Small molecule drug dynamic analysis system based on spiking neural network

Abstract

The invention discloses a small molecule drug dynamic analysis system based on a spiking neural network, and relates to the field of drug analysis, and the system comprises a management module which is used for controlling the editing and sending of control instructions of all function modules and units; the data collection module is used for collecting known drug compound structure data and corresponding in-vivo performance data; the model construction module is used for constructing a pulse graph neural network model architecture, and taking the drug compound structure data collected by the data collection module and the corresponding in-vivo performance data as training samples for training; the pulse calculation of the biological neural network can better simulate the characteristics of a biological system and the dynamic process of the biological system, so that the prediction precision of complex drug molecule behaviors is improved, training is performed on input data distribution through dynamic adjustment, the pulse graph neural network has higher generalization ability, error sources are captured, and the prediction precision is improved. Model structures and parameters are continuously adjusted from model identification output, and dynamic optimization is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of drug analysis, in particular to a small molecule drug dynamic analysis system using a pulse neural network. Background Art

[0002] Small molecule drugs play an important role in the treatment of many diseases, especially in the treatment of cancer, infection and chronic diseases. The ability to quickly and accurately analyze the structure of small molecule drugs is of great significance for the development of new drugs. Pulsing neural networks are a cutting-edge artificial intelligence technology that mimics the behavior of biological nervous systems. Compared with traditional artificial neural networks, they can better process time series information and perform calculations with lower energy consumption, which makes them show potential application prospects in fields such as dynamic analysis of small molecule drug structures. The modern drug discovery process is complex and time-consuming, involving a large amount of data analysis and structure prediction. Traditional drug design methods often rely on physical models and experience, making it difficult to efficiently select and optimize drugs. Existing technologies usually rely on traditional machine learning models, such as decision trees or regression models, which are often insufficient in modeling complex structures. These methods are difficult to effectively capture the nonlinear relationship between small molecule structure and its drug activity, and cannot fully utilize graph structure data to process molecular features. They often simplify compounds into fixed feature vectors, making it difficult to capture the topological information of molecules and their dynamic changes. They often perform poorly when faced with new molecules, especially unprecedented compounds. Overfitting problems are common during model training. After errors are identified, there is often a lack of effective feedback and iterative update mechanisms, making it difficult to form a closed-loop system improvement solution. Summary of the invention

[0003] In view of the above-mentioned shortcomings of the prior art, the present invention provides a small molecule drug dynamic analysis system based on a pulse neural network, which can effectively solve the problems of the prior art.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: The present invention discloses a small molecule drug dynamic analysis system of a pulse neural network, comprising: Management module, used to control the editing and sending of control instructions of each functional module and unit; A data collection module is used to collect known drug compound structure data and corresponding in vivo performance data; The model building module is used to build the pulse graph neural network model architecture, select the graph convolution layer and the pulse transfer mechanism, define the number of nodes and activation function of the input layer, hidden layer and output layer, determine the connection weight of the network, and set the hyperparameters. The training is based on the drug compound structure data collected by the data collection module and the corresponding in vivo performance data as training samples; The dynamic recognition module is used to input new molecules into the trained pulse graph neural network model for recognition. The model predicts and outputs the in vivo performance of the new molecules, including drug distribution, metabolism and effect. Based on the prediction results output by the model, candidate drug compounds that meet the preset demand threshold are screened out, the candidate drug compounds are evaluated, and the model evaluation value is output; The demand setting module is used to customize the demand threshold. After setting, the configuration is applied to the dynamic recognition module; A data repository for organizing candidate drug compounds that meet the required threshold into a compound library, with associated prediction data and model evaluation values ​​attached to each candidate compound; A validation acquisition module is used to acquire experimental data obtained from compound data experimental validation in the data repository to obtain actual in vivo performance data; A judgment unit, used to compare the data predicted by the model with the results displayed by the actual verification data to determine whether there is an error; The error identification module is used to trigger when the judgment unit determines that an error occurs, calculate the error between the model prediction and the actual result, and trace the source based on the input data, model structure and parameter settings; The error adjustment module is used to adjust the hyperparameters, network structure or input data characteristics of the model construction module according to the tracing results of the error identification module.

[0005] Furthermore, the drug compound structure data attributes collected by the data collection module include: molecular formula, functional group and three-dimensional conformation, and the in vivo performance data attributes include: distribution status, metabolic pathway, excretion mode, biological activity, therapeutic effect and toxicity coefficient.

[0006] Furthermore, the data collection module is interactively connected to a preprocessing module via a wireless network, and the preprocessing module is used to standardize the drug structure and biological performance data collected by the data collection module, convert the compound structure information into a graph representation that matches the neural network processing, use the nodes and edges of the molecular graph to represent the atoms and chemical bonds in the compound structure molecules, and normalize the biological performance data.

[0007] Furthermore, the pulse graph neural network model constructed by the model building module converts the new drug molecular structure into a graph structure, enters the pulse graph neural network, and uses the network's multi-layer transmission mechanism to perform deep learning on the input features to extract structural features.

[0008] Furthermore, the operation logic of the impulse graph neural network model in the dynamic recognition module is: ; In the formula, represents the drug performance results predicted by the model, represents the feedforward function of the impulse graph neural network, which indicates the model's processing of the input data X, where θ represents the model parameters, Represents the total number of nodes in the network, Represents to Node The weight of Represents the activation function, which enables the model to learn complex features through nonlinear mapping. Representative Node The set of neighbor nodes of Representative Node and neighbor nodes The connection relationship between them can be the elements of the adjacency matrix. Representative Node In the The hidden state of the layer represents the feature representation of the node. Representative Node The bias term.

[0009] Furthermore, the dynamic identification module screens out candidate drug compounds that meet the preset demand threshold based on the drug performance results predicted by the model. The specific process is as follows: K = {X | Y > T}; In the formula, K represents the set of molecules that meet the conditions, Y represents the drug performance results predicted by the model, T represents the preset demand threshold, and X represents the data currently input into the model.

[0010] Furthermore, the judgment unit is deployed with submodules at a lower level, and the submodules include: a definition module, a threshold module and a configuration module, the definition module and the threshold module are interactively connected through a wireless network, and the threshold module and the configuration module are interactively connected through a wireless network, wherein: The definition module is used to define the standards and basis for judging data attributes, specify the format, scope and data type of data, set judgment rules for each data attribute, and build a standardized evaluation framework; The threshold module is used to formulate several standard thresholds for judging data based on the standardized evaluation framework constructed by the threshold module, and set different priorities or weights according to different attributes; The configuration module is used to actually apply the standard threshold value, compare the current data to be judged with the corresponding standard threshold value, and judge whether the current data meets the requirements.

[0011] Furthermore, in the process of tracing the source of the error based on the input data, model structure and parameter settings, the error identification module: The traceability items of input data are: data outlier status, distribution status of training data and validation data; The traceability items of the model structure are: training error and validation error gap status, learning curve stability status, and feature extraction status; The traceability items of parameter settings are: error fluctuation status, convergence speed status and optimization algorithm effect status during training.

[0012] Furthermore, the error adjustment module, during the adjustment of the input data, network structure and hyperparameters of the model building module: The input data adjustment strategies are: data cleaning and supplementation, data balancing, and data set enhancement; The strategies for adjusting the network structure are: increasing or decreasing the number of layers or neurons in the graph neural network, changing the network structure, and regularization; The hyperparameter adjustment strategy is to adjust the learning rate, adjust the batch size of training data, and select the optimization algorithm.

[0013] Furthermore, the management module is interactively connected to the data collection module through a wireless network, the data collection module is interactively connected to the model construction module through a wireless network, the dynamic identification module is interactively connected to the model construction module, the demand setting module and the data storage repository through a wireless network, the verification acquisition module is interactively connected to the data storage repository and the judgment unit through a wireless network, the judgment unit is interactively connected to the error identification module through a wireless network, the error identification module is interactively connected to the error adjustment module through a wireless network, and the error adjustment module is interactively connected to the model construction module through a wireless network.

[0014] Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects: 1. Through the pulse calculation of biological neural networks, the characteristics of biological systems and their dynamic processes can be better simulated, thereby improving the prediction accuracy of complex drug molecular behaviors. By converting molecular structures into graph representations, the connectivity and interactions between molecules can be more comprehensively considered, and the understanding of the internal structure of molecules can be improved. In addition to the structure of the drug, the system can also comprehensively consider the distribution, metabolism, excretion, and effects of the drug in the body, providing a comprehensive analytical perspective.

[0015] 2. By dynamically adjusting the input data distribution for training, the pulse graph neural network has stronger generalization ability, enabling it to better adapt to new data input, and has self-learning ability. It can self-adjust and optimize after receiving new data, improve the model's adaptability to new molecular reactions, and enhance the model's robustness.

[0016] 3. By introducing deep analysis, combining statistical methods and comparison of model output data, it is possible to capture more detailed sources of errors, provide a basis for subsequent parameter adjustment and model design, build a feedback mechanism, and continuously adjust the model structure and parameters from the model recognition output to achieve dynamic optimization and improve the model prediction ability. By dynamically adjusting the model and retraining based on new data, it is ensured that the model is always consistent with the latest research results, thereby improving its accuracy in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 It is a schematic diagram of the overall framework of the present invention; Figure 2 It is a schematic diagram of the framework of the judgment unit in the present invention.

[0019] The numbers in the figure represent, respectively, 1. Management module; 2. Data collection module; 3. Model building module; 4. Dynamic identification module; 5. Demand setting module; 6. Data repository; 7. Verification and acquisition module; 8. Judgment unit; 81. Definition module; 82. Threshold module; 83. Configuration module; 9. Error identification module; 10. Error adjustment module; 11. Preprocessing module. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0021] The present invention will be further described below in conjunction with the embodiments.

[0022] Example 1: A pulse neural network small molecule drug dynamic analysis system of this embodiment, such as Figure 1 As shown, including: Management module 1, used to control the editing and sending of control instructions of each functional module and unit; Data collection module 2 is used to collect known drug compound structure data and corresponding in vivo performance data; the drug compound structure data attributes collected by data collection module 2 include: molecular formula, functional group and three-dimensional conformation, and the in vivo performance data attributes include: distribution state, metabolic pathway, excretion mode, biological activity, efficacy performance and toxicity coefficient; The data collection module 2 is interactively connected to the preprocessing module 11 through a wireless network. The preprocessing module 11 is used to perform standardization processing on the drug structure and biological performance data collected by the data collection module 2, convert the compound structure information into a graph representation matching the neural network processing, use the nodes and edges of the molecular graph to represent the atoms and chemical bonds in the compound structure molecules, and perform normalization processing on the biological performance data; convert the drug structure information into a graph representation suitable for neural network processing, and perform normalization processing on the biological performance data. This process lays the foundation for subsequent analysis, so that the data can be processed in a unified format; Model building module 3 is used to build the pulse graph neural network model architecture, select the graph convolution layer and the pulse transfer mechanism, define the number of nodes and activation function of the input layer, hidden layer and output layer, determine the connection weight of the network, and set the hyperparameters. The drug compound structure data collected by the data collection module 2 and the corresponding in vivo performance data are used as training samples for training to ensure that the model can learn the mapping between the internal laws of drug molecules and biological performance; Dynamic identification module 4 is used to input new molecules into the trained pulse graph neural network model for identification. The model predicts and outputs the in vivo performance of the new molecules, including drug distribution, metabolism and effect. Candidate drug compounds that meet the preset demand threshold are screened out according to the prediction results output by the model, and the candidate drug compounds are evaluated and the model evaluation value is output. Based on the prediction results output by the model, candidate drug compounds that meet the preset demand threshold can be screened out and further evaluated. This module can efficiently identify potentially effective drugs and improve research and development efficiency; The demand setting module 5 is used to customize the demand threshold. After the setting is completed, the configuration is applied to the dynamic identification module 4, allowing the user to customize the demand threshold and indicate the screening criteria for candidate drugs; A data repository 6, which is used to organize candidate drug compounds that meet the required threshold into a compound library, attach associated prediction data and model evaluation values ​​to each candidate compound, so as to facilitate subsequent review and in-depth analysis, and form a structured database; The verification collection module 7 is used to collect experimental data obtained from the compound data experimental verification of the data repository 6, obtain actual in vivo performance data, and provide empirical support for the accuracy of the model by collecting the actual in vivo performance data obtained by the experiment; The judgment unit 8 is used to compare the data predicted by the model with the results displayed by the actual verification data to determine whether there is an error, thereby providing a data basis for the optimization of the model; The error identification module 9 is used to be triggered when the judgment unit 8 judges that an error occurs, calculates the error between the model prediction and the actual result, and traces the source based on the input data, model structure and parameter settings; in the process of tracing the source based on the input data, model structure and parameter settings, the error identification module 9: The traceability items of input data are: data outlier status, distribution status of training data and validation data; The traceability items of the model structure are: training error and validation error gap status, learning curve stability status, and feature extraction status; The traceability items of parameter settings are: error fluctuation status, convergence speed status and optimization algorithm effect status during training; The error adjustment module 10 is used to adjust the hyperparameters, network structure or input data features of the model construction module 3 according to the tracing results of the error identification module 9; during the adjustment process of the input data, network structure and hyperparameters of the model construction module 3, the error adjustment module 10: The input data adjustment strategy is: data cleaning and supplementation, data balancing and data set enhancement, denoising the data, filling missing values ​​and removing outliers to ensure data quality, and generating more training samples through data balancing and enhancement technology to improve the model's ability to learn different data patterns; The adjustment strategies for the network structure are: increase or decrease the number of layers or neurons of the graph neural network, change the network structure and regularization. According to the performance of the model, appropriately increase or decrease the number of layers or neurons of the graph neural network, balance the complexity and training ability of the model, try different types of graph neural network architectures, and different activation functions, improve the representation ability of the model, introduce regularization technology to avoid overfitting, and improve the generalization ability of the model; The hyperparameter adjustment strategy is: adjust the learning rate, adjust the batch size of training data and select the optimization algorithm, use learning rate decay or learning rate scheduler to dynamically adjust the learning rate, choose the appropriate batch size to balance training efficiency and stability, try different optimization algorithms, and choose the optimizer that best suits the current model and data; The management module 1 is interactively connected to the data collection module 2 through a wireless network, the data collection module 2 is interactively connected to the model building module 3 through a wireless network, the dynamic identification module 4 is interactively connected to the model building module 3, the demand setting module 5 and the data storage library 6 through a wireless network, the verification collection module 7 is interactively connected to the data storage library 6 and the judgment unit 8 through a wireless network, the judgment unit 8 is interactively connected to the error identification module 9 through a wireless network, the error identification module 9 is interactively connected to the error adjustment module 10 through a wireless network, and the error adjustment module 10 is interactively connected to the model building module 3 through a wireless network.

[0023] Compared with the existing technology, the dynamic analysis system using pulse neural network technology, especially in dynamic identification and error source identification and adjustment, provides an effective solution to the problems of insufficient structural feature processing, poor model generalization ability, incomplete error analysis and lack of feedback mechanism in the existing technology. It can not only improve the understanding and prediction accuracy of small molecule drugs, but also accelerate the drug screening and development process, and improve the overall efficiency and accuracy.

[0024] Embodiment 2: In other aspects, this embodiment also provides an operation logic of a pulse graph neural network model, specifically: ; In the formula, represents the drug performance results predicted by the model, represents the feedforward function of the impulse graph neural network, which indicates the model's processing of the input data X, where θ represents the model parameters, Represents the total number of nodes in the network, Represents to Node The weight of Represents the activation function, which enables the model to learn complex features through nonlinear mapping. Representative Node The set of neighbor nodes of Representative Node and neighbor nodes The connection relationship between them can be the elements of the adjacency matrix. Representative Node In the The hidden state of the layer represents the feature representation of the node. Representative Node The bias term.

[0025] In the specific implementation of this example, the pulse graph neural network model converts the new drug molecular structure into a graph structure so that it can enter the pulse graph neural network for processing. , using the network's multi-layer transmission mechanism to perform deep learning on the input features and extract structural features. These features will help the model predict the biological performance of molecules. The calculated results represent the model's predictive performance for new molecules; The dynamic identification module 4 screens out candidate drug compounds that meet the preset demand threshold based on the drug performance results predicted by the model. The specific process is as follows: K = {X | Y > T}; In the formula, K represents the set of molecules that meet the conditions, Y represents the drug performance results predicted by the model, T represents the preset demand threshold, and X represents the data currently input into the model; The model prediction result Y is compared with the preset requirement threshold T to identify candidate molecules that meet the requirements. The requirement threshold T here can be set according to the actual research objectives, for example, the minimum effective concentration or the expected biological activity value. New molecules that meet the condition Y>T are included in the candidate drug set. These molecules are considered to have potential drug activity. Finally, the screened candidate compounds are organized into a compound library, and the corresponding prediction data and model evaluation values ​​are attached to facilitate subsequent experimental verification and in-depth analysis.

[0026] Embodiment 3: In this embodiment, as Figure 2 As shown, the judgment unit 8 has submodules deployed at the lower level, including: a definition module 81, a threshold module 82 and a configuration module 83. The definition module 81 and the threshold module 82 are interactively connected through a wireless network, and the threshold module 82 and the configuration module 83 are interactively connected through a wireless network, wherein: A definition module 81 is used to define the standards and basis for judging data attributes, specify the format, range and data type of data, set judgment rules for each data attribute, and build a standardized evaluation framework; A threshold module 82 is used to formulate a number of standard thresholds for judging data based on the standardized evaluation framework constructed by the threshold module 82, and to set different priorities or weights according to different attributes; The configuration module 83 is used to actually apply the standard threshold value, and to determine whether the current data meets the requirements based on the comparison between the current data to be determined and the corresponding standard threshold value.

[0027] It ensures the accuracy and consistency of data attributes and provides a standardized framework for data analysis. It sets reasonable standards based on drug development goals and scientific basis, provides a clear basis for data evaluation, screens and classifies candidate drugs according to the set standard thresholds, and dynamically feeds back and optimizes thresholds to improve the accuracy and efficiency of the system. It ensures that all steps in data analysis follow scientific logic and the actual needs of drug development, and ultimately helps to accurately screen out drug molecules that meet the standards.

[0028] Working principle: in the specific implementation of the present invention, the management module 1 sends the control instruction, the data collection module 2 collects the raw data of the drug compound structure, the preprocessing module 11 preprocesses the raw data, the model building module 3 builds the pulse graph neural network model, and the pulse neural network model is trained by the known drug compound structure data and its corresponding in vivo performance data to capture the relationship between the drug compound structure and the distribution, metabolism, excretion and effect performance data in the organism. The demand setting module 5 sets the demand threshold, and the dynamic recognition module 4 uses the trained model to dynamically recognize new molecules, screens out candidate drug compounds that meet the demand threshold, and builds a compound library through the data storage library 6; The verification acquisition module 7 obtains the actual verification data of the new molecule, the error identification module 9 evaluates based on the analysis results, analyzes whether there is an error, identifies the source of the error, and the error adjustment module 10 adjusts the pulse graph neural network model parameters corresponding to the error source; Through the collaborative work of various modules, dynamic analysis of small molecule drugs can be efficiently performed, improving efficiency and accuracy in the drug development and screening process. This structured system can make drug development more accurate and provide solid data support for drug verification before clinical application.

[0029] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A pulse neural network small molecule drug dynamic analysis system, characterized in that: include: A management module (1), used to control the editing and sending of control instructions of each functional module and unit; A data collection module (2), used to collect known drug compound structure data and corresponding in vivo performance data; A model building module (3) is used to build a pulse graph neural network model architecture, select a graph convolution layer and a pulse transfer mechanism, define the number of nodes and activation functions of the input layer, hidden layer and output layer, determine the connection weights of the network, and set hyperparameters, and perform training based on the drug compound structure data collected by the data collection module (2) and the corresponding in vivo performance data as training samples; A dynamic recognition module (4) is used to input new molecules into the trained pulse graph neural network model for recognition. The model predicts and outputs the in vivo performance of the new molecules, including drug distribution, metabolism and effect. Candidate drug compounds that meet the preset demand threshold are screened out based on the prediction results output by the model, and the candidate drug compounds are evaluated and the model evaluation value is output; A demand setting module (5) is used to customize the demand threshold, and after the setting is completed, the configuration is applied to the dynamic recognition module (4); A data repository (6) for organizing candidate drug compounds that meet the required threshold into a compound library, and attaching associated prediction data and model evaluation values ​​to each candidate compound; A verification acquisition module (7) is used to acquire experimental data obtained from the compound data experimental verification in the data repository (6) to obtain actual in vivo performance data; A judgment unit (8) is used to compare the data predicted by the model with the results displayed by the actual verification data to determine whether there is an error; An error identification module (9) is used to be triggered when the judgment unit (8) judges that an error occurs, calculate the error between the model prediction and the actual result, and trace the source based on the input data, model structure and parameter settings; The error adjustment module (10) is used to adjust the hyperparameters, network structure or input data features of the model construction module (3) according to the tracing result of the error identification module (9).

2. A pulse neural network small molecule drug dynamic analysis system according to claim 1, characterized in that: The drug compound structure data attributes collected by the data collection module (2) include: molecular formula, functional group and three-dimensional conformation, and the in vivo performance data attributes include: distribution state, metabolic pathway, excretion mode, biological activity, therapeutic effect and toxicity coefficient.

3. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1 is characterized in that: The data collection module (2) is interactively connected to a preprocessing module (11) via a wireless network. The preprocessing module (11) is used to perform standardization processing on the drug structure and biological performance data collected by the data collection module (2), convert the compound structure information into a graph representation form that matches the neural network processing, use the nodes and edges of the molecular graph to represent the atoms and chemical bonds in the compound structure molecules, and perform normalization processing on the biological performance data.

4. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: The pulse graph neural network model constructed by the model construction module (3) converts the new drug molecule structure into a graph structure, enters the pulse graph neural network, and uses the network's multi-layer transmission mechanism to perform deep learning on the input features, extract structural features, and the calculated results represent the model's prediction performance for the new molecule.

5. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: The operating logic of the impulse graph neural network model in the dynamic recognition module (4) is: ； In the formula, represents the drug performance results predicted by the model, represents the feedforward function of the impulse graph neural network, which indicates the model's processing of the input data X, where θ represents the model parameters, Represents the total number of nodes in the network, Represents to Node The weight of Represents the activation function, which enables the model to learn complex features through nonlinear mapping. Representative Node The set of neighbor nodes of Representative Node and neighbor nodes The connection relationship between them can be the elements of the adjacency matrix. Representative Node In the The hidden state of the layer represents the feature representation of the node. Representative Node The bias term.

6. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: The dynamic identification module (4) screens out candidate drug compounds that meet the preset demand threshold based on the drug performance results predicted by the model. The specific process is as follows: K = {X | Y > T}; In the formula, K represents the set of molecules that meet the conditions, Y represents the drug performance results predicted by the model, T represents the preset demand threshold, and X represents the data currently input into the model.

7. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: The judgment unit (8) is provided with submodules at a lower level, and the submodules include: a definition module (81), a threshold module (82) and a configuration module (83), wherein the definition module (81) and the threshold module (82) are interactively connected via a wireless network, and the threshold module (82) and the configuration module (83) are interactively connected via a wireless network, wherein: A definition module (81) is used to define the standards and basis for judging data attributes, specify the format, range and data type of data, set judgment rules for each data attribute, and build a standardized evaluation framework; A threshold module (82) is used to formulate a number of standard thresholds for judging data based on the standardized evaluation framework constructed by the threshold module (82), and to set different priorities or weights according to different attributes; The configuration module (83) is used to actually apply the standard threshold value, compare the current data to be judged with the corresponding standard threshold value, and judge whether the current data meets the requirements.

8. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: In the process of tracing the source of the error based on the input data, model structure and parameter settings, the error identification module (9) The traceability items of input data are: data outlier status, distribution status of training data and validation data; The traceability items of the model structure are: training error and validation error gap status, learning curve stability status, and feature extraction status; The traceability items of parameter settings are: error fluctuation status, convergence speed status and optimization algorithm effect status during training.

9. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: The error adjustment module (10) is used in the process of adjusting the input data, network structure and hyperparameters of the model building module (3): The input data adjustment strategies are: data cleaning and supplementation, data balancing, and data set enhancement; The strategies for adjusting the network structure are: increasing or decreasing the number of layers or neurons in the graph neural network, changing the network structure, and regularization; The hyperparameter adjustment strategy is to adjust the learning rate, adjust the batch size of training data, and select the optimization algorithm.

10. The small molecule drug dynamic analysis system of a pulse neural network according to claim 1, characterized in that: The management module (1) is interactively connected to the data collection module (2) via a wireless network, the data collection module (2) is interactively connected to the model construction module (3) via a wireless network, the dynamic identification module (4) is interactively connected to the model construction module (3), the demand setting module (5) and the data repository (6) via a wireless network, the verification collection module (7) is interactively connected to the data repository (6) and the judgment unit (8) via a wireless network, the judgment unit (8) is interactively connected to the error identification module (9) via a wireless network, the error identification module (9) is interactively connected to the error adjustment module (10) via a wireless network, and the error adjustment module (10) is interactively connected to the model construction module (3) via a wireless network.

Images