CN-116486898-B - Method and device for analyzing relationship between medicine and cancer based on graph neural network

CN116486898BCN 116486898 BCN116486898 BCN 116486898BCN-116486898-B

Abstract

The invention relates to the field of intelligent decision making and digital medical treatment and discloses a method, a device, electronic equipment and a storage medium for analyzing the relation between medicines and cancers based on a graph neural network, wherein the method comprises the steps of inquiring a plurality of groups of chemical data and gene regulation networks of cancer information, inquiring a molecular map of the medicine information, determining cancer vectors of the cancer information by utilizing a graph neural network model, and determining the medicine vectors of the medicine information by utilizing the graph neural network model; the method comprises the steps of constructing a cancer-drug bipartite graph of cancer information and drug information, calculating a first loss value of a graph neural network model, calculating a second loss value between the cancer information and the drug information, carrying out model training on the graph neural network model by using the first loss value and the second loss value to obtain a trained graph neural network model, and identifying a cancer-drug relationship between the cancer information and the drug information by using the trained graph neural network model. The invention can improve the analysis effect of the relation between the medicine and the cancer.

Inventors

LIU XIAOSHUANG

Assignees

平安科技（深圳）有限公司

Dates

Publication Date: 20260505
Application Date: 20230420

Claims (9)

1. A method for analyzing a relationship between a drug and cancer based on a graph neural network, the method comprising: Acquiring cancer information and drug information, querying a plurality of groups of chemical data and gene regulation networks of the cancer information, querying a molecular map of the drug information, determining a cancer vector of the cancer information by using a graph neural network model based on the plurality of groups of chemical data and the gene regulation networks, and determining a drug vector of the drug information by using the graph neural network model; constructing a cancer-drug bipartite graph between the cancer information and the drug information, and calculating a first loss value of the graph neural network model by using the cancer vector, the drug vector and the cancer-drug bipartite graph; Calculating a second loss value between the cancer information and the drug information from the cancer vector, the drug vector, and the cancer-drug bipartite graph; Performing model training on the graph neural network model by using the first loss value and the second loss value to obtain a trained graph neural network model, and identifying a cancer-drug relationship between the cancer information and the drug information by using the trained graph neural network model; The method comprises the steps of carrying out bipartite graph coding on the cancer-drug bipartite graph to obtain a bipartite graph vector, constructing a node initial vector of the cancer-drug bipartite graph by using the cancer vector and the drug vector, carrying out node vector updating on the node initial vector to obtain an updated node vector, carrying out structure scrambling processing on the cancer-drug bipartite graph to obtain a scrambled bipartite graph, calculating a scrambling node vector of the scrambled bipartite graph, calculating a first vector distance between the scrambled node vector and the bipartite graph vector, calculating a second vector distance between the updated node vector and the bipartite graph vector, and calculating the second loss value according to the first vector distance and the second vector distance by using the following formula: Wherein, the Representing the value of the second loss in question, Representing the first vector distance, i.e. the first The first of the disarranged node vector and bipartite graph vector The distance values between the vectors of the individual nodes, Representing the second vector distance, i.e. the first Update node vector and the second node vector The distance values between the vectors of the individual nodes, Representing the number of nodes in the bipartite graph, Representing the node numbers in the bipartite graph.
2. The method for analyzing the relationship between drugs and cancers based on the graphic neural network according to claim 1, wherein the querying of the multiple sets of chemical data and gene regulation network of the cancer information comprises: identifying a query target for the cancer information; Querying multiple sets of mathematical data of the cancer information based on the query targets; Constructing a regulatory network prior distribution of the cancer information according to the multiple sets of chemical data; And constructing a gene regulation network model of the cancer information by using the regulation network prior distribution, and taking the gene regulation network model as the gene regulation network.
3. The method for analyzing the relationship between a drug and cancer based on a graph neural network according to claim 1, wherein the querying the molecular map of the drug information comprises: determining a medication name of the medication information; querying a drug composition molecule of the drug name; Extracting the molecular topology of the drug composition molecules; and determining a molecular map of the drug information by using the molecular topology structure.
4. The method for analyzing the relationship between drugs and cancers based on the graph neural network according to claim 1, wherein the construction of the cancer-drug bipartite graph between the cancer information and the drug information comprises: determining a cancer vertex of the cancer information and a drug vertex of the drug information; querying historical relationships between the cancer vertices and the drug vertices; constructing a relationship edge between the cancer vertex and the drug vertex based on the historical relationship; And determining a cancer-drug bipartite graph between the cancer information and the drug information according to the relation edge, the cancer vertex and the drug vertex.
5. The method of claim 1, wherein calculating a first loss value for the graph neural network model using the cancer vector, the drug vector, and the cancer-drug bipartite graph comprises: Calculating a vector inner product between the cancer vector and the drug vector; Calculating an activation probability value of the vector inner product; Querying the cancer-drug bipartite graph for a true relationship between the cancer vector and the drug vector; determining a true relationship probability of the true relationship; And calculating a cross entropy loss value between the true relation probability and the activation probability value, and taking the cross entropy loss value as a first loss value of the graph neural network model.
6. The method for analyzing a relationship between a drug and cancer based on a graph neural network according to claim 1, wherein the model training the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model comprises: performing back propagation processing on the graph neural network model by using the first loss value and the second loss value to obtain a back propagation model; Calculating an output loss value of the back propagation model; Returning to the step of performing back propagation processing on the graph neural network model by using the first loss value and the second loss value when the output loss value is not smaller than a preset loss value, so as to obtain a back propagation model; and when the output loss value is smaller than a preset loss value, obtaining the trained graphic neural network model.
7. A graph neural network-based relationship analysis apparatus between a drug and cancer for realizing the graph neural network-based relationship analysis method as set forth in any one of claims 1 to 6, characterized in that the apparatus comprises: The vector determining module is used for acquiring cancer information and drug information, inquiring a plurality of groups of chemical data and gene regulation networks of the cancer information, inquiring a molecular map of the drug information, determining a cancer vector of the cancer information by using a graph neural network model based on the plurality of groups of chemical data and the gene regulation networks, and determining a drug vector of the drug information by using the graph neural network model; A first loss calculation module for constructing a cancer-drug bipartite graph between the cancer information and the drug information, calculating a first loss value of the graph neural network model using the cancer vector, the drug vector, and the cancer-drug bipartite graph; a second loss calculation module for calculating a second loss value between the cancer information and the drug information from the cancer vector, the drug vector, and the cancer-drug bipartite graph; And the relation recognition module is used for carrying out model training on the graph neural network model by utilizing the first loss value and the second loss value to obtain a trained graph neural network model, and recognizing the cancer-drug relation between the cancer information and the drug information by utilizing the trained graph neural network model.
8. An electronic device, the electronic device comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the graph neural network-based drug and cancer relationship analysis method of any one of claims 1 to 6.
9. A computer-readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the method for analyzing a relationship between a drug based on a neural network and cancer according to any one of claims 1 to 6.

Description

Method and device for analyzing relationship between medicine and cancer based on graph neural network Technical Field The invention relates to the field of intelligent decision making and digital medical treatment, in particular to a method and a device for analyzing the relationship between a medicine and cancer based on a graph neural network. Background Analysis of the relationship between drugs and cancer based on the graph neural network refers to the process of outputting the relationship between drugs and cancer using the graph neural network model for predicting the response of different tumor cells to drugs. At present, with the rise of machine learning technology, the modern digital medical neighborhood can support functions of disease auxiliary diagnosis, health management, remote consultation and the like, for example, in the cancer treatment scene, the response of different tumor cells to medicines can be predicted through a neural network model, but in the prior art, most of text data or structural information of cancer data and medicine data are used as input data of the neural network model, due to the lack of hidden associated information in an image structure, the hidden information extraction of the relation between medicines and cancers by the analysis method is insufficient, and secondly, in the prior art, the training of the neural network model is realized through the historical data of the cancer data and the medicine data, and the training of the novel relation between the cancer data and the medicine data is lack by utilizing the model. Therefore, the analysis of the relationship between the drug and the cancer is not effective. Disclosure of Invention The invention provides a method and a device for analyzing the relationship between a medicine and cancer based on a graph neural network, which mainly aim to obtain a new cancer vector and a new medicine vector by updating a historical data vector, and utilize a graph structure of cancer information and medicine information as input data of a graph neural network model so as to improve the effect of analyzing the relationship between the medicine and the cancer. In order to achieve the above object, the present invention provides a method for analyzing a relationship between a drug and cancer based on a graph neural network, comprising: Acquiring cancer information and drug information, querying a plurality of groups of chemical data and gene regulation networks of the cancer information, querying a molecular map of the drug information, determining a cancer vector of the cancer information by using a graph neural network model based on the plurality of groups of chemical data and the gene regulation networks, and determining a drug vector of the drug information by using the graph neural network model; constructing a cancer-drug bipartite graph between the cancer information and the drug information, and calculating a first loss value of the graph neural network model by using the cancer vector, the drug vector and the cancer-drug bipartite graph; Calculating a second loss value between the cancer information and the drug information from the cancer vector, the drug vector, and the cancer-drug bipartite graph; And carrying out model training on the graph neural network model by using the first loss value and the second loss value to obtain a trained graph neural network model, and identifying the cancer-drug relationship between the cancer information and the drug information by using the trained graph neural network model. Optionally, the querying the multiple sets of chemical data and gene regulation network for cancer information comprises: identifying a query target for the cancer information; Querying multiple sets of mathematical data of the cancer information based on the query targets; Constructing a regulatory network prior distribution of the cancer information according to the multiple sets of chemical data; And constructing a gene regulation network model of the cancer information by using the regulation network prior distribution, and taking the gene regulation network model as the gene regulation network. Optionally, the querying the molecular map of the drug information includes: determining a medication name of the medication information; querying a drug composition molecule of the drug name; Extracting the molecular topology of the drug composition molecules; and determining a molecular map of the drug information by using the molecular topology structure. Optionally, the constructing a cancer-drug bipartite graph between the cancer information and the drug information includes: determining a cancer vertex of the cancer information and a drug vertex of the drug information; querying historical relationships between the cancer vertices and the drug vertices; constructing a relationship edge between the cancer vertex and the drug vertex based on the historical relationship; And determining a cancer-drug bipartite gr