.4), the linear weighted fusion algorithm is used to integrate the pulse influence parameter and the blood oxygen influence parameter into a weighted value, and the hierarchical remodeling index of the neural function regulation in rehabilitation training is distributed by the weighted value. The hierarchical remodeling index is divided into three levels: when the weighted value is less than 0.3, it is in a low demand state, 0.3-0.6 is in a medium demand state, and more than 0.6 is in a high demand state.

[0064] In step 103, the motor evoked potential of the target patient muscle cortex in the electromyogram signal of the motor cortex is extracted, the excitability of the patient's cortex is evaluated through the motor evoked potential, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device, and the cortex excitability index of the patient in rehabilitation training is obtained.

[0065] In some embodiments, the motor evoked potential of the target patient muscle cortex in the electromyogram signal of the motor cortex can be implemented by the following steps:

[0066] The signal peak amplitude and the peak latency interval of the target patient muscle cortex are obtained from the electromyogram signal of the motor cortex.

[0067] The motor evoked potential of the target patient muscle cortex is screened from the electromyogram signal of the motor cortex through the signal peak amplitude and the peak latency interval.

[0068] It should be noted that in the present application, the motor evoked potential represents the compound muscle action potential generated by motor neurons under the action of electrical stimulation or magnetic stimulation, reflecting the conduction function and excitability of the motor nerve pathway; the signal peak amplitude represents the degree of synchronization of the recruited discharge of motor neurons, and the peak latency interval represents the conduction velocity of nerve impulses in the motor pathway.

[0069] In a specific implementation, the signal peak amplitude and the peak latency interval of the target patient muscle cortex can be obtained from the electromyogram signal of the motor cortex in the following manner: a sliding time window analysis technique (window width 100 ms, step size 10 ms) is used to detect the potential peak value in the electromyogram signal of the motor cortex whose signal amplitude exceeds 3 times the standard deviation of the resting state as the signal peak amplitude, and the time interval between adjacent peaks is measured as the peak latency interval; then, the motor evoked potential of the target patient muscle cortex is screened from the electromyogram signal of the motor cortex by using the signal peak amplitude and the peak latency interval in the following manner: a double-threshold screening mechanism is established: first, the signal peak amplitude reaches a preset response threshold (default 50 μV), and second, the peak latency interval is within the normal physiological range (8-30 ms for upper limb muscles and 15-40 ms for lower limb muscles), and the potential signal that meets the signal peak amplitude and peak latency interval standards at the same time is screened from the electromyogram signal of the motor cortex using the above-mentioned double-threshold screening mechanism, and is time-locked with the stimulation pulse after eliminating spontaneous muscle activity and environmental noise interference, and is regarded as an effective motor evoked potential.

[0070] In some embodiments, the excitability of the patient's cortex is evaluated by the motor evoked potential, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device, and the cortex excitability index of the patient in the rehabilitation training is obtained, which is used as a reference for the rehabilitation training of the patient. Figure 3 The figure is a flowchart for determining the cortex excitability index in some embodiments of the present application, and the cortex excitability index can be determined in the present embodiment by using the following steps:

[0071] In step 1031, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device are monitored in real time.

[0072] In step 1032, the basic cortex excitability of the patient's cortex is determined by the signal peak amplitude of the motor evoked potential and the stimulation intensity of the stimulation pulse in the rehabilitation training device.

[0073] In step 1033, the basic cortex excitability is fitted and adjusted according to the stimulation frequency of the stimulation pulse in the rehabilitation training device, and the cortex excitability index of the patient in the rehabilitation training is obtained.

[0074] It should be noted that in the present application, the cortical excitability index represents the activation degree of neurons; the stimulation frequency represents the number of pulses per unit time, the stimulation intensity represents the energy size of a single pulse; the basal cortical excitability represents the neural electrical activity level under the current stimulation intensity.

[0075] In a specific implementation, first, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device can be monitored in real time in the following manner: the stimulation frequency (unit: Hz) and the stimulation intensity (unit: mA) of the stimulation pulse are collected and recorded in real time by the sensors built in the rehabilitation training device, and the sampling interval is not greater than 100 ms to ensure real-time data; then, the basal cortical excitability of the patient's cortex can be determined by the signal peak amplitude of the motor evoked potential and the stimulation intensity of the stimulation pulse in the rehabilitation training device in the following manner: the ratio of the signal peak amplitude of the motor evoked potential to the stimulation intensity of the stimulation pulse in the rehabilitation training device is taken as the basal cortical excitability of the patient's cortex, and this ratio reflects the response efficiency of the neural pathway under unit stimulation intensity; finally, the basal cortical excitability can be fitted and adjusted according to the stimulation frequency of the stimulation pulse in the rehabilitation training device to obtain the cortical excitability index of the patient in the rehabilitation training in the following manner: the stimulation frequency is input into the normalization function built in the rehabilitation training device to generate an adjustment coefficient, and the product of the adjustment coefficient and the basal cortical excitability can be taken as the final cortical excitability index.

[0076] In step 104, the cortical excitability index and the hierarchical remodeling indicator of the neural function regulation are multi-modally and cooperatively integrated to obtain the neural regulation target parameter of the remodeling indicator in the rehabilitation training device, and then the stimulation parameter scheme of the rehabilitation training device is dynamically adjusted based on the neural regulation target parameter, and a visual rehabilitation treatment recommendation report is output to the patient.

[0077] In some embodiments, the multi-modal and cooperative integration of the cortical excitability index and the hierarchical remodeling indicator of the neural function regulation can be realized in the following manner: the cortical excitability index is matched with the hierarchical remodeling indicator of the neural function regulation, and the weighting value corresponding to the matched remodeling indicator is taken as the neural regulation target parameter of the remodeling indicator in the rehabilitation training device.

[0078] In some embodiments, the dynamic adjustment of the stimulation parameter scheme of the rehabilitation training device based on the neural regulation target parameter can be realized in the following steps:

[0079] An online policy optimization model based on deep learning is initialized;

[0080] The neural regulation target parameter is taken as the target value in the online policy optimization model;

[0081] the cortex excitability index as a current state quantity in a proximal policy optimization model;

[0082] The proximal policy optimization model using the input target value and the current state quantity is used to perform gradient adjustment on the stimulation frequency and the stimulation intensity in the rehabilitation training device, so as to adjust the stimulation parameter scheme of the rehabilitation training device, which will not be described herein.

[0083] It should be noted that the proximal policy optimization model is a reinforcement learning algorithm based on policy gradient, which ensures training stability by limiting the amplitude of policy update. In the rehabilitation training system, the proximal policy optimization model is composed of a policy network and a value network. The policy network receives the difference between the current state quantity (cortex excitability index) and the target value (neurological function remodeling level indicator) as input and outputs the adjustment gradient of the stimulation parameter. The value network evaluates the value of the current state to calculate the advantage function. Through the clipping mechanism and the control of the policy update step, the update amplitude is automatically limited when the difference between the new and old policies exceeds the threshold, avoiding system shock caused by drastic adjustment. The proximal policy optimization model is trained in a multi-round iteration manner. Each time, the policy gradient is calculated according to the real-time feedback data (for example: change of motor evoked potential), and the parameter adjustment strategy is gradually optimized, so as to finally realize precise and progressive adjustment of the stimulation frequency and intensity, ensuring that the neuroregulation process is effective and safe.

[0084] In addition, while adjusting the stimulation parameter scheme of the rehabilitation training device, the application can also output a visual rehabilitation treatment recommendation report to the patient. In specific implementation, for example, the dynamically adjusted stimulation parameter scheme is converted into a clinically readable recommendation of treatment intensity and frequency; a treatment progress quantitative evaluation indicator is generated based on the neuroregulation target parameter and the cortex excitability index; the clinically readable recommendation and the quantitative evaluation indicator are presented in a visual report through chart combination, in which the spatiotemporal changes of blood oxygen concentration and rhythm power in different brain regions can be displayed through a heat map, the excitability index is automatically marked red and a voice prompt is played through a warning module when the excitability index exceeds a safety threshold, and the rehabilitation rate can be intuitively displayed through a historical comparison view: superimposing this time and previous training data, and a fold line slope, which will not be described herein.

[0085] In addition, another aspect of the application, in some embodiments, the application provides an artificial intelligence-based neurology rehabilitation training system, which comprises a rehabilitation training control unit, which is described with reference to Figure 4 The figure is a structural schematic diagram of a rehabilitation training control unit according to some embodiments of the application, which comprises a collection module 201, a processing module 202 and an execution module 203, which are described as follows:

[0086] The collection module 201 is mainly used for starting the rehabilitation training equipment to perform rehabilitation training on the patient with a nervous system disease, and collecting physiological signals of the target nerve target point of the patient in the rehabilitation training equipment and electromyography signals of the motor cortex under a specified stimulation pulse during the rehabilitation training;

[0087] The processing module 202 is used for extracting the rhythm power of the stimulation pulse and the blood oxygen concentration of the target nerve target point of the patient from the physiological signals, performing efficiency grading on the remodeling level of the neural function in the rehabilitation training through the blood oxygen concentration and the rhythm power, and obtaining a graded remodeling index of the neural function regulation in the rehabilitation training;

[0088] It should be noted that the processing module 202 is also used for extracting a motor evoked potential of the target patient muscle cortex in the electromyography signals of the motor cortex, evaluating the excitability of the cortex of the patient through the motor evoked potential, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training equipment, and obtaining a cortex excitability index of the patient in the rehabilitation training;

[0089] The execution module 203 is mainly used for multi-modal collaborative integration of the cortex excitability index and the graded remodeling index of the neural function regulation, obtaining a neural regulation target parameter of the remodeling index in the rehabilitation training equipment, dynamically adjusting the stimulation parameter scheme of the rehabilitation training equipment based on the neural regulation target parameter, and outputting a visual rehabilitation treatment recommendation report to the patient.

[0090] The above describes an example of the neural internal medicine rehabilitation training system and regulation method based on artificial intelligence provided by the embodiment of the application. It can be understood that the corresponding device includes the hardware structure and / or software module corresponding to the execution of each function in order to achieve the above functions. Those skilled in the art should easily realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present application can be realized in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in hardware or computer software driven hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0091] In some embodiments, the present application also provides a computer device, which comprises a memory and a processor, the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device executes the above-mentioned neural internal medicine rehabilitation training regulation method based on artificial intelligence.

[0092] In some embodiments, with reference toFigure 5 The dashed line in the figure indicates that the unit or the module is optional, and the figure is a structural schematic diagram of a computer device for implementing the method for regulating and controlling rehabilitation training in neurology based on artificial intelligence provided by the embodiments of the present application. The method for regulating and controlling rehabilitation training in neurology based on artificial intelligence described in the above embodiments can be implemented by the computer device shown in the figure, which includes at least one processor 301, a memory 302, and at least one communication unit 305. The computer device can be a terminal device or a server or a chip. Figure 5 The computer device shown in the figure can be a terminal device or a server or a chip.

[0093] The processor 301 can be a general-purpose processor or a special-purpose processor. For example, the processor 301 can be a central processing unit (CPU), which can be used to control the computer device, execute software programs, and process data of the software programs. The computer device can further include a communication unit 305 to realize input (reception) and output (transmission) of signals.

[0094] For example, the computer device can be a chip, and the communication unit 305 can be an input and / or output circuit of the chip, or the communication unit 305 can be a communication interface of the chip. The chip can be a component of a terminal device or a network device or other devices.

[0095] For another example, the computer device can be a terminal device or a server, and the communication unit 305 can be a transceiver of the terminal device or the server, or the communication unit 305 can be a transceiver circuit of the terminal device or the server.

[0096] The computer device can include one or more memories 302, which have programs 304 stored thereon. The programs 304 can be run by the processor 301 to generate instructions 303, so that the processor 301 executes the method described in the above method embodiments according to the instructions 303. Optionally, the memory 302 can also store data (such as a target review model). Optionally, the processor 301 can also read the data stored in the memory 302. The data can be stored in the same storage address as the program 304, or the data can be stored in a different storage address from the program 304.

[0097] The processor 301 and the memory 302 can be separately arranged or integrated together, for example, integrated on a system on chip (SOC) of the terminal device.

[0098] It should be understood that each step of the above method embodiments can be accomplished by logic circuits in the form of hardware or instructions in the form of software in the processor 301, which can be a CPU, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof.

[0099] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0100] For example, in some embodiments, the present application also provides a computer readable storage medium, wherein instructions or codes are stored in the computer readable storage medium, and when the instructions or codes are run on a computer, the computer is caused to perform the above-mentioned artificial intelligence-based rehabilitation training regulation method for neurology.

[0101] Although preferred embodiments of the application have been described, those skilled in the art will appreciate that additional modifications and variations to the preferred embodiments can be made without departing from the spirit and scope of the application. Accordingly, it is intended that the appended claims be interpreted as including all such alterations and modifications as fall within the scope of the application.

[0102] Obviously, various modifications and changes can be made to the present application by those skilled in the art without departing from the spirit and scope of the present application. Thus, it is intended that the present application cover the modifications and changes as fall within the scope of the claims and their equivalents.

Claims

1. An artificial intelligence-based neurology rehabilitation training system comprising a rehabilitation training regulation unit, characterized in that, The rehabilitation training regulation unit comprises: The acquisition module is used for acquiring physiological signals and motor cortex electromyography signals of a target nerve point of a patient under a specified stimulation pulse collected by a rehabilitation training device during rehabilitation training of the patient after starting the rehabilitation training device for the rehabilitation training of the patient for a nervous system disease; The processing module is used for extracting a rhythm power of the stimulation pulse and a blood oxygen concentration of the target nerve point of the patient from the physiological signals, performing performance grading on a remodeling level of a neural function in the rehabilitation training through the blood oxygen concentration and the rhythm power, and obtaining a graded remodeling index of the neural function regulation in the rehabilitation training; The processing module is also used for extracting a motor evoked potential of a target muscle cortex of the patient from the motor cortex electromyography signals, evaluating an excitability of a cortex of the patient through the motor evoked potential, a stimulation frequency and a stimulation intensity of the stimulation pulse in the rehabilitation training device, and obtaining a cortex excitability index of the patient in the rehabilitation training; The execution module is used for performing multi-modal collaborative integration of the cortex excitability index and the graded remodeling index of the neural function regulation, obtaining a neural regulation target parameter of the remodeling index in the rehabilitation training device, and then performing dynamic regulation on a stimulation parameter in the rehabilitation regulation device based on the neural regulation target parameter; The performance grading on the remodeling level of the neural function in the rehabilitation training through the blood oxygen concentration and the rhythm power, and obtaining the graded remodeling index of the neural function regulation in the rehabilitation training specifically comprises: Performing demand influence normalization on the blood oxygen concentration and the rhythm power to obtain a pulse influence parameter and a blood oxygen influence parameter; Performing parameter weighted fusion on the pulse influence parameter and the blood oxygen influence parameter to obtain the graded remodeling index of the neural function regulation in the rehabilitation training; The evaluation of the excitability of the cortex of the patient through the motor evoked potential, the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device, and obtaining the cortex excitability index of the patient in the rehabilitation training specifically comprises: Real-time monitoring of the stimulation frequency and the stimulation intensity of the stimulation pulse in the rehabilitation training device; Determining a basic cortex excitability of the cortex of the patient through a signal peak amplitude of the motor evoked potential and the stimulation intensity of the stimulation pulse in the rehabilitation training device; Fitting and adjusting the basic cortex excitability according to the stimulation frequency of the stimulation pulse in the rehabilitation training device to obtain the cortex excitability index of the patient in the rehabilitation training; The dynamic regulation on the stimulation parameter in the rehabilitation regulation device based on the neural regulation target parameter specifically comprises: Initializing a proximal policy optimization model based on deep learning; Taking the neural regulation target parameter as a target value in the proximal policy optimization model; Taking the cortex excitability index as a current state quantity in the proximal policy optimization model; Using the proximal policy optimization model with the input target value and the current state quantity to perform gradient adjustment on the stimulation frequency and the stimulation intensity in the rehabilitation training device; The multi-modal collaborative integration of the cortex excitability index and the graded remodeling index of the neural function regulation, and obtaining the neural regulation target parameter of the remodeling index in the rehabilitation training device specifically comprises: The cortical excitability index is matched with the remodeling index of the neural function regulation, so that the weighted value corresponding to the matched remodeling index is taken as the neural regulation target parameter of the remodeling index in the rehabilitation training device.

2. The system of claim 1, wherein, The processing module extracts the rhythm power of the stimulation pulse and the blood oxygen concentration of the target nerve of the patient from the physiological signal, and specifically includes: The electroencephalogram and near-infrared spectrum of the patient in the rehabilitation training of the neurology department are obtained from the physiological signal; The rhythm power of the target frequency band in the stimulation pulse is extracted from the electroencephalogram; The oxygenated hemoglobin concentration change amount of the target area is extracted from the near-infrared spectrum as the blood oxygen concentration of the target nerve of the patient.

3. The system of claim 1, wherein, The processing module extracts the motor evoked potential of the target patient muscle cortex in the electromyogram signal of the motor cortex, and specifically includes: The signal peak amplitude and peak latency interval of the target patient muscle cortex are obtained from the electromyogram signal of the motor cortex; The motor evoked potential of the target patient muscle cortex is screened from the electromyogram signal of the motor cortex through the signal peak amplitude and the peak latency interval.

4. The system of claim 1, wherein, The physiological signal includes electroencephalogram and near-infrared spectrum.

Images