A method, system, device, and medium for classifying a neurotransmitter
By employing dynamic weight fusion and multi-model collaboration, the problem of poor accuracy in neurotransmitter classification was solved, and the accuracy of neurotransmitter classification was improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUN YAT SEN UNIV
- Filing Date
- 2026-02-13
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies struggle to accurately distinguish between different neurotransmitters, resulting in poor accuracy in neurotransmitter classification.
The predicted classification vector is fused by combining dynamic first classification weights with second classification weights. The time series and local features of neurotransmitter electrochemical curves are extracted through time series extraction module and dimensionality increase module. The feature distribution is learned using Naive Bayes model. The predicted classification is fused by combining the outputs of feature classification model and sequence classification model.
It improves the accuracy of neurotransmitter classification, makes full use of the advantages of different models, ensures that the sequence classification model can play its role in dynamic data processing and integrate rich feature information.
Smart Images
Figure CN122153639A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of machine learning technology, and in particular to a method, system, device, and medium for classifying neurotransmitters. Background Technology
[0002] Neurotransmitters are crucial molecules responsible for signal transmission in the central nervous system. Their concentrations are extremely low and fluctuate dramatically, playing a decisive role in neuronal activity, brain function regulation, and the pathogenesis of various neuropsychiatric disorders. Clinical and basic research has shown that a range of neurological diseases, such as Parkinson's disease, depression, and epilepsy, exhibit abnormal expression of specific neurotransmitters. Accurate detection and classification of neurotransmitter levels are essential scientific foundations for early disease screening, disease progression monitoring, and personalized medication.
[0003] Currently, related technologies are usually based on electrochemical detection techniques to quantitatively detect and classify neurotransmitters. However, due to the wide variety of neurotransmitters and their significant structural differences, this method often fails to accurately distinguish between different neurotransmitters, resulting in poor accuracy in neurotransmitter classification.
[0004] Therefore, the problems with the relevant technologies still need to be solved and optimized. Summary of the Invention
[0005] The purpose of this invention is to at least partially solve one of the technical problems existing in the related art.
[0006] Therefore, one objective of this invention is to provide a method, system, device, and medium for classifying neurotransmitters, wherein the method can effectively improve the accuracy of neurotransmitter classification.
[0007] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of this application include: In a first aspect, embodiments of this application provide a method for classifying neurotransmitters, including: Obtain target feature data of the target neurotransmitter, as well as a first classification weight corresponding to the trained feature classification model and a second classification weight corresponding to the trained sequence classification model, wherein the first classification weight is less than the second classification weight; the first classification weight is positively correlated with the number of feature types contained in the target feature data; The target feature data is input into the trained feature classification model for prediction and classification to obtain a first predicted classification vector. The target feature data is input into the trained sequence classification model for prediction and classification to obtain a second predicted classification vector. Based on the first classification weight and the first predicted classification vector, the second classification weight and the second predicted classification vector are fused to obtain the classification result of the target neurotransmitter.
[0008] In addition, the method according to the above embodiments of this application may also have the following additional technical features: Furthermore, in one embodiment of this application, the trained feature classification model is obtained through the following steps: Obtain neurotransmitter training data for the target object and construct a Naive Bayes model; the neurotransmitter training data includes several neurotransmitter category groups with different feature types; Based on the Naive Bayes model, feature distribution learning processing is performed on the neurotransmitter training data to obtain feature distribution data for each neurotransmitter category group; Based on the feature distribution data and the Naive Bayes model, the trained feature classification model is obtained.
[0009] Furthermore, in one embodiment of this application, obtaining the neurotransmitter training data of the target object includes: Obtain the electrochemical curve data of the target object; Multidimensional feature extraction is performed on the electrochemical curve data to obtain intermediate feature data; The electrochemical curve data is subjected to feature dimensionality reduction to obtain dimensionality-reduced feature data; The neurotransmitter training data is obtained based on the intermediate feature data and the dimensionality reduction feature data.
[0010] Furthermore, in one embodiment of this application, the sequence classification model includes a time-series extraction module, a dimensionality-upgrading module, and a dimensionality-reducing classification module. The step of inputting the target feature data into the trained sequence classification model for prediction and classification to obtain a second predicted classification vector includes: The target feature data is extracted temporally using the temporal extraction module to obtain temporal information features. The time-series information features are upgraded and fused using the upgrade module to obtain upgraded and fused features. The second predicted classification vector is obtained by performing dimensionality reduction classification on the dimensionality-upgraded fusion features through the dimensionality reduction classification module.
[0011] Furthermore, in one embodiment of this application, the step of performing time-series extraction on the target feature data through the time-series extraction module to obtain time-series information features includes: The target feature data is processed by a first temporal network to obtain intermediate temporal features; The intermediate temporal features are processed by a second temporal network to obtain the temporal information features. The first time-series network and the second time-series network have the same network structure.
[0012] Furthermore, in one embodiment of this application, the dimensionality-upgrading module includes several cascaded residual dimensionality-upgrading networks. The step of performing dimensionality-upgrading fusion on the time-series information features through the dimensionality-upgrading module to obtain dimensionality-upgraded fused features includes: The time-series information features are input into several cascaded residual upscaling networks for step-by-step upscaling and fusion, resulting in the upscaling fusion features output by the last residual upscaling network. The residual upscaling network is used to perform the following steps: Obtain the input features of the current feature level, wherein the input features are the time-series information features or the output features of the previous feature level; The input features are convolved and increased in dimensionality to obtain convolved and increased dimensionality features; Based on the input features, feature fusion is performed on the convolutional up-dimensional features to obtain the output features of the current feature level.
[0013] Further, in this embodiment of the application, the step of performing dimensionality reduction classification on the dimensionality-upgraded fusion features through the dimensionality reduction classification module to obtain the second predicted classification vector includes: The dimensionality-upgraded fusion features are then subjected to pooling dimensionality reduction to obtain pooling dimensionality-reduced features; The pooled dimensionality reduction features are then used for prediction and classification to obtain the second prediction and classification vector.
[0014] Secondly, embodiments of this application provide a neurotransmitter classification system, including: The first processing unit is used to acquire target feature data of the target neurotransmitter, as well as a first classification weight corresponding to the trained feature classification model and a second classification weight corresponding to the trained sequence classification model, wherein the first classification weight is less than the second classification weight; the first classification weight is positively correlated with the number of feature types contained in the target feature data. The second processing unit is used to input the target feature data into the trained feature classification model for prediction and classification, and obtain the first prediction classification vector. The third processing unit is used to input the target feature data into the trained sequence classification model for prediction and classification, and obtain the second prediction classification vector. The fourth processing unit is used to perform predictive classification fusion on the second classification weight and the second predicted classification vector based on the first classification weight and the first predicted classification vector to obtain the classification result of the target neurotransmitter.
[0015] Thirdly, embodiments of this application also provide an electronic device, including: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.
[0016] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a processor-executable program, which, when executed by the processor, is used to implement the above-described method.
[0017] Fifthly, embodiments of this application also provide a computer program product, which includes a computer program stored in a computer-readable storage medium. A processor of an electronic device reads the computer program from the computer-readable storage medium and executes the computer program, causing the electronic device to perform the method described above.
[0018] The advantages and beneficial effects of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application: This application discloses a method, system, device, and medium for classifying neurotransmitters. The method acquires target feature data of a target neurotransmitter, a first classification weight corresponding to a trained feature classification model, and a second classification weight corresponding to a trained sequence classification model. The first classification weight is less than the second classification weight. The first classification weight is positively correlated with the number of feature types contained in the target feature data. The target feature data is input into the trained feature classification model for prediction classification to obtain a first predicted classification vector. The target feature data is then input into the trained sequence classification model for prediction classification to obtain a second predicted classification vector. Based on the first classification weight and the first predicted classification vector, the second classification weight and the second predicted classification vector are fused to obtain the classification result of the target neurotransmitter. This method effectively improves the accuracy of neurotransmitter classification by dynamically fusing the predicted classification vector using the first classification weight and the second classification weight. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of this application or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions in this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0020] Figure 1 A flowchart illustrating a method for classifying neurotransmitters provided in an embodiment of this application; Figure 2 A schematic diagram of the framework of a neurotransmitter classification system provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0021] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.
[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.
[0023] The technical terms used in the embodiments of this application are explained below: Neurotransmitters are chemical substances that transmit information between neurons or between neurons and effector cells (such as muscle cells and glandular cells). Based on their chemical composition, neurotransmitters mainly include cholinergic neurotransmitters (such as acetylcholine), monoamine neurotransmitters (such as norepinephrine, dopamine, and serotonin), amino acids (excitatory neurotransmitters such as glutamate and aspartate; inhibitory neurotransmitters such as gamma-aminobutyric acid, glycine, and taurine), and neuropeptides.
[0024] Currently, relevant technologies typically rely on electrochemical detection techniques to quantitatively detect and classify neurotransmitters. Specifically, this involves collecting and recording parameters such as current and potential during the electrochemical reactions of neurotransmitters to generate electrochemical curves. A single model is then used to process these curves to classify neurotransmitters. However, due to the vast number of neurotransmitters and their significant structural differences, a single model struggles to effectively handle complex electrochemical curves and cannot comprehensively capture key features. Consequently, this approach often fails to accurately distinguish between different neurotransmitters, resulting in poor accuracy in neurotransmitter classification.
[0025] It should be noted that the aforementioned related technologies are only used to assist in understanding the technical solutions of this application and do not mean that they belong to the publicly disclosed prior art.
[0026] In view of this, embodiments of this application provide a method, system, device, and medium for classifying neurotransmitters. The method fuses predicted classification vectors using a dynamic first classification weight combined with a second classification weight, which can effectively improve the accuracy of neurotransmitter classification. Specifically: This method performs temporal extraction and feature fusion on target feature data through a temporal extraction module and a dimensionality enhancement module. The temporal extraction module (e.g., LSTM network) effectively captures the time-series features (i.e., temporal information features) of neurotransmitter electrochemical curves, while the dimensionality enhancement module (e.g., convolutional neural network) extracts local curve features. Through the synergistic effect of the temporal extraction and dimensionality enhancement modules, it can fully extract the temporal variation information of complex electrochemical curves and uncover potential patterns. The method classifies the target feature data using a feature classification model trained on a Naive Bayes model. This model can describe the electrochemical curve features from multiple perspectives (i.e., describe the target feature data from multiple feature types), providing rich information for prediction classification. Based on the principle that the second classification weight is greater than the first classification weight, this method fuses the predicted classification vectors output by the two models through classification weights. This ensures that the dynamic data processing capabilities of the sequence classification model are fully utilized during the prediction classification process, while simultaneously fusing the rich information in the predicted classification vectors of the feature classification model. This fully integrates the advantages of the feature classification model and the sequence classification model, effectively improving the classification accuracy of neurotransmitters.
[0027] Furthermore, experimental data on neurotransmitter classification shows that the classification accuracy of the feature classification model is positively correlated with the number of input feature types, meaning the output reliability of the feature classification model is positively correlated with the number of input feature types. This method, based on a first classification weight positively correlated with the number of feature types in the target feature data, dynamically provides classification weights to the first predicted classification vector output by the feature classification model based on the input target feature data. This allows the importance of the first predicted classification vector to the classification result to change with the positive correlation of the target feature data's feature types. Combined with subsequent predictive classification fusion, it can suppress the impact on the obtained classification result when the reliability of the first predicted classification vector is low; conversely, it can increase the impact on the obtained classification result when the reliability of the first predicted classification vector is high, thereby improving the classification accuracy of neurotransmitters.
[0028] Reference Figure 1 In this application embodiment, a method for classifying neurotransmitters includes: Step 110: Obtain the target feature data of the target neurotransmitter, as well as the first classification weight corresponding to the trained feature classification model and the second classification weight corresponding to the trained sequence classification model. The first classification weight is less than the second classification weight. The first classification weight is positively correlated with the number of feature types contained in the target feature data. In this embodiment, the target neurotransmitter can be a neurotransmitter to be classified, and its target feature data can be electrochemical curve data of the target neurotransmitter collected based on electrochemical detection technology. The first classification weight is used to characterize the importance of the predicted classification vector output by the corresponding feature classification model in the final classification result; while the content of the second classification weight is similar to that of the first classification weight and can be easily deduced by analogy.
[0029] It is understandable that the feature types in the target feature data can be peak width features, half-maximum width features, peak height features, peak potential features (such as peak potential features and valley potential features) in electrochemical curve data. The specific value of the first classification weight can be determined based on experimental data. For example, it can be determined based on the statistical positive correlation between the number of feature types in the target feature data and the classification accuracy of the feature classification model in the experimental data, under the constraint that the first classification weight is less than the second classification weight.
[0030] For ease of understanding, this application embodiment uses an example where the number of feature types of the target feature data is less than or equal to 3. The content where the number of feature types of the target feature data is greater than or equal to 4 can be simply deduced from the example of this application. For example, in the first embodiment, if the number of feature types of the target feature data is 1, the specific value of the first classification weight can be 0.3, and the specific value of the second classification weight can be 0.7; or, if the number of feature types of the target feature data is 2, the specific value of the first classification weight can be 0.4, and the specific value of the second classification weight can be 0.7; or, if the number of feature types of the target feature data is 3, the specific value of the first classification weight can be 0.5, and the specific value of the second classification weight can be 0.7.
[0031] In the second embodiment, if the number of feature types of the target feature data is 1, the specific value of the first classification weight can be 0.2, and the specific value of the second classification weight can be 0.8; or, if the number of feature types of the target feature data is 2, the specific value of the first classification weight can be 0.3, and the specific value of the second classification weight can be 0.7; or, if the number of feature types of the target feature data is 3, the specific value of the first classification weight can be 0.4, and the specific value of the second classification weight can be 0.6.
[0032] Step 120: Input the target feature data into the trained feature classification model for prediction and classification to obtain the first predicted classification vector; In this embodiment, target feature data can be input into a trained feature classification model. The trained feature classification model can then predict and classify the feature information of each feature type in the target feature data, thereby obtaining the first predicted classification vector output by the trained feature classification model.
[0033] In some embodiments, the trained feature classification model is obtained through the following steps: Obtain neurotransmitter training data for the target object and construct a Naive Bayes model; the neurotransmitter training data includes several neurotransmitter category groups with different feature types; Furthermore, acquiring the neurotransmitter training data of the target object includes: Obtain the electrochemical curve data of the target object; Multidimensional feature extraction is performed on the electrochemical curve data to obtain intermediate feature data; The electrochemical curve data is subjected to feature dimensionality reduction to obtain dimensionality-reduced feature data; The neurotransmitter training data is obtained based on the intermediate feature data and the dimensionality reduction feature data.
[0034] In this embodiment, the Naive Bayes model can specifically be a Gaussian Naive Bayes model; the electrochemical curve data of the target object can be the electrochemical curve data of the target neurotransmitter, or the electrochemical curve data of the neurotransmitter used to train the Naive Bayes model. Multidimensional feature extraction can involve performing preprocessing operations such as normalization on the voltage-current data in the electrochemical curve data, extracting and integrating several feature types from the preprocessed electrochemical curve data to obtain intermediate feature data in vector form. For example, peak features, valley features, peak slope features, peak width features, and peak shape features can be extracted.
[0035] Specifically, peak and valley characteristics can be obtained by using the `scipy.signal.find_peaks` function in the Scipy software tool to detect the peaks and estimated values in the preprocessed electrochemical curve data; peak slope characteristics can be obtained by calculating the slope of each peak in the preprocessed electrochemical curve data, which can be defined as the absolute value of the difference between the peak current and the adjacent valley current divided by the corresponding voltage difference, reflecting the steepness of the peak; peak width characteristics can be obtained by calculating the width between the valleys on both sides of the peak; and peak shape characteristics can be obtained by calculating the vertical distance from the peak top to the valley bottom and measuring the peak shape using geometric methods.
[0036] Understandably, feature reduction can be performed on preprocessed electrochemical curve data. Specifically, it can be achieved by using Principal Component Analysis (PCA) to reduce the dimensionality of the preprocessed electrochemical curve data, resulting in vector-based feature data. For neurotransmitter training data, this can be achieved by concatenating the intermediate feature data and the reduced feature data.
[0037] Based on the Naive Bayes model, feature distribution learning processing is performed on the neurotransmitter training data to obtain feature distribution data for each neurotransmitter category group; Based on the feature distribution data and the Naive Bayes model, the trained feature classification model is obtained.
[0038] In this embodiment, the feature distribution learning process can involve inputting neurotransmitter training data into a Naive Bayes model. The Naive Bayes model then analyzes the probability distribution parameters of all data features (i.e., neurotransmitter category groups) under each feature type in the neurotransmitter training data. For example, it analyzes the probability distribution parameters (such as mean and variance) of peak features in the neurotransmitter training data, thereby obtaining the feature distribution data (i.e., probability distribution parameters) of the neurotransmitter category groups corresponding to all peak features. For a trained feature classification model, the trained feature classification model can be easily obtained based on the Naive Bayes model and the feature distribution data of each neurotransmitter category group determined by the Naive Bayes model.
[0039] Step 130: Input the target feature data into the trained sequence classification model for prediction and classification to obtain a second predicted classification vector; In this embodiment of the application, target feature data can be input into a trained sequence classification model, and the trained sequence classification model can be used to predict and classify the target feature data, thereby obtaining a second predicted classification vector output by the trained sequence classification model.
[0040] In some embodiments, the sequence classification model includes a time-series extraction module, a dimensionality-upgrading module, and a dimensionality-reducing classification module. The step of inputting the target feature data into the trained sequence classification model for prediction and classification to obtain a second predicted classification vector includes: The target feature data is extracted temporally using the temporal extraction module to obtain temporal information features. Further, the step of performing time-series extraction on the target feature data through the time-series extraction module to obtain time-series information features includes: The target feature data is processed by a first temporal network to obtain intermediate temporal features; The intermediate temporal features are processed by a second temporal network to obtain the temporal information features. In this embodiment, the temporal extraction module may include a cascaded first temporal network and a second temporal network, wherein both the first and second temporal networks may be Long Short-Term Memory (LSTM) networks. Temporal extraction may involve inputting target feature data into the temporal extraction module, and then extracting the core information features from this vector sequence of target feature data to obtain temporal information features.
[0041] Specifically, the first time-series processing may involve inputting the target feature data into a first time-series network to obtain the intermediate time-series features output by the first time-series network; while the second time-series processing may involve inputting the intermediate time-series features output by the first time-series network into a second time-series network to obtain the time-series information features output by the second time-series network.
[0042] The time-series information features are upgraded and fused using the upgrade module to obtain upgraded and fused features. Furthermore, the dimensionality-upgrading module includes several cascaded residual dimensionality-upgrading networks. The step of performing dimensionality-upgrading fusion on the time-series information features through the dimensionality-upgrading module to obtain dimensionality-upgraded fused features includes: The time-series information features are input into several cascaded residual upscaling networks for step-by-step upscaling and fusion, resulting in the upscaling fusion features output by the last residual upscaling network. The residual upscaling network is used to perform the following steps: Obtain the input features of the current feature level, wherein the input features are the time-series information features or the output features of the previous feature level; The input features are convolved and increased in dimensionality to obtain convolved and increased dimensionality features; Based on the input features, feature fusion is performed on the convolutional up-dimensional features to obtain the output features of the current feature level.
[0043] In this embodiment, the dimensionality-upgrading module includes several cascaded residual dimensionality-upgrading networks, each of which can be a convolutional neural network (CNN) containing residual structures. For ease of understanding, this embodiment uses two cascaded residual dimensionality-upgrading networks as an example, and the content when the number of cascaded residual dimensionality-upgrading networks is greater than or equal to 3 can be easily deduced by analogy.
[0044] Specifically, for the first residual dimensionality-upgrading network in two cascaded residual dimensionality-upgrading networks, the input feature of the current feature level can be a temporal information feature, that is, the input feature of the first feature level; convolutional dimensionality-upgrading can be achieved by inputting the temporal information feature into the convolutional neural network, and then upgrading the feature dimension of the temporal information feature through the convolutional neural network to obtain the convolutional dimensionality-upgrading feature; while feature fusion can be based on the residual structure, connecting the temporal information feature residual to the convolutional dimensionality-upgrading feature output by the convolutional neural network, and then fusing the temporal information feature and the convolutional dimensionality-upgrading feature through a concat operation to obtain the output feature of the first feature level.
[0045] For the second residual dimensionality-upgrading network in the cascaded two residual dimensionality-upgrading networks, the input feature of the current feature level can be the output feature of the first feature level; convolutional dimensionality-upgrading can be achieved by inputting the output feature of the first feature level into the convolutional neural network, and then using the convolutional neural network to upgrade the feature dimension of the output feature of the first feature level to obtain the convolutional dimensionality-upgrading feature; while feature fusion can be based on the residual structure, connecting the temporal information feature residual to the convolutional dimensionality-upgrading feature output by the convolutional neural network, and then fusing the temporal information feature and the convolutional dimensionality-upgrading feature through a concat operation to obtain the output feature of the second feature level, and then determining the output feature of the second feature level as the dimensionality-upgrading fusion feature.
[0046] The second predicted classification vector is obtained by performing dimensionality reduction classification on the dimensionality-upgraded fusion features through the dimensionality reduction classification module.
[0047] Further, the step of performing dimensionality reduction classification on the dimensionality-upgraded fusion features through the dimensionality reduction classification module to obtain the second predicted classification vector includes: The dimensionality-upgraded fusion features are then subjected to pooling dimensionality reduction to obtain pooling dimensionality-reduced features; The pooled dimensionality reduction features are then used for prediction and classification to obtain the second prediction and classification vector.
[0048] In this embodiment, a Global Average Pooling (GAP) operation can be performed on the dimensionality-upgraded fusion features to compress this high-dimensional data into one dimension, thereby achieving dimensionality reduction through pooling and obtaining pooled dimensionality-reduced features. The prediction classification can then be performed by inputting the pooled dimensionality-reduced features into a Multilayer Perceptron (MLP), which predicts the pooled dimensionality-reduced features to obtain a second predicted classification vector output by the sequence classification model.
[0049] Step 140: Based on the first classification weight and the first predicted classification vector, perform prediction classification fusion on the second classification weight and the second predicted classification vector to obtain the classification result of the target neurotransmitter.
[0050] In this embodiment, the classification result of the target neurotransmitter can be obtained by fusing the first classification weight, the first predicted classification vector, the second classification weight, and the second predicted classification vector through a weighted summation method. Specifically, the classification result can be expressed as:
[0051] in, The classification results; This is the first predicted classification vector; The first category weight; The weight for the second category; This is the second predicted classification vector.
[0052] Reference Figure 2 The neurotransmitter classification system proposed in this application includes: The first processing unit 101 is used to acquire target feature data of the target neurotransmitter, as well as a first classification weight corresponding to the trained feature classification model and a second classification weight corresponding to the trained sequence classification model, wherein the first classification weight is less than the second classification weight; the first classification weight is positively correlated with the number of feature types contained in the target feature data. The second processing unit 102 is used to input the target feature data into the trained feature classification model for prediction and classification, and obtain a first prediction classification vector. The third processing unit 103 is used to input the target feature data into the trained sequence classification model for prediction and classification, and obtain a second prediction classification vector. The fourth processing unit 104 is used to perform predictive classification fusion on the second classification weight and the second predicted classification vector according to the first classification weight and the first predicted classification vector to obtain the classification result of the target neurotransmitter.
[0053] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.
[0054] Reference Figure 3 This application also provides an electronic device, including: At least one processor 201; At least one memory 202 is used to store at least one program; When the at least one program is executed by the at least one processor 201, the at least one processor 201 implements the method embodiment described above.
[0055] Similarly, it can be understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0056] This application also provides a computer-readable storage medium storing a program executable by a processor 201, which, when executed by the processor 201, is used to implement the above-described method embodiments.
[0057] Similarly, the content of the above method embodiments is applicable to the present computer-readable storage medium embodiments. The specific functions implemented by the present computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0058] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0059] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods.
[0060] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this application are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.
[0061] Furthermore, although this application is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding this application. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional technology for an engineer. Therefore, those skilled in the art can implement the application set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of this application, which is determined by the full scope of the appended claims and their equivalents.
[0062] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0063] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0064] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.
[0065] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0066] In the foregoing description of this specification, the references to terms such as "one embodiment," "another embodiment," or "some embodiments," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0067] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.
[0068] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.
Claims
1. A method for classifying neurotransmitters, characterized in that, include: Obtain target feature data of the target neurotransmitter, as well as a first classification weight corresponding to the trained feature classification model and a second classification weight corresponding to the trained sequence classification model, wherein the first classification weight is less than the second classification weight; the first classification weight is positively correlated with the number of feature types contained in the target feature data; The target feature data is input into the trained feature classification model for prediction and classification to obtain a first predicted classification vector. The target feature data is input into the trained sequence classification model for prediction and classification to obtain a second predicted classification vector. Based on the first classification weight and the first predicted classification vector, the second classification weight and the second predicted classification vector are fused to obtain the classification result of the target neurotransmitter.
2. The method according to claim 1, characterized in that, The trained feature classification model is obtained through the following steps: Obtain neurotransmitter training data for the target object and construct a Naive Bayes model; the neurotransmitter training data includes several neurotransmitter category groups with different feature types; Based on the Naive Bayes model, feature distribution learning processing is performed on the neurotransmitter training data to obtain feature distribution data for each neurotransmitter category group; Based on the feature distribution data and the Naive Bayes model, the trained feature classification model is obtained.
3. The method according to claim 2, characterized in that, The acquisition of neurotransmitter training data for the target object includes: Obtain the electrochemical curve data of the target object; Multidimensional feature extraction is performed on the electrochemical curve data to obtain intermediate feature data; The electrochemical curve data is subjected to feature dimensionality reduction to obtain dimensionality-reduced feature data; The neurotransmitter training data is obtained based on the intermediate feature data and the dimensionality reduction feature data.
4. The method according to any one of claims 1-3, characterized in that, The sequence classification model includes a time-series extraction module, a dimensionality-upgrading module, and a dimensionality-reducing classification module. The step of inputting the target feature data into the trained sequence classification model for prediction and classification to obtain a second predicted classification vector includes: The target feature data is extracted temporally using the temporal extraction module to obtain temporal information features. The time-series information features are upgraded and fused using the upgrade module to obtain upgraded and fused features. The second predicted classification vector is obtained by performing dimensionality reduction classification on the dimensionality-upgraded fusion features through the dimensionality reduction classification module.
5. The method according to claim 4, characterized in that, The step of extracting time-series information features from the target feature data using the time-series extraction module includes: The target feature data is processed by a first temporal network to obtain intermediate temporal features; The intermediate temporal features are processed by a second temporal network to obtain the temporal information features. The first time-series network and the second time-series network have the same network structure.
6. The method according to claim 4, characterized in that, The dimensionality-upgrading module includes several cascaded residual dimensionality-upgrading networks. The dimensionality-upgrading fusion of the time-series information features through the dimensionality-upgrading module to obtain dimensionality-upgraded fused features includes: The time-series information features are input into several cascaded residual upscaling networks for step-by-step upscaling and fusion, resulting in the upscaling fusion features output by the last residual upscaling network. The residual upscaling network is used to perform the following steps: Obtain the input features of the current feature level, wherein the input features are the time-series information features or the output features of the previous feature level; The input features are convolved and increased in dimensionality to obtain convolved and increased dimensionality features; Based on the input features, feature fusion is performed on the convolutional up-dimensional features to obtain the output features of the current feature level.
7. The method according to claim 4, characterized in that, The step of performing dimensionality reduction classification on the dimensionality-upgraded fusion features through the dimensionality reduction classification module to obtain the second predicted classification vector includes: The dimensionality-upgraded fusion features are then subjected to pooling dimensionality reduction to obtain pooling dimensionality-reduced features; The pooled dimensionality reduction features are then used for prediction and classification to obtain the second prediction and classification vector.
8. A classification system for neurotransmitters, characterized in that, include: The first processing unit is used to acquire target feature data of the target neurotransmitter, as well as a first classification weight corresponding to the trained feature classification model and a second classification weight corresponding to the trained sequence classification model, wherein the first classification weight is less than the second classification weight; the first classification weight is positively correlated with the number of feature types contained in the target feature data. The second processing unit is used to input the target feature data into the trained feature classification model for prediction and classification, and obtain the first prediction classification vector. The third processing unit is used to input the target feature data into the trained sequence classification model for prediction and classification, and obtain the second prediction classification vector. The fourth processing unit is used to perform predictive classification fusion on the second classification weight and the second predicted classification vector based on the first classification weight and the first predicted classification vector to obtain the classification result of the target neurotransmitter.
9. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in any one of claims 1-7.
10. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to implement the method as described in any one of claims 1-7.