Traditional Chinese medicine syndrome human body magnetic induction signal acquisition and recognition system and method

Abstract

The invention discloses a traditional Chinese medicine syndrome human body magnetic induction signal acquisition and identification system and method, magnetic induction signals at different parts of a human body are collected to serve as identification signal sources, and compared with a traditional technology, the signal sources are numerous, and information contained in the signals is rich; meanwhile, a magnetic induction signal is introduced, higher signal acquisition stability can be provided, and the influence of individual difference on data analysis is reduced through the learning ability of the neural network, so that the stability and consistency of data acquisition are ensured; besides, through the neural network model, the complex nonlinear mapping relation between the virtual and real evidence and the magnetic induction signal can be better captured, so that a more accurate signal classification result can be provided, and a more accurate auxiliary basis can be provided for medical staff; therefore, the magnetic induction signals are combined with neural network modeling to classify the virtual and actual syndrome characteristics of the magnetic induction signals, and an accurate auxiliary basis can be provided for traditional Chinese medicine virtual and actual syndrome classification.

Description

Technical Field

[0001] The present invention belongs to the technical field of magnetic induction signal processing, and in particular relates to a system and method for acquiring and identifying human body magnetic induction signals of traditional Chinese medicine syndromes. Background Art

[0002] Traditional Chinese medicine relies on the experience of Chinese medicine practitioners to judge the patient's pathological properties through the combination of observation, auscultation, inquiry and palpation. With the development of modern science and technology, researchers have gradually tried to combine modern technology with Chinese medicine theory and developed a variety of auxiliary diagnosis systems and equipment. The following are several typical existing technologies:

[0003] Pulse detector, tongue analysis system, and TCM constitution identification instrument. Among them, the pulse detector collects the patient's pulse signal, uses sensors to extract relevant parameters of the pulse (such as waveform, strength, rhythm, etc.), and uses algorithms to analyze the patient's deficiency or excess syndrome; the pulse detector aims to standardize the pulse diagnosis process and reduce the influence of subjective factors; the tongue analysis system uses image processing technology to analyze the tongue image, and judges the patient's deficiency or excess syndrome through tongue color, thickness of tongue coating, cracks and other characteristics. This technology improves the objectivity of tongue diagnosis through computer image processing; and the TCM constitution identification instrument judges the patient's deficiency or excess syndrome type based on the patient's physical signs, medical interview information, etc. through data collection and algorithm analysis. This type of equipment combines medical interview and physical examination information, and attempts to improve the standardization of TCM syndrome differentiation through data-based methods.

[0004] At the same time, with the development of biomedical signal processing technology, more and more biological signals (such as electroencephalogram, electrocardiogram, electromyogram, skin conductance, etc.) are used in medical diagnosis and health monitoring. These signals are collected through sensors and analyzed using computer algorithms. They aim to reflect the internal health status of the human body in a non-invasive way. Therefore, feature recognition technologies based on biological signals have also been derived, such as electroencephalogram (EEG) recognition and electrocardiogram (ECG) analysis.

[0005] However, the above-mentioned prior art has the following shortcomings:

[0006] (1) Single signal source and insufficient information dimension: The pathological changes of deficiency and excess syndromes in traditional Chinese medicine are complex, and a single pulse, tongue or biological signal is difficult to fully reflect the patient's deficiency or excess state.

[0007] (2) Poor stability and consistency of data acquisition: Existing TCM diagnostic instruments are easily interfered by external factors during the data acquisition process, making it difficult to ensure signal stability.

[0008] (3) The algorithm is not complex enough and it is difficult to handle nonlinear relationships: The algorithms in the existing technology are mostly simple statistical analysis or elementary machine learning algorithms, which are difficult to handle nonlinear complex relationships such as virtual and real evidence.

[0009] Therefore, based on the aforementioned deficiencies, how to provide an auxiliary system that uses magnetic induction signals with sufficient information and high stability to classify signals and derive signal categories, so as to assist medical staff in accurately classifying virtual and real evidence, has become a problem that needs to be solved urgently. Summary of the invention

[0010] The purpose of the present invention is to provide a system and method for acquiring and identifying human magnetic induction signals of TCM syndromes, so as to solve the problems of existing auxiliary systems, such as single signal source, insufficient information dimension, poor stability and consistency of data acquisition, and difficulty in handling nonlinear relationships.

[0011] In order to achieve the above object, the present invention adopts the following technical solutions:

[0012] In the first aspect, a system for acquiring and identifying human magnetic induction signals of TCM syndromes is provided, comprising:

[0013] A magnetic induction signal acquisition device is used to acquire magnetic induction signals at different parts of the target human body, obtain the target human body magnetic induction signal, and send the target human body magnetic induction signal to the medical workbench;

[0014] A medical workbench, electrically connected to the magnetic induction signal acquisition device, for receiving a target human body magnetic induction signal transmitted by the magnetic induction signal acquisition device, and performing data preprocessing on the target human body magnetic induction signal to obtain a preprocessed magnetic induction signal;

[0015] The medical workbench is further used to input the pre-processed magnetic induction signal into a signal classification model, obtain the signal category of the target human body magnetic induction signal, and visualize the signal category, wherein the signal category is that the target human body magnetic induction signal belongs to a deficiency-type magnetic induction signal, or the target human body magnetic induction signal belongs to an excess-type magnetic induction signal, and the signal classification model adopts a neural network model;

[0016] The signal classification model comprises: a first feature extraction layer, a second feature extraction layer, a pooling layer and a classification layer;

[0017] The first feature extraction layer is used to perform a first feature extraction process on the pre-processed magnetic induction signal to obtain first feature information, and input the first feature information into the second feature extraction layer;

[0018] A second feature extraction layer is used to perform a second feature extraction process on the first feature information to obtain second feature information, wherein the dimension of the second feature information is higher than the dimension of the first feature information;

[0019] A pooling layer, used for performing maximum pooling processing on the second feature information to obtain third feature information;

[0020] The classification layer is used to use an activation function to classify the third feature information to obtain the classification probability of the third feature information, so as to derive the signal category of the target human body magnetic induction signal based on the classification probability.

[0021] Based on the above disclosed content, the acquisition and identification system of human body magnetic induction signals of TCM syndromes provided by the present invention collects magnetic induction signals at different parts of the target human body through magnetic induction signal acquisition equipment, and transmits the magnetic induction signals as signal sources to the medical workbench, so that the medical workbench can classify the signals based on this, thereby assisting medical staff in making a virtual and real diagnosis; wherein, the medical workbench is pre-set with a signal classification model based on a neural network, therefore, in actual use, the magnetic induction signal of the target human body is input into the signal classification model, and the signal category of the target human body magnetic induction signal can be obtained; finally, the signal category is visualized, which can assist medical staff in making a virtual and real diagnosis.

[0022] Through the above design, the present invention collects magnetic induction signals at different parts of the human body as the identification signal source. Compared with traditional technologies, the signal sources are numerous and the signals contain rich information, thus making up for the deficiency of the single signal source in the prior art; at the same time, the present invention introduces magnetic induction signals, which can provide higher signal acquisition stability, and through the learning ability of neural networks, reduces the influence of individual differences on data analysis, thereby ensuring the stability and consistency of data acquisition; in addition, through the neural network model, it can better capture the complex nonlinear mapping relationship between virtual and real syndromes and magnetic induction signals, thereby providing more accurate signal classification results, thereby providing more accurate auxiliary basis for medical staff; therefore, the present invention combines magnetic induction signals with neural network modeling to perform virtual and real syndrome feature classification of magnetic induction signals, filling this technical gap, and can provide accurate auxiliary basis for the classification of virtual and real syndromes in traditional Chinese medicine, therefore, it is very suitable for large-scale application and promotion.

[0023] In a possible design, the signal classification model is trained in the following manner:

[0024] A medical workbench, used for obtaining an initial training set, wherein the initial training set includes sample magnetic induction signals of multiple sample human bodies, each sample magnetic induction signal corresponds to a signal category, the signal category includes magnetic induction signals of deficiency syndrome and magnetic induction signals of excess syndrome, and the sample magnetic induction signals of any sample human body include magnetic induction signals of different parts of the any sample human body;

[0025] A medical workbench, used to perform data enhancement processing on each sample magnetic induction signal in the initial training set to obtain an enhanced initial training set;

[0026] A medical workbench, used for performing data preprocessing on the enhanced initial training set to obtain a preprocessed initial training set;

[0027] The medical workbench is used to divide the preprocessed initial training set into training set, validation set and test set;

[0028] A medical workbench, used for training an initial signal classification model by taking each piece of training data in the training set as input and the signal category corresponding to each piece of training data as output, to obtain a pre-trained signal classification model;

[0029] A medical workbench, used to perform model verification processing on the pre-trained signal classification model using the verification set and a cross-validation method, so as to optimize the pre-trained signal classification model based on the model verification result to obtain an optimized pre-trained signal classification model;

[0030] The medical workbench is also used to use the test set to perform model testing on the optimized pre-trained signal classification model, so as to obtain the signal classification model that meets the accuracy of the classification task after the model test passes.

[0031] In a possible design, the medical workstation is used to perform time reversal processing on the initial training set to obtain a first enhanced training set;

[0032] A medical workbench, used for performing noise processing on the first enhanced training set to obtain a second enhanced training set;

[0033] The medical workbench is also used to use the initial training set and the second enhanced training set to form the enhanced initial training set.

[0034] In a possible design, a medical workbench is used to train the initial signal classification model based on an Adam optimizer and adopt a learning rate decay strategy and an early stopping strategy;

[0035] Among them, the medical workbench is used to adopt a regularization strategy, and use a grid search or a random search method to optimize the pre-trained signal classification model to obtain the optimized pre-trained signal classification model.

[0036] In a possible design, the first feature extraction layer includes three first convolutional layers, wherein the number of convolution kernels in each first convolutional layer is 64, the size of the convolution kernel is 5, and the activation function used in each first convolutional layer is the ReLU function.

[0037] In a possible design, the second feature extraction layer includes three second convolutional layers, wherein the number of convolution kernels in each second convolutional layer is 128, the convolution kernel size is 5, and the activation function used in each second convolutional layer is the ReLU function.

[0038] In a possible design, the classification layer includes a fully connected layer, and the activation function corresponding to the fully connected layer is a Softmax function.

[0039] In a possible design, the medical workbench is used to extract the time domain, frequency domain or time-frequency domain features of the target human body magnetic induction signal, so as to use the time domain, frequency domain or time-frequency domain features as the pre-processed magnetic induction signal.

[0040] In a second aspect, a working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes in the first aspect or any possible design of the first aspect is provided, comprising:

[0041] The magnetic induction signal acquisition device acquires magnetic induction signals at different parts of the target human body, obtains the target human body magnetic induction signal, and sends the target human body magnetic induction signal to the medical workbench;

[0042] The medical workbench receives the target human body magnetic induction signal transmitted by the magnetic induction signal acquisition device, and performs data preprocessing on the target human body magnetic induction signal to obtain a preprocessed magnetic induction signal;

[0043] The medical workbench inputs the pre-processed magnetic induction signal into a signal classification model, obtains the signal category of the target human body magnetic induction signal, and visualizes the signal category, wherein the signal category is that the target human body magnetic induction signal belongs to a deficiency-type magnetic induction signal, or the target human body magnetic induction signal belongs to an excess-type magnetic induction signal, and the signal classification model adopts a neural network model;

[0044] The signal classification model comprises: a first feature extraction layer, a second feature extraction layer, a pooling layer and a classification layer;

[0045] The first feature extraction layer is used to perform a first feature extraction process on the pre-processed magnetic induction signal to obtain first feature information, and input the first feature information into the second feature extraction layer;

[0046] A second feature extraction layer is used to perform a second feature extraction process on the first feature information to obtain second feature information, wherein the dimension of the second feature information is higher than the dimension of the first feature information;

[0047] A pooling layer, used for performing maximum pooling processing on the second feature information to obtain third feature information;

[0048] The classification layer is used to use an activation function to classify the third feature information to obtain the classification probability of the third feature information, so as to derive the signal category of the target human body magnetic induction signal based on the classification probability.

[0049] In a third aspect, an electronic device is provided, comprising a memory, a processor and a transceiver which are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes as described in the second aspect.

[0050] In a fourth aspect, a storage medium is provided, on which instructions are stored. When the instructions are run on a computer, the working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes as described in the second aspect is executed.

[0051] In a fifth aspect, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute the working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes as described in the second aspect.

[0052] Beneficial effects:

[0053] (1) The present invention collects magnetic induction signals from different parts of the human body as the identification signal source. Compared with traditional technologies, the present invention has many signal sources and rich signals with rich information, thus making up for the deficiency of single signal source in the prior art. At the same time, the present invention introduces magnetic induction signals to provide higher signal acquisition stability, and reduces the influence of individual differences on data analysis through the learning ability of neural networks, thereby ensuring the stability and consistency of data acquisition. In addition, the complex nonlinear mapping relationship between deficiency and excess syndrome and magnetic induction signals can be better captured through the neural network model, thereby providing more accurate signal classification results, and thus providing more accurate auxiliary basis for medical staff. Therefore, the present invention combines magnetic induction signals with neural network modeling to classify the deficiency and excess syndrome characteristics of magnetic induction signals, thus filling this technical gap and providing accurate auxiliary basis for the classification of deficiency and excess syndrome in traditional Chinese medicine. Therefore, it is very suitable for large-scale application and promotion.

[0054] (2) By training and optimizing a large amount of data through deep learning technology, the neural network model provided by the present invention can extract effective features from complex magnetic induction signals, thereby achieving high-precision signal classification; at the same time, after rigorous verification and optimization, the model shows a high classification accuracy and has good robustness to different individual differences, thus ensuring its wide applicability in different clinical scenarios.

[0055] (3) The present invention uses a non-invasive magnetic induction signal acquisition device, which will not cause any harm or discomfort to the patient's body and has extremely high safety. At the same time, the non-invasive feature also enables the technology to be widely used in patients of different age groups and different health conditions, with a wide range of applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A schematic diagram of the architecture of a system for acquiring and identifying human body magnetic induction signals of TCM syndromes provided by an embodiment of the present invention;

[0057] Figure 2 A schematic diagram of the structure of a signal classification model provided by an embodiment of the present invention;

[0058] Figure 3 A flowchart of the steps of the working method of the system for acquiring and identifying the human body magnetic induction signals of TCM syndromes provided by an embodiment of the present invention;

[0059] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0061] It should be understood that although the terms first, second, etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another unit. For example, a first unit can be referred to as a second unit, and similarly, a second unit can be referred to as a first unit without departing from the scope of the exemplary embodiments of the present invention.

[0062] It should be understood that the term "and / or" that may appear in this article is merely a description of the association relationship between associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, B exists alone, and A and B exist at the same time. The term " / and" that may appear in this article describes another type of association object relationship, indicating that two relationships may exist. For example, A / and B can represent two situations: A exists alone, and A and B exist alone. In addition, the character " / " that may appear in this article generally indicates that the previous and next associated objects are in an "or" relationship.

[0063] Example:

[0064] See also Figure 1 As shown, the system for acquiring and identifying the human body magnetic induction signals of TCM syndromes provided in this embodiment may include, but is not limited to, a magnetic induction signal acquisition device and a medical workbench, wherein the magnetic induction signal acquisition device is used to perform non-invasive whole-body magnetic field signal acquisition on the patient, that is, to acquire magnetic induction signals at different parts of the target human body, obtain the target human body magnetic induction signal, and send the target human body magnetic induction signal to the medical workbench.

[0065] The medical workbench is electrically connected to the magnetic induction signal acquisition device, and is used to receive the target human body magnetic induction signal transmitted by the magnetic induction signal acquisition device, and perform data preprocessing on the target human body magnetic induction signal to obtain the preprocessed magnetic induction signal; wherein, in this embodiment, the medical workbench can be, but is not limited to, a computer or a server, and at the same time, the aforementioned data preprocessing is mainly to extract the time domain, frequency domain or time-frequency domain characteristics of the target human body magnetic induction signal (such as the signal's average value, standard deviation, spectral energy, etc.), so as to use the time domain, frequency domain or time-frequency domain characteristics as the preprocessed magnetic induction signal.

[0066] Furthermore, for example, a pre-trained signal classification model is deployed in the medical workbench. Therefore, in actual use, the aforementioned target human body magnetic induction signal can be directly input into the signal classification model to obtain the corresponding signal category; finally, a visual display of it can assist medical staff in classifying virtual and real evidence.

[0067] Optionally, the aforementioned signal classification model is a neural network model (such as a CNN network), see Figure 2 As shown, it mainly includes a first feature extraction layer, a second feature extraction layer, a pooling layer and a classification layer; further, the aforementioned first feature extraction layer and the second feature extraction layer mainly include convolutional layers, and the classification layer is a fully connected layer.

[0068] Among them, the following discloses one training process of the signal classification model.

[0069] In specific applications, the signal classification model uses the medical workbench as a carrier to implement model training, namely:

[0070] A medical workbench is used to obtain an initial training set, wherein the initial training set contains sample magnetic induction signals of multiple sample human bodies, each sample magnetic induction signal corresponds to a signal category, the signal category includes magnetic induction signals of deficiency syndrome and magnetic induction signals of excess syndrome, and the sample magnetic induction signals of any sample human body include magnetic induction signals on different parts of the any sample human body; in specific implementation, the initial training set can be obtained through an expert system, that is, for any sample human body, an expert system is used to classify deficiency syndrome and excess syndrome, that is, three deputy chief Chinese medicine doctors with more than 10 years of work experience classify the deficiency syndrome and excess syndrome of any sample human body respectively, and each doctor independently judges the disease and syndrome attributes of any sample human body. When the judgments of the three experts are consistent, the case of any sample human body is included in the next step of data collection process.

[0071] Among them, when the three doctors' classification results of deficiency and excess syndrome for any sample human body are consistent, the case of any sample human body is determined to be a consistent case, that is, the three doctors upload their own deficiency and excess syndrome classification results to the expert system respectively, and when the expert system determines that the three results are consistent, the case of any sample human body is marked as a consistent case; in this way, this step ensures that the data used for modeling has high accuracy and consistency, and reduces the errors caused by subjectivity in traditional Chinese medicine diagnosis.

[0072] Next, the human body magnetic induction signal can be collected for any sample human body corresponding to the consistent case, that is, the magnetic induction signals on different parts of any sample human body can be collected using the aforementioned magnetic induction signal acquisition equipment, so as to obtain the sample magnetic induction signal of any sample human body; at the same time, the consistent case results in the expert system (that is, the classification results of the false and real syndromes) are used as the signal category (that is, label data) of the sample magnetic induction signal. Based on this, the aforementioned method can be used to construct an initial training set for model training.

[0073] After obtaining the initial training set, in order to increase the diversity of training data, this embodiment also performs data enhancement processing on the initial training set, that is: a medical workbench is used to perform data enhancement processing on each sample magnetic induction signal in the initial training set to obtain an enhanced initial training set; wherein, the data enhancement processing may include but is not limited to: time reversal and noise addition processing; specifically, the initial training set is firstly subjected to time reversal processing to obtain a first enhanced training set; then, the first enhanced training set is subjected to noise addition processing to obtain a second enhanced training set; finally, the initial training set and the second enhanced training set can be used to form the enhanced initial training set.

[0074] In specific implementation, time reversal of the initial training set is to time-flip the magnetic induction signals of each sample, that is, if a signal is sampled from time point 0 to N-1 (where N is the length of the signal), then time axis flipping is to exchange the sample at time point 0 with the sample at time point N-1, the sample at time point 1 with the sample at time point N-2, and so on, until the sample in the middle is exchanged; in this way, different signal acquisition sequences can be simulated; and signal noise addition is to add random noise to each data in the first enhanced training set, thereby enhancing the robustness of the model to noise.

[0075] Therefore, after the aforementioned data enhancement processing, the diversity of training data is effectively increased and the generalization ability of the model is improved.

[0076] After obtaining the enhanced initial training set, data preprocessing can be performed to extract corresponding time domain, frequency domain or time-frequency domain features, that is: the medical workbench is used to perform data preprocessing on the enhanced initial training set to obtain a preprocessed initial training set; wherein, the data preprocessing process of the enhanced initial training set is the same as the data preprocessing process of the aforementioned target human body magnetic induction signal, and will not be repeated here.

[0077] After obtaining the preprocessed initial training set, the initial signal classification model (i.e., the untrained signal classification model) can be trained, wherein the specific training process is as follows:

[0078] The medical workbench is used to divide the preprocessed initial training set into a training set, a validation set and a test set; in this embodiment, the data ratios of the aforementioned training set, validation set and test set are 70%, 15% and 15% as examples; of course, the specific ratios can be set according to actual use and are not limited to the aforementioned examples.

[0079] After completing the division of the training set, the validation set, and the test set, the training set can be used to train the initial classification model, that is, the medical workbench is used to train the initial signal classification model with each training data in the training set as input and the signal category corresponding to each training data as output to obtain a pre-trained signal classification model; in this embodiment, supervised learning is used to optimize the model weights and parameters during the training process.

[0080] In which, during the training process, after any training data (which can be expressed as (n, c), n represents the length of any training data, and c represents signals in different parts or different components of multidimensional signals) is input into the initial classification model, it is first processed by the first feature extraction layer in the model to perform a first feature extraction process to obtain the first sample feature information corresponding to the any training data; in this embodiment, the first feature extraction layer extracts the local time features of the any training data, that is, the convolution kernel sliding window moves along the time axis to extract the local signal pattern.

[0081] Then, the extracted first sample feature information is input into the second feature extraction layer for feature re-extraction, that is, the second feature extraction layer is used to perform second feature extraction processing on the first sample feature information to obtain second sample feature information. In this way, the second feature extraction layer still uses a convolutional layer to extract more complex signal features (that is, the dimension of the second sample feature information is higher than the dimension of the first sample feature information).

[0082] Next, the pooling layer is used to perform maximum pooling processing on the second sample feature information to reduce the feature dimension while retaining the most important features; in this way, the third sample feature information can be obtained.

[0083] Finally, the classification layer is used to use the corresponding activation function (such as Softmax) to classify the third sample feature information, and obtain the classification probability of the third sample feature information, so as to derive the signal category of any training data based on the classification probability; wherein, the classification layer is the output layer, and the number of neurons is 2, representing two categories of "deficiency syndrome signal" and "excess syndrome signal", and the activation function is used to output the probability distribution of each category; therefore, the category corresponding to the maximum value can be taken as the final classification result.

[0084] Optionally, for example, during the training process, this embodiment is based on the Adam optimizer, and adopts a learning rate decay strategy and an early stopping strategy to train the initial signal classification model; wherein, the initial learning rate during the training process can be set to 0.001, and the learning rate decay strategy is used during the training process, and the learning rate can be gradually reduced during the model convergence process, thereby achieving the purpose of improving the convergence speed and stability of the model.

[0085] At the same time, the specific implementation method of the early stopping strategy is: use the validation set to verify the model after each training cycle, and monitor the loss function of the validation set. When the loss on the validation set no longer decreases, stop training in advance; after many experiments, it was found that after 50 training cycles, the accuracy and loss of the model on the validation set tended to be stable, so the training cycle was finally set to 50 times; in this way, overfitting of the model can be prevented.

[0086] Furthermore, for example, the loss function of the initial signal classification model may be, but is not limited to, a cross entropy loss function to measure the difference between the model output and the actual label.

[0087] Thus, after completing the training of the initial signal classification model, in order to ensure the model training effect, this embodiment also uses a validation set to evaluate the performance of the model; wherein the model validation process is:

[0088] The medical workbench is used to use the verification set and the cross-validation method to perform model verification processing on the pre-trained signal classification model, so as to optimize the pre-trained signal classification model based on the model verification result to obtain the optimized pre-trained signal classification model; in specific applications, model verification and optimization are key steps to ensure that the signal classification model can accurately classify magnetic induction signals. By verifying and optimizing the trained model, the generalization ability of the model can be improved to ensure that its performance on different patient data is stable and reliable.

[0089] Among them, examples can include but are not limited to using a regularization strategy, and using a grid search or random search method to optimize the pre-trained signal classification model to obtain an optimized pre-trained signal classification model; at the same time, a five-fold cross-validation method is used to evaluate the model to avoid overfitting of the model and ensure that the model performs consistently on different data partitions.

[0090] Specifically, the following hyperparameters of the model were optimized through grid search or random search:

[0091] Convolution kernel size: We tried different convolution kernel sizes (3, 5, 7), and finally selected a convolution kernel of size 5, which performed best in extracting local temporal features.

[0092] The number of convolutional layers is increased from the initial two convolutional layers to three convolutional layers, which further enriches the feature extraction capability. The three convolutional layers can capture more complex signal patterns.

[0093] Learning rate: By dynamically adjusting the learning rate, the initial learning rate is finally determined to be 0.001. During the training process, the learning rate decay strategy is used to gradually reduce the learning rate during the model convergence process, thereby improving the convergence speed and stability of the model.

[0094] The regularization strategies include:

[0095] Dropout: Dropout is introduced after the fully connected layer (i.e., classification layer), randomly discarding 50% of the neurons to enhance the robustness of the model.

[0096] L2 regularization: L2 regularization is added to the second feature extraction layer and the fully connected layer to limit the size of the weight and reduce the risk of overfitting.

[0097] In this way, after the above model optimization, the optimized pre-trained signal classification model can be obtained; then, the test set can be used to finally evaluate the classification effect of the model. The process is as follows:

[0098] The medical workbench is also used to use the test set to perform model testing on the optimized pre-trained signal classification model, so as to obtain the signal classification model that meets the accuracy of the classification task after the model test passes.

[0099] In this embodiment, a test set is used to test the model, and accuracy (Accuracy, the proportion of correct model predictions); recall (Reca ll, the proportion of correctly identifying deficiency syndrome or excess syndrome); precision (Precision, the accuracy of the model in classifying deficiency syndrome or excess syndrome signals); F1 value (F1Score, the harmonic mean of precision and recall, a comprehensive measure of the classification performance of the model) are used to comprehensively evaluate the classification effect of the optimized pre-trained signal classification model, and a confusion matrix is ​​used to visualize the performance of the model in the classification of deficiency syndrome and excess syndrome signals.

[0100] Among them, the final evaluation results of the optimized pre-trained signal classification model are as follows:

[0101] Accuracy: 94.5%.

[0102] Recall rate: 94.1% for deficiency syndrome and 96.8% for excess syndrome.

[0103] Precision: 96.3% for deficiency syndrome and 94.5% for excess syndrome.

[0104] F1 value (F1Score): 95.2% for deficiency syndrome and 95.6% for excess syndrome.

[0105] These results show that the model performs well in the classification task of virtual and real evidence signals, with an accuracy rate of nearly 95%. At the same time, the F1 value shows that the model has a good balance when dealing with virtual and real evidence categories, and the classification effect of virtual and real evidence signals is relatively ideal.

[0106] In this way, when the optimized pre-trained signal classification model passes the model test, it can be used as the signal classification model that meets the accuracy of the classification task.

[0107] Based on this, through the above model training, this embodiment obtains the optimal model structure of the signal classification model, namely:

[0108] See also Figure 2 As shown, the first feature extraction layer includes three first convolutional layers, wherein the number of convolution kernels in each first convolutional layer is 64, the size of the convolution kernel is 5, and the activation function used in each first convolutional layer is the ReLU function; similarly, the second feature extraction layer includes three second convolutional layers, the number of convolution kernels in each second convolutional layer is 128, the size of the convolution kernel is 5, and the activation function used in each second convolutional layer is also the ReLU function; further, the pooling window of the pooling layer is set to 2×2, and the classification layer includes a fully connected layer, and the activation function corresponding to the fully connected layer is the Softmax function.

[0109] Through the above explanation, the training of the signal classification model can be completed, and then the signal classification model can be deployed on a medical workbench to classify and process the human body magnetic induction signal of the target human body.

[0110] The classification process is:

[0111] The medical workbench is also used to input the preprocessed magnetic induction signal into a signal classification model to obtain the signal category of the target human body magnetic induction signal, and visualize the signal category, wherein the signal category is that the target human body magnetic induction signal belongs to a deficiency-type magnetic induction signal, or the target human body magnetic induction signal belongs to an excess-type magnetic induction signal.

[0112] Of course, the processing process of the pre-processed magnetic induction signal in the signal classification model is the same as any of the aforementioned training data, that is, firstly, the first feature extraction process is performed on the first feature extraction layer to obtain the first feature information, and the first feature information is input into the second feature extraction layer; and the second feature extraction layer is used to perform the second feature extraction process on the first feature information to obtain the second feature information (the dimension of the second feature information is higher than the dimension of the first feature information); then, the pooling layer is used to perform the maximum pooling process on the second feature information to obtain the third feature information; finally, the fully connected layer can be used to output the classification result, that is, the classification layer is used to use the activation function to classify the third feature information to obtain the classification probability of the third feature information, so as to derive the signal category of the target human body magnetic induction signal based on the classification probability.

[0113] Based on this, after obtaining the signal category of the target human body's human magnetic induction signal, it is visualized and displayed, which can provide accurate auxiliary basis for medical staff to classify the virtual and real syndromes.

[0114] Therefore, through the above detailed description of the acquisition and recognition system of human body magnetic induction signals of TCM syndromes, the present invention has the following advantages.

[0115] Beneficial effects:

[0116] (1) The classification results of human magnetic induction signals are used to assist medical staff in classifying the deficiency and excess syndromes. In this way, biophysical signals are combined with traditional Chinese medicine theory for the first time, providing a new technical approach for the objectification and dataization of deficiency and excess syndromes. This innovation enables traditional Chinese medicine to transform from subjective sensory judgment to analysis based on objective biological signals.

[0117] (2) The present invention introduces deep learning models such as neural network models to extract features and classify magnetic induction signals. Among them, neural network models are good at extracting local features from complex signals and can identify key characteristic patterns of deficiency and excess syndrome signals. In this way, by training a large number of patient magnetic induction signal samples, the model can automatically classify deficiency and excess syndrome signals, thereby greatly reducing the subjectivity of human judgment; this deep learning-based automated classification method is a core innovation to assist in the classification of deficiency and excess syndromes in traditional Chinese medicine.

[0118] (3) The present invention adopts a non-invasive magnetic induction signal acquisition device, which can non-invasively and safely acquire magnetic field signals from multiple parts of the human body. This technology not only ensures the comfort of the patient, but also can obtain valuable biophysical signals without interfering with the patient's body. Therefore, this non-invasive signal acquisition method is suitable for patients of various physical conditions, ages and health conditions, and has broad application prospects.

[0119] like Figure 3 As shown, the second aspect of this embodiment provides a working method of the system for acquiring and identifying the human body magnetic induction signal of TCM syndrome described in the first aspect of the embodiment, comprising:

[0120] S1. The magnetic induction signal acquisition device acquires magnetic induction signals at different parts of the target human body, obtains the target human body magnetic induction signal, and sends the target human body magnetic induction signal to the medical workbench.

[0121] S2. The medical workbench receives the target human body magnetic induction signal transmitted by the magnetic induction signal acquisition device, and performs data preprocessing on the target human body magnetic induction signal to obtain a preprocessed magnetic induction signal.

[0122] S3. The medical workstation inputs the preprocessed magnetic induction signal into the signal classification model to obtain the signal category of the target human body magnetic induction signal, and visualizes the signal category, wherein the signal category is that the target human body magnetic induction signal belongs to the deficiency syndrome magnetic induction signal, or the target human body magnetic induction signal belongs to the excess syndrome magnetic induction signal, and the signal classification model adopts a neural network model.

[0123] The signal classification model includes: a first feature extraction layer, a second feature extraction layer, a pooling layer and a classification layer.

[0124] The first feature extraction layer is used to perform a first feature extraction process on the preprocessed magnetic induction signal to obtain first feature information, and input the first feature information into the second feature extraction layer.

[0125] The second feature extraction layer is used to perform second feature extraction processing on the first feature information to obtain second feature information, wherein the dimension of the second feature information is higher than the dimension of the first feature information.

[0126] The pooling layer is used to perform maximum pooling processing on the second feature information to obtain third feature information.

[0127] The classification layer is used to use an activation function to classify the third feature information to obtain the classification probability of the third feature information, so as to derive the signal category of the target human body magnetic induction signal based on the classification probability.

[0128] The working process, working details and technical effects of the working method of the system provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0129] like Figure 4 As shown, the third aspect of this embodiment provides an electronic device, comprising: a memory, a processor and a transceiver that are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes as described in the second aspect of the embodiment.

[0130] For example, the memory may include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out memory (FIFO) and / or first-in-last-out memory (FILO), etc. Specifically, the processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). At the same time, the processor may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.

[0131] In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. For example, the processor may not be limited to a microprocessor of the STM32F105 series, a reduced instruction set computer (RISC) microprocessor, an X86 or other architecture processor, or a processor with an integrated embedded neural network processor (NPU); the transceiver may be, but is not limited to, a wireless fidelity (WIFI) wireless transceiver, a Bluetooth wireless transceiver, a general packet radio service technology (Genera l Packet Rad ioService, GPRS) wireless transceiver, a ZigBee protocol (a low-power LAN protocol based on the IEEE 802.15.4 standard, ZigBee) wireless transceiver, a 3G transceiver, a 4G transceiver and / or a 5G transceiver, etc. In addition, the device may also include, but is not limited to, a power module, a display screen and other necessary components.

[0132] The working process, working details and technical effects of the electronic device provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0133] The fourth aspect of this embodiment provides a storage medium that stores instructions for the working method of the system for acquiring and identifying human body magnetic induction signals of Traditional Chinese Medicine syndromes as described in the second aspect of the embodiment, that is, the storage medium stores instructions, and when the instructions are run on a computer, the working method of the system for acquiring and identifying human body magnetic induction signals of Traditional Chinese Medicine syndromes as described in the second aspect of the embodiment is executed.

[0134] The storage medium refers to a carrier for storing data, which may include but is not limited to a floppy disk, a CD, a hard disk, a flash memory, a USB flash drive and / or a memory stick, etc. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0135] The working process, working details and technical effects of the storage medium provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0136] The fifth aspect of this embodiment provides a computer program product containing instructions. When the instructions are executed on a computer, the computer is caused to execute the working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes as described in the second aspect of the embodiment, wherein the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0137] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A system for acquiring and identifying human magnetic induction signals of TCM syndromes, characterized in that: include: A magnetic induction signal acquisition device is used to acquire magnetic induction signals at different parts of the target human body, obtain the target human body magnetic induction signal, and send the target human body magnetic induction signal to the medical workbench; A medical workbench, electrically connected to the magnetic induction signal acquisition device, for receiving a target human body magnetic induction signal transmitted by the magnetic induction signal acquisition device, and performing data preprocessing on the target human body magnetic induction signal to obtain a preprocessed magnetic induction signal; The medical workbench is further used to input the pre-processed magnetic induction signal into a signal classification model, obtain the signal category of the target human body magnetic induction signal, and visualize the signal category, wherein the signal category is that the target human body magnetic induction signal belongs to a deficiency-type magnetic induction signal, or the target human body magnetic induction signal belongs to an excess-type magnetic induction signal, and the signal classification model adopts a neural network model; The signal classification model comprises: a first feature extraction layer, a second feature extraction layer, a pooling layer and a classification layer; The first feature extraction layer is used to perform a first feature extraction process on the pre-processed magnetic induction signal to obtain first feature information, and input the first feature information into the second feature extraction layer; A second feature extraction layer is used to perform a second feature extraction process on the first feature information to obtain second feature information, wherein the dimension of the second feature information is higher than the dimension of the first feature information; A pooling layer, used for performing maximum pooling processing on the second feature information to obtain third feature information; The classification layer is used to use an activation function to classify the third feature information to obtain the classification probability of the third feature information, so as to derive the signal category of the target human body magnetic induction signal based on the classification probability.

2. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 1, characterized in that: The signal classification model is trained in the following way: A medical workbench, used for obtaining an initial training set, wherein the initial training set includes sample magnetic induction signals of multiple sample human bodies, each sample magnetic induction signal corresponds to a signal category, the signal category includes magnetic induction signals of deficiency syndrome and magnetic induction signals of excess syndrome, and the sample magnetic induction signals of any sample human body include magnetic induction signals of different parts of the any sample human body; A medical workbench, used to perform data enhancement processing on each sample magnetic induction signal in the initial training set to obtain an enhanced initial training set; A medical workbench, used for performing data preprocessing on the enhanced initial training set to obtain a preprocessed initial training set; The medical workbench is used to divide the preprocessed initial training set into training set, validation set and test set; A medical workbench, used for training an initial signal classification model by taking each piece of training data in the training set as input and the signal category corresponding to each piece of training data as output, to obtain a pre-trained signal classification model; A medical workbench, used to perform model verification processing on the pre-trained signal classification model using the verification set and a cross-validation method, so as to optimize the pre-trained signal classification model based on the model verification result to obtain an optimized pre-trained signal classification model; The medical workbench is also used to use the test set to perform model testing on the optimized pre-trained signal classification model, so as to obtain the signal classification model that meets the accuracy of the classification task after the model test passes.

3. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 2, characterized in that: A medical workbench, used for performing time reversal processing on the initial training set to obtain a first enhanced training set; A medical workbench, used for performing noise processing on the first enhanced training set to obtain a second enhanced training set; The medical workbench is also used to use the initial training set and the second enhanced training set to form the enhanced initial training set.

4. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 2, characterized in that: A medical workbench, used for training the initial signal classification model based on an Adam optimizer and adopting a learning rate decay strategy and an early stopping strategy; Among them, the medical workbench is used to adopt a regularization strategy, and use a grid search or a random search method to optimize the pre-trained signal classification model to obtain the optimized pre-trained signal classification model.

5. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 1, characterized in that: The first feature extraction layer includes three first convolutional layers, wherein the number of convolution kernels of each first convolutional layer is 64, the size of the convolution kernel is 5, and the activation function used in each first convolutional layer is the ReLU function.

6. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 1, characterized in that: The second feature extraction layer includes three second convolutional layers, wherein the number of convolution kernels in each second convolutional layer is 128, the size of the convolution kernel is 5, and the activation function used in each second convolutional layer is the ReLU function.

7. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 1, characterized in that: The classification layer includes a fully connected layer, and the activation function corresponding to the fully connected layer is a Softmax function.

8. A system for acquiring and identifying human body magnetic induction signals of TCM syndromes according to claim 1, characterized in that: The medical workbench is used to extract the time domain, frequency domain or time-frequency domain features of the target human body magnetic induction signal, so as to use the time domain, frequency domain or time-frequency domain features as the pre-processed magnetic induction signal.

9. A working method of the system for acquiring and identifying the human body magnetic induction signals of TCM syndromes according to any one of claims 1 to 8, characterized in that: include: The magnetic induction signal acquisition device acquires magnetic induction signals at different parts of the target human body, obtains the target human body magnetic induction signal, and sends the target human body magnetic induction signal to the medical workbench; The medical workbench receives the target human body magnetic induction signal transmitted by the magnetic induction signal acquisition device, and performs data preprocessing on the target human body magnetic induction signal to obtain a preprocessed magnetic induction signal; The medical workbench inputs the pre-processed magnetic induction signal into a signal classification model, obtains the signal category of the target human body magnetic induction signal, and visualizes the signal category, wherein the signal category is that the target human body magnetic induction signal belongs to a deficiency-type magnetic induction signal, or the target human body magnetic induction signal belongs to an excess-type magnetic induction signal, and the signal classification model adopts a neural network model; The signal classification model comprises: a first feature extraction layer, a second feature extraction layer, a pooling layer and a classification layer; The first feature extraction layer is used to perform a first feature extraction process on the pre-processed magnetic induction signal to obtain first feature information, and input the first feature information into the second feature extraction layer; A second feature extraction layer is used to perform a second feature extraction process on the first feature information to obtain second feature information, wherein the dimension of the second feature information is higher than the dimension of the first feature information; A pooling layer, used for performing maximum pooling processing on the second feature information to obtain third feature information; The classification layer is used to use an activation function to classify the third feature information to obtain the classification probability of the third feature information, so as to derive the signal category of the target human body magnetic induction signal based on the classification probability.

10. An electronic device, characterized in that: include: A memory, a processor and a transceiver which are sequentially connected in communication, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program, and execute the working method of the system for acquiring and identifying human body magnetic induction signals of TCM syndromes as described in claim 9.

Images