Processing method and device for performing brain efficacy prediction based on multi-dimensional indexes

Abstract

The embodiment of the invention relates to a processing method and device for performing brain efficacy prediction based on multi-dimensional indexes. The method comprises the following steps: setting a multi-dimensional index data set for brain efficacy prediction to obtain a corresponding multi-dimensional index set; constructing a deep learning model used for performing brain efficacy prediction as a corresponding brain efficacy prediction model; constructing a model data set and recording the model data set as a corresponding first data set in a manner of performing traditional Chinese and western medicine examination and brain efficacy expert evaluation on pre-recruited volunteer crowds; training a brain efficacy prediction model based on the first data set; and after the model training is finished, performing brain efficacy index prediction according to the multi-dimensional index set of any testee based on the brain efficacy prediction model. According to the invention, prediction efficiency and prediction accuracy can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a processing method and device for predicting brain performance based on multidimensional indicators. Background Art

[0002] Mental Efficacy is a concept that comprehensively reflects the efficiency and effectiveness of brain information processing, encompassing multiple dimensions including cognition, emotion, physical function, and social participation. Traditional mental efficiency prediction mechanisms often focus on a single dimension, such as isolated analyses using neuropsychological scales or neuroimaging indicators. A comprehensive, multidimensional prediction system encompassing physiological, psychological, and Traditional Chinese Medicine (TCM) syndromes has yet to be established. Due to the limitations of single-dimensional data, the accuracy of traditional mental efficiency prediction schemes is difficult to improve upon once it reaches a certain level. Summary of the Invention

[0003] The purpose of the present invention is to provide a processing method, device, electronic device and computer-readable storage medium for predicting brain performance based on multidimensional indicators in response to the defects of the prior art. The present invention first customizes a multidimensional (self-evaluation, neurology, cognitive psychology, and traditional Chinese medicine symptoms) indicator data set for brain performance prediction, and constructs a brain performance prediction model that can predict the brain performance index based on the multidimensional indicator set; then, a model data set is constructed by conducting Chinese and Western medicine examinations and brain performance expert evaluations on a pre-recruited volunteer population, and the brain performance prediction model is trained based on the data set; after the model training is completed, the brain performance index of any subject is predicted based on the brain performance prediction model. The present invention improves the feature richness through a multidimensional comprehensive prediction system that integrates subjective and objective evaluation systems and Chinese and Western medicine evaluation systems, thereby improving prediction accuracy; improves prediction performance and prediction efficiency through the brain performance prediction model; and improves model robustness and generalization through the masking strategy in model training.

[0004] To achieve the above objectives, a first aspect of an embodiment of the present invention provides a method for predicting brain performance based on multidimensional indicators, the method comprising:

[0005] A multidimensional indicator data set for brain performance prediction is set to obtain a corresponding multidimensional indicator set; the multidimensional indicator set is composed of four sub-data sets, namely a self-evaluation indicator set, a neurological indicator set, a cognitive psychology indicator set, and a traditional Chinese medicine syndrome indicator set;

[0006] Constructing a deep learning model for predicting brain performance as a corresponding brain performance prediction model; the brain performance prediction model is used to predict brain performance based on the multidimensional indicator set input into the model and output a corresponding prediction performance index;

[0007] The model data set is constructed by conducting Chinese and Western medicine examinations and brain performance expert evaluation on the pre-recruited volunteer population and recorded as the corresponding first data set;

[0008] Training the brain performance prediction model based on the first data set;

[0009] After the model training is completed, the multidimensional indicator set of any subject input by the user is input into the brain efficiency prediction model for prediction, and the prediction efficiency index output by this prediction is fed back to the current user.

[0010] Preferably, the self-evaluation indicator set is composed of multiple categories of self-evaluation indicators, including at least: a self-evaluation indicator of learning ability, a self-evaluation indicator of memory, a self-evaluation indicator of attention, a self-evaluation indicator of energy and physical strength, a self-evaluation indicator of emotional stability, a self-evaluation indicator of sleep and repair, a self-evaluation indicator of willpower, a self-evaluation indicator of control, a self-evaluation indicator of execution, a self-evaluation indicator of social vitality, a self-evaluation indicator of life vitality, a self-confidence self-evaluation indicator, and a self-awareness self-evaluation indicator; each category of self-evaluation indicator is a self-evaluation score; the score range of all self-evaluation indicators is the same, which is a preset self-score threshold range;

[0011] The neurological indicator set is composed of multiple types of neurological indicators, including at least: brain region volume index, brain white matter integrity index, brain cortical thickness index, functional connectivity density index, neural oscillation synchronization index, and causal interaction index; the brain region volume index is specifically the volume of the hippocampus; the white matter integrity index is specifically the fractional anisotropy score of the corpus callosum; the cortical thickness index is specifically the thickness of the prefrontal cortex; the functional connectivity density index is specifically the global connection strength of the brain's default mode network; the neural oscillation synchronization index is specifically the alpha wave rhythm phase locking value of the electroencephalogram; the causal interaction index is specifically the average strength of the causal flow strength of a preset activated brain region to another or more other preset brain regions;

[0012] The cognitive psychology index set is composed of multiple categories of cognitive psychology indicators, including at least: MMSE evaluation index, MoCA evaluation index, AVLT evaluation index, WMS evaluation index, VFT evaluation index, BNT evaluation index, TMT-A evaluation index, TMT-B evaluation index, CDT evaluation index, Stroop test evaluation index, ADL evaluation index, NPI evaluation index, HAMA evaluation index, HAMD evaluation index; the MMSE evaluation index is specifically the Mini-Mental State Examination evaluation score; the MoCA evaluation index is specifically the Montreal Cognitive Assessment score; the AVLT evaluation index is specifically the Auditory Verbal Learning Test evaluation score; the WMS evaluation index is specifically the Wei The VFT evaluation index is specifically the verbal fluency test evaluation score; the BNT evaluation index is specifically the Boston Naming Test evaluation score; the TMT-A and B evaluation indexes are specifically the line drawing test A and B evaluation scores; the CDT evaluation index is specifically the clock drawing test evaluation score; the Stroop test evaluation index is specifically the Stroop test evaluation score; the ADL evaluation index is specifically the daily living ability scale evaluation score; the NPI evaluation index is specifically the neuropsychiatric questionnaire evaluation score; the HAMA evaluation index is specifically the Hamilton Anxiety Rating Scale evaluation score; the HAMD evaluation index is specifically the Hamilton Depression Rating Scale evaluation score;

[0013] The TCM syndrome index set is composed of multiple categories of TCM syndrome indicators, including at least: marrow sea gradually emptying syndrome indicator, spleen and kidney deficiency syndrome indicator, qi and blood deficiency syndrome indicator, phlegm turbidity obstructing the orifices syndrome indicator, blood stasis blocking the brain meridians syndrome indicator, heart and liver fire hyperactivity syndrome indicator, and toxicity and extreme deficiency syndrome indicator; each category of syndrome indicator is a binary indicator, specifically yes or no;

[0014] The first data set includes multiple first data records; the first data record includes a first training indicator set and a first label index; the first training indicator set is a multidimensional indicator set; the self-evaluation indicator set of the first training indicator set maintains data integrity and has no invalid indicators; if there are preset invalid indicators in the first training indicator set, the ratio of the total number of invalid indicators to the total number of all indicators meets the preset first masking rate, and all the invalid indicators are randomly distributed in the neurological indicator set, the cognitive psychology indicator set and the traditional Chinese medicine syndrome indicator set; the total number of invalid indicators is the total number of invalid indicators, and the total number of all indicators is the total number of indicators in the multidimensional indicator set.

[0015] Preferably, the model input end of the brain performance prediction model is used to receive the multidimensional indicator set, and the model output end is used to output the corresponding prediction performance index;

[0016] The brain performance prediction model includes a preprocessing module, a self-evaluation feature coding module, a neurological feature coding module, a cognitive psychology feature coding module, a traditional Chinese medicine syndrome feature coding module, a Western medicine feature fusion module, a traditional Chinese and Western medicine feature fusion module, a subjective and objective feature fusion module, a feature mapping module and an index prediction module;

[0017] The input end of the preprocessing module is connected to the input end of the model, and the first, second, third and fourth output ends are respectively connected to the input ends of the self-evaluation feature coding module, the neurological feature coding module, the cognitive psychological feature coding module and the traditional Chinese medicine syndrome feature coding module; the output ends of the neurological feature coding module and the cognitive psychological feature coding module are connected to the first and second input ends of the Western medicine feature fusion module; the output ends of the traditional Chinese medicine syndrome feature coding module and the Western medicine feature fusion module are connected to the first and second input ends of the traditional Chinese and Western medicine feature fusion module; the output ends of the self-evaluation feature coding module and the traditional Chinese and Western medicine feature fusion module are connected to the first and second input ends of the subjective and objective feature fusion module; the output end of the subjective and objective feature fusion module is connected to the input end of the feature mapping module; the output end of the feature mapping module is connected to the input end of the index prediction module; the output end of the index prediction module is connected to the output end of the model;

[0018] The preprocessing module is used to use the self-evaluation index set, the neurological index set, the cognitive psychology index set and the traditional Chinese medicine syndrome index set of the multidimensional index set as the corresponding self-evaluation index vector X 1 , neurological index vector X 2 , cognitive psychological index vector X 3 and TCM syndrome index vector X 4 The self-evaluation index vector X is sent to the corresponding self-evaluation feature coding module, the neurological feature coding module, the cognitive psychology feature coding module, and the TCM syndrome feature coding module; 1 The shape is C1×1, the neurological index vector X 2 The shape is C2×1, the cognitive psychological index vector X 3 The shape is C3×1, the TCM syndrome index vector X 4 The shape is C4×1, where C1, C2, C3, and C4 are the preset first, second, third, and fourth feature dimensions;

[0019] The self-evaluation feature encoding module is implemented based on the MLP model; the self-evaluation feature encoding module is used to encode the self-evaluation index vector X 1 Perform feature encoding processing to obtain the corresponding feature tensor H 1 Sending to the subjective and objective feature fusion module;

[0020] The feature tensor H 1 The reasoning process is:

[0021] h 11 =GELU(W 11 X 1 +b 11 ),

[0022] h 12 =GELU(W 12 h 11 +b 12 ),

[0023] H 1 =Reshape(h 12 ,C1,C7), C7=C6 / C1;

[0024] Among them, W 11 、W 12 are the two weight matrix parameters of the self-evaluation feature encoding module, b 11 、b 12 are two offset vector parameters of the self-evaluation feature encoding module; h 11 、h 12 are the two process feature vectors of the self-evaluation feature encoding module; GELU() is the GELU activation function; the weight matrix parameter W 11 The shape is C5×C1, the offset vector parameter b 11 The shape is C5×1; the weight matrix parameter W 12 The shape is C6×C5, the offset vector parameter b 12 The shape is C6×1; the process feature vector h 11 、h 12 The shape is C5×1, C6×1; Reshape() is a vector shape reshaping function used to reshape the process feature vector h into a shape of C6×1. 12 Converted into a feature tensor H of shape C1×C7 1 ; C5, C6, C7 are the preset fifth, sixth and seventh characteristic dimensions, C1<C5<C6, C7=C6 / C1;

[0025] The neurological feature encoding module is implemented based on the MLP model; the neurological feature encoding module is used to encode the neurological index vector X 2 Perform feature encoding processing to obtain the corresponding feature tensor H 2 Sending to the Western medicine feature fusion module;

[0026] The feature tensor H 2 The reasoning process is:

[0027] h 21 =GELU(W 21 X 2 +b 21 ),

[0028] h 22 =GELU(W 22 h 21 +b 22 ),

[0029] H 2 =Reshape(h 22 ,C1,C7);

[0030] Among them, W 21 、W 22 are the two weight matrix parameters of the neural feature encoding module, b 21 、b 22 are two offset vector parameters of the neurological feature encoding module; h 21 、h 22 are the two process feature vectors of the neurological feature encoding module; the weight matrix parameter W 21 The shape is C8×C2, the offset vector parameter b 21 The shape is C8×1; the weight matrix parameter W 22 The shape is C9×C8, the offset vector parameter b 22 The shape is C9×1; the process feature vector h 21 、h 22 The shape of the vector is C8×1, C9×1; the vector reshape function Reshape() is used to reshape the process feature vector h into a shape of C9×1 22 Converted into a feature tensor H of shape C1×C7 2 ; C8 and C9 are the preset eighth and ninth characteristic dimensions, C2<C8<C9, C9=C1×C7;

[0031] The cognitive psychological feature encoding module is implemented based on the MLP model; the cognitive psychological feature encoding module is used to encode the cognitive psychological index vector X 3 Perform feature encoding processing to obtain the corresponding feature tensor H 3 Sending to the Western medicine feature fusion module;

[0032] The feature tensor H 3 The reasoning process is:

[0033] h 31 =GELU(W 31 X 3 +b 31 ),

[0034] h32 =GELU(W 32 h 31 +b 32 ),

[0035] H 3 =Reshape(h 32 ,C1,C7);

[0036] Among them, W 31 、W 32 are the two weight matrix parameters of the cognitive psychological feature encoding module, b 31 、b 32 are two offset vector parameters of the cognitive psychological feature encoding module; h 31 、h 32 are the two process feature vectors of the cognitive psychological feature coding module; the weight matrix parameter W 31 The shape is C 10 ×C3, offset vector parameter b 31 The shape is C 10 ×1; weight matrix parameter W 32 The shape is C 11 ×C 10 , offset vector parameter b 32 The shape is C 11 ×1; process characteristic vector h 31 、h 32 The shape is C 10 ×1, C 11 ×1; the vector reshape function Reshape() is used to reshape the vector into C 11 ×1 process characteristic vector h 32 Converted into a feature tensor H of shape C1×C7 3 ; C 10 、C 11 For the preset tenth and eleventh characteristic dimensions, C3<C 10 <C 11 、C 11 =C1×C7;

[0037] The TCM syndrome feature encoding module is implemented based on the MLP model; the TCM syndrome feature encoding module is used to encode the TCM syndrome index vector X 4 Perform feature encoding processing to obtain the corresponding feature tensor H 4 Sending to the Chinese and Western medicine feature fusion module;

[0038] The feature tensor H 4 The reasoning process is:

[0039] h 41 =GELU(W41 X 4 +b 41 ),

[0040] h 42 =GELU(W 42 h 41 +b 42 ),

[0041] H 4 =Reshape(h 42 ,C4,C 16 ), C 16 =C 15 / C4=C7;

[0042] Among them, W 41 、W 42 are the two weight matrix parameters of the TCM syndrome feature encoding module, b 41 、b 42 are two offset vector parameters of the TCM syndrome feature encoding module; 41 、h 42 are the two process feature vectors of the TCM syndrome feature encoding module; the weight matrix parameter W 41 The shape is C 12 ×C4, offset vector parameter b 41 The shape is C 12 ×1; weight matrix parameter W 42 The shape is C 13 ×C 12 , offset vector parameter b 42 The shape is C 13 ×1; process characteristic vector h 41 、h 42 The shape is C 12 ×1, C 13 ×1; the vector reshape function Reshape() is used to reshape the vector into C 13 ×1 process characteristic vector h 42 Converted into a feature tensor H of shape C1×C7 4 ; C 12 、C 13 For the preset twelfth and thirteenth characteristic dimensions, C4<C 12 <C 13 、C 13 =C1×C7;

[0043] The Western medicine feature fusion module is used to 2 Generate the corresponding query tensor Q, based on the feature tensor H 3Generate corresponding key and value tensors K and V; and perform attention feature extraction based on the query, key and value tensors Q, K and V to obtain the corresponding attention tensor A; and extract the feature tensor H 2 Perform residual connection with the attention tensor A to obtain the corresponding feature tensor H 5 Sending to the Chinese and Western medicine feature fusion module;

[0044] The feature tensor H 5 The reasoning process is:

[0045] Q=H 2 W Q , K=H 3 W K , V=H 3 W V ,

[0046]

[0047] H 5 =H 2 +A;

[0048] Among them, W Q 、W K 、W V are the three weight matrix parameters of the Western medicine feature fusion module; the weight matrix parameter W Q 、W K 、W V The shape is C7×C qkv , the preset feature dimension C qkv =C7; the query tensor Q, the key tensor K, the value tensor V, the attention tensor A and the feature tensor H 5 The shape of is C1×C7; Softmax() is the Softmax function;

[0049] The Chinese and Western medicine feature fusion module is used to combine the feature tensor H 4 and the feature tensor H 5 Perform feature concatenation to obtain a shape of C1×C 14 The characteristic tensor H 6 , C 14 is the preset fourteenth characteristic dimension, C 14 =2×C7; and by the feature tensor H 6 Perform a full connection operation to obtain a gated weight tensor G with a shape of C1×1; and form a gated weight tensor G with a shape of C1×C7 by C7 of the gated weight tensors G ’ ; and based on the gated weight tensor G ’ , the feature tensor H 4and the feature tensor H 5 Perform feature fusion processing to obtain the corresponding feature tensor H 7 Sending to the subjective and objective feature fusion module;

[0050] The feature tensor H 7 The reasoning process is:

[0051] G=Sigmoid(H 6 W G +b G ),

[0052] H 7 =G′⊙H 4 +(IG′)⊙H 5 ;

[0053] Among them, W G 、b G The weight matrix parameters and offset vector parameters of the Chinese and Western medicine feature fusion module are as follows: G The shape is C 14 ×1, offset vector parameter b G The shape of is C1×1; Sigmoid() is the Sigmoid activation function; the shape of the gated weight tensor G is C1×1; I is a preset all-1 tensor with a shape of C1×C7; ⊙ is the Hadamard product; the feature tensor H 7 The shape of all is C1×C7;

[0054] The subjective and objective feature fusion module is used to fusion the feature tensor H 1 and the feature tensor H 7 Perform feature fusion processing to obtain the corresponding feature tensor H 8 Sending to the feature mapping module;

[0055] The feature tensor H 8 The reasoning process is:

[0056] H 8 =H 1 +H 7 ;

[0057] The feature tensor H 8 The shape is C1×C7;

[0058] The feature mapping module is used to use a 1×1 convolution kernel to map the feature tensor H 8 Perform feature dimensionality reduction to obtain a shape of C1×C 15 The characteristic tensor H 9 , C 15 is the preset fifteenth characteristic dimension, C1<C15 <C7; and the characteristic tensor H 9 Perform global maximum pooling along the indicator dimension to obtain a shape of 1×C 15 The eigenvector H 10 ; and the eigenvector H 10 Perform full connection operation to obtain the corresponding feature scalar H 11 Sending to the index prediction module;

[0059] The characteristic scalar H 11 The reasoning process is:

[0060] H 11 =H 10 W 51 +b 51 ;

[0061] Among them, W 51 、b 51 The weight matrix parameter and offset vector parameter of the feature mapping module; the weight matrix parameter W 51 The shape is C 15 ×C 16 , offset vector parameter b G The shape is 1×C 16 ; C 16 is the preset sixteenth characteristic dimension, C 16 =1;

[0062] The index prediction module is used to calculate the weight scalar W according to the preset 61 , offset scalar b 61 For the characteristic scalar H 11 Perform linear scaling to obtain the corresponding prediction performance index and output it; prediction performance index = W 61 ×H 11 +b 61 .

[0063] Preferably, the model dataset constructed by conducting Chinese and Western medicine examinations and brain performance expert evaluations on the pre-recruited volunteer population is recorded as the corresponding first dataset, specifically including:

[0064] Taking each volunteer in the volunteer group as a corresponding current volunteer;

[0065] And all the self-assessment indicators obtained by the current volunteer through self-assessment according to the preset brain efficiency self-assessment scale and the corresponding scale self-assessment rules constitute a corresponding self-assessment indicator set; the brain efficiency self-assessment scale includes learning ability self-assessment items, memory self-assessment items, attention self-assessment items, energy and physical strength self-assessment items, emotional stability self-assessment items, sleep repair ability self-assessment items, willpower self-assessment items, control ability self-assessment items, execution ability self-assessment items, social vitality self-assessment items, life vitality self-assessment items, self-confidence self-assessment items, and self-awareness self-assessment items; the self-assessment items of the brain efficiency self-assessment scale correspond one to one with the self-assessment indicators of the self-assessment indicator set; The self-assessment rules of the scale include self-assessment rules for learning ability, memory, attention, energy and physical strength, emotional stability, sleep and repair ability, willpower, control, execution, social vitality, life vitality, self-confidence, and self-awareness; the self-assessment rules of the scale correspond one-to-one to the self-assessment items of the brain efficiency self-assessment scale; each self-assessment rule of the scale is used to set the score of the current self-assessment item in a graded manner, and the scoring rule corresponding to each grade is explained;

[0066] and medical experts from a preset first expert group perform resting-state brain structural magnetic resonance imaging, diffusion tensor imaging, and functional magnetic resonance imaging on the current volunteer to obtain corresponding sMRI images, DTI images, and fMRI images; and perform hippocampal brain region segmentation and volume measurement based on the sMRI images based on FreeSurfer, FSL, or SPM analysis tools, and use the obtained hippocampal volume as the corresponding brain region volume index; and perform cerebral cortical surface reconstruction based on the sMRI images based on FreeSurfer or CAT12 analysis tools, and measure the prefrontal cortex thickness on the reconstructed cerebral cortical surface, and use the obtained prefrontal cortex thickness as the corresponding cerebral cortical thickness index; and calculate the partial anisotropy score of the corpus callosum based on the DTI image based on the FSL analysis tool, and use the calculated score as the corresponding brain white matter integrity index; and analyze the global connectivity strength of the brain's default mode network based on the fMRI images based on the CONN or GRETNA analysis tool, and use the analysis result as the corresponding functional connectivity density index; the first expert group is a medical imaging expert group;

[0067] A second group of medical experts, comprising experts in neuroelectrophysiology and neuroscience, performs an electroencephalogram (EEG) on the current volunteer to obtain a corresponding first EEG graph; and performs an alpha wave rhythm phase-locking value analysis based on the first EEG graph using EEGLAB, FieldTrip, or Chronux analysis tools, and uses the analysis result as the corresponding neural oscillation synchronization indicator.

[0068] Medical experts from a preset third expert group perform transcranial magnetic stimulation on the activated brain area of ​​the current volunteer, and simultaneously perform electroencephalogram (EEG) examinations on the activated brain area of ​​the current volunteer and one or more other brain areas to obtain corresponding second EEG graphs and third EEG atlases; and according to the Granger causality analysis method, analyze the causal flow strength of the activated brain area of ​​the current volunteer and each of the other brain areas based on the second EEG graph and the third EEG atlas, calculate the mean of all the obtained causal flow strengths, and use the calculation result as the corresponding causal interaction indicator; the second EEG graph corresponds to the activated brain area; the third EEG atlas consists of one or more third EEG graphs, and the third EEG graphs correspond one-to-one with the other brain areas; the third expert group is a neuroelectrophysiology and neuroscience expert group;

[0069] and having medical experts from a preset fourth expert group evaluate all cognitive psychological indicators of the cognitive psychological indicator set of the current volunteer; the fourth expert group is an expert group in neuropsychology, behavioral neurology, psychiatry, and neurology;

[0070] and having medical experts from a preset fifth expert group evaluate all cognitive psychological indicators of the TCM syndrome index set of the current volunteer; the fifth expert group is an expert group in the field of TCM;

[0071] A joint expert group consisting of the first, second, third, fourth and fifth expert groups assesses and scores the current brain performance status of the volunteer and uses the obtained assessment score as the corresponding first label index;

[0072] The brain area volume index, the white matter integrity index, the cerebral cortical thickness index, the functional connectivity density index, the neural oscillation synchronization index, and the causal interaction index corresponding to the current volunteer form a corresponding neurological index set; and the self-evaluation index set, the neurological index set, the cognitive psychology index set, and the traditional Chinese medicine syndrome index set corresponding to the current volunteer form a corresponding first training index set;

[0073] and performing a random replacement process on the indicators of the neurological indicator set, the cognitive psychology indicator set, and the TCM syndrome indicator set of the current first training indicator set based on the first masking rate and the invalid indicator to obtain a derived first training indicator set, and repeating the random replacement process N times to obtain N derived first training indicator sets; N is a preset positive integer;

[0074] Each first training indicator set and the corresponding first tag index corresponding to the current volunteer form a corresponding first data record; and N+1 first data records corresponding to the current volunteer form a corresponding first data record set;

[0075] And all the first data record sets corresponding to all the volunteers in the volunteer group form the corresponding first data set.

[0076] Preferably, the training of the brain performance prediction model based on the first data set specifically includes:

[0077] Step 51: Split the first data set into two sub-data sets based on a preset first split ratio and record them as a corresponding first training set and a first evaluation set;

[0078] Wherein, both the first training set and the first evaluation set are composed of a plurality of the first data records; the ratio of the total number of records in the first training set and the first evaluation set satisfies the first segmentation ratio;

[0079] Step 52: Perform a round of traversal on all the first data records of the first training set; and during this round of traversal, use the first data record currently traversed as the corresponding current training record; and input the first training indicator set of the current training record as the current multidimensional indicator set into the brain performance prediction model for prediction, and use the prediction efficiency index output by this prediction as the corresponding first prediction index; and form a corresponding first prediction-label pair from the first prediction index and the first label index of the current training record; and at the end of this round of traversal, bring all the first prediction-label pairs obtained in this round of traversal into the preset first model loss function to calculate and obtain the corresponding first loss value;

[0080] Wherein, the first model loss function is implemented based on the L1 loss function or the L2 loss function;

[0081] Step 53: Identify whether the first loss value satisfies a preset first loss value range; if so, proceed to step 54; if not, perform a round of modulation on the model parameters of the brain performance prediction model in a direction that minimizes the first model loss function based on a preset first model optimizer, and return to step 52 to continue training at the end of this round of modulation;

[0082] Wherein, the first model optimizer includes at least an Adam optimizer and an SGD optimizer;

[0083] Step 54: perform a round of traversal on all the first data records of the first evaluation set; and during this round of traversal, use the first data record currently traversed as the corresponding current evaluation record; and input the first training indicator set of the current evaluation record as the current multidimensional indicator set into the brain performance prediction model for prediction, and use the prediction efficiency index output by this prediction as the corresponding second prediction index; and form a corresponding second prediction-label pair from the second prediction index and the first label index of the current evaluation record; and at the end of this round of traversal, bring all the second prediction-label pairs obtained in this round of traversal into the preset first model evaluation function to calculate and obtain the corresponding first evaluation value;

[0084] Wherein, the first model evaluation function is implemented based on MAE function, MSE function or RMSE function;

[0085] Step 55 , identifying whether the first evaluation value meets a preset first evaluation value range; if not, returning to step 52 to continue training; if so, confirming that the model training is completed.

[0086] A second aspect of an embodiment of the present invention provides a device for implementing the processing method for predicting brain performance based on multidimensional indicators described in the first aspect, the device comprising: a multidimensional indicator system construction module, a model construction module, a data set construction module, a model training module, and a model prediction module;

[0087] The multidimensional indicator system construction module is used to set the multidimensional indicator data set for brain performance prediction to obtain a corresponding multidimensional indicator set; the multidimensional indicator set is composed of sub-data sets of four dimensions, namely a self-evaluation indicator set, a neurological indicator set, a cognitive psychology indicator set, and a traditional Chinese medicine syndrome indicator set;

[0088] The model building module is used to build a deep learning model for predicting brain performance as a corresponding brain performance prediction model; the brain performance prediction model is used to predict brain performance based on the multidimensional indicator set input by the model and output a corresponding prediction performance index;

[0089] The data set construction module is used to construct a model data set by performing Chinese and Western medicine examinations and brain performance expert evaluation on a pre-recruited volunteer group, which is recorded as the corresponding first data set;

[0090] The model training module is used to train the brain performance prediction model based on the first data set;

[0091] The model prediction module is used to input the multidimensional indicator set of any subject input by the user into the brain performance prediction model for prediction after the model training is completed, and to feed back the prediction performance index output by this prediction to the current user.

[0092] A third aspect of an embodiment of the present invention provides an electronic device, including: a memory, a processor, and a transceiver;

[0093] The processor is configured to be coupled to the memory, read and execute instructions in the memory, so as to implement the method steps described in the first aspect above;

[0094] The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0095] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a computer, the computer executes the instructions of the method described in the first aspect above.

[0096] The embodiment of the present invention provides a processing method, device, electronic device and computer-readable storage medium for predicting brain performance based on multidimensional indicators. As can be seen from the above content, the embodiment of the present invention first customizes the multidimensional (self-evaluation, neurology, cognitive psychology, and traditional Chinese medicine symptoms) indicator data set for brain performance prediction, and constructs a brain performance prediction model that can predict the brain performance index based on the multidimensional indicator set; then, a model data set is constructed by conducting Chinese and Western medicine examinations and brain performance expert evaluations on a pre-recruited volunteer population, and the brain performance prediction model is trained based on the data set; after the model training is completed, the brain performance index of any subject is predicted based on the brain performance prediction model. The embodiment of the present invention improves the richness of data features and improves prediction accuracy through a multidimensional comprehensive prediction system that integrates subjective and objective evaluation systems and Chinese and Western medicine evaluation systems; improves prediction performance and prediction efficiency by using a brain performance prediction model; and improves the robustness and generalization of the model through a masking strategy set in model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Figure 1 A schematic diagram of a processing method for predicting brain performance based on multidimensional indicators provided in Example 1 of the present invention;

[0098] Figure 2 A schematic diagram of the data structure of a multi-dimensional indicator set provided in the first embodiment of the present invention;

[0099] Figure 3 This is a schematic diagram of the modules of the brain performance prediction model provided in Example 1 of the present invention;

[0100] Figure 4 This is a module structure diagram of a processing device for predicting brain performance based on multi-dimensional indicators provided in the second embodiment of the present invention;

[0101] Figure 5 This is a structural diagram of an electronic device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0102] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the embodiments described herein are merely some, rather than all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0103] The first embodiment of the present invention provides a method for predicting brain performance based on multi-dimensional indicators, such as Figure 1 A schematic diagram of a method for predicting brain performance based on multidimensional indicators provided in the first embodiment of the present invention is shown. The method mainly includes the following steps:

[0104] Step 1: Set the multidimensional indicator data set for brain performance prediction to obtain the corresponding multidimensional indicator set.

[0105] Here, the multidimensional index set of the embodiment of the present invention is composed of sub-data sets of four dimensions, namely, a self-evaluation index set, a neurological index set, a cognitive psychology index set, and a traditional Chinese medicine syndrome index set. Figure 2 This is a schematic diagram of the data structure of the multidimensional indicator set provided in Example 1 of the present invention.

[0106] The self-evaluation index set of the embodiment of the present invention is composed of multiple categories of self-evaluation indicators, including at least: learning ability self-evaluation index, memory self-evaluation index, attention self-evaluation index, energy and physical strength self-evaluation index, emotional stability self-evaluation index, sleep repair ability self-evaluation index, willpower self-evaluation index, control self-evaluation index, execution self-evaluation index, social vitality self-evaluation index, life vitality self-evaluation index, self-confidence self-evaluation index, and self-awareness self-evaluation index. Each category of self-evaluation indicator is a self-evaluation score; the score range of all self-evaluation indicators is the same, which is a preset self-evaluation threshold range. The self-evaluation threshold range here is a preset score range, such as a score range of 0-10.

[0107] The neurological indicator set of the embodiment of the present invention is composed of multiple types of neurological indicators, including at least: brain region volume index, brain white matter integrity index, brain cortical thickness index, functional connectivity density index, neural oscillation synchronization index, and causal interaction index. Among them, the brain region volume index is specifically the volume of the hippocampus; the white matter integrity index is specifically the fractional anisotropy (FA) score of the corpus callosum; the cortical thickness index is specifically the thickness of the prefrontal cortex; the functional connectivity density index is specifically the global connection strength of the brain's default mode network (DMN); the neural oscillation synchronization index is specifically the phase locking value (PLV) of the alpha wave rhythm of the electroencephalogram (EEG); and the causal interaction index is specifically the average strength of the causal flow strength of a preset activated brain region to another or more other preset brain regions.

[0108] The cognitive psychology indicator set consists of multiple categories of cognitive psychology indicators, including at least: MMSE evaluation indicators, MoCA evaluation indicators, AVLT evaluation indicators, WMS evaluation indicators, VFT evaluation indicators, BNT evaluation indicators, TMT-A evaluation indicators, TMT-B evaluation indicators, CDT evaluation indicators, Stroop test evaluation indicators, ADL evaluation indicators, NPI evaluation indicators, HAMA evaluation indicators, and HAMD evaluation indicators. The MMSE assessment index is specifically the Mini-Mental State Examination (MMSE) assessment score; the MoCA assessment index is specifically the Montreal Cognitive Assessment (MoCA) assessment score; the AVLT assessment index is specifically the Auditory Verbal Learning Test (AVLT) assessment score; the WMS assessment index is specifically the Wechsler Memory Scale (WMS) test assessment score; the VFT assessment index is specifically the Verbal Fluency Test (VFT) assessment score; the BNT assessment index is specifically the Boston Naming Test (BNT) assessment score; the TMT-A, B assessment index is specifically the Trail Making Test-A, B (TMT-A, B) assessment score; the CDT assessment index is specifically the Clock Drawing Test (CDT) assessment score; the Stroop test assessment index is specifically the Stroop test assessment score; and the ADL assessment index is specifically the Activity of Daily Living Scale (ADL). The NPI assessment index is specifically the assessment score of the Neuropsychiatric Inventory (NPI); the HAMA assessment index is specifically the assessment score of the Hamilton Anxiety Rating Scale (HAMA); and the HAMD assessment index is specifically the assessment score of the Hamilton Depression Rating Scale (HAMD).

[0109] The TCM syndrome indicator set is composed of multiple categories of TCM syndrome indicators, including at least: marrow sea gradually emptying syndrome indicator, spleen and kidney deficiency syndrome indicator, qi and blood deficiency syndrome indicator, phlegm and turbidity obstructing the orifices syndrome indicator, blood stasis blocking the brain meridians syndrome indicator, heart and liver fire hyperactivity syndrome indicator, and toxic excess and extreme deficiency syndrome indicator; each category of syndrome indicator is a binary indicator, specifically yes or no.

[0110] Step 2: Build a deep learning model for brain performance prediction as the corresponding brain performance prediction model.

[0111] Here, the brain performance prediction model of the embodiment of the present invention is used to predict brain performance based on the multi-dimensional indicator set input by the model and output a corresponding prediction performance index.

[0112] like Figure 3 As shown in the module diagram of the brain performance prediction model provided in the first embodiment of the present invention, the model input end of the brain performance prediction model is used to receive a multi-dimensional indicator set, and the model output end is used to output a corresponding prediction performance index.

[0113] The model components of the brain performance prediction model include: preprocessing module, self-evaluation feature encoding module, neurological feature encoding module, cognitive psychological feature encoding module, traditional Chinese medicine syndrome feature encoding module, Western medicine feature fusion module, traditional Chinese and Western medicine feature fusion module, subjective and objective feature fusion module, feature mapping module and index prediction module.

[0114] The connection relationship of the model components of the brain efficiency prediction model is as follows: the input end of the preprocessing module is connected to the model input end, and the first, second, third and fourth output ends are connected to the input ends of the self-evaluation feature coding module, the neurological feature coding module, the cognitive psychological feature coding module, and the traditional Chinese medicine syndrome feature coding module respectively; the output ends of the neurological feature coding module and the cognitive psychological feature coding module are connected to the first and second input ends of the Western medicine feature fusion module; the output ends of the traditional Chinese medicine syndrome feature coding module and the Western medicine feature fusion module are connected to the first and second input ends of the traditional Chinese and Western medicine feature fusion module; the output ends of the self-evaluation feature coding module and the traditional Chinese and Western medicine feature fusion module are connected to the first and second input ends of the subjective and objective feature fusion module; the output end of the subjective and objective feature fusion module is connected to the input end of the feature mapping module; the output end of the feature mapping module is connected to the input end of the index prediction module; and the output end of the index prediction module is connected to the output end of the model.

[0115] The model components of the brain performance prediction model are shown below.

[0116] 1) Preprocessing module:

[0117] The preprocessing module is used to take the self-evaluation index set, neurological index set, cognitive psychology index set and TCM syndrome index set of the multidimensional index set as the corresponding self-evaluation index vector X 1 , neurological index vector X 2 , cognitive psychological index vector X 3 and TCM syndrome index vector X 4 Send to the corresponding self-evaluation feature coding module, neurological feature coding module, cognitive psychology feature coding module, and traditional Chinese medicine syndrome feature coding module.

[0118] Here, the self-evaluation index vector X 1 The shape is C1×1, the neurological indicator vector X 2 The shape is C2×1, the cognitive psychological index vector X 3 The shape is C3×1, and the TCM syndrome index vector X 4 The shape is C4×1, where C1, C2, C3, and C4 are the preset first, second, third, and fourth feature dimensions, for example, C1=13, C2=6, C3=14, and C4=7.

[0119] 2) Self-evaluation feature encoding module:

[0120] The self-evaluation feature encoding module is implemented based on the MLP model; the self-evaluation feature encoding module is used to encode the self-evaluation index vector X 1 Perform feature encoding processing to obtain the corresponding feature tensor H 1 Send to the subjective and objective feature fusion module.

[0121] Here, the feature tensor H 1 The reasoning process is:

[0122] h 11 =GELU(W 11 X 1 +b 11 ),

[0123] h 12 =GELU(W 12 h 11 +b 12 ),

[0124] H 1 =Reshape(h 12 ,C1,C7), C7=C6 / C1;

[0125] Among them, W 11 、W 12 are the two weight matrix parameters of the self-evaluation feature encoding module, b 11 、b 12 are the two offset vector parameters of the self-evaluation feature encoding module; h 11 、h 12 are the two process feature vectors of the self-evaluation feature encoding module; GELU() is the GELU activation function; the weight matrix parameter W 11 The shape is C5×C1, the offset vector parameter b 11 The shape is C5×1; the weight matrix parameter W 12 The shape is C6×C5, the offset vector parameter b 12 The shape is C6×1; the process feature vector h 11 、h12 The shape is C5×1, C6×1; Reshape() is a vector shape reshaping function used to reshape the process feature vector h into a shape of C6×1. 12 Converted into a feature tensor H of shape C1×C7 1 ; C5, C6, C7 are the preset fifth, sixth and seventh characteristic dimensions, C1<C5<C6, C7=C6 / C1, for example, C5=256, C6=832, C7=64.

[0126] 3) Neurological feature encoding module:

[0127] The neurological feature encoding module is implemented based on the MLP model; the neurological feature encoding module is used to encode the neurological index vector X 2 Perform feature encoding processing to obtain the corresponding feature tensor H 2 Send to the Western medicine feature fusion module.

[0128] Here, the feature tensor H 2 The reasoning process is:

[0129] h 21 =GELU(W 21 X 2 +b 21 ),

[0130] h 22 =GELU(W 22 h 21 +b 22 ),

[0131] H 2 =Reshape(h 22 ,C1,C7);

[0132] Among them, W 21 、W 22 are the two weight matrix parameters of the neural feature encoding module, b 21 、b 22 are the two offset vector parameters of the neural feature encoding module; h 21 、h 22 are the two process feature vectors of the neurological feature encoding module; the weight matrix parameter W 21 The shape is C8×C2, the offset vector parameter b 21 The shape is C8×1; the weight matrix parameter W 22 The shape is C9×C8, the offset vector parameter b 22 The shape is C9×1; the process feature vector h 21 、h 22The shape is C8×1, C9×1; the vector shape reshaping function Reshape() is used to reshape the process feature vector h into a shape of C9×1 22 Converted into a feature tensor H of shape C1×C7 2 ; C8 and C9 are the preset eighth and ninth characteristic dimensions, C2<C8<C9, C9=C1×C7, for example, C8=256, C9=832.

[0133] 4) Cognitive psychological feature encoding module:

[0134] The cognitive psychological feature encoding module is implemented based on the MLP model; the cognitive psychological feature encoding module is used to encode the cognitive psychological index vector X 3 Perform feature encoding processing to obtain the corresponding feature tensor H 3 Send to the Western medicine feature fusion module;

[0135] Here, the feature tensor H 3 The reasoning process is:

[0136] h 31 =GELU(W 31 X 3 +b 31 ),

[0137] h 32 =GELU(W 32 h 31 +b 32 ),

[0138] H 3 =Reshape(h 32 ,C1,C7);

[0139] Among them, W 31 、W 32 are the two weight matrix parameters of the cognitive psychological feature encoding module, b 31 、b 32 are the two offset vector parameters of the cognitive psychological feature encoding module; h 31 、h 32 are the two process feature vectors of the cognitive psychological feature encoding module; the weight matrix parameter W 31 The shape is C 10 ×C3, offset vector parameter b 31 The shape is C 10 ×1; weight matrix parameter W 32 The shape is C 11 ×C 10 , offset vector parameter b 32 The shape is C 11 ×1; process characteristic vector h 31 、h32 The shape is C 10 ×1, C 11 ×1; the vector shape reshaping function Reshape() is used to reshape the shape to C 11 ×1 process characteristic vector h 32 Converted into a feature tensor H of shape C1×C7 3 ; C 10 、C 11 For the preset tenth and eleventh characteristic dimensions, C3<C 10 <C 11 、C 11 =C1×C7, for example, C 10 =256, C 11 =832.

[0140] 5) Traditional Chinese Medicine Syndrome Characteristic Coding Module:

[0141] The TCM syndrome feature encoding module is implemented based on the MLP model; the TCM syndrome feature encoding module is used to encode the TCM syndrome index vector X 4 Perform feature encoding processing to obtain the corresponding feature tensor H 4 Send to the Chinese and Western medicine feature fusion module.

[0142] Here, the feature tensor H 4 The reasoning process is:

[0143] h 41 =GELU(W 41 X 4 +b 41 ),

[0144] h 42 =GELU(W 42 h 41 +b 42 ),

[0145] H 4 =Reshape(h 42 ,C4,C 16 ), C 16 =C 15 / C4=C7;

[0146] Among them, W 41 、W 42 are the two weight matrix parameters of the TCM syndrome feature encoding module, b 41 、b 42 are two offset vector parameters of the TCM syndrome feature encoding module; h 41 、h 42 are the two process feature vectors of the TCM syndrome feature encoding module; the weight matrix parameter W 41The shape is C 12 ×C4, offset vector parameter b 41 The shape is C 12 ×1; weight matrix parameter W 42 The shape is C 13 ×C 12 , offset vector parameter b 42 The shape is C 13 ×1; process characteristic vector h 41 、h 42 The shape is C 12 ×1, C 13 ×1; the vector shape reshaping function Reshape() is used to reshape the shape to C 13 ×1 process characteristic vector h 42 Converted into a feature tensor H of shape C1×C7 4 ; C 12 、C 13 For the preset twelfth and thirteenth characteristic dimensions, C4<C 12 <C 13 、C 13 =C1×C7, for example, C 12 =256, C 13 =832.

[0147] 6) Western medicine feature fusion module:

[0148] Western medicine feature fusion module is used to 2 Generate the corresponding query tensor Q, based on the feature tensor H 3 Generate the corresponding key and value tensors K and V; and perform attention feature extraction based on the query, key, and value tensors Q, K, and V to obtain the corresponding attention tensor A; and perform feature extraction on the feature tensor H 2 Perform residual connection with the attention tensor A to obtain the corresponding feature tensor H 5 Send to the Chinese and Western medicine feature fusion module.

[0149] Here, the feature tensor H 5 The reasoning process is:

[0150] Q=H 2 W Q , K=H 3 W K , V=H 3 W V ,

[0151]

[0152] H 5 =H 2 +A;

[0153] Among them, W Q 、W K 、W V are the three weight matrix parameters of the Western medicine feature fusion module; the weight matrix parameter W Q 、W K 、W V The shape is C7×C qkv , the preset feature dimension C qkv = C7; query tensor Q, key tensor K, value tensor V, attention tensor A and feature tensor H 5 The shape of is C1×C7; Softmax() is the Softmax function.

[0154] 7) Module for integrating Chinese and Western medicine features:

[0155] The Chinese and Western medicine feature fusion module is used to fusion the feature tensor H 4 and the feature tensor H 5 Perform feature concatenation to obtain a shape of C1×C 14 The characteristic tensor H 6 , C 14 is the preset fourteenth characteristic dimension, C 14 =2×C7; and by the feature tensor H 6 Perform a full connection operation to obtain a gated weight tensor G with a shape of C1×1; and C7 gated weight tensors G form a gated weight tensor G with a shape of C1×C7 ’ ; and based on the gated weight tensor G ’ , feature tensor H 4 and the feature tensor H 5 Perform feature fusion processing to obtain the corresponding feature tensor H 7 Send to the subjective and objective feature fusion module.

[0156] Here, the feature tensor H 7 The reasoning process is:

[0157] G=Sigmoid(H 6 W G +b G ),

[0158] H 7 =G′⊙H 4 +(IG′)⊙H 5 ;

[0159] Among them, W G 、b G The weight matrix parameters and offset vector parameters of the Chinese and Western medicine feature fusion module; the weight matrix parameter W G The shape is C 14×1, offset vector parameter b G The shape of is C1×1; Sigmoid() is the Sigmoid activation function; the shape of the gated weight tensor G is C1×1; I is a preset all-1 tensor with a shape of C1×C7; ⊙ is the Hadamard product; the feature tensor H 7 The shape is C1×C7.

[0160] 8) Subjective and objective feature fusion module:

[0161] The subjective and objective feature fusion module is used to fusion the feature tensor H 1 and the feature tensor H 7 Perform feature fusion processing to obtain the corresponding feature tensor H 8 Send to the feature mapping module.

[0162] Here, the feature tensor H 8 The reasoning process is:

[0163] H 8 =H 1 +H 7 ;

[0164] Among them, the feature tensor H 8 The shape is C1×C7.

[0165] 9) Feature Mapping Module:

[0166] The feature mapping module is used to use a 1×1 convolution kernel to map the feature tensor H 8 Perform feature dimensionality reduction to obtain a shape of C1×C 15 The characteristic tensor H 9 , C 15 is the preset fifteenth characteristic dimension, C1<C 15 <C7;and for the feature tensor H 9 Perform global maximum pooling along the indicator dimension to obtain a shape of 1×C 15 The eigenvector H 10 ; and for the eigenvector H 10 Perform full connection operation to obtain the corresponding feature scalar H 11 Send to the index prediction module.

[0167] Here, the characteristic scalar H 11 The reasoning process is:

[0168] H 11 =H 10 W 51 +b 51 ;

[0169] Among them, W 51 、b 51is the weight matrix parameter and offset vector parameter of the feature mapping module; the weight matrix parameter W 51 The shape is C 15 ×C 16 , offset vector parameter b G The shape is 1×C 16 ; C 16 is the preset sixteenth characteristic dimension, C 16 =1.

[0170] 10) Index prediction module:

[0171] The index prediction module is used to predict the index based on the preset weight scalar W 61 , offset scalar b 61 For the characteristic scalar H 11 Perform linear scaling to obtain the corresponding prediction performance index and output it.

[0172] Here, the prediction efficiency index = W 61 ×H 11 +b 61 ;

[0173] Weight scalar W 61 , offset scalar b 61 are two learnable scalar parameters of the exponential prediction module.

[0174] Step 3: construct a model data set by conducting Chinese and Western medicine examinations and brain performance expert evaluations on the pre-recruited volunteer population, which is recorded as the corresponding first data set.

[0175] Here, the pre-recruited volunteer population is a volunteer population covering multiple age groups; and when recruiting volunteers, the volunteers' brain efficiency is pre-evaluated, and the number of four types of pre-evaluated volunteers (volunteers with suspected severe decline in brain efficiency, volunteers with suspected moderate decline in brain efficiency, volunteers with suspected mild decline in brain efficiency, and volunteers with suspected good brain efficiency) is balanced so that the ratio of the four types of pre-evaluated volunteers is approximately 1:1:1:1.

[0176] The first data set includes multiple first data records; the first data record includes a first training indicator set and a first label index; the first training indicator set is a multidimensional indicator set; the self-evaluation indicator set of the first training indicator set maintains data integrity and has no invalid indicators; if there are preset invalid indicators in the first training indicator set, the ratio of the total number of invalid indicators to the total number of all indicators meets the preset first masking rate, and all invalid indicators are randomly distributed in the neurological indicator set, the cognitive psychology indicator set and the traditional Chinese medicine syndrome indicator set; the total number of invalid indicators is the total number of invalid indicators, and the total number of all indicators is the total number of indicators in the multidimensional indicator set; the first masking rate is a preset ratio parameter, and the maximum value of the ratio parameter cannot exceed 40%, for example 30%, 40%.

[0177] The current step 3 specifically includes:

[0178] Step 3-1, taking each volunteer in the volunteer group as the corresponding current volunteer;

[0179] Step 3-2: All self-evaluation indicators obtained by the current volunteer through self-evaluation according to the preset brain efficiency self-evaluation scale and the corresponding scale self-evaluation rules form a corresponding self-evaluation indicator set;

[0180] Here, the brain performance self-assessment scale of the embodiment of the present invention is a customized assessment scale, consisting of multiple self-assessment items, namely: learning ability self-assessment item, memory self-assessment item, attention self-assessment item, energy and physical strength self-assessment item, emotional stability self-assessment item, sleep and repair ability self-assessment item, willpower self-assessment item, control self-assessment item, execution self-assessment item, social vitality self-assessment item, life vitality self-assessment item, self-confidence self-assessment item, and self-awareness self-assessment item; the self-assessment items of the brain performance self-assessment scale correspond one-to-one to the self-assessment indicators of the self-assessment indicator set;

[0181] The self-assessment rules for the Brain Efficacy Scale consist of multiple self-assessment rules, namely: self-assessment rules for learning ability, self-assessment rules for memory, self-assessment rules for attention, self-assessment rules for energy and physical strength, self-assessment rules for emotional stability, self-assessment rules for sleep and repair, self-assessment rules for willpower, self-assessment rules for control, self-assessment rules for execution, self-assessment rules for social vitality, self-assessment rules for life vitality, self-confidence, and self-awareness. The self-assessment rules of the scale correspond one-to-one to the self-assessment items of the Brain Efficacy Scale.

[0182] Each self-assessment rule of the scale is used to set the score of the current self-assessment item in a graded manner, and to explain the scoring rules corresponding to each grade;

[0183] Step 3-3, the medical experts of the first expert group are assigned to perform resting-state brain structural magnetic resonance imaging (sMRI), diffusion tensor imaging (DTI) and functional magnetic resonance imaging (FMRI) on the current volunteers. Corresponding sMRI images, DTI images and fMRI images were obtained by using FreeSurfer, FSL or SPM analysis tools; hippocampal brain region segmentation and volume measurement were performed based on the sMRI images, and the obtained hippocampal volume was used as the corresponding brain region volume index; cortical surface reconstruction was performed based on the sMRI images based on FreeSurfer or CAT12 analysis tools, and the prefrontal cortex thickness was measured on the reconstructed cortical surface, and the obtained prefrontal cortex thickness was used as the corresponding cortical thickness index; partial anisotropy scores of the corpus callosum were calculated based on the DTI images based on the FSL analysis tool, and the calculated scores were used as the corresponding white matter integrity index; global connectivity strength of the default mode network of the brain was analyzed based on the fMRI images based on the CONN or GRETNA analysis tools, and the analysis results were used as the corresponding functional connectivity density index;

[0184] Here, the first expert group in the embodiment of the present invention is a medical imaging expert group;

[0185] In step 3-4, a second group of medical experts, who are pre-set, performs an electroencephalogram (EEG) on the current volunteer to obtain a corresponding first EEG graph; and performs an alpha wave rhythm phase-locking value analysis based on the first EEG graph using EEGLAB, FieldTrip, or Chronux analysis tools, and uses the analysis results as the corresponding neural oscillation synchronization index;

[0186] Here, the second expert group of the embodiment of the present invention is a neuroelectrophysiology and neuroscience expert group;

[0187] In steps 3-5, medical experts from a preset third expert group perform transcranial magnetic stimulation (TMS) on the activated brain area of ​​the current volunteer, and simultaneously perform electroencephalogram (EEG) examinations on the activated brain area of ​​the current volunteer and one or more other brain areas to obtain corresponding second EEG graphs and third EEG atlases; and analyze the causal flow strength of the activated brain area of ​​the current volunteer and each other brain area based on the second EEG graph and the third EEG atlas according to the Granger causality analysis method, calculate the mean of all the obtained causal flow strengths, and use the calculated result as the corresponding causal interaction indicator;

[0188] Here, the second EEG map corresponds to the activated brain area; the third EEG map set consists of one or more third EEG maps, and the third EEG maps correspond one-to-one to other brain areas;

[0189] The third expert group of the embodiment of the present invention is the neuroelectrophysiology and neuroscience expert group;

[0190] Steps 3-6, and the medical experts of the preset fourth expert group evaluate all cognitive psychological indicators of the current volunteer's cognitive psychological indicator set;

[0191] Here, the fourth expert group of the embodiment of the present invention is the expert group of neuropsychology, behavioral neurology, psychiatry and neurology;

[0192] Steps 3-7, and the medical experts of the preset fifth expert group will evaluate all cognitive psychological indicators of the current volunteer's TCM syndrome index set;

[0193] Here, the fifth expert group in the embodiment of the present invention is an expert group in the field of traditional Chinese medicine;

[0194] Step 3-8: A joint expert group consisting of the first, second, third, fourth and fifth expert groups assesses and scores the current brain performance status of the volunteer and uses the resulting assessment score as the corresponding first label index;

[0195] Steps 3-9, the brain volume index, white matter integrity index, cortical thickness index, functional connectivity density index, neural oscillation synchronization index, and causal interaction index corresponding to the current volunteer are used to form a corresponding neurological index set; and the self-evaluation index set, neurological index set, cognitive psychology index set, and traditional Chinese medicine syndrome index set corresponding to the current volunteer are used to form a corresponding first training index set;

[0196] Step 3-10, performing a random replacement process on the indicators of the neurological indicator set, the cognitive psychological indicator set, and the traditional Chinese medicine syndrome indicator set of the current first training indicator set based on the first masking rate and the invalid indicator to obtain a derived first training indicator set, and repeating the random replacement process N times to obtain N derived first training indicator sets;

[0197] Here, N is a preset positive integer. The random replacement process in the embodiment of the present invention is actually a random masking mechanism. Based on this random masking mechanism, on the one hand, the sample size expansion effect of data augmentation is achieved, and on the other hand, the diversity of test data is improved. Training a model based on such data helps improve the model's predictive performance, predictive robustness, and model generalization.

[0198] Step 3-11, each first training indicator set corresponding to the current volunteer and the corresponding first label index form a corresponding first data record; and the N+1 first data records corresponding to the current volunteer form a corresponding first data record set;

[0199] Step 3-12, and forming a corresponding first data set from all first data record sets corresponding to all volunteers in the volunteer population.

[0200] Step 4: training a brain performance prediction model based on the first data set;

[0201] Specifically comprising: step 41, dividing the first data set into two sub-data sets based on a preset first division ratio and recording them as a corresponding first training set and a first evaluation set;

[0202] The first split ratio is a preset ratio parameter, such as 8:2; the first training set and the first evaluation set are both composed of a plurality of first data records; the ratio of the total number of records in the first training set to the total number of records in the first evaluation set satisfies the first split ratio;

[0203] Step 42: Perform a round of traversal on all first data records of the first training set; during this round of traversal, use the currently traversed first data record as the corresponding current training record; input the first training indicator set of the current training record into the brain performance prediction model as the current multidimensional indicator set for prediction, and use the prediction performance index output by this prediction as the corresponding first prediction index; and form a corresponding first prediction-label pair from the first prediction index and the first label index of the current training record; and at the end of this round of traversal, bring all the first prediction-label pairs obtained in this round of traversal into the preset first model loss function to calculate the corresponding first loss value;

[0204] Wherein, the first model loss function is implemented based on the L1 loss function or the L2 loss function;

[0205] Step 43: Identify whether the first loss value satisfies a preset first loss value range; if so, proceed to step 44; if not, perform a round of modulation on the model parameters of the brain performance prediction model in a direction that minimizes the first model loss function based on a preset first model optimizer, and return to step 42 to continue training at the end of this round of modulation;

[0206] Here, the first loss value range is a preset numerical range; the first model optimizer includes at least an Adam optimizer and an SGD optimizer;

[0207] Step 44: Perform a round of traversal on all first data records of the first evaluation set; during this round of traversal, use the currently traversed first data record as the corresponding current evaluation record; input the first training indicator set of the current evaluation record into the brain performance prediction model as the current multidimensional indicator set for prediction, and use the predicted performance index output by this prediction as the corresponding second prediction index; and form a corresponding second prediction-label pair from the second prediction index and the first label index of the current evaluation record; and at the end of this round of traversal, bring all the second prediction-label pairs obtained in this round of traversal into the preset first model evaluation function to calculate and obtain the corresponding first evaluation value;

[0208] Wherein, the first model evaluation function is implemented based on the MAE function, the MSE function or the RMSE function;

[0209] Step 45 , identifying whether the first evaluation value meets the preset first evaluation value range; if not, returning to step 42 to continue training; if so, confirming that the model training is completed.

[0210] Here, the first evaluation value range is a preset numerical range.

[0211] Step 5: After the model training is completed, the multidimensional indicator set of any subject input by the user is input into the brain performance prediction model for prediction and the prediction performance index output by this prediction is fed back to the current user.

[0212] Here, in the multidimensional indicator set of the subjects, the self-evaluation indicator set must maintain complete data and no invalid indicators; the neurological indicator set, cognitive psychology indicator set and traditional Chinese medicine syndrome indicator set can have invalid indicators, as long as the ratio of the total number of invalid indicators to the total number of all indicators does not exceed the first masking rate.

[0213] It should be noted that for the multidimensional indicator set of the embodiment of the present invention, it is also possible to perform indicator addition and deletion operations in each sub-indicator set (self-evaluation indicator set, neurological indicator set, cognitive psychology indicator set, and traditional Chinese medicine syndrome indicator set) based on actual application needs. After each indicator addition and deletion operation is completed, it is necessary to adaptively adjust the training data in the first data set, adaptively adjust the input / output structure of each level of the brain performance prediction model module, and re-train the adjusted brain performance prediction model based on the adjusted first data set. In this way, a new version of the prediction model adapted to the new version of the multidimensional indicator set can be obtained.

[0214] Figure 4This is a module structure diagram of a processing device for predicting brain performance based on multi-dimensional indicators provided in the second embodiment of the present invention. The device is a terminal device or server that implements the aforementioned method embodiment, or it can be a device that enables the aforementioned terminal device or server to implement the aforementioned method embodiment. For example, the device can be a device or chip system of the aforementioned terminal device or server. Figure 4 As shown, the device includes: a multidimensional indicator system construction module 201, a model construction module 202, a data set construction module 203, a model training module 204 and a model prediction module 205.

[0215] The multidimensional indicator system construction module 201 is used to set the multidimensional indicator data set for brain performance prediction to obtain the corresponding multidimensional indicator set; the multidimensional indicator set is composed of sub-data sets of four dimensions, namely, a self-evaluation indicator set, a neurological indicator set, a cognitive psychology indicator set and a traditional Chinese medicine syndrome indicator set.

[0216] The model building module 202 is used to build a deep learning model for brain performance prediction as a corresponding brain performance prediction model; the brain performance prediction model is used to predict brain performance based on the multidimensional indicator set input by the model and output the corresponding prediction performance index.

[0217] The data set construction module 203 is used to construct a model data set by performing Chinese and Western medicine examinations and brain performance expert evaluations on a pre-recruited volunteer group, which is recorded as the corresponding first data set.

[0218] The model training module 204 is used to train the brain performance prediction model based on the first data set.

[0219] The model prediction module 205 is used to input the multidimensional indicator set of any subject input by the user into the brain performance prediction model for prediction after the model training is completed, and to feed back the prediction performance index output by this prediction to the current user.

[0220] An embodiment of the present invention provides a processing device for predicting brain performance based on multidimensional indicators, which can execute the method steps in the above method embodiment. Its implementation principles and technical effects are similar and will not be repeated here.

[0221] It should be noted that the division of the modules of the above devices is merely a division of logical functions. In actual implementation, they can be fully or partially integrated into a single physical entity or physically separated. Furthermore, these modules can be implemented entirely in the form of software called by a processing element; or entirely in the form of hardware; or some modules can be implemented in the form of software called by a processing element, and some modules can be implemented in the form of hardware. For example, the multidimensional indicator system construction module can be a separate processing element, or it can be integrated into a chip of the above device. Furthermore, it can be stored in the form of program code in the memory of the above device, and called by a processing element of the above device to perform the functions of the above-mentioned module. The implementation of other modules is similar. Furthermore, these modules can be fully or partially integrated together, or implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. During implementation, each step of the above method or each of the above modules can be completed by hardware integrated logic circuits in the processor element or by software instructions.

[0222] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs). For another example, when a module is implemented by scheduling program code through a processing element, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0223] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the above method embodiments are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The above-mentioned computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the above-mentioned computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, Bluetooth, microwave, etc.) means. The above-mentioned computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The above-mentioned available medium can be a magnetic medium (such as a floppy disk, hard disk, tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0224] Figure 5 This is a schematic diagram of the structure of an electronic device provided in the third embodiment of the present invention. The electronic device can be a terminal device or server that implements the method of the aforementioned embodiment, or it can be a terminal device or server that implements the method of the aforementioned embodiment connected to the aforementioned terminal device or server. Figure 5 As shown, the electronic device may include: a processor 301 (such as a CPU), a memory 302, and a transceiver 303; the transceiver 303 is coupled to the processor 301, and the processor 301 controls the transceiver 303's transceiver actions. Various instructions may be stored in the memory 302 for completing various processing functions and implementing the processing steps described in the aforementioned embodiment method. Preferably, the electronic device involved in the embodiment of the present invention further includes: a power supply 304, a system bus 305, and a communication port 306. The system bus 305 is used to realize communication connections between components. The above-mentioned communication port 306 is used for connecting and communicating between the electronic device and other peripherals.

[0225] exist Figure 5The system bus 305 mentioned in the figure can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used to realize communication between the database access device and other devices (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM) and may also include non-volatile memory (Non-Volatile Memory), such as at least one disk storage.

[0226] The above-mentioned processors can be general-purpose processors, including central processing units (CPUs), network processors (NPs), graphics processing units (GPUs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0227] It should be noted that an embodiment of the present invention further provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the methods and processing procedures provided in the above embodiments.

[0228] The embodiment of the present invention provides a processing method, device, electronic device and computer-readable storage medium for predicting brain performance based on multidimensional indicators. As can be seen from the above content, the embodiment of the present invention first customizes the multidimensional (self-evaluation, neurology, cognitive psychology, and traditional Chinese medicine symptoms) indicator data set for brain performance prediction, and constructs a brain performance prediction model that can predict the brain performance index based on the multidimensional indicator set; then, a model data set is constructed by conducting Chinese and Western medicine examinations and brain performance expert evaluations on a pre-recruited volunteer population, and the brain performance prediction model is trained based on the data set; after the model training is completed, the brain performance index of any subject is predicted based on the brain performance prediction model. The embodiment of the present invention improves the richness of data features and improves prediction accuracy through a multidimensional comprehensive prediction system that integrates subjective and objective evaluation systems and Chinese and Western medicine evaluation systems; improves prediction performance and prediction efficiency by using a brain performance prediction model; and improves the robustness and generalization of the model through a masking strategy set in model training.

[0229] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0230] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for predicting brain performance based on multidimensional indicators, characterized in that: The method comprises: A multidimensional indicator data set for brain performance prediction is set to obtain a corresponding multidimensional indicator set; the multidimensional indicator set is composed of four sub-data sets, namely a self-evaluation indicator set, a neurological indicator set, a cognitive psychology indicator set, and a traditional Chinese medicine syndrome indicator set; Constructing a deep learning model for predicting brain performance as a corresponding brain performance prediction model; the brain performance prediction model is used to predict brain performance based on the multidimensional indicator set input into the model and output a corresponding prediction performance index; The model data set is constructed by conducting Chinese and Western medicine examinations and brain performance expert evaluation on the pre-recruited volunteer population and recorded as the corresponding first data set; Training the brain performance prediction model based on the first data set; After the model training is completed, the multidimensional indicator set of any subject input by the user is input into the brain efficiency prediction model for prediction, and the prediction efficiency index output by this prediction is fed back to the current user.

2. The method for predicting brain performance based on multidimensional indicators according to claim 1, characterized in that: The self-evaluation indicator set is composed of multiple categories of self-evaluation indicators, including at least: a self-evaluation indicator of learning ability, a self-evaluation indicator of memory, a self-evaluation indicator of attention, a self-evaluation indicator of energy and physical strength, a self-evaluation indicator of emotional stability, a self-evaluation indicator of sleep and repair, a self-evaluation indicator of willpower, a self-evaluation indicator of control, a self-evaluation indicator of execution, a self-evaluation indicator of social vitality, a self-evaluation indicator of life vitality, a self-confidence self-evaluation indicator, and a self-awareness self-evaluation indicator; each category of self-evaluation indicator is represented by a self-evaluation score; the score range of all self-evaluation indicators is the same, which is a preset self-evaluation score threshold range; The neurological indicator set is composed of multiple types of neurological indicators, including at least: brain region volume index, brain white matter integrity index, brain cortical thickness index, functional connectivity density index, neural oscillation synchronization index, and causal interaction index; the brain region volume index is specifically the volume of the hippocampus; the white matter integrity index is specifically the fractional anisotropy score of the corpus callosum; the cortical thickness index is specifically the thickness of the prefrontal cortex; the functional connectivity density index is specifically the global connection strength of the brain's default mode network; the neural oscillation synchronization index is specifically the alpha wave rhythm phase locking value of the electroencephalogram; the causal interaction index is specifically the average strength of the causal flow strength of a preset activated brain region to another or more other preset brain regions; The cognitive psychology index set is composed of multiple categories of cognitive psychology indicators, including at least: MMSE evaluation index, MoCA evaluation index, AVLT evaluation index, WMS evaluation index, VFT evaluation index, BNT evaluation index, TMT-A evaluation index, TMT-B evaluation index, CDT evaluation index, Stroop test evaluation index, ADL evaluation index, NPI evaluation index, HAMA evaluation index, HAMD evaluation index; the MMSE evaluation index is specifically the Mini-Mental State Examination evaluation score; the MoCA evaluation index is specifically the Montreal Cognitive Assessment score; the AVLT evaluation index is specifically the Auditory Verbal Learning Test evaluation score; the WMS evaluation index is specifically the Wei The VFT evaluation index is specifically the verbal fluency test evaluation score; the BNT evaluation index is specifically the Boston Naming Test evaluation score; the TMT-A and B evaluation indexes are specifically the line drawing test A and B evaluation scores; the CDT evaluation index is specifically the clock drawing test evaluation score; the Stroop test evaluation index is specifically the Stroop test evaluation score; the ADL evaluation index is specifically the daily living ability scale evaluation score; the NPI evaluation index is specifically the neuropsychiatric questionnaire evaluation score; the HAMA evaluation index is specifically the Hamilton Anxiety Rating Scale evaluation score; the HAMD evaluation index is specifically the Hamilton Depression Rating Scale evaluation score; The TCM syndrome index set is composed of multiple categories of TCM syndrome indicators, including at least: marrow sea gradually emptying syndrome indicator, spleen and kidney deficiency syndrome indicator, qi and blood deficiency syndrome indicator, phlegm turbidity obstructing the orifices syndrome indicator, blood stasis blocking the brain meridians syndrome indicator, heart and liver fire hyperactivity syndrome indicator, and toxicity and extreme deficiency syndrome indicator; each category of syndrome indicator is a binary indicator, specifically yes or no; The first data set includes multiple first data records; the first data record includes a first training indicator set and a first label index; the first training indicator set is a multidimensional indicator set; the self-evaluation indicator set of the first training indicator set maintains data integrity and has no invalid indicators; if there are preset invalid indicators in the first training indicator set, the ratio of the total number of invalid indicators to the total number of all indicators meets the preset first masking rate, and all the invalid indicators are randomly distributed in the neurological indicator set, the cognitive psychology indicator set and the traditional Chinese medicine syndrome indicator set; the total number of invalid indicators is the total number of invalid indicators, and the total number of all indicators is the total number of indicators in the multidimensional indicator set.

3. The method for predicting brain performance based on multidimensional indicators according to claim 1, characterized in that: The model input end of the brain performance prediction model is used to receive the multi-dimensional indicator set, and the model output end is used to output the corresponding prediction performance index; The brain performance prediction model includes a preprocessing module, a self-evaluation feature coding module, a neurological feature coding module, a cognitive psychology feature coding module, a traditional Chinese medicine syndrome feature coding module, a Western medicine feature fusion module, a traditional Chinese and Western medicine feature fusion module, a subjective and objective feature fusion module, a feature mapping module and an index prediction module; The input end of the preprocessing module is connected to the input end of the model, and the first, second, third and fourth output ends are respectively connected to the input ends of the self-evaluation feature coding module, the neurological feature coding module, the cognitive psychological feature coding module and the traditional Chinese medicine syndrome feature coding module; the output ends of the neurological feature coding module and the cognitive psychological feature coding module are connected to the first and second input ends of the Western medicine feature fusion module; the output ends of the traditional Chinese medicine syndrome feature coding module and the Western medicine feature fusion module are connected to the first and second input ends of the traditional Chinese and Western medicine feature fusion module; the output ends of the self-evaluation feature coding module and the traditional Chinese and Western medicine feature fusion module are connected to the first and second input ends of the subjective and objective feature fusion module; the output end of the subjective and objective feature fusion module is connected to the input end of the feature mapping module; the output end of the feature mapping module is connected to the input end of the index prediction module; the output end of the index prediction module is connected to the output end of the model; The preprocessing module is used to use the self-evaluation index set, the neurological index set, the cognitive psychology index set and the traditional Chinese medicine syndrome index set of the multidimensional index set as the corresponding self-evaluation index vector X 1 , neurological index vector X 2 , cognitive psychological index vector X 3 and TCM syndrome index vector X 4 The self-evaluation index vector X is sent to the corresponding self-evaluation feature coding module, the neurological feature coding module, the cognitive psychology feature coding module, and the TCM syndrome feature coding module; 1 The shape is C1×1, the neurological index vector X 2 The shape is C2×1, the cognitive psychological index vector X 3 The shape is C3×1, the TCM syndrome index vector X 4 The shape is C4×1, where C1, C2, C3, and C4 are the preset first, second, third, and fourth feature dimensions; The self-evaluation feature encoding module is implemented based on the MLP model; the self-evaluation feature encoding module is used to encode the self-evaluation index vector X 1 Perform feature encoding processing to obtain the corresponding feature tensor H 1 Sending to the subjective and objective feature fusion module; The feature tensor H 1 The reasoning process is: h 11 =GELU(W 11 X 1 +b 11 ), h 12 =GELU(W 12 h 11 +b 12 ), H 1 =Reshape(h 12 ,C1,C7), C7=C6 / C1; Among them, W 11 、W 12 are the two weight matrix parameters of the self-evaluation feature encoding module, b 11 、b 12 are two offset vector parameters of the self-evaluation feature encoding module; h 11 、h 12 are the two process feature vectors of the self-evaluation feature encoding module; GELU() is the GELU activation function; the weight matrix parameter W 11 The shape is C5×C1, the offset vector parameter b 11 The shape is C5×1; the weight matrix parameter W 12 The shape is C6×C5, the offset vector parameter b 12 The shape is C6×1; the process feature vector h 11 、h 12 The shape is C5×1, C6×1; Reshape() is a vector shape reshaping function used to reshape the process feature vector h into a shape of C6×1. 12 Converted into a feature tensor H of shape C1×C7 1 ; C5, C6, C7 are the preset fifth, sixth and seventh characteristic dimensions, C1<C5<C6, C7=C6 / C1; The neurological feature encoding module is implemented based on the MLP model; the neurological feature encoding module is used to encode the neurological index vector X 2 Perform feature encoding processing to obtain the corresponding feature tensor H 2 Sending to the Western medicine feature fusion module; The feature tensor H 2 The reasoning process is: h 21 =GELU(W 21 X 2 +b 21 ), h 22 =GELU(W 22 h 21 +b 22 ), H 2 =Reshape(h 22 ,C1,C7); Among them, W 21 、W 22 are the two weight matrix parameters of the neural feature encoding module, b 21 、b 22 are two offset vector parameters of the neurological feature encoding module; h 21 、h 22 are the two process feature vectors of the neurological feature encoding module; the weight matrix parameter W 21 The shape is C8×C2, the offset vector parameter b 21 The shape is C8×1; the weight matrix parameter W 22 The shape is C9×C8, the offset vector parameter b 22 The shape is C9×1; the process feature vector h 21 、h 22 The shape of the vector is C8×1, C9×1; the vector reshape function Reshape() is used to reshape the process feature vector h into a shape of C9×1 22 Converted into a feature tensor H of shape C1×C7 2 ; C8 and C9 are the preset eighth and ninth characteristic dimensions, C2<C8<C9, C9=C1×C7; The cognitive psychological feature encoding module is implemented based on the MLP model; the cognitive psychological feature encoding module is used to encode the cognitive psychological index vector X 3 Perform feature encoding processing to obtain the corresponding feature tensor H 3 Sending to the Western medicine feature fusion module; The feature tensor H 3 The reasoning process is: h 31 =GELU(W 31 X 3 +b 31 ), h 32 =GELU(W 32 h 31 +b 32 ), H 3 =Reshape(h 32 ,C1,C7); Among them, W 31 、W 32 are the two weight matrix parameters of the cognitive psychological feature encoding module, b 31 、b 32 are two offset vector parameters of the cognitive psychological feature encoding module; h 31 、h 32 are the two process feature vectors of the cognitive psychological feature coding module; the weight matrix parameter W 31 The shape is C 10 ×C3, offset vector parameter b 31 The shape is C 10 ×1; weight matrix parameter W 32 The shape is C 11 ×C 10 , offset vector parameter b 32 The shape is C 11 ×1; process characteristic vector h 31 、h 32 The shape is C 10 ×1, C 11 ×1; the vector reshape function Reshape() is used to reshape the vector into C 11 ×1 process characteristic vector h 32 Converted into a feature tensor H of shape C1×C7 3 ; C 10 、C 11 For the preset tenth and eleventh characteristic dimensions, C3<C 10 <C 11 、C 11 =C1×C7; The TCM syndrome feature encoding module is implemented based on the MLP model; the TCM syndrome feature encoding module is used to encode the TCM syndrome index vector X 4 Perform feature encoding processing to obtain the corresponding feature tensor H 4 Sending to the Chinese and Western medicine feature fusion module; The feature tensor H 4 The reasoning process is: h 41 =GELU(W 41 X 4 +b 41 ), h 42 =GELU(W 42 h 41 +b 42 ), H 4 =Reshape(h 42 ,C4,C 16 ),C 16 =C 15 / C4=C7; Among them, W 41 、W 42 are the two weight matrix parameters of the TCM syndrome feature encoding module, b 41 、b 42 are two offset vector parameters of the TCM syndrome feature encoding module; 41 、h 42 are the two process feature vectors of the TCM syndrome feature encoding module; the weight matrix parameter W 41 The shape is C 12 ×C4, offset vector parameter b 41 The shape is C 12 ×1; weight matrix parameter W 42 The shape is C 13 ×C 12 , offset vector parameter b 42 The shape is C 13 ×1; process characteristic vector h 41 、h 42 The shape is C 12 ×1, C 13 ×1; the vector reshape function Reshape() is used to reshape the vector into C 13 ×1 process characteristic vector h 42 Converted into a feature tensor H of shape C1×C7 4 ; C 12 、C 13 For the preset twelfth and thirteenth characteristic dimensions, C4<C 12 <C 13 、C 13 =C1×C7; The Western medicine feature fusion module is used to 2 Generate the corresponding query tensor Q, based on the feature tensor H 3 Generate corresponding key and value tensors K and V; and perform attention feature extraction based on the query, key and value tensors Q, K and V to obtain the corresponding attention tensor A; and extract the feature tensor H 2 Perform residual connection with the attention tensor A to obtain the corresponding feature tensor H 5 Sending to the Chinese and Western medicine feature fusion module; The feature tensor H 5 The reasoning process is: Q=H 2 W Q ,K=H 3 W K ,V=H 3 W V , H 5 =H 2 +A; Among them, W Q 、W K 、W V are the three weight matrix parameters of the Western medicine feature fusion module; the weight matrix parameter W Q 、W K 、W V The shape is C7×C qkv , the preset feature dimension C qkv =C7; the query tensor Q, the key tensor K, the value tensor V, the attention tensor A and the feature tensor H 5 The shape of is C1×C7; Softmax() is the Softmax function; The Chinese and Western medicine feature fusion module is used to combine the feature tensor H 4 and the feature tensor H 5 Perform feature concatenation to obtain a shape of C1×C 14 The characteristic tensor H 6 , C 14 is the preset fourteenth characteristic dimension, C 14 =2×C7; and by the feature tensor H 6 Perform a full connection operation to obtain a gated weight tensor G with a shape of C1×1; and form a gated weight tensor G with a shape of C1×C7 by C7 of the gated weight tensors G ’ ; and based on the gated weight tensor G ’ , the feature tensor H 4 and the feature tensor H 5 Perform feature fusion processing to obtain the corresponding feature tensor H 7 Sending to the subjective and objective feature fusion module; The feature tensor H 7 The reasoning process is: G=Sigmoid(H 6 W G +b G ), H 7 =G′⊙H 4 +(I-G′)⊙H 5 ; Among them, W G 、b G The weight matrix parameters and offset vector parameters of the Chinese and Western medicine feature fusion module are as follows: G The shape is C 14 ×1, offset vector parameter b G The shape of is C1×1; Sigmoid() is the Sigmoid activation function; the shape of the gated weight tensor G is C1×1; I is a preset all-1 tensor with a shape of C1×C7; ⊙ is the Hadamard product; the feature tensor H 7 The shape of all is C1×C7; The subjective and objective feature fusion module is used to fusion the feature tensor H 1 and the feature tensor H 7 Perform feature fusion processing to obtain the corresponding feature tensor H 8 Sending to the feature mapping module; The feature tensor H 8 The reasoning process is: H 8 =H 1 +H 7 ; The feature tensor H 8 The shape is C1×C7; The feature mapping module is used to use a 1×1 convolution kernel to map the feature tensor H 8 Perform feature dimensionality reduction to obtain a shape of C1×C 15 The characteristic tensor H 9 , C 15 is the preset fifteenth characteristic dimension, C1<C 15 <C7; and the characteristic tensor H 9 Perform global maximum pooling along the indicator dimension to obtain a shape of 1×C 15 The eigenvector H 10 ; and the eigenvector H 10 Perform full connection operation to obtain the corresponding feature scalar H 11 Sending to the index prediction module; The characteristic scalar H 11 The reasoning process is: H 11 =H 10 W 51 +b 51 ; Among them, W 51 、b 51 The weight matrix parameter and offset vector parameter of the feature mapping module; the weight matrix parameter W 51 The shape is C 15 ×C 16 , offset vector parameter b G The shape is 1×C 16 ; C 16 is the preset sixteenth characteristic dimension, C 16 =1; The index prediction module is used to calculate the weight scalar W according to the preset 61 , offset scalar b 61 For the characteristic scalar H 11 Perform linear scaling to obtain the corresponding prediction performance index and output it; prediction performance index = W 61 ×H 11 +b 61 .

4. The method for predicting brain performance based on multidimensional indicators according to claim 2, characterized in that: The model dataset constructed by performing Chinese and Western medicine examinations and brain performance expert evaluation on the pre-recruited volunteer population is recorded as the corresponding first dataset, specifically including: Taking each volunteer in the volunteer group as a corresponding current volunteer; And all the self-assessment indicators obtained by the current volunteer through self-assessment according to the preset brain efficiency self-assessment scale and the corresponding scale self-assessment rules constitute a corresponding self-assessment indicator set; the brain efficiency self-assessment scale includes learning ability self-assessment items, memory self-assessment items, attention self-assessment items, energy and physical strength self-assessment items, emotional stability self-assessment items, sleep repair ability self-assessment items, willpower self-assessment items, control ability self-assessment items, execution ability self-assessment items, social vitality self-assessment items, life vitality self-assessment items, self-confidence self-assessment items, and self-awareness self-assessment items; the self-assessment items of the brain efficiency self-assessment scale correspond one to one with the self-assessment indicators of the self-assessment indicator set; The self-assessment rules of the scale include self-assessment rules for learning ability, memory, attention, energy and physical strength, emotional stability, sleep and repair ability, willpower, control, execution, social vitality, life vitality, self-confidence, and self-awareness; the self-assessment rules of the scale correspond one-to-one to the self-assessment items of the brain efficiency self-assessment scale; each self-assessment rule of the scale is used to set the score of the current self-assessment item in a graded manner, and the scoring rule corresponding to each grade is explained; and medical experts from a preset first expert group perform resting-state brain structural magnetic resonance imaging, diffusion tensor imaging, and functional magnetic resonance imaging on the current volunteer to obtain corresponding sMRI images, DTI images, and fMRI images; and perform hippocampal brain region segmentation and volume measurement based on the sMRI images based on FreeSurfer, FSL, or SPM analysis tools, and use the obtained hippocampal volume as the corresponding brain region volume index; and perform cerebral cortical surface reconstruction based on the sMRI images based on FreeSurfer or CAT12 analysis tools, and measure the prefrontal cortex thickness on the reconstructed cerebral cortical surface, and use the obtained prefrontal cortex thickness as the corresponding cerebral cortical thickness index; and calculate the partial anisotropy score of the corpus callosum based on the DTI image based on the FSL analysis tool, and use the calculated score as the corresponding brain white matter integrity index; and analyze the global connectivity strength of the brain's default mode network based on the fMRI images based on the CONN or GRETNA analysis tool, and use the analysis result as the corresponding functional connectivity density index; the first expert group is a medical imaging expert group; A second group of medical experts, comprising experts in neuroelectrophysiology and neuroscience, performs an electroencephalogram (EEG) on the current volunteer to obtain a corresponding first EEG graph; and performs an alpha wave rhythm phase-locking value analysis based on the first EEG graph using EEGLAB, FieldTri p, or Chronux analysis tools, and uses the analysis result as the corresponding neural oscillation synchronization indicator. Medical experts from a preset third expert group perform transcranial magnetic stimulation on the activated brain area of ​​the current volunteer, and simultaneously perform electroencephalogram (EEG) examinations on the activated brain area of ​​the current volunteer and one or more other brain areas to obtain corresponding second EEG graphs and third EEG atlases; and according to the Granger causality analysis method, analyze the causal flow strength of the activated brain area of ​​the current volunteer and each of the other brain areas based on the second EEG graph and the third EEG atlas, calculate the mean of all the obtained causal flow strengths, and use the calculation result as the corresponding causal interaction indicator; the second EEG graph corresponds to the activated brain area; the third EEG atlas consists of one or more third EEG graphs, and the third EEG graphs correspond one-to-one with the other brain areas; the third expert group is a neuroelectrophysiology and neuroscience expert group; and having medical experts from a preset fourth expert group evaluate all cognitive psychological indicators of the cognitive psychological indicator set of the current volunteer; the fourth expert group is an expert group in neuropsychology, behavioral neurology, psychiatry, and neurology; and having medical experts from a preset fifth expert group evaluate all cognitive psychological indicators of the TCM syndrome index set of the current volunteer; the fifth expert group is an expert group in the field of TCM; A joint expert group consisting of the first, second, third, fourth and fifth expert groups assesses and scores the current brain performance status of the volunteer and uses the obtained assessment score as the corresponding first label index; The brain area volume index, the white matter integrity index, the cerebral cortical thickness index, the functional connectivity density index, the neural oscillation synchronization index, and the causal interaction index corresponding to the current volunteer form a corresponding neurological index set; and the self-evaluation index set, the neurological index set, the cognitive psychology index set, and the traditional Chinese medicine syndrome index set corresponding to the current volunteer form a corresponding first training index set; and performing a random replacement process on the indicators of the neurological indicator set, the cognitive psychology indicator set, and the TCM syndrome indicator set of the current first training indicator set based on the first masking rate and the invalid indicator to obtain a derived first training indicator set, and repeating the random replacement process N times to obtain N derived first training indicator sets; N is a preset positive integer; Each first training indicator set and the corresponding first tag index corresponding to the current volunteer form a corresponding first data record; and N+1 first data records corresponding to the current volunteer form a corresponding first data record set; And all the first data record sets corresponding to all the volunteers in the volunteer group form the corresponding first data set.

5. The method for predicting brain performance based on multidimensional indicators according to claim 2, characterized in that: The training of the brain performance prediction model based on the first data set specifically includes: Step 51: Split the first data set into two sub-data sets based on a preset first split ratio and record them as a corresponding first training set and a first evaluation set; Wherein, both the first training set and the first evaluation set are composed of a plurality of the first data records; the ratio of the total number of records in the first training set and the first evaluation set satisfies the first segmentation ratio; Step 52: Perform a round of traversal on all the first data records of the first training set; and during this round of traversal, use the first data record currently traversed as the corresponding current training record; and input the first training indicator set of the current training record as the current multidimensional indicator set into the brain performance prediction model for prediction, and use the prediction efficiency index output by this prediction as the corresponding first prediction index; and form a corresponding first prediction-label pair from the first prediction index and the first label index of the current training record; and at the end of this round of traversal, bring all the first prediction-label pairs obtained in this round of traversal into the preset first model loss function to calculate and obtain the corresponding first loss value; Wherein, the first model loss function is implemented based on the L1 loss function or the L2 loss function; Step 53: Identify whether the first loss value satisfies a preset first loss value range; if so, proceed to step 54; if not, perform a round of modulation on the model parameters of the brain performance prediction model in a direction that minimizes the first model loss function based on a preset first model optimizer, and return to step 52 to continue training at the end of this round of modulation; Wherein, the first model optimizer includes at least an Adam optimizer and an SGD optimizer; Step 54: perform a round of traversal on all the first data records of the first evaluation set; and during this round of traversal, use the first data record currently traversed as the corresponding current evaluation record; and input the first training indicator set of the current evaluation record as the current multidimensional indicator set into the brain performance prediction model for prediction, and use the prediction efficiency index output by this prediction as the corresponding second prediction index; and form a corresponding second prediction-label pair from the second prediction index and the first label index of the current evaluation record; and at the end of this round of traversal, bring all the second prediction-label pairs obtained in this round of traversal into the preset first model evaluation function to calculate and obtain the corresponding first evaluation value; Wherein, the first model evaluation function is implemented based on MAE function, MSE function or RMSE function; Step 55 , identifying whether the first evaluation value meets a preset first evaluation value range; if not, returning to step 52 to continue training; if so, confirming that the model training is completed.

6. A device for executing the method for predicting brain performance based on multidimensional indicators according to any one of claims 1 to 5, characterized in that: The device includes: a multidimensional indicator system construction module, a model construction module, a data set construction module, a model training module and a model prediction module; The multidimensional indicator system construction module is used to set the multidimensional indicator data set for brain performance prediction to obtain a corresponding multidimensional indicator set; the multidimensional indicator set is composed of sub-data sets of four dimensions, namely a self-evaluation indicator set, a neurological indicator set, a cognitive psychology indicator set, and a traditional Chinese medicine syndrome indicator set; The model building module is used to build a deep learning model for predicting brain performance as a corresponding brain performance prediction model; the brain performance prediction model is used to predict brain performance based on the multidimensional indicator set input by the model and output a corresponding prediction performance index; The data set construction module is used to construct a model data set by performing Chinese and Western medicine examinations and brain performance expert evaluation on a pre-recruited volunteer group, which is recorded as the corresponding first data set; The model training module is used to train the brain performance prediction model based on the first data set; The model prediction module is used to input the multidimensional indicator set of any subject input by the user into the brain performance prediction model for prediction after the model training is completed, and to feed back the prediction performance index output by this prediction to the current user.

7. An electronic device, characterized in that: include: memory, processors, and transceivers; The processor is configured to be coupled to the memory, read and execute instructions in the memory, so as to implement the method according to any one of claims 1 to 5; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a computer, the computer is caused to execute the method according to any one of claims 1 to 5.

Images