Auxiliary identification method, device and equipment for early schizophrenia and storage medium

Abstract

The present application relates to the technical field of data analysis, and discloses an auxiliary identification method, device and equipment for early schizophrenia and a storage medium, which are used to improve the auxiliary identification accuracy of schizophrenia. The auxiliary identification method for early schizophrenia comprises: classifying multidimensional data of a target individual, including a topology structure type, a sequence type and a type without special structure; using a graph neural network, a recurrent neural network and a convolutional neural network to extract feature information of the three types of data respectively; fusing the feature information to form a target vector; processing the target vector through a preset early schizophrenia identification model to obtain an identification result, so as to determine the risk degree of the target individual in the early stage of schizophrenia.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to an auxiliary identification method, device, equipment, and storage medium for early-stage schizophrenia. Background Technology

[0002] Currently, methods for assisting in the early identification of schizophrenia typically rely on a single or limited data source, such as brain imaging data or genetic data alone, to identify the risk level of an individual in the early stages of schizophrenia through feature extraction and classification algorithms.

[0003] However, schizophrenia, as a complex mental illness, exhibits high heterogeneity in its pathogenesis and clinical manifestations, involving multiple levels of factors including genetics, environment, neurobiology, and sociopsychology. Therefore, relying on a single data source for identification may overlook other important factors closely related to the onset of schizophrenia, limiting the model's generalization ability and making accurate identification difficult. Summary of the Invention

[0004] This invention provides an auxiliary identification method, device, equipment, and storage medium for early-stage schizophrenia, to solve the technical problem in the prior art where the model's generalization ability is limited due to a single data source, making it difficult to achieve accurate identification.

[0005] The first aspect of this invention provides an auxiliary identification method for early-stage schizophrenia, comprising: classifying multi-dimensional comprehensive data collected from a target individual to obtain topological structure information, sequence information, and information without special structure, wherein the multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota microbiota, and cognitive function test levels; extracting features from the topological structure information, sequence information, and information without special structure based on a pre-set target graph neural network, target recurrent neural network, and target convolutional neural network, respectively, to obtain first feature information, second feature information, and third feature information; fusing the first feature information, second feature information, and third feature information to obtain a target vector; and processing the target vector based on a pre-set early schizophrenia identification model to obtain an identification result, wherein the identification result is used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0006] In one feasible implementation, classifying the multi-dimensional comprehensive data of the collected target individuals to obtain topological structure information, sequence information, and information without special structure includes: classifying data with node and edge connection patterns in the multi-dimensional comprehensive data as topological structure information; identifying sequence information based on the sequence and time dependency of the multi-dimensional comprehensive data; and classifying data that does not belong to the topological structure information or the sequence information as information without special structure.

[0007] In one feasible implementation, before extracting features from the topological structure information, the sequence information, and the information without special structure based on a pre-set target graph neural network, target recurrent neural network, and target convolutional neural network to obtain first feature information, second feature information, and third feature information respectively, the method further includes: grouping a large amount of pre-collected data to be processed into data of topological structure to be processed, data of sequence to be processed, and data of data without special structure to be processed; constructing a target graph neural network based on the data of topological structure to be processed; constructing a target recurrent neural network based on the data of sequence to be processed; and constructing a target convolutional neural network based on the data of data without special structure to be processed.

[0008] In one feasible implementation, the step of constructing a target graph neural network based on the topology data to be processed includes: preprocessing the topology data to be processed; selecting the architecture of the graph neural network and configuring the corresponding number of graph convolutional layers, activation functions, and output layer configuration; training the model based on the configured graph neural network using the preprocessed topology data to be processed, and adjusting the network parameters through the backpropagation algorithm until a preset convergence condition is reached, thereby constructing a target graph neural network capable of extracting topology features.

[0009] In one feasible implementation, constructing a target recurrent neural network based on the sequence data to be processed includes: analyzing the temporal characteristics of the sequence data to be processed, determining the sequence length, time step, and potential periodic or trend components; selecting the architecture of the recurrent neural network, and configuring the corresponding number of hidden layers, unit size, and parameters of the forget gate, input gate, and output gate; based on the configured recurrent neural network, segmenting and batch processing the analysis structure of the sequence data to be processed, and using a preset training strategy, combined with the cross-entropy loss function to optimize the network weights, to construct a target recurrent neural network capable of capturing the dynamic features of the sequence.

[0010] In one feasible implementation, constructing a target convolutional neural network based on the unstructured data to be processed includes: preprocessing the unstructured data to be processed; selecting the architecture of the convolutional neural network and configuring the corresponding number of convolutional layers, kernel size, stride, pooling method, and fully connected layer configuration; training the model based on the configured convolutional neural network using the preprocessed unstructured data to be processed, adjusting the network weights through backpropagation algorithm and optimizer until the model reaches a stable accuracy or loss value on the validation set, thereby constructing a target convolutional neural network capable of extracting features from unstructured data.

[0011] In one feasible implementation, the step of fusing the first feature information, the second feature information, and the third feature information to obtain a target vector includes: calculating the contribution of all features in the first feature information, the second feature information, and the third feature information; removing features with a contribution less than a preset threshold to obtain processed first feature information, second feature information, and third feature information; and concatenating the processed first feature information, second feature information, and third feature information to obtain the target vector.

[0012] A second aspect of the present invention provides an auxiliary identification device for early-stage schizophrenia, comprising: a classification module for classifying multi-dimensional comprehensive data collected from a target individual to obtain topological structure information, sequence information, and information without special structure, wherein the multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota microbiota, and cognitive function test levels; a feature extraction module for extracting features from the topological structure information, the sequence information, and the information without special structure based on a preset target graph neural network, a target recurrent neural network, and a target convolutional neural network, respectively, to obtain first feature information, second feature information, and third feature information; a feature fusion module for fusing the first feature information, the second feature information, and the third feature information to obtain a target vector; and an identification module for processing the target vector based on a preset early-stage schizophrenia identification model to obtain an identification result, wherein the identification result is used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0013] In one feasible implementation, the classification module is specifically used to: classify data with node and edge connection patterns in the multi-dimensional integrated data into topological structure information; identify sequence information based on the sequence and time dependency of the multi-dimensional integrated data; and classify data that does not belong to the topological structure information or the sequence information into information without special structure.

[0014] In one feasible implementation, the auxiliary identification device for early schizophrenia further includes: a processing module for grouping a large amount of pre-collected data to be processed into data of topological structure to be processed, data of sequence to be processed, and data of no special structure to be processed; a first construction module for constructing a target graph neural network based on the data of topological structure to be processed; a second construction module for constructing a target recurrent neural network based on the data of sequence to be processed; and a third construction module for constructing a target convolutional neural network based on the data of no special structure to be processed.

[0015] In one feasible implementation, the first construction module is specifically used for: preprocessing the topology data to be processed; selecting the architecture of the graph neural network and configuring the corresponding number of graph convolutional layers, activation functions and output layer configuration; based on the configured graph neural network, using the preprocessed topology to be processed to train the model, adjusting the network parameters through the backpropagation algorithm until the preset convergence condition is reached, thereby constructing a target graph neural network that can extract topology features.

[0016] In one feasible implementation, the second construction module is specifically used to: analyze the temporal characteristics of the sequence data to be processed, determine the sequence length, time step, and potential periodic or trend components; select the architecture of the recurrent neural network, and configure the corresponding number of hidden layers, unit size, and parameters of the forget gate, input gate, and output gate; based on the configured recurrent neural network, segment and batch process the analysis structure of the sequence data to be processed, and use a preset training strategy, combined with the cross-entropy loss function to optimize the network weights, to construct a target recurrent neural network capable of capturing the dynamic features of the sequence.

[0017] In one feasible implementation, the third building module is specifically used for: preprocessing the unstructured data to be processed; selecting the architecture of the convolutional neural network and configuring the corresponding number of convolutional layers, kernel size, stride, pooling method, and fully connected layer configuration; based on the configured convolutional neural network, using the preprocessed unstructured data to be processed to train the model, adjusting the network weights through the backpropagation algorithm and optimizer until the model reaches a stable accuracy or loss value on the validation set, so as to construct a target convolutional neural network that can extract features from unstructured data.

[0018] In one feasible implementation, the feature fusion module is specifically used to: calculate the contribution of all features in the first feature information, the second feature information, and the third feature information; remove features with a contribution less than a preset threshold to obtain the processed first feature information, the second feature information, and the third feature information; and concatenate the processed first feature information, the second feature information, and the third feature information to obtain the target vector.

[0019] A third aspect of the present invention provides an auxiliary identification device for early-stage schizophrenia, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the auxiliary identification device for early-stage schizophrenia to perform the aforementioned auxiliary identification method for early-stage schizophrenia.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the aforementioned auxiliary identification method for early schizophrenia.

[0021] In the technical solution provided by this invention, multi-dimensional comprehensive data of a target individual is classified to obtain topological structure information, sequence information, and information without special structure. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota microbiota, and cognitive function test levels. Based on a pre-set target graph neural network, target recurrent neural network, and target convolutional neural network, features are extracted from the topological structure information, sequence information, and information without special structure, respectively, to obtain first feature information, second feature information, and third feature information. The first feature information, second feature information, and third feature information are fused to obtain a target vector. The target vector is processed based on a pre-set early schizophrenia identification model to obtain an identification result, which is used to indicate the risk level of the target individual in the early stage of schizophrenia. In this embodiment of the invention, by classifying and processing the multi-dimensional comprehensive data of the target individual, the target graph neural network, the target recurrent neural network, and the target convolutional neural network are used to extract features for topological structure information, sequence information, and information without special structure, respectively, thereby achieving accurate feature capture. Then, a comprehensive target vector is obtained through feature fusion, and identification is performed based on a pre-set early schizophrenia identification model, which improves the accuracy of auxiliary identification of early schizophrenia. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of an embodiment of the auxiliary identification method for early-stage schizophrenia in this invention;

[0023] Figure 2 This is a schematic diagram of another embodiment of the auxiliary identification method for early-stage schizophrenia in this invention;

[0024] Figure 3 This is a schematic diagram of one embodiment of the auxiliary identification device for early-stage schizophrenia in this invention;

[0025] Figure 4 This is a schematic diagram of another embodiment of the auxiliary identification device for early-stage schizophrenia in this invention.

[0026] Figure 5 This is a schematic diagram of one embodiment of an auxiliary identification device for early-stage schizophrenia in this invention. Detailed Implementation

[0027] This invention provides an auxiliary identification method, device, equipment, and storage medium for early-stage schizophrenia. By classifying and integrating features from multi-dimensional data, it improves the accuracy of early-stage schizophrenia identification.

[0028] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] It is understood that the executing entity of this invention can be an auxiliary identification device for early-stage schizophrenia, or it can be a terminal or a server; the specific implementation is not limited here. This embodiment of the invention will be described using a server as an example.

[0030] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the auxiliary identification method for early-stage schizophrenia in this invention includes:

[0031] 101. Classify the multi-dimensional comprehensive data collected from the target individuals to obtain topological structure information, sequence information, and information without special structure. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota, and cognitive function test levels.

[0032] Basic information includes age, gender, height, family medical history, etc.; environmental information includes life stress, living environment, social relationships, etc.; candidate genes include genetic variations related to schizophrenia; proteomics includes the expression level of proteins in the blood; brain complex networks include the connection patterns and activity status between different brain regions; gut microbiota microbiota includes the structure and function of the gut microbiota; cognitive function testing includes assessments of memory, attention, and mental flexibility.

[0033] Basic data, environmental information, proteomics, and cognitive function test levels typically present as unstructured text or numerical data. These data lack clear topological structure or sequence characteristics and are therefore classified as information without specific structure. Candidate gene data exhibits obvious sequence characteristics, meaning the bases in the gene sequence are arranged in a specific order. This linear structure makes candidate gene data suitable for sequence-based information processing methods. Brain complex networks and gut microbiota analysis data, on the other hand, exhibit complex topological structures. Brain complex networks describe the connection patterns and activity states between different brain regions, forming a vast network where nodes represent brain regions and edges represent connections between regions. Gut microbiota analysis data also possesses network characteristics, describing the interactions and relationships among gut microbiota communities. This type of data is suitable for graph-based methods to capture the correlations and information flow between nodes. During the classification process, specific algorithms or tools are used to identify and label each type of data.

[0034] 102. Based on the pre-set target graph neural network, target recurrent neural network, and target convolutional neural network, feature extraction is performed on topological structure information, sequence information, and information without special structure, respectively, to obtain the first feature information, the second feature information, and the third feature information;

[0035] Topological information is transformed into graph structure data, where nodes represent different entities, such as brain regions or microbial species, and edges represent connections or relationships between entities. Topological information includes brain complex networks and gut microbiota microbiome data. Using convolution, pooling, and graph attention mechanisms in the target graph neural network, node features and global structural features in the graph structure data are learned and extracted. Then, through the transmission and fusion of multiple layers of networks, key features that can characterize the topological information are obtained, namely the first feature information.

[0036] When processing complex brain network data, it is first constructed as a graph structure, where nodes represent different brain regions, such as cortical regions, hippocampus, and amygdala, and edges represent functional or structural connections between these regions, such as functional connection strength and protein fiber bundles. Then, using a pre-built target graph neural network, information from each node and its neighbors is aggregated through graph convolutional layers. This process considers not only the node's own attributes, such as the activity level and metabolic level of local brain regions, but also integrates information from other connected nodes, thereby learning and extracting local features for each node. These features may include node activity, connection strength, and information flow direction. Simultaneously, pooling layers in the target graph neural network further reduce the graph size by aggregating features from adjacent nodes while retaining key information, thus extracting higher-level global features. These features may reveal interaction patterns between brain regions, the topological structure of the functional network, and the dynamic processes of information processing. This process is iterative, with each layer further abstracting and refining information based on the previous layer, until finally extracting features that comprehensively reflect the characteristics of complex brain networks. These features include, but are not limited to, functional connection patterns between brain regions, network modularity, and information flow efficiency.

[0037] When processing gut microbiota microbiome data, it is first converted into a graph structure, where nodes represent different gut microbial species and edges reflect the interactions or symbiotic relationships between microorganisms. Then, a pre-built target graph neural network is applied, and weights are dynamically assigned to different neighboring nodes of each node through a graph attention mechanism. This process considers the interaction strength between microorganisms, interspecies dependence, and the influence of environmental factors, thereby more accurately capturing the mutual influence between microorganisms. The multi-layer structure of the target graph neural network gradually integrates node features and the overall structural information of the graph on this basis. In each layer, node features not only contain basic information about microbial species, such as abundance and metabolic type, but also integrate information about connected neighboring nodes and weight information calculated through the graph attention mechanism. As the network deepens, these features are further abstracted and refined to form higher-level feature representations. These features may include, but are not limited to: the relative abundance of microbial species, the structural diversity of the microbial community, the interaction network pattern between microorganisms, community stability indicators, and potential functional associations, such as metabolic pathways and immune regulation.

[0038] The target sequence data is input into the target recurrent neural network. The network processes each element in the sequence step by step through its internal recurrent connection mechanism, while retaining and updating the hidden state to capture the temporal dependence and contextual information in the sequence. As the sequence progresses, the network gradually accumulates and extracts key features, namely the second feature information.

[0039] When processing candidate gene data, the sequence data of the candidate genes is used as input. The sequence data includes DNA sequences, RNA expression sequences, etc. Each nucleotide or expression value in the sequence is processed step by step through a recurrent neural network. At the same time, the hidden state is used to capture the temporal dependence and structural features in the gene sequence. In this process, key features of the gene sequence are learned and extracted. These features include gene expression patterns, the presence and distribution of regulatory elements, the location of mutation sites and their potential effects, etc.

[0040] When extracting features from unstructured information based on a pre-built model, the original unstructured data is first transformed into a format suitable for model processing. Then, the model is used to learn the intrinsic representation of the data, and the data features are gradually abstracted and refined through multi-layer nonlinear transformations, which is the third feature information.

[0041] For basic data, it can be transformed into numerical feature vectors, and then feature extraction can be performed using a pre-built convolutional neural network. The network learns local features and patterns in the basic data through convolutional layers, and reduces feature dimensionality while retaining key features through pooling layers. These features may include key information related to a specific research or prediction task, such as age, gender, and medical history.

[0042] For environmental information, it can be transformed into two-dimensional images or sequence data, and then a pre-built convolutional neural network can be applied for feature extraction. The network captures the spatial features and temporal dependencies in the environmental information through convolution operations, and gradually extracts key features through multi-layer convolution and pooling operations. These features may include key information such as life stress level, quality of living environment, and complexity of social relationships.

[0043] For proteomics, protein expression data can be transformed into two-dimensional matrices or images, and then feature extraction can be performed using a pre-built convolutional neural network. The network learns local patterns and differential expression features in the protein expression profile through convolutional kernels, and reduces the dimensionality of the feature map while retaining key features through pooling layers. These features may include the expression level of specific proteins, the interaction relationships between proteins, and the overall pattern of the protein expression profile.

[0044] For data on cognitive function test levels, the data is transformed into sequential data or one-dimensional vector form, and a pre-built convolutional neural network is applied for feature extraction. The network learns the temporal dependence and pattern changes in the test data to extract key features related to cognitive function. These features may include the changing trends and patterns of test indicators such as reaction time, accuracy, and error rate.

[0045] 103. Perform feature fusion on the first feature information, the second feature information, and the third feature information to obtain the target vector;

[0046] Graph embedding technology is used to transform graph structure features into low-dimensional vectors, and recurrent neural networks are used to transform sequence features into fixed-length feature vectors. At the same time, unstructured data features are transformed into feature vectors according to data type through numerical transformation, two-dimensional matrix or image representation, and convolutional neural network / recurrent neural network processing. A feature concatenation strategy is adopted, that is, these feature vectors from different sources are directly concatenated end to end to form a long vector that integrates various feature information, which serves as the final target vector. The target vector integrates topological structure information, sequence information, and features without special structure information extracted from multi-dimensional integrated data. These features together constitute a comprehensive description of the mental health status of the target individual.

[0047] 104. The target vector is processed based on a pre-set early schizophrenia identification model to obtain identification results, which are used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0048] The target vector is input into the model, which performs nonlinear transformation and feature extraction on the target vector through a multi-layer neural network structure. Then, a classification algorithm is used to analyze and judge the extracted features, and finally outputs a recognition result. This result is usually expressed in the form of probability or score, which is used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0049] The construction method of the early identification model of schizophrenia can be as follows: collect and preprocess multi-dimensional comprehensive data of patients and non-patients with early schizophrenia, and then use multiple neural network architectures, including graph neural networks, recurrent neural networks and convolutional neural networks, to process topological structure, sequence and no special structure information respectively, perform feature extraction and fusion, and then train a machine learning model through optimization strategies such as backpropagation algorithm and gradient descent. During the iteration process, the network weights are continuously adjusted to minimize the prediction error until the model achieves stable identification performance on the validation set. Finally, the output can quantify the probability or score of the target individual's early risk of schizophrenia.

[0050] In this embodiment of the invention, by classifying and processing the multi-dimensional comprehensive data of the target individual, the target graph neural network, the target recurrent neural network, and the target convolutional neural network are used to extract features for topological structure information, sequence information, and information without special structure, respectively, thereby achieving accurate feature capture. Then, a comprehensive target vector is obtained through feature fusion, and the risk level of the target individual in the early stage of schizophrenia is identified based on a pre-set early schizophrenia identification model, thereby improving the accuracy of auxiliary identification of early schizophrenia.

[0051] Please see Figure 2 Another embodiment of the auxiliary identification method for early-stage schizophrenia in this invention includes:

[0052] 201. Classify the multi-dimensional comprehensive data collected from the target individuals to obtain topological structure information, sequence information, and information without special structure. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota, and cognitive function test levels.

[0053] Data with node and edge connection patterns in multi-dimensional integrated data are classified as topological structure information; sequence information is identified based on the sequentiality and time dependency of multi-dimensional integrated data; data that does not belong to topological structure information or sequence information is classified as information without special structure.

[0054] First, data containing clear node and edge connection patterns, such as brain complex networks and gut microbiota microbiome data, are identified. These data exhibit typical topological structure features and are therefore classified as topological structure information. Second, data with sequence and time dependence, such as candidate genes, are identified as sequence information. Finally, other data that do not belong to the topological structure category or meet the sequence characteristics, such as basic data, environmental information, proteomics, and cognitive function test levels, do not have specific structures or sequence patterns and are therefore classified as information without special structure.

[0055] 202. The large amount of pre-collected data to be processed is grouped into data of topological structure to be processed, data of sequence to be processed, and data of no special structure to be processed.

[0056] A large amount of data was pre-collected for processing, including multi-dimensional comprehensive information from early-stage schizophrenia patients and non-patients. This multi-dimensional comprehensive data included basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota, and cognitive function test levels. This data was organized and grouped into three categories: topological structure data, sequence data, and data without specific structure. Specifically, the complex brain networks and gut microbiota of early-stage schizophrenia patients and non-patients were classified as topological structure data; candidate genes were classified as sequence data; and basic information, environmental information, proteomics, and cognitive function test levels were classified as data without specific structure.

[0057] 203. Construct a target graph neural network based on the topological structure data to be processed;

[0058] Preprocess the topological data to be processed; select the architecture of the graph neural network and configure the corresponding number of graph convolutional layers, activation functions and output layer configuration; based on the configured graph neural network, train the model using the preprocessed topological structure to be processed, and adjust the network parameters through the backpropagation algorithm until the preset convergence condition is reached, thereby constructing a target graph neural network that can extract topological features.

[0059] The process begins with preprocessing the topological data, including data cleaning, noise removal, and structuring the data into a graph. Nodes represent brain regions or gut microbiota species, while edges represent connections or interactions between them. Next, a suitable graph neural network architecture, such as a Graph Convolutional Network (GCN) or a Graph Attention Network (GAT), is selected based on the data characteristics and task requirements. When configuring the graph neural network, the number of convolutional layers is adjusted to capture topological features at different levels, appropriate activation functions such as ReLU or Sigmoid are chosen to introduce non-linearity, and the output layer is designed to match the needs of the classification or regression task. Then, the model is trained using the preprocessed topological data. Forward propagation is used to calculate predicted values, and backpropagation is used to adjust network parameters to minimize the loss function. During training, model performance is monitored, including accuracy, recall, and F1 score, and hyperparameters such as learning rate and batch size are adjusted as needed until the model reaches the preset convergence condition. Finally, a target graph neural network capable of accurately extracting topological features is constructed.

[0060] 204. Construct a target recurrent neural network based on the sequence data to be processed;

[0061] Analyze the temporal characteristics of the sequence data to be processed, determine the sequence length, time step, and potential periodic or trend components; select the architecture of the recurrent neural network, and configure the corresponding number of hidden layers, unit size, and parameters of the forget gate, input gate, and output gate; based on the configured recurrent neural network, segment and batch process the analysis structure of the sequence data to be processed, and use a preset training strategy, combined with the cross-entropy loss function to optimize the network weights, to construct a target recurrent neural network that can capture the dynamic features of the sequence.

[0062] Choose a recurrent neural network architecture, such as a long short-term memory network or a gated recurrent unit (GRU), as these network structures are particularly adept at handling long-term dependencies in long sequence data. When configuring the recurrent neural network, adjust the number of hidden layers, unit size, and parameters of the forget gate, input gate, and output gate to optimize the model's memory and expressive capabilities. Next, split the sequence data to be processed into training, validation, and test sets, and perform batch processing to accelerate the training process. During training, adopt preset training strategies, such as gradually reducing the learning rate or using early stopping to avoid overfitting, and combine the cross-entropy loss function to optimize the network weights. Through continuous iterative training, a target recurrent neural network that can accurately identify the dynamic features in candidate gene sequences is finally constructed.

[0063] 205. Construct a target convolutional neural network based on the unstructured data to be processed;

[0064] Preprocess the data of unstructured classes to be processed; select the architecture of the convolutional neural network and configure the corresponding number of convolutional layers, kernel size, stride, pooling method and fully connected layer configuration; based on the configured convolutional neural network, train the model using the preprocessed unstructured data of unstructured classes to be processed, and adjust the network weights through the backpropagation algorithm and optimizer until the model reaches a stable accuracy or loss value on the validation set, so as to construct a target convolutional neural network that can extract features from unstructured data.

[0065] The data to be processed, which is unstructured data, includes basic information, environmental information, proteomics data, and cognitive function test levels of patients and non-patients in the early stages of schizophrenia. Preprocessing includes data cleaning, numericalizing categorical variables, and standardizing or normalizing numerical variables. Then, the architecture of the convolutional neural network is selected and configured, including determining the number and configuration of convolutional layers, selecting the pooling method, and designing the structure of fully connected layers. Next, the preprocessed data is used to train the configured convolutional neural network. The network weights are continuously adjusted through the backpropagation algorithm and a suitable optimizer until the accuracy of the model on the validation set tends to stabilize or the loss value reaches a preset threshold. This constructs a target convolutional neural network that can effectively extract and characterize key features in unstructured data.

[0066] 206. Based on the pre-set target graph neural network, target recurrent neural network, and target convolutional neural network, feature extraction is performed on topological structure information, sequence information, and information without special structure, respectively, to obtain the first feature information, the second feature information, and the third feature information;

[0067] Step 206 is similar to the execution process of step 102 above, and will not be described again here.

[0068] 207. Perform feature fusion on the first feature information, the second feature information, and the third feature information to obtain the target vector;

[0069] Calculate the contribution of all features in the first, second, and third feature information; remove features with a contribution less than a preset threshold to obtain the processed first, second, and third feature information; concatenate the processed first, second, and third feature information to obtain the target vector.

[0070] Contribution is a quantitative metric used to evaluate the importance or effectiveness of each feature in a schizophrenia identification task. Feature importance assessment methods can be used to calculate the contribution of each feature. Statistical feature selection methods can be employed; specifically, by using a GBDT model to accumulate the contribution of each feature to the loss function during iteration, the importance of the features is ranked. Simultaneously, the chi-square statistic or mutual information value between each feature and the schizophrenia label can be calculated to quantify the strength of their association. These methods allow for the measurement of the contribution of each feature to schizophrenia identification.

[0071] 208. The target vector is processed based on a pre-set early schizophrenia identification model to obtain identification results, which are used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0072] The construction method of the early identification model of schizophrenia can be as follows: collect and process multi-dimensional comprehensive data of patients and non-patients in the early stage of schizophrenia, extract features by using target graph neural network, target recurrent neural network and target convolutional neural network for topological structure information, sequence information and no special structure information respectively, and then perform feature fusion. Use the fused feature vector to train the machine learning model, and continuously optimize the model parameters through continuous iteration to finally obtain a predictive model that can accurately assess the risk level of the target individual in the early stage of schizophrenia.

[0073] In this embodiment of the invention, multi-dimensional comprehensive data is classified into topological structure class, sequence class, and no special structure class information. Target graph neural network, target recurrent neural network, and target convolutional neural network are constructed in a targeted manner for feature extraction. This fully utilizes multi-source data such as basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota microbiology, and cognitive function test levels. Furthermore, feature fusion technology is used to integrate the feature information extracted from different networks into a comprehensive target vector. Based on a pre-set early schizophrenia identification model, the risk level of the target individual in the early stage of schizophrenia is identified, improving the accuracy and reliability of early schizophrenia auxiliary identification.

[0074] The above describes the auxiliary identification method for early-stage schizophrenia in embodiments of the present invention. The following describes the auxiliary identification device for early-stage schizophrenia in embodiments of the present invention. Please refer to [link to relevant documentation]. Figure 3 One embodiment of the auxiliary identification device for early-stage schizophrenia in this invention includes:

[0075] The classification module 301 is used to classify the multi-dimensional comprehensive data of the collected target individuals to obtain topological structure information, sequence information and no special structure information. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, brain complex networks, gut microbiota, and cognitive function test levels.

[0076] Feature extraction module 302 is used to extract features from the topological structure information, the sequence information and the information without special structure based on a preset target graph neural network, target recurrent neural network and target convolutional neural network, respectively, to obtain first feature information, second feature information and third feature information respectively;

[0077] Feature fusion module 303 is used to fuse the first feature information, the second feature information and the third feature information to obtain a target vector;

[0078] The identification module 304 is used to process the target vector based on a preset early schizophrenia identification model to obtain an identification result, which is used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0079] In this embodiment of the invention, by classifying and processing the multi-dimensional comprehensive data of the target individual, the target graph neural network, the target recurrent neural network, and the target convolutional neural network are used to extract features for topological structure information, sequence information, and information without special structure, respectively, thereby achieving accurate feature capture. Then, a comprehensive target vector is obtained through feature fusion, and the risk level of the target individual in the early stage of schizophrenia is identified based on a pre-set early schizophrenia identification model, thereby improving the accuracy of auxiliary identification of early schizophrenia.

[0080] Please see Figure 4 Another embodiment of the auxiliary identification device for early-stage schizophrenia in this invention includes:

[0081] The classification module 301 is used to classify the multi-dimensional comprehensive data of the collected target individuals to obtain topological structure information, sequence information and no special structure information. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, brain complex networks, gut microbiota, and cognitive function test levels.

[0082] Feature extraction module 302 is used to extract features from the topological structure information, the sequence information and the information without special structure based on a preset target graph neural network, target recurrent neural network and target convolutional neural network, respectively, to obtain first feature information, second feature information and third feature information respectively;

[0083] Feature fusion module 303 is used to fuse the first feature information, the second feature information and the third feature information to obtain a target vector;

[0084] The identification module 304 is used to process the target vector based on a preset early schizophrenia identification model to obtain an identification result, which is used to indicate the risk level of the target individual in the early stage of schizophrenia.

[0085] Optionally, the classification module 301 can be specifically used for:

[0086] Data with node and edge connection patterns in multi-dimensional integrated data are classified as topological structure information; sequence information is identified based on the sequentiality and time dependency of multi-dimensional integrated data; data that does not belong to topological structure information or sequence information is classified as information without special structure.

[0087] Optional aids for early-stage schizophrenia identification may also include:

[0088] The processing module 305 is used to group a large amount of pre-collected data to be processed into data of topological structure, data of sequence, and data of no special structure.

[0089] The first construction module 306 is used to construct a target graph neural network based on the topological structure data to be processed.

[0090] The second construction module 307 is used to construct a target recurrent neural network based on the sequence data to be processed.

[0091] The third building module 308 is used to build a target convolutional neural network based on the data to be processed that has no special structure.

[0092] Optionally, the first building block 306 can be specifically used for:

[0093] Preprocess the topological data to be processed; select the architecture of the graph neural network and configure the corresponding number of graph convolutional layers, activation functions and output layer configuration; based on the configured graph neural network, train the model using the preprocessed topological structure to be processed, and adjust the network parameters through the backpropagation algorithm until the preset convergence condition is reached, thereby constructing a target graph neural network that can extract topological features.

[0094] Optionally, the second building block 307 can be specifically used for:

[0095] Analyze the temporal characteristics of the sequence data to be processed, determine the sequence length, time step, and potential periodic or trend components; select the architecture of the recurrent neural network, and configure the corresponding number of hidden layers, unit size, and parameters of the forget gate, input gate, and output gate; based on the configured recurrent neural network, segment and batch process the analysis structure of the sequence data to be processed, and use a preset training strategy, combined with the cross-entropy loss function to optimize the network weights, to construct a target recurrent neural network that can capture the dynamic features of the sequence.

[0096] Optionally, the third building block 308 can be specifically used for:

[0097] Preprocess the data of unstructured classes to be processed; select the architecture of the convolutional neural network and configure the corresponding number of convolutional layers, kernel size, stride, pooling method and fully connected layer configuration; based on the configured convolutional neural network, train the model using the preprocessed unstructured data of unstructured classes to be processed, and adjust the network weights through the backpropagation algorithm and optimizer until the model reaches a stable accuracy or loss value on the validation set, so as to construct a target convolutional neural network that can extract features from unstructured data.

[0098] Optionally, the feature fusion module 303 can be specifically used to: calculate the contribution of all features in the first feature information, the second feature information, and the third feature information; remove features with a contribution less than a preset threshold to obtain the processed first feature information, the second feature information, and the third feature information; and concatenate the processed first feature information, the second feature information, and the third feature information to obtain the target vector.

[0099] In this embodiment of the invention, multi-dimensional comprehensive data is classified into topological structure class, sequence class, and no special structure class information. Target graph neural network, target recurrent neural network, and target convolutional neural network are constructed in a targeted manner for feature extraction. This fully utilizes multi-source data such as basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota microbiology, and cognitive function test levels. Furthermore, feature fusion technology is used to integrate the feature information extracted from different networks into a comprehensive target vector. Based on a pre-set early schizophrenia identification model, the risk level of the target individual in the early stage of schizophrenia is identified, improving the accuracy and reliability of early schizophrenia auxiliary identification.

[0100] above Figure 3 and Figure 4The auxiliary identification device for early schizophrenia in this embodiment of the invention is described in detail from the perspective of modular functional entities. The auxiliary identification device for early schizophrenia in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0101] See Figure 5 As shown, the auxiliary identification device for early-stage schizophrenia includes a processor 500 and a memory 501. The memory 501 stores machine-executable instructions that can be executed by the processor 500. The processor 500 executes the machine-executable instructions to implement the aforementioned auxiliary identification method for early-stage schizophrenia.

[0102] Furthermore, Figure 5 The early schizophrenia aid for identification also includes a bus 502 and a communication interface 503. The processor 500, the communication interface 503, and the memory 501 are connected via the bus 502.

[0103] The memory 501 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 503 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 502 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0104] The processor 500 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 500 or by instructions in software form. The processor 500 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 501. The processor 500 reads the information in memory 501 and, in conjunction with its hardware, completes the method steps of the aforementioned embodiment.

[0105] The present invention also provides an auxiliary identification device for early-stage schizophrenia. The computer device includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor performs the steps of the auxiliary identification method for early-stage schizophrenia in the above embodiments.

[0106] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the auxiliary identification method for early schizophrenia.

[0107] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0108] If the integrated unit 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 the present invention, in essence, or the part that contributes to the prior art, or all or 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 described in the various embodiments of the present invention. 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.

[0109] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do 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 method for assisting in the identification of early-stage schizophrenia, characterized in that, The auxiliary identification methods for early-stage schizophrenia include: The multi-dimensional comprehensive data collected from the target individuals is classified into topological structure information, sequence information, and information without special structure. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, complex brain networks, gut microbiota, and cognitive function test levels. Based on the preset target graph neural network, target recurrent neural network and target convolutional neural network, feature extraction is performed on the topological structure information, the sequence information and the information without special structure, respectively, to obtain the first feature information, the second feature information and the third feature information; The first feature information, the second feature information, and the third feature information are fused to obtain the target vector; The target vector is processed based on a pre-set early schizophrenia identification model to obtain identification results, which are used to indicate the risk level of the target individual in the early stage of schizophrenia. The step of classifying the multi-dimensional comprehensive data of the collected target individuals to obtain topological structure information, sequence information, and information without special structure includes: classifying data with node and edge connection patterns in the multi-dimensional comprehensive data of the collected target individuals as topological structure information; identifying sequence information based on the sequence and time dependency of the multi-dimensional comprehensive data; and classifying data that does not belong to the topological structure information or the sequence information as information without special structure.

2. The auxiliary identification method for early schizophrenia according to claim 1, characterized in that, Before extracting features from the topological structure information, the sequence information, and the information without special structure based on a pre-set target graph neural network, target recurrent neural network, and target convolutional neural network to obtain the first feature information, the second feature information, and the third feature information, respectively, the process further includes: The large amount of pre-collected data to be processed is grouped into data of topological structure, data of sequence, and data of no special structure. Construct a target graph neural network based on the topology data to be processed; Construct a target recurrent neural network based on the sequence data to be processed; Construct a target convolutional neural network based on the unstructured data to be processed.

3. The auxiliary identification method for early-stage schizophrenia according to claim 2, characterized in that, The step of constructing a target graph neural network based on the topology data to be processed includes: The topology data to be processed is preprocessed; Choose the architecture of the graph neural network and configure the corresponding number of graph convolutional layers, activation functions, and output layer configuration; Based on the configured graph neural network, the model is trained using the preprocessed topology to be processed. The network parameters are adjusted through the backpropagation algorithm until the preset convergence condition is reached, thereby constructing a target graph neural network that can extract topological features.

4. The auxiliary identification method for early-stage schizophrenia according to claim 2, characterized in that, The step of constructing a target recurrent neural network based on the sequence data to be processed includes: Analyze the temporal characteristics of the sequence data to be processed to determine the sequence length, time step, and potential periodic or trend components; Choose the architecture of the recurrent neural network and configure the corresponding number of hidden layers, unit size, and parameters for the forget gate, input gate, and output gate; Based on the configured recurrent neural network, the analysis structure of the sequence data to be processed is segmented and batch-processed. A preset training strategy is adopted, and the network weights are optimized by combining the cross-entropy loss function to construct a target recurrent neural network that can capture the dynamic features of the sequence.

5. The auxiliary identification method for early-stage schizophrenia according to claim 2, characterized in that, The step of constructing a target convolutional neural network based on the unstructured data to be processed includes: Preprocess the data to be processed that has no special structure; Choose the architecture of the convolutional neural network and configure the corresponding number of convolutional layers, kernel size, stride, pooling method, and fully connected layer configuration; Based on the configured convolutional neural network, the model is trained using preprocessed data of unstructured classes. The network weights are adjusted through backpropagation algorithm and optimizer until the model reaches a stable accuracy or loss value on the validation set, so as to construct a target convolutional neural network that can extract features from unstructured data.

6. The auxiliary identification method for early-stage schizophrenia according to any one of claims 1-5, characterized in that, The step of fusing the first feature information, the second feature information, and the third feature information to obtain the target vector includes: Calculate the contribution of all features in the first feature information, the second feature information, and the third feature information; Features with a contribution value less than a preset threshold are removed to obtain the processed first feature information, second feature information, and third feature information. The processed first feature information, second feature information, and third feature information are concatenated to obtain the target vector.

7. An auxiliary identification device for early-stage schizophrenia, characterized in that, The auxiliary identification device for early-stage schizophrenia includes: The classification module is used to classify the multi-dimensional comprehensive data of the collected target individuals to obtain topological structure information, sequence information, and information without special structure. The multi-dimensional comprehensive data includes basic information, environmental information, candidate genes, proteomics, brain complex networks, gut microbiota, and cognitive function test levels. The feature extraction module is used to extract features from the topological structure information, the sequence information and the information without special structure based on the preset target graph neural network, target recurrent neural network and target convolutional neural network respectively, to obtain the first feature information, the second feature information and the third feature information respectively; The feature fusion module is used to fuse the first feature information, the second feature information and the third feature information to obtain the target vector; The identification module is used to process the target vector based on a pre-set early schizophrenia identification model to obtain an identification result, which is used to indicate the risk level of the target individual in the early stage of schizophrenia. The classification module is specifically used to: classify data with node and edge connection patterns in the multi-dimensional integrated data into topological structure information; identify sequence information based on the sequence and time dependency of the multi-dimensional integrated data; and classify data that does not belong to the topological structure information or the sequence information into information without special structure.

8. An auxiliary identification device for early-stage schizophrenia, characterized in that, The auxiliary identification device for early schizophrenia includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the early schizophrenia assistive identification device to perform the early schizophrenia assistive identification method as described in any one of claims 1-6.

9. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the auxiliary identification method for early-stage schizophrenia as described in any one of claims 1-6.

Images