A transcranial light-based closed-loop neuromodulation method and system

Abstract

The application provides a closed-loop neuroregulation method and system based on transcranial light, which comprises the following steps: obtaining scale data, brain blood oxygen information and brain electrical information of a target user, inputting the information into a pre-trained target brain function rehabilitation prediction model, outputting a brain function rehabilitation prediction result of the target user by the target brain function rehabilitation prediction model based on the information, and adjusting a light regulation scheme of the target user based on the brain function rehabilitation prediction result. Since the scale data, brain blood oxygen information and brain electrical information contain brain function information of the target user from multiple aspects, the target brain function rehabilitation prediction model obtains sufficient comprehensive input information, and can output an accurate brain function rehabilitation prediction result, thereby improving the accuracy of the light regulation scheme obtained based on the brain function rehabilitation prediction result, and further improving the gain of transcranial light regulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical equipment, and in particular to a closed-loop neural regulation method and system based on transcranial light. BACKGROUND

[0002] Stroke is the leading cause of death and disability among residents, and presents the characteristics of high incidence, high disability rate, high mortality and high recurrence rate. Although rehabilitation treatment can be accepted, about 60%-80% of stroke patients still have obvious motor dysfunction, which brings heavy care costs to families and society.

[0003] Through physical stimulation such as acousto-optic electromagnetic, non-invasive stimulation is performed on the brain, central, muscle and the like of a patient, and after a period of stimulation, the cortex activity can be improved to promote the reconstruction of neural circuit and motor function, which is an important development direction of neural regulation. Correspondingly, how to stimulate the patient by transcranial light and then improve the accuracy of transcranial light stimulation is an important problem in neural regulation. SUMMARY

[0004] Therefore, the embodiments of the present application provide a closed-loop neural regulation method and system based on transcranial light to improve the accuracy of transcranial light stimulation.

[0005] According to an aspect of the present application, a closed-loop neural regulation method based on transcranial light is provided, and the method comprises:

[0006] Obtaining detection data of a target user, wherein the detection data comprises scale data, cerebral oxygen data and electroencephalogram data of the target user;

[0007] Inputting the detection data into a pre-trained target brain function rehabilitation prediction model, and obtaining a target brain function rehabilitation result output by the target brain function rehabilitation prediction model based on the detection data;

[0008] Adjusting a target light regulation scheme of the target user based on the target brain function rehabilitation result, wherein the target light regulation scheme comprises a position of transcranial light stimulation;

[0009] The target brain function rehabilitation prediction model is pre-trained by the following steps:

[0010] Obtaining a training data set, wherein the training data set comprises a plurality of training data, each training data comprises scale data, cerebral oxygen data and electroencephalogram data of a user, each training data corresponds to a brain function rehabilitation label, and the brain function rehabilitation label is an actual brain function rehabilitation result of the user;

[0011] inputting the training data set into an initial brain function rehabilitation prediction model, so that the initial brain function rehabilitation prediction model outputs a brain function rehabilitation prediction result corresponding to each of the training data based on the training data set;

[0012] adjusting parameters of the brain function rehabilitation prediction model based on a target difference between the brain function rehabilitation prediction result corresponding to each of the training data and the brain function rehabilitation label of each of the training data, until the target difference converges;

[0013] taking the brain function rehabilitation prediction model corresponding to the time when the target difference converges as a target brain function rehabilitation prediction model.

[0014] In a possible embodiment, before the inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model, the method further includes:

[0015] extracting a brain hemisphere activation index, a brain lateralization index, and a hemisphere functional connectivity index based on the brain oxygen data of the target user;

[0016] extracting a global efficiency index and a local efficiency index of different frequency bands based on the electroencephalogram data of the target user;

[0017] The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model includes:

[0018] inputting the brain hemisphere activation index, the brain lateralization index, and the hemisphere functional connectivity index, and the global efficiency index and the local efficiency index of different frequency bands into the target brain function rehabilitation prediction model.

[0019] In a possible embodiment, the method further includes:

[0020] normalizing the scale data and filtering noise of the brain oxygen data and the electroencephalogram data for the to-be-detected data;

[0021] The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model includes:

[0022] inputting the normalized scale data and the filtered brain oxygen data and electroencephalogram data into the pre-trained target brain function rehabilitation prediction model.

[0023] In a possible embodiment, the adjusting the target light regulation scheme of the target user based on the target brain function rehabilitation result includes:

[0024] in a case where the target brain function rehabilitation result is higher than a first preset threshold, performing transcranial light stimulation on a first brain function area of the target user; in a case where the target brain function rehabilitation result is higher than a first preset threshold, performing transcranial light stimulation on a first brain function area of the target user;

[0025] in a case where the target brain function rehabilitation result is lower than a second preset threshold, performing transcranial light stimulation on a second brain function area of the target user, wherein the second preset threshold is smaller than the first preset threshold;

[0026] in a case where the target brain function rehabilitation result is between the second preset threshold and the first preset threshold, performing transcranial light stimulation on the first brain function area and the second brain function area of the target user.

[0027] In a possible embodiment, the method further comprises:

[0028] During the transcranial light stimulation on the target user according to the target light regulation scheme, real-time electroencephalogram data of the target user are continuously monitored, and light stimulation parameters of the transcranial light are adjusted based on the real-time electroencephalogram data, the light stimulation parameters including light stimulation intensity.

[0029] According to another aspect of the present application, a closed-loop neural regulation system based on transcranial light is provided, the system comprising:

[0030] an acquisition module configured to acquire to-be-detected data of a target user, wherein the to-be-detected data comprises scale data, cerebral oxygen data and electroencephalogram data of the target user;

[0031] an input module configured to input the to-be-detected data into a pre-trained target brain function rehabilitation prediction model, and acquire a target brain function rehabilitation result output by the target brain function rehabilitation prediction model based on the to-be-detected data;

[0032] an adjustment module configured to adjust a target light regulation scheme of the target user based on the target brain function rehabilitation result, wherein the target light regulation scheme comprises a position of transcranial light stimulation;

[0033] wherein the target brain function rehabilitation prediction model is pre-trained by the following steps:

[0034] acquiring a training data set, wherein the training data set comprises a plurality of training data, each of the training data comprises scale data, cerebral oxygen data and electroencephalogram data of a user, and each of the training data corresponds to a brain function rehabilitation label, the brain function rehabilitation label being an actual brain function rehabilitation result of the user;

[0035] inputting the training data set into an initial brain function rehabilitation prediction model, so that the initial brain function rehabilitation prediction model outputs a brain function rehabilitation prediction result corresponding to each of the training data based on the training data set;

[0036] adjust parameters of the brain function rehabilitation prediction model based on a target difference between a brain function rehabilitation prediction result corresponding to each of the training data and a brain function rehabilitation label of each of the training data, until the target difference converges;

[0037] use the brain function rehabilitation prediction model corresponding to the time when the target difference converges as a target brain function rehabilitation prediction model.

[0038] In a possible embodiment, the system further includes:

[0039] an extraction module configured to extract a brain hemisphere activation indicator, a brain lateralization indicator, and a hemisphere functional connectivity indicator based on the brain oxygen data of the target user, and extract a global efficiency indicator and a local efficiency indicator of different frequency bands based on the electroencephalogram data of the target user;

[0040] The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model includes:

[0041] The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model includes:

[0042] In a possible embodiment, the system further includes:

[0043] a preprocessing module configured to normalize the scale data and filter noise from the brain oxygen data and the electroencephalogram data for the to-be-detected data;

[0044] The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model includes:

[0045] The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model includes:

[0046] In a possible embodiment, the adjusting the target light regulation scheme of the target user based on the target brain function rehabilitation result includes:

[0047] In a case where the target brain function rehabilitation result is higher than a first preset threshold, performing transcranial light stimulation on a first brain function area of the target user;

[0048] In a case where the target brain function rehabilitation result is lower than a second preset threshold, performing transcranial light stimulation on a second brain function area of the target user, where the second preset threshold is lower than the first preset threshold;

[0049] In a case where the target brain function rehabilitation result is between the second preset threshold and the first preset threshold, the first brain function area and the second brain function area of the target user are subjected to transcranial light stimulation.

[0050] In a possible embodiment, the system further comprises:

[0051] a feedback module configured to continuously monitor real-time electroencephalogram data of the target user during the transcranial light stimulation of the target user according to the target light regulation scheme, and adjust a light stimulation parameter of the transcranial light based on the real-time electroencephalogram data, the light stimulation parameter including a light stimulation intensity.

[0052] The one or more technical solutions provided in the embodiments of the present application obtain scale data, brain blood oxygen information and electroencephalogram information of a target user, input the information into a pre-trained target brain function rehabilitation prediction model, output a brain function rehabilitation prediction result of the target user based on the information by the target brain function rehabilitation prediction model, and adjust a light regulation scheme of the target user based on the brain function rehabilitation prediction result. Since the scale data, brain blood oxygen information and electroencephalogram information contain brain function information of the target user in multiple aspects, the target brain function rehabilitation prediction model obtains sufficient comprehensive input information, and can thus output an accurate brain function rehabilitation prediction result, improving the accuracy of the light regulation scheme obtained based on the brain function rehabilitation prediction result, and further improving the gain of transcranial light regulation. In addition, the target brain function rehabilitation training model is trained based on a large amount of brain function data of users in multiple aspects, so that the target brain function rehabilitation prediction model can learn the correlation between a large amount of brain function data and brain function rehabilitation results, further improving the accuracy of the output brain function rehabilitation prediction result and the gain of transcranial light regulation. BRIEF DESCRIPTION OF DRAWINGS

[0053] In the following description of exemplary embodiments in conjunction with the accompanying drawings, more details, features and advantages of the present application are disclosed, in which:

[0054] Figure 1 A flowchart of a transcranial light-based closed-loop neural regulation method provided by an embodiment of the present application is shown in FIG. 2;

[0055] Figure 2 A training flowchart of a target brain function rehabilitation prediction model in a transcranial light-based closed-loop neural regulation method provided by an embodiment of the present application is shown in FIG. 3;

[0056] Figure 3 A logical structure diagram of a transcranial light-based closed-loop neural regulation system provided by an embodiment of the present application is shown in FIG. 4;

[0057] Figure 4Another logical structure schematic diagram of the transcranial light-based closed-loop neuromodulation system provided by the embodiments of the present application. DETAILED DESCRIPTION

[0058] Embodiments of the present application will be described in more detail with reference to the drawings. While the present application is shown in the drawings and described as being implemented in one or more specific embodiments, it will be understood that the present application can be embodied in various forms and should not be construed as being limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and fully convey the scope of the application to those skilled in the art. It should be understood that the drawings and the embodiments are only for illustrative purposes and should not be used to limit the scope of protection of the present application.

[0059] It should be understood that the various steps in the method embodiments of the present application can be performed in different orders and / or in parallel. In addition, the method embodiments can include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0060] The term "comprising" and variations thereof as used herein are used inclusively, i.e., "comprising, but not limited to". The term "based on" is "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Related definitions will be given in the description below. It should be noted that the concepts "first", "second", etc. mentioned in the present application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0061] It should be noted that the modification of "one" or "multiple" mentioned in the present application is illustrative and not restrictive, and those skilled in the art should understand that unless otherwise explicitly indicated in the context, it should be understood as "one or more".

[0062] The names of the messages or information exchanged between the devices in the embodiments of the present application are only for illustrative purposes, and are not used to limit the scope of the messages or information.

[0063] In order to improve the accuracy of transcranial light stimulation, the present application provides a transcranial light-based closed-loop neuromodulation method and system. The transcranial light-based closed-loop neuromodulation method provided by the present application can be applied to any electronic device with closed-loop neuromodulation function, which can be a computer, a mobile terminal, etc. The scheme of the present application will be described below with reference to the drawings:

[0064] Figure 1 A flowchart of the transcranial light-based closed-loop neuromodulation method provided by the embodiments of the present application,

[0065] S101, acquire the target user's to be detected data, wherein the to be detected data includes the target user's scale data, brain blood oxygen data and brain electrical data;

[0066] S102, input the target user's to be detected data into the pre-trained target brain function rehabilitation prediction model, and acquire the target brain function rehabilitation result output by the target brain function rehabilitation prediction model based on the to be detected data;

[0067] S103, adjust the target light regulation scheme of the target user based on the target brain function rehabilitation result, wherein the target light regulation scheme includes the position of transcranial light stimulation;

[0068] Wherein, the target brain function rehabilitation prediction model is pre-trained by the following steps:

[0069] S201, acquire the training data set, the training data set contains multiple training data, each training data contains the user's scale data, brain blood oxygen data and brain electrical data; each training data corresponds to a brain function rehabilitation label, and the brain function rehabilitation label is the actual brain function rehabilitation result of the user;

[0070] S202, input the training data set into the initial brain function rehabilitation prediction model, so that the initial brain function rehabilitation prediction model outputs the brain function rehabilitation prediction result corresponding to each training data based on the training data set;

[0071] S203, based on the target difference between the brain function rehabilitation prediction result corresponding to each training data and the brain function rehabilitation label of each training data, adjust the parameters of the brain function rehabilitation prediction model until the target difference converges;

[0072] S204, the brain function rehabilitation prediction model corresponding to the target difference when the target difference converges is taken as the target brain function rehabilitation prediction model.

[0073] With the embodiment of the present application, the scale data, the cerebral blood oxygen information and the brain electrical information of the target user are acquired, and the information is input into the target brain function rehabilitation prediction model pre-trained, the target brain function rehabilitation prediction model outputs the brain function rehabilitation prediction result of the target user based on the above information, and adjusts the light regulation scheme of the target user based on the brain function rehabilitation prediction result. Since the scale data, the cerebral blood oxygen information and the brain electrical information contain the brain function information of the target user in multiple aspects, the target brain function rehabilitation prediction model obtains sufficient comprehensive input information, and can output accurate brain function rehabilitation prediction results, thereby improving the accuracy of the light regulation scheme based on the brain function rehabilitation prediction result, and further improving the gain of transcranial light regulation. In addition, the target brain function rehabilitation training model is trained based on a large amount of brain function data of users in multiple aspects, so that the target brain function rehabilitation prediction model can learn the correlation between a large amount of brain function data and brain function rehabilitation results, further improving the accuracy of the output brain function rehabilitation prediction result and the gain of transcranial light regulation.

[0074] The above S101-S103 and S201-S204 are exemplarily described as follows:

[0075] In the present application, the target user refers to a user who needs to be subjected to transcranial light neural regulation. In one possible embodiment, the detection data of the target user can be collected before the target user is subjected to transcranial light neural regulation, so as to generate or adjust the light regulation scheme of the target user. In one possible embodiment, the detection data of the target user can be collected during the rehabilitation training of the target user. The rehabilitation training can include any rehabilitation training scene, such as upper limb rehabilitation, lower limb rehabilitation, etc.

[0076] As one possible implementation, the detection data of the target user can be collected in a resting state and an upper limb rehabilitation training state. The resting state refers to a state in which the target user is inactive and at rest, and in this state, the body and brain of the target user are relatively relaxed and static. The upper limb rehabilitation training state refers to a state in which the target user is undergoing upper limb rehabilitation training, and in this state, the body of the user is in a state of motion. By collecting the detection data of the target user in the resting state and the motion state, the detection data can more comprehensively reflect the brain function of the user.

[0077] The to-be-detected data can include clinical scale data, cerebral oxygen data, and electroencephalogram data of the target user, wherein the clinical scale data refers to physical information and laboratory examination data of the target user, and can include upper limb movement rating and lower limb movement rating of the target user. For example, various clinical scale data generated by the target user can be stored in correspondence with a target user identifier, which can be a code or identity information of the target user. Correspondingly, the corresponding clinical scale data can be obtained based on the target user identifier. As a possible implementation, the cerebral oxygen data of the target user can be collected by using a near-infrared brain function imaging device.

[0078] After obtaining the to-be-detected data of the target user, the to-be-detected data can be input into a pre-trained target brain function rehabilitation prediction model, and the target brain function rehabilitation prediction model can output the target brain function rehabilitation result of the target user based on the to-be-detected data.

[0079] The target brain function rehabilitation prediction model can be pre-trained based on a preset training data set, and each piece of training data in the training data set can include clinical scale information, cerebral oxygen information, and electroencephalogram information of a user. As a possible implementation, the clinical scale data, cerebral oxygen data, and electroencephalogram data of each user who needs to be subjected to transcranial photic neural regulation can be stored in a historical database in correspondence with a user identifier, and the clinical scale data, cerebral oxygen data, and electroencephalogram data of each user can be obtained from the historical database based on the user identifier to form a training data set during model training.

[0080] In a possible embodiment, each piece of user data stored in the historical database can be divided according to a preset ratio to obtain a training data set and a verification data set, and the preset ratio can be set according to an actual application scenario. For example, the user data in the historical database can be divided according to a ratio of 7:3 to obtain the training data set and the verification data set.

[0081] Each piece of training data in the training data set corresponds to a specific brain function rehabilitation label, and the brain function rehabilitation label identifies the brain function recovery result of the user. The specific content of the brain function recovery result can be set according to an actual application scenario. For example, the brain function rehabilitation label can identify good recovery, medium recovery, and poor recovery, and the specific content of the brain function rehabilitation label can be 1, 0.5, and 0, respectively.

[0082] The structure of the target brain function rehabilitation model can be flexibly selected according to an actual application scenario, and can be a CNN (Convolutional Neural Network), an RNN (Recurrent Neural Network), or the like.

[0083] In a possible embodiment, the training data set can be output to an initial brain function rehabilitation prediction model, the brain function rehabilitation prediction model can perform feature extraction on each piece of training data, and output a predicted brain function improvement prediction result corresponding to each piece of training data based on the extracted features.

[0084] Then, a target difference between the predicted brain function improvement result corresponding to each piece of training data and the brain function rehabilitation label corresponding to the training data can be calculated. The target difference can be calculated based on a preset loss function, which can be a cross-entropy loss function, a variance loss function, etc.

[0085] After obtaining the target difference, the parameters of the brain function improvement model can be adjusted based on the target difference. Specifically, the model parameters can be adjusted by gradient descent, gradient ascent, etc., until the target difference converges. The target difference converges means that the difference between the target differences obtained in two consecutive iterations is less than a preset difference threshold, or the target difference is less than a preset difference threshold.

[0086] In a possible embodiment, before the training data set is input to the brain function rehabilitation prediction model, the training data can be preprocessed and index data can be extracted. The preprocessing process can include normalizing the clinical scale data. Specifically, the scale data can be normalized by maximum normalization, Z-core normalization, etc. The preprocessing process can also include filtering the brain oxygen signal and the brain electrical signal to remove noise and artifacts. For example, the brain oxygen signal and the brain electrical signal can be filtered by low-pass filtering, high-pass filtering, etc. according to the actual application scenario.

[0087] The index data extraction can specifically include extracting the brain hemisphere activation index, lateralization, and hemisphere functional connectivity index from the brain oxygen signal; and extracting the global efficiency and local efficiency index of different frequency bands from the brain electrical signal.

[0088] The brain hemisphere activation index is used to evaluate the activity of the two hemispheres of the brain under a specific task or state. In the rehabilitation task, the two hemispheres of the brain are usually divided into the affected side and the non-affected side. The affected side usually refers to the brain hemisphere that is activated to a lower degree under a specific task or state, while the non-affected side refers to the brain hemisphere that is activated to a higher degree. Accordingly, the brain hemisphere activation index can be divided into the affected side brain activation index, the non-affected side brain activation index, and the whole brain activation index.

[0089] For example, the affected side brain activation index can be obtained by the following formula:

[0090]

[0091] Wherein, GC SL is the affected side brain activation index, p is the number of channels of the affected side brain hemisphere in the task state, W RWi is the brain blood oxygen wave amplitude of the i th channel of the affected side brain hemisphere in the task state, g RWi is the proportion of the number of connections of the i th channel of the affected side brain hemisphere in the task state to the number of all channels of the affected side hemisphere, q is the number of channels of the affected side brain hemisphere in the task state, W JXi is the brain blood oxygen wave amplitude of the i th channel of the affected side brain hemisphere in the resting state, g RWi is the proportion of the number of connections of the i th channel of the affected side brain hemisphere in the resting state to the number of all channels of the affected side hemisphere.

[0092] The non-affected side brain activation index can be calculated by the following formula:

[0093]

[0094] Wherein, GC FSL is the non-affected side brain activation value, a is the number of channels of the non-affected side brain hemisphere in the task state, W RWj is the brain blood oxygen wave amplitude of the j th channel of the affected side brain hemisphere in the task state, g RWj is the proportion of the number of connections of the j th channel of the affected side brain hemisphere in the task state to the number of all channels of the affected side hemisphere, b is the number of channels of the non-affected side brain hemisphere in the task state, W JXj' is the brain blood oxygen wave amplitude of the j th channel of the non-affected side brain hemisphere in the resting state, g RWj is the proportion of the number of connections of the j th channel of the non-affected side brain hemisphere in the resting state to the number of all channels of the non-affected side hemisphere.

[0095] The whole brain activation index can be calculated by the following formula:

[0096]

[0097] Wherein, GC NL is the whole brain activation value, m is the number of channels of the whole brain in the task state, W RWk is the brain blood oxygen wave amplitude of the k th channel of the whole brain in the task state, g RWk is the proportion of the number of connections of the k th channel of the whole brain in the task state to the number of all channels, n is the number of channels of the whole brain in the task state, W JXk' is the brain blood oxygen wave amplitude of the k th channel of the whole brain in the resting state, g RWk is the proportion of the number of connections of the k th channel of the whole brain in the resting state to the number of all channels of the whole brain.

[0098] Lateralization refers to the asymmetry in function between the two hemispheres of the brain, with one side potentially dominant in certain functions. For example, for most right-handed people, the left brain typically controls language and logical functions, while the right brain is more involved in spatial processing and facial recognition. The brain lateralization index in embodiments of the present invention can be calculated by the following formula:

[0099]

[0100] wherein PI is the lateralization index, RC SRW is the number of channel connections in the non-affected hemisphere in the task state, RC SJX is the number of channel connections in the non-affected hemisphere in the resting state, RC RW is the number of channel connections in the affected hemisphere in the task state, RC JX is the number of channel connections in the affected hemisphere in the resting state, L ij is the information transmission efficiency of channel i and channel j.

[0101] The hemisphere functional connectivity index is used to quantify the degree of functional connectivity between different regions of the two hemispheres of the brain. High functional connectivity indicates strong interaction and information flow between two regions. The hemisphere functional connectivity index can also include the affected hemisphere functional connectivity index, the non-affected hemisphere functional connectivity index, and the whole brain functional connectivity index. For example, the affected hemisphere functional connectivity value can be calculated by the following formula:

[0102]

[0103] wherein KM SLJ is the affected hemisphere functional connectivity value, DZ SRWLJ is the number of channels with functional connectivity in the affected hemisphere in the task state, DZ SJXLJ is the number of channels with functional connectivity in the affected hemisphere in the resting state, FS RWij is the transmission efficiency of the i-th channel and the j-th channel in the affected hemisphere in the task state, FS JXij is the transmission efficiency of the i-th channel and the j-th channel in the affected hemisphere in the resting state.

[0104] The non-affected hemisphere functional connectivity index can be calculated by the following formula:

[0105]

[0106] wherein KM FSLJ is the non-affected hemisphere functional connectivity value, DZ FSRWLJ is the number of channels with functional connectivity in the non-affected hemisphere in the task state, DZ FSJXLJis the number of channels with functional connection in the non-affected cerebral hemisphere in the resting state, FS FRWij is the transmission efficiency of the i th channel and the j th channel in the non-affected cerebral hemisphere in the task state, FS FJXij is the transmission efficiency of the i th channel and the j th channel in the non-affected cerebral hemisphere in the resting state.

[0107] The whole brain functional connection index can be calculated by the following formula:

[0108]

[0109] Wherein, KM LJ is the whole brain functional connection value, DZ RWLJ is the number of channels with functional connection in the whole brain in the task state, DZ JXLJ is the number of channels with functional connection in the whole brain in the resting state, FS RWij is the transmission efficiency of the i th channel and the j th channel in the whole brain in the task state, FS JXij is the transmission efficiency of the i th channel and the j th channel in the whole brain in the resting state.

[0110] The electroencephalogram information includes four frequency bands: theta band (4-8HZ), alpha band (8-13HZ), beta band (13-30HZ), and gamma band (30-40HZ), therefore, the information of the above different frequency bands can be extracted from each channel of the electroencephalogram information, so as to calculate the global efficiency and local efficiency indexes of different frequency bands. The above channel refers to the conductive path connecting the scalp and the electroencephalogram acquisition device, which can reflect the electrical activity of different parts of the brain. The global efficiency and local efficiency indexes of the above different frequency bands can reflect the connection characteristics of the brain network function, and the specific calculation formula is as follows:

[0111] Global efficiency index in theta band:

[0112]

[0113] Wherein, FN θ is the global efficiency of the electroencephalogram signal in theta band; QB RWθ is the number of connected node channels of the electroencephalogram signal in theta band in the task state; QB JXθ is the number of connected node channels of the electroencephalogram signal in theta band in the resting state; K ij is the shortest path length between node i and node j, i.e. connection efficiency.

[0114] Local efficiency in theta band:

[0115]

[0116] Wherein, HL θis the local efficiency of the θ frequency band of the electroencephalogram; N is the number of nodes; f RWijθ is the number of channels connected between node i and node j in the θ frequency band of the electroencephalogram in the task state and passing through node k; l RWijθ is the shortest path length when node i and node j are connected in the θ frequency band of the electroencephalogram in the task state and passing through node h; d RWijθ is the shortest path length between node i and node j in the θ frequency band of the electroencephalogram in the task state; f JXijθ is the number of channels connected between node i and node j in the θ frequency band of the electroencephalogram in the resting state and passing through node k; l JXijθ is the shortest path length when node i and node j are connected in the θ frequency band of the electroencephalogram in the resting state and passing through node h; d JXijθ is the shortest path length between node i and node j in the θ frequency band of the electroencephalogram in the resting state; G RWijθ is the total number of connections of node i with other nodes in the θ frequency band of the electroencephalogram in the task state.

[0117] Global efficiency in the α frequency band:

[0118]

[0119] wherein, FN a is the global efficiency of the electroencephalogram in the α frequency band; QB RWa is the number of node channels connected in the α frequency band of the electroencephalogram in the task state; QB JXa is the number of node channels connected in the α frequency band of the electroencephalogram in the resting state; K ij is the shortest path length between node i and node j, i.e., the connection efficiency.

[0120] Local efficiency in the α frequency band:

[0121]

[0122] wherein, HL a is the local efficiency of the electroencephalogram in the α frequency band; N is the number of nodes;

[0123] f RWija is the number of channels connected between node i and node j in the α frequency band of the electroencephalogram in the task state and passing through node k; l RWija is the shortest path length when node i and node j are connected in the α frequency band of the electroencephalogram in the task state and passing through node h; d RWia is the shortest path length between node i and node j in the α frequency band of the electroencephalogram in the task state; f JXija is the number of channels connected between node i and node j in the α frequency band of the electroencephalogram in the resting state and passing through node k; l JXija is the shortest path length when node i and node j are connected in the α frequency band of the electroencephalogram in the resting state and passing through node h; dJXija is the shortest path length between node i and node j in the resting state of the a frequency band of the electroencephalogram; G RWija is the total number of connections between node i and other nodes in the task state of the a frequency band of the electroencephalogram.

[0124] Global efficiency in the beta frequency band:

[0125]

[0126] FN = QB - K β is the global efficiency of the beta frequency band of the electroencephalogram; QB RWβ is the number of node channels connected in the beta frequency band of the electroencephalogram in the task state; QB JXβ is the number of node channels connected in the beta frequency band of the electroencephalogram in the resting state; K ij is the shortest path length between node i and node j, i.e., the connection efficiency.

[0127] Local efficiency in the beta frequency band:

[0128]

[0129] HL = N - f β is the local efficiency of the beta frequency band of the electroencephalogram; N is the number of nodes; f RWijβ is the number of channels connected between node i and node j in the beta frequency band of the electroencephalogram in the task state and passing through node k; l RWijβ is the shortest path length when node i and node j are connected in the beta frequency band of the electroencephalogram in the task state and passing through node h; d RWiβ is the shortest path length between node i and node j in the beta frequency band of the electroencephalogram in the task state; f JXijβ is the number of channels connected between node i and node j in the beta frequency band of the electroencephalogram in the resting state and passing through node k; l JXijβ is the shortest path length when node i and node j are connected in the beta frequency band of the electroencephalogram in the resting state and passing through node h; d JXijβ is the shortest path length between node i and node j in the resting state of the beta frequency band of the electroencephalogram; G RWijβ is the total number of connections between node i and other nodes in the beta frequency band of the electroencephalogram in the task state.

[0130] Global efficiency in the gamma frequency band:

[0131]

[0132] FN = QB - K γ is the global efficiency of the gamma frequency band of the electroencephalogram; QB RWγ is the number of node channels connected in the gamma frequency band of the electroencephalogram in the task state; QB JXγ is the number of node channels connected in the gamma frequency band of the electroencephalogram in the resting state; Kij is the shortest path length between node i and node j, i.e., the connection efficiency.

[0133] Local efficiency in the gamma band:

[0134]

[0135] wherein, HL γ is the local efficiency of the gamma band of the electroencephalogram; N is the number of nodes; f RWijγ is the number of channels connecting node i and node j in the gamma band of the electroencephalogram in the task state and passing through node k; l RWijγ is the shortest path length when node i and node j are connected in the gamma band of the electroencephalogram in the task state and passing through node h; d RWiγ is the shortest path length between node i and node j in the gamma band of the electroencephalogram in the task state; f JXijγ is the number of channels connecting node i and node j in the gamma band of the electroencephalogram in the resting state and passing through node k; l JXijγ is the shortest path length when node i and node j are connected in the gamma band of the electroencephalogram in the resting state and passing through node h; d JXijγ is the shortest path length between node i and node j in the gamma band of the electroencephalogram in the resting state; G RWijγ is the total number of connections of node i with other nodes in the gamma band of the electroencephalogram in the task state.

[0136] By extracting various indicators of the brain oxygen and electroencephalogram information and outputting them to the brain function recovery prediction model, the brain function recovery prediction model can more comprehensively and deeply learn the brain oxygen and electroencephalogram information contained in the training data, and the prediction accuracy of the model is improved.

[0137] Correspondingly, in actual application, before the to-be-detected data of the target user is output to the target brain function recovery prediction model, the to-be-detected data can be preprocessed and indicators can be extracted. Specifically, in one possible embodiment, the above method can further include the following steps:

[0138] extracting a brain hemisphere activation indicator, a brain lateralization indicator, and a hemisphere functional connectivity indicator based on the brain oxygen data of the target user;

[0139] extracting global efficiency indicators and local efficiency indicators of different frequency bands based on the electroencephalogram data of the target user.

[0140] Correspondingly, the brain hemisphere activation indicator, the brain lateralization indicator, and the hemisphere functional connectivity indicator, the global efficiency indicators and the local efficiency indicators of different frequency bands can be input to the target brain function recovery prediction model.

[0141] In a possible embodiment, the method can further comprise normalizing the scale data, and filtering noise from the cerebral oxygenation data and the electroencephalogram data for the to-be-detected data; and the normalized scale data and the filtered cerebral oxygenation data and electroencephalogram data can be input into the pre-trained target brain function recovery result prediction model.

[0142] After obtaining the target brain function recovery result of the target user output by the target brain function recovery result prediction model, the transcranial light regulation scheme of the target user can be adjusted based on the recovery result. As described above, the brain function recovery result can be represented by a numerical value, and therefore, as a possible implementation, in a case where the target brain function recovery result is higher than a first preset threshold, the first brain function area of the target user is subjected to transcranial light stimulation; in a case where the target brain function recovery result is lower than a second preset threshold, the second brain function area of the target user is subjected to transcranial light stimulation, wherein the second preset threshold is lower than the first preset threshold; and in a case where the target brain function recovery result is between the second preset threshold and the first threshold, the first brain function area and the second brain function area of the target user are subjected to transcranial light stimulation.

[0143] The first preset threshold is used to distinguish between good brain function recovery and moderate brain function recovery, and the second preset threshold is used to distinguish between moderate brain function recovery and poor brain function recovery. The first brain function area and the second brain function area can be the affected side cerebral hemisphere and the non-affected side cerebral hemisphere, respectively.

[0144] For example, when the target brain function recovery result is poor, it indicates that the recovery of the affected side cerebral hemisphere of the user is poor, and therefore, it is recommended to perform light stimulation on the affected side cerebral hemisphere of the user; when the target brain function recovery result is moderate, it indicates that the affected side cerebral hemisphere and the non-affected side cerebral hemisphere of the user are recovering simultaneously, and therefore, it is recommended to perform light stimulation on the affected side cerebral hemisphere and the non-affected side cerebral hemisphere of the user simultaneously; and when the target brain function recovery result is good, it indicates that the recovery of the affected side cerebral hemisphere of the user is good, and therefore, it is recommended to perform light stimulation on the non-affected side cerebral hemisphere of the user.

[0145] In a possible embodiment, real-time brain blood oxygen information and real-time brain electrical information of the target user can be continuously monitored during transcranial light stimulation of the target user, and the light stimulation parameters of the transcranial light can be adjusted based on the real-time brain blood oxygen information and the real-time brain electrical information. The light stimulation parameters can include light stimulation frequency, light stimulation wavelength, light stimulation intensity, and the like. The parameter adjustment process can be determined according to the actual application scenario. For example, during the transcranial light regulation of the target user according to the target light stimulation scheme, if the brain electrical signal intensity of the target user increases, it can be determined that the target user has a better response to the current light stimulation parameters, and therefore the light stimulation intensity can be increased to achieve greater light regulation gain.

[0146] By the above technical means, the feedback mechanism is applied to adjust the light stimulation parameters, so that the light stimulation parameters can change in time with the change of the brain electrical information of the user, which helps to find greater light regulation gain.

[0147] In a possible embodiment, the model output process, the target transcranial light regulation scheme, and the feedback process can be displayed in a display device to improve the operation convenience of relevant personnel.

[0148] By applying the embodiments of the present application, the brain function improvement of the patient is predicted through the multi-level information changes of the global efficiency and the local efficiency of the brain electrical signals of different frequency bands, and the activation index, the lateralization index, and the functional connection index of the brain blood oxygen signals of different brain hemispheres. The synergistic control effect of the brain electrical signals of different frequency bands and the brain blood oxygen signals of different brain hemispheres is fully considered, and the brain function state of the patient can be comprehensively evaluated and predicted. Further, a personalized adaptive light stimulation regulation scheme can be provided according to the prediction result of the brain function improvement of the patient, so that the light stimulation assisted rehabilitation training can achieve the maximum gain effect, and the rehabilitation training efficiency and effect of the patient can be improved.

[0149] The brain electrical signals of the patient are monitored at the same time of the light stimulation, and the light regulation stimulation parameters of the patient are adjusted in real time according to the change of the brain electrical signals, so as to promote the collaborative optimization and real-time feedback of the brain and the light stimulation data.

[0150] Based on the same inventive concept, the embodiments of the present application also provide a closed-loop neural regulation system based on transcranial light, as shown in Figure 3 The system 300 can include:

[0151] The acquisition module 301 is configured to acquire the to-be-detected data of the target user, wherein the to-be-detected data includes scale data, brain blood oxygen data, and brain electrical data of the target user.

[0152] The input module 302 is configured to input the to-be-detected data into a pre-trained target brain function rehabilitation prediction model, and obtain a target brain function rehabilitation result output by the target brain function rehabilitation prediction model based on the to-be-detected data.

[0153] The adjustment module 303 is configured to adjust a target light regulation scheme of the target user based on the target brain function rehabilitation result, wherein the target light regulation scheme comprises a position of transcranial light stimulation.

[0154] The target brain function rehabilitation prediction model is pre-trained through the following steps.

[0155] A training data set is obtained, the training data set comprising a plurality of training data, each piece of training data comprising scale data, brain blood oxygen data and electroencephalogram data of a user, and each piece of training data corresponding to a brain function rehabilitation label, the brain function rehabilitation label being an actual brain function rehabilitation result of the user.

[0156] The training data set is input into an initial brain function rehabilitation prediction model, so that the initial brain function rehabilitation prediction model outputs a brain function rehabilitation prediction result corresponding to each piece of training data based on the training data set.

[0157] Parameters of the brain function rehabilitation prediction model are adjusted based on a target difference between the brain function rehabilitation prediction result corresponding to each piece of training data and the brain function rehabilitation label of each piece of training data, until the target difference converges.

[0158] The brain function rehabilitation prediction model corresponding to the converged target difference is taken as a target brain function rehabilitation prediction model.

[0159] In a possible embodiment, the system further comprises:

[0160] The extraction module is configured to extract a brain hemisphere activation index, a brain lateralization index and a hemisphere functional connectivity index based on the brain blood oxygen data of the target user, and extract a global efficiency index and a local efficiency index of different frequency bands based on the electroencephalogram data of the target user.

[0161] The inputting of the to-be-detected data into the pre-trained target brain function rehabilitation prediction model comprises:

[0162] The brain hemisphere activation index, the brain lateralization index and the hemisphere functional connectivity index, and the global efficiency index and the local efficiency index of different frequency bands are input into the target brain function rehabilitation prediction model.

[0163] In a possible embodiment, the system further comprises:

[0164] a preprocessing module configured to normalize the scale data, and filter noise from the cerebral oxygenation data and the electroencephalogram data, for the to-be-detected data;

[0165] The inputting of the to-be-detected data into the pre-trained target brain function rehabilitation prediction model comprises:

[0166] The inputting of the normalized scale data, the filtered cerebral oxygenation data, and the filtered electroencephalogram data into the pre-trained target brain function rehabilitation prediction model.

[0167] In a possible embodiment, the adjusting of the target light regulation scheme of the target user based on the target brain function rehabilitation result comprises:

[0168] In a case where the target brain function rehabilitation result is higher than a first preset threshold, transcranial light stimulation is performed on a first brain function area of the target user.

[0169] In a case where the target brain function rehabilitation result is lower than a second preset threshold, transcranial light stimulation is performed on a second brain function area of the target user, wherein the second preset threshold is lower than the first preset threshold.

[0170] In a case where the target brain function rehabilitation result is between the second preset threshold and the first preset threshold, transcranial light stimulation is performed on the first brain function area and the second brain function area of the target user.

[0171] In a possible embodiment, the system further comprises:

[0172] a feedback module configured to continuously monitor real-time electroencephalogram data of the target user during the transcranial light stimulation of the target user according to the target light regulation scheme, and adjust a light stimulation parameter of the transcranial light based on the real-time electroencephalogram data, the light stimulation parameter comprising a light stimulation intensity.

[0173] In a possible embodiment, the system can further comprise a display device configured to display a feedback process of the feedback module, a model output result, and a target light regulation scheme.

[0174] As Figure 4 shown, Figure 4Another kind of logical structure schematic diagram of the transcranial light closed-loop neuromodulation system provided by the embodiment of the present application can comprise: an information acquisition and analysis module, a brain function rehabilitation prediction module, a light regulation recommendation module, a light stimulation module, an electroencephalogram synchronous monitoring module, and a feedback module, wherein the information acquisition and analysis module is used to acquire and analyze the brain blood oxygen and electroencephalogram information of a user in a resting state and an upper limb rehabilitation training task state; the brain function rehabilitation prediction module is used to predict the brain function improvement of the user according to the information of the information acquisition and analysis module; the light regulation recommendation module recommends light stimulation parameters according to the prediction result of the brain function rehabilitation prediction module; the light stimulation module is used to perform light stimulation according to the light stimulation parameters recommended by the light regulation recommendation module; the electroencephalogram synchronous monitoring module is used to perform electroencephalogram information synchronous monitoring while the light regulation is performed; and the feedback module is used to adjust the parameters of the light stimulation module according to the electroencephalogram information monitoring and analysis result of the electroencephalogram synchronous monitoring module.

[0175] In the present application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information comply with relevant laws and regulations and do not violate public order and good customs.

Claims

1. A closed-loop neuromodulation method based on transcranial light, characterized in that, The method comprises: obtaining target user's to-be-detected data, wherein the to-be-detected data comprises the target user's scale data, cerebral oxygen data and electroencephalogram data; inputting the to-be-detected data into a pre-trained target brain function rehabilitation prediction model, and obtaining a target brain function rehabilitation result output by the target brain function rehabilitation prediction model based on the to-be-detected data; adjusting a target light regulation scheme of the target user based on the target brain function rehabilitation result, wherein the target light regulation scheme comprises a position of transcranial light stimulation; wherein the target brain function rehabilitation prediction model is pre-trained by the following steps: obtaining a training data set, wherein the training data set comprises a plurality of training data, each piece of training data comprises scale data, cerebral oxygen data and electroencephalogram data of a user, each piece of training data corresponds to a brain function rehabilitation label, and the brain function rehabilitation label is an actual brain function rehabilitation result of the user; inputting the training data set into an initial brain function rehabilitation prediction model, so that the initial brain function rehabilitation prediction model outputs a brain function rehabilitation prediction result corresponding to each piece of training data based on the training data set; adjusting parameters of the brain function rehabilitation prediction model based on a target difference between the brain function rehabilitation prediction result corresponding to each piece of training data and the brain function rehabilitation label of each piece of training data until the target difference converges; taking the brain function rehabilitation prediction model corresponding to the target difference when the target difference converges as the target brain function rehabilitation prediction model; before inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model, comprising: extracting a brain hemisphere activation index, a brain lateralization index and a hemisphere functional connectivity index based on the cerebral oxygen data of the target user; extracting a global efficiency index and a local efficiency index of different frequency bands based on the electroencephalogram data of the target user; inputting the brain hemisphere activation index, the brain lateralization index and the hemisphere functional connectivity index, the global efficiency index of different frequency bands and the local efficiency index into the target brain function rehabilitation prediction model. The method further comprises:

2. The method of claim 1, wherein, normalizing the scale data and filtering the noise of the cerebral oxygen data and the electroencephalogram data for the to-be-detected data; inputting the normalized scale data and the filtered cerebral oxygen data and electroencephalogram data into the pre-trained target brain function rehabilitation prediction model. The method further comprises: in the case that the target brain function rehabilitation result is higher than a first preset threshold, performing transcranial light stimulation on a first brain function area of the target user.

3. The method of claim 1, wherein, ​ ​ In a case where the target brain function rehabilitation result is lower than a second preset threshold, performing transcranial light stimulation on a second brain function area of the target user, wherein the second preset threshold is smaller than the first preset threshold; In a case where the target brain function rehabilitation result is between the second preset threshold and the first preset threshold, performing transcranial light stimulation on the first brain function area and the second brain function area of the target user.

4. The method of claim 1, wherein, The method further comprises: During the transcranial light stimulation on the target user according to the target light regulation scheme, real-time electroencephalogram data of the target user is continuously monitored, and a light stimulation parameter of the transcranial light is adjusted based on the real-time electroencephalogram data, the light stimulation parameter including a light stimulation intensity.

5. A closed-loop neuroregulation system based on transcranial light, characterized in that, The system comprises: An acquisition module configured to acquire to-be-detected data of a target user, wherein the to-be-detected data includes scale data, brain blood oxygen data, and electroencephalogram data of the target user; An input module configured to input the to-be-detected data into a pre-trained target brain function rehabilitation prediction model, and acquire a target brain function rehabilitation result output by the target brain function rehabilitation prediction model based on the to-be-detected data; An adjustment module configured to adjust a target light regulation scheme of the target user based on the target brain function rehabilitation result, wherein the target light regulation scheme includes a position of transcranial light stimulation. The target brain function rehabilitation prediction model is pre-trained through the following steps: acquiring a training data set, wherein the training data set contains a plurality of training data, each piece of training data contains scale data, brain blood oxygen data, and electroencephalogram data of a user, and each piece of training data corresponds to a brain function rehabilitation label, which is an actual brain function rehabilitation result of the user; inputting the training data set into an initial brain function rehabilitation prediction model, so that the initial brain function rehabilitation prediction model outputs a brain function rehabilitation prediction result corresponding to each piece of training data based on the training data set; adjusting parameters of the brain function rehabilitation prediction model based on a target difference between the brain function rehabilitation prediction result corresponding to each piece of training data and the brain function rehabilitation label of each piece of training data until the target difference converges; taking the brain function rehabilitation prediction model corresponding to the target difference when the target difference converges as the target brain function rehabilitation prediction model; The system further comprises: an extraction module configured to extract a brain hemisphere activation index, a brain lateralization index, and a hemisphere functional connectivity index based on the brain blood oxygen data of the target user, and extract a global efficiency index and a local efficiency index of different frequency bands based on the electroencephalogram data of the target user; the inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model comprises: inputting the brain hemisphere activation index, the brain lateralization index, the hemisphere functional connectivity index, the global efficiency index of different frequency bands, and the local efficiency index into the target brain function rehabilitation prediction model.

6. The system of claim 5, wherein, The system further comprises: The preprocessing module is configured to normalize the scale data, and to filter noise from the cerebral oxygen data and the electroencephalogram data, for the to-be-detected data; The inputting the to-be-detected data into the pre-trained target brain function rehabilitation prediction model comprises: The inputting the normalized scale data, the filtered cerebral oxygen data and the filtered electroencephalogram data into the pre-trained target brain function rehabilitation prediction model.

7. The system of claim 5, wherein, The adjusting the target light regulation scheme of the target user based on the target brain function rehabilitation result comprises: In a case where the target brain function rehabilitation result is higher than a first preset threshold, performing transcranial light stimulation on a first brain function area of the target user; In a case where the target brain function rehabilitation result is lower than a second preset threshold, performing transcranial light stimulation on a second brain function area of the target user, wherein the second preset threshold is less than the first preset threshold; In a case where the target brain function rehabilitation result is between the second preset threshold and the first preset threshold, performing transcranial light stimulation on the first brain function area and the second brain function area of the target user.

8. The system of claim 5, wherein, The system further comprises: The feedback module is configured to continuously monitor real-time electroencephalogram data of the target user in the process of performing transcranial light stimulation on the target user according to the target light regulation scheme, and to adjust a light stimulation parameter of the transcranial light based on the real-time electroencephalogram data, the light stimulation parameter comprising a light stimulation intensity.

Images