Graph construction and retrieval method, device, electronic device and storage medium

Abstract

The present invention provides a method, device, electronic device and storage medium for graph construction and retrieval, wherein the method includes: constructing a disease signal pathway graph based on medical research literature; correlating the disease signal pathway graphs of each disease based on the same molecular entities in the disease signal pathway graphs of each disease to obtain a disease relationship graph; constructing a drug-disease relationship graph based on the relationship between the disease entity and the target entity in the disease relationship graph, and the relationship between the predetermined drug and the target. The graph construction and retrieval method, device, electronic device and storage medium provided by the present invention can mine the potential association between drugs and diseases based on the constructed drug-disease relationship graph, thereby helping users to efficiently and accurately find new indications for drugs.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a graph construction and retrieval method, device, electronic device and storage medium. Background Art

[0002] With the rapid development of the pharmaceutical R&D market, competition in new drug R&D is becoming increasingly fierce. The R&D of both innovative and generic drugs faces a lot of challenges and requires a long R&D process, a lot of R&D costs and human and material resources before they can be marketed. Compared with the R&D of innovative and generic drugs, the R&D of new drug indications is a lower-risk and more profitable option. Unlike the R&D of innovative and generic drugs, the R&D of new drug indications does not require R&D from scratch or huge R&D costs, so the risk is lower.

[0003] At present, since drug R&D information is scattered in massive medical literature, the mining of new drug indication information is mostly achieved through manual sorting, that is, collecting various drug R&D information from massive literature data, and screening, sorting and classifying it. This method is not only time-consuming and inefficient, but also limited by data integrity or personal cognition, resulting in poor reliability and accuracy of drug information mining. Summary of the invention

[0004] The present invention provides a graph construction and retrieval method, device, electronic device and storage medium to solve the defects of low efficiency, poor reliability and accuracy of drug information mining in the prior art.

[0005] The present invention provides a method for constructing a graph, comprising:

[0006] Based on medical research literature, construct a disease signaling pathway map based on disease;

[0007] Based on the same molecular entities in the disease signal pathway maps of each disease, the disease signal pathway maps of each disease are associated to obtain a disease relationship map, wherein the molecular entities are biomarker entities and / or target entities;

[0008] Based on the relationship between disease entities and target entities in the disease relationship map, and the predetermined relationship between drugs and targets, a drug-disease relationship map is constructed.

[0009] According to a method for constructing a graph provided by the present invention, the step of determining the relationship between the disease entity and the target entity in the disease relationship graph comprises:

[0010] Obtaining a signal pathway for connecting each disease entity with each target entity in the disease relationship map;

[0011] Based on the relationship between the nodes in the signal pathway, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined.

[0012] According to a graph construction method provided by the present invention, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined based on the relationship between the nodes in the signal pathway, including:

[0013] Counting the number of negatively regulated relationships between nodes in the signaling pathway;

[0014] Based on the number of relationships, it is determined whether the relationship between the disease entity and the target entity corresponding to the signal pathway is a positive regulation relationship or a negative regulation relationship.

[0015] According to a graph construction method provided by the present invention, the step of determining the relationship between the disease entity and the target entity in the disease relationship graph further includes:

[0016] If the signal pathway exists in the disease signal pathway map, determining that the relationship between the disease entity and the target entity corresponding to the signal pathway is an actual association relationship;

[0017] Otherwise, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined to be a potential association relationship.

[0018] According to a method for constructing a map provided by the present invention, the method constructs a disease signaling pathway map based on diseases as units based on medical research literature, comprising:

[0019] Performing entity recognition on sentences in the medical research document to obtain entity pairs contained in the medical research document, wherein the entity pairs include entities and entity relationships between the entities;

[0020] Based on the entities and the entity relationships, the disease signal pathway map based on diseases is constructed.

[0021] According to a graph construction method provided by the present invention, the entity recognition is performed on the sentences in the medical research document to obtain the entity pairs contained in the medical research document, including:

[0022] Obtaining the title text and abstract text of the medical research document;

[0023] Inputting the title text and the summary text into a sentence classifier to obtain the sentence type of each sentence in the title text and the summary text output by the sentence classifier;

[0024] Entity recognition is performed on the sentence type to be recognized to obtain entity pairs contained in the medical research document.

[0025] According to a graph construction method provided by the present invention, the entity recognition is performed to obtain entity pairs contained in the medical research document, including:

[0026] Performing entity recognition on the sentence to obtain entities and pronouns in the sentence;

[0027] Inputting the entity and the pronoun into a reference relationship classifier, and obtaining a reference relationship between the entity and the pronoun output by the reference relationship classifier;

[0028] In the case where the reference relationship is yes, the pronouns in the sentence are replaced with corresponding entities to obtain an optimized sentence;

[0029] The optimization statement and entities in the optimization statement are input into an entity relationship classifier, and entity relationships between the entities output by the entity relationship classifier are obtained.

[0030] The present invention also provides a retrieval method, comprising:

[0031] Get the target entity to be queried;

[0032] Based on the drug-disease relationship map, the relevant knowledge information of the target entity is determined, and the drug-disease relationship map is determined based on any of the map construction methods described above.

[0033] The present invention also provides a map construction device, comprising:

[0034] The initial map construction unit is used to construct a disease signaling pathway map based on disease based on medical research literature;

[0035] A map association unit, for associating the disease signal pathway maps of each disease based on the same molecular entities in the disease signal pathway maps of each disease to obtain a disease relationship map, wherein the molecular entities are biomarker entities and / or target entities;

[0036] The target map construction unit is used to construct a drug-disease relationship map based on the relationship between disease entities and target entities in the disease relationship map, and the relationship between the predetermined drugs and targets.

[0037] The present invention also provides a retrieval device, comprising:

[0038] An acquisition unit, used for acquiring a target entity to be queried;

[0039] A retrieval unit is used to determine the relevant knowledge information of the target entity based on a drug-disease relationship map, wherein the drug-disease relationship map is determined based on any of the map construction methods described above.

[0040] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the graph construction method or retrieval method described above is implemented.

[0041] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the graph construction methods or retrieval methods described above.

[0042] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the graph construction methods or retrieval methods described above.

[0043] The present invention provides a map construction and retrieval method, device, electronic device and storage medium. By constructing a disease signal pathway map based on diseases based on medical research literature, the information in massive medical research literature can be comprehensively and accurately extracted into the disease signal pathway map of each disease, thereby improving the efficiency of information mining and extraction. By associating the disease signal pathway maps of each disease, the potential correlation between the disease and the target can be mined, and then based on the predetermined relationship between the drug and the target, the potential correlation between the drug and the disease can be further mined. The drug-disease relationship map thus constructed can help users efficiently and accurately mine new indications for drugs, thereby improving the efficiency, reliability and accuracy of drug information mining. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 It is a schematic diagram of the process of the atlas construction method provided by the present invention;

[0046] Figure 2 It is a flow chart of the steps of determining the relationship between disease entities and target entities in the disease relationship map provided by the present invention;

[0047] Figure 3is a schematic flow chart of step 220 in the atlas construction method provided by the present invention;

[0048] Figure 4 is a schematic flow chart of step 110 in the atlas construction method provided by the present invention;

[0049] Figure 5 is a schematic flow chart of step 111 in the atlas construction method provided by the present invention;

[0050] Figure 6 It is a schematic diagram of the structure of the disease signal pathway map provided by the present invention;

[0051] Figure 7 It is a schematic diagram of the structure of the disease relationship map provided by the present invention;

[0052] Figure 8 It is a schematic diagram of the structure of the target-disease relationship map provided by the present invention;

[0053] Fig. 9 It is a schematic diagram of the structure of the drug-target association relationship and drug-disease relationship map provided by the present invention;

[0054] Fig.10 It is a schematic diagram of the flow of the retrieval method provided by the present invention;

[0055] Fig.11 It is a structural schematic diagram of the atlas construction device provided by the present invention;

[0056] Fig.12 It is a schematic diagram of the structure of the retrieval device provided by the present invention;

[0057] Fig.13 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings 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 creative work are within the scope of protection of the present invention.

[0059] Drug R&D is a challenging, risky and costly task. The R&D of innovative drugs requires several years of research and development, huge amounts of money, manpower and resources, and is also closely monitored by regulatory agencies. The R&D market for generic drugs also faces great challenges. Generic drugs have a short product life cycle, and the time from listing to expiration is very short. Requirements such as consistency evaluation have led to a significant increase in the R&D costs of generic drugs, which has caused companies to face great competitive pressure and the dilemma of rising R&D costs in the generic drug market.

[0060] Compared with the R&D of innovative drugs and generic drugs, the R&D of new drug indications is a lower-risk and more profitable option. Exploring drug indications means finding a patient population that has both potential disease treatment needs and can accept the drug based on the known pharmacological effects and drug safety, thereby expanding the scope of drug application. Exploring new drug indications does not require R&D from scratch, nor does it require huge R&D costs, so the risk is relatively low.

[0061] At present, the research and development information of various drugs is mostly scattered in a large amount of medical literature. In this case, mining new drug indication information needs to rely on professionals, that is, collecting the research and development information of various drugs from massive data, and screening, sorting and classifying it to obtain the drug indication information. This method relies too much on manual experience and is inefficient, making it difficult to ensure the reliability and accuracy of drug information mining. In this regard, an embodiment of the present invention provides a graph construction method, which can help users efficiently and accurately mine new drug indications by constructing a drug-disease relationship graph, thereby improving drug sales and increasing corporate revenue and profits.

[0062] Figure 1 It is a schematic diagram of the process of the map construction method provided by the present invention, such as Figure 1 As shown, the method includes:

[0063] Step 110, constructing a disease signal pathway map based on diseases based on medical research literature;

[0064] Specifically, the research directions of medicine include basic medicine, clinical medicine, preventive medicine, health care medicine, rehabilitation medicine, etc., and the medical research literature can be literature related to the above-mentioned various types of medical research. The medical research literature can be obtained from a literature retrieval database including a large number of relevant literature on medicine, biology, health or nursing, for example, the medical research literature can be obtained from a literature retrieval database such as PubMed, Web of Science, MedPeer, etc., and the embodiment of the present invention does not specifically limit this.

[0065] It should be understood that the disease signaling pathway map is essentially a semantic network composed of nodes representing entities and edges representing the relationships between entities. Therefore, after obtaining a large amount of medical research literature, the biological entities related to the disease pathogenesis in the medical research literature and the entity relationships between entities can be mined, and then a disease signaling pathway map based on diseases can be constructed. Here, in order to efficiently, comprehensively and accurately mine biological entities and entity relationships related to disease pathogenesis from medical research literature, entities can be defined in advance. For example, the above-mentioned biological entities related to disease pathogenesis can include disease entities, molecular entities, and pathophysiological mechanism entities, etc. The disease signaling pathway map is the disease pathogenesis map.

[0066] After obtaining numerous entities based on medical research literature mining, the relationships between entities can be identified based on the text information of the medical research literature, and entities can be associated with entities to obtain entity relationships between entities. In order to ensure accurate entity relationships, entity relationships can be defined in advance based on expert experience, and the entity relationships can include but are not limited to: positive (upregulation, activation, promotion) regulatory relationships, negative (downregulation, inactivation, inhibition) regulatory relationships, etc. For example, the entity relationship between molecule A and disease B can be positive regulation or negative regulation; the entity relationship between molecule A and pathophysiological mechanism C can be activation regulation, etc.

[0067] After mining entities and entity relationships based on medical research literature, a disease signaling pathway map can be constructed based on entities and entity relationships. Each entity constitutes a node in the disease signaling pathway map, each entity relationship constitutes an edge connection between nodes in the disease signaling pathway map, and the specific relationship between entities (i.e., the specific content of the entity relationship) constitutes the attribute of the edge. The disease signaling pathway map thus constructed is based on disease entities, which can provide conditions for the subsequent construction of a drug-disease relationship map.

[0068] Step 120, based on the same molecular entities in the disease signal pathway maps of each disease, the disease signal pathway maps of each disease are associated to obtain a disease relationship map, where the molecular entities are biomarker entities and / or target entities;

[0069] Specifically, entities in the disease signaling pathway map may include disease entities, molecular entities, pathophysiological mechanism entities, etc., wherein molecular entities are biomarker entities and / or target entities, and targets and / or biomarkers may be marked at the molecular entity nodes in the disease signaling pathway map for distinction. Here, biomarker entities refer to biological molecular entities such as nucleic acids, proteins (peptides), metabolites, etc. that can reflect the physiological and pathological states of the body; target entities refer to specific molecular entities that exist inside and outside tissue cells and interact with drugs and give drug effects, including receptors, enzymes, ion channels, nucleic acids and other biological macromolecules. Since there is an action relationship between targets and drugs, in order to realize the mining of new drug indications information, the potential association between diseases and targets can be first mined, and then the drugs and diseases can be associated through targets, thereby mining the potential association between drugs and diseases, and then mining to obtain new drug indications information.

[0070] It is understandable that the disease signal pathway map is based on the disease, that is, the disease signal pathway map corresponds to the disease one-to-one. Due to the large number of diseases, the number of constructed disease signal pathway maps is also large. Therefore, in order to realize the mining of the potential relationship between diseases and targets, it is necessary to associate the disease signal pathway maps of each disease. Since the target entity is included in the molecular entity, the disease signal pathway maps of each disease can be associated based on the same molecular entity to generate a relationship map between diseases.

[0071] When associating the disease signal pathway maps of various diseases, a certain disease can be determined as a target disease, and the disease signal pathway map corresponding to the target disease is used as the target disease signal pathway map, and other disease signal pathway maps are used as disease signal pathway maps to be associated; all molecular entities in the target disease signal pathway map are obtained as target molecular entities, and each target molecular entity is compared with each molecular entity in the disease signal pathway map to be associated in turn; when the target molecular entity exists in the disease signal pathway map to be associated, the disease signal pathway map to be associated is associated with the target disease signal pathway map based on the target molecular entity, thereby generating a relationship map between diseases.

[0072] Step 130, constructing a drug-disease relationship map based on the relationship between disease entities and target entities in the disease relationship map, and the predetermined relationship between drugs and targets.

[0073] Specifically, the disease-to-disease relationship map is obtained by associating the disease signal pathway maps of each disease. Therefore, the entities in the disease-to-disease relationship map are consistent with the entities in the disease signal pathway maps of each disease, and may include disease entities, target entities, biomarker entities, pathophysiological mechanism entities, etc. Based on the disease-to-disease relationship map, disease entities, target entities, and the relationship between disease entities and target entities can be mined. Here, the relationship between disease entities and target entities includes both the specific entity relationship between disease entities and target entities, and the actual association relationship and potential association relationship between disease entities and target entities. Based on disease entities, target entities, and the relationship between disease entities and target entities, a target-disease relationship map can be constructed. Based on the target-disease relationship map and the predetermined relationship between drugs and targets, a drug-disease relationship map can be constructed.

[0074] Here, the predetermined relationship between drugs and targets refers to the association between drugs and targets that is pre-constructed based on the already identified drug targets. A key link in modern new drug research and development is to select and determine novel and effective drug targets, and drug indication mining is based on new drug research and development. Therefore, before mining drug indications, the relationship between drugs and targets is determined, based on which the relationship between drugs and targets can be pre-constructed.

[0075] In order to construct a drug-disease relationship map, the target-disease relationship map and the relationship between the drug and the target can be associated based on the same target entity. Specifically, the association can be performed in the following manner: all target entities are obtained based on the target-disease relationship map, and these target entities are used as target target entities; for each target target entity, the disease entity that has an association relationship with the target target entity is first obtained based on the target-disease relationship map, and the disease entity is used as the target disease entity; then, based on the relationship between the drug and the target, the drug entity that has an association relationship with the target target entity is searched, and the drug entity is used as the target drug entity; and the above target disease entity and target drug entity are associated.

[0076] It is understandable that the above-mentioned target disease entity and target drug entity can be one or more. When associating, since the target disease entity and the target drug entity are obtained based on the same target point entity, it is necessary to associate each target drug entity with each target disease entity. After the association relationship between all target disease entities and target drug entities is obtained based on all target target entities, a drug-disease relationship map can be constructed.

[0077] The map construction method provided in the embodiment of the present invention, by constructing a disease signal pathway map based on diseases based on medical research literature, can enable the information in massive medical research literature to be comprehensively and accurately extracted into the disease signal pathway map of each disease, thereby improving the efficiency of information mining and extraction. By associating the disease signal pathway maps of each disease, the potential correlation between the disease and the target can be mined, and then based on the predetermined relationship between the drug and the target, the potential correlation between the drug and the disease can be further mined. The drug-disease relationship map thus constructed can help users efficiently and accurately mine new indications for drugs, thereby improving the efficiency, reliability and accuracy of drug information mining.

[0078] Based on the above embodiments, Figure 2 It is a flowchart of the steps for determining the relationship between disease entities and target entities in the disease relationship map provided by the present invention. Based on the disease relationship map, the relationship between disease entities and target entities can be mined. The relationship here includes both the specific entity relationship between disease entities and target entities and the actual association relationship and potential association relationship between disease entities and target entities. In the embodiment of the present invention, by executing the following steps 210 and 220, the specific entity relationship between disease entities and target entities can be determined based on the disease relationship map. Figure 2 As shown, the steps for determining the relationship between the disease entity and the target entity in the disease relationship map include:

[0079] Step 210, obtaining a signal pathway for connecting each disease entity with each target entity in the disease relationship map;

[0080] Specifically, the signal pathway refers to the information pathway between the disease entity and the target entity in the disease relationship map. For example, molecule 1 is connected to disease A through pathophysiological mechanism 1 and molecule 2 in sequence, where molecule 1 is marked as a target and molecule 2 is marked as a biomarker. Then the signal pathway between disease A (i.e., disease entity) and molecule 1 (i.e., target entity) is: molecule 1-pathophysiological mechanism 1-molecule 2-disease A; for another example, molecule 3 is directly connected to disease A and molecule 3 is marked as a target, then the signal pathway between disease A (i.e., disease entity) and molecule 3 (i.e., target entity) is: molecule 3-disease A.

[0081] Step 220, based on the relationship between the nodes in the signal pathway, determine the relationship between the disease entity and the target entity corresponding to the signal pathway.

[0082] Specifically, each node here refers to each entity in the signal pathway, and the relationship between each node and the relationship between the disease entity and the target entity all refer to entity relationships, which may include but are not limited to: positive (upregulation, activation, promotion) regulatory relationship, negative (downregulation, inactivation, inhibition) regulatory relationship, etc. In the case of obtaining the signal pathway of the disease entity and the target entity, it indicates that there is an association relationship between the target entity and the disease entity. Therefore, the specific entity relationship between the corresponding disease entity and the target entity can be determined based on the signal pathway.

[0083] When determining the relationship between the disease entity and the target entity, it can be determined based on the relationship between the nodes in the signal pathway. First, the entity relationship between each two adjacent nodes in the signal pathway is obtained in turn, and then the relationship between the disease entity and the target entity is determined based on all the entity relationships obtained in the signal pathway. For example, the signal pathway between disease A (i.e., disease entity) and molecule 1 (i.e., target entity) is: molecule 1-pathophysiological mechanism 1-molecule 2-disease A, where the entity relationship between molecule 1 and pathophysiological mechanism 1 is a promotion relationship, the entity relationship between pathophysiological mechanism 1 and molecule 2 is an inhibition relationship, and the entity relationship between molecule 2 and disease A is a positive regulation relationship; based on the signal pathway, the relationship between each node can be obtained as a promotion relationship, an inhibition relationship, and a positive regulation relationship, respectively. Based on the relationship between the above nodes, it can be determined that the relationship between disease A (i.e., disease entity) and molecule 1 (i.e., target entity) corresponding to the signal pathway is a negative regulation relationship. For example, the signal pathway between disease A (i.e., disease entity) and molecule 3 (i.e., target entity) is: molecule 3-disease A, where the entity relationship between molecule 3 and disease A is a positive regulatory relationship. Based on this signal pathway, it can be determined that the relationship between disease A (i.e., disease entity) and molecule 3 (i.e., target entity) is a positive regulatory relationship.

[0084] In an embodiment of the present invention, the signal pathway between each disease entity and each target entity is obtained based on the disease relationship map, so that based on the relationship between each node in the signal pathway, the entity relationship between the disease entity and the target entity can be determined, which provides conditions for subsequent mining of the entity relationship between the drug entity and the disease entity, so that a drug-disease relationship map can be constructed.

[0085] Based on the above embodiments, Figure 3 is a flow chart of step 220 in the method for constructing a map provided by the present invention, such as Figure 3 As shown, step 220 specifically includes:

[0086] Step 221, counting the number of negatively regulated relationships between nodes in the signal pathway;

[0087] Step 222, based on the number of relationships, determine whether the relationship between the disease entity corresponding to the signal pathway and the target entity is a positive regulatory relationship or a negative regulatory relationship.

[0088] Specifically, the relationship between each node in the signal pathway can be a positive (upregulation, activation, promotion) regulatory relationship or a negative (downregulation, inactivation, inhibition) regulatory relationship. When determining the relationship between the disease entity and the target entity based on the relationship between the nodes in the signal pathway, the entity relationship between each two adjacent nodes in the signal pathway can be obtained in turn, and then the number of negative regulatory relationships among all entity relationships obtained in the signal pathway can be counted. When the number of relationships is an even number, it can be determined that the relationship between the disease entity and the target entity corresponding to the signal pathway is a positive regulatory relationship; when the number of relationships is an odd number, it can be determined that the relationship between the disease entity and the target entity corresponding to the signal pathway is a negative regulatory relationship.

[0089] Exemplarily, the signal pathway between disease A (i.e., disease entity) and molecule 1 (i.e., target entity) is obtained as: molecule 1-pathophysiological mechanism 1-molecule 2-disease A, wherein the entity relationship between molecule 1 and pathophysiological mechanism 1 is a promotion relationship, the entity relationship between pathophysiological mechanism 1 and molecule 2 is an inhibition relationship, and the entity relationship between molecule 2 and disease A is a positive regulation relationship; since the promotion relationship belongs to a positive regulation relationship and the inhibition relationship belongs to a negative regulation relationship, therefore, by counting the relationship between the nodes in the signal pathway, it can be obtained that the number of negative regulation relationships is 1, that is, the number of negative regulation relationships is an odd number, thereby determining that the relationship between disease A (i.e., disease entity) and molecule 1 (i.e., target entity) is a negative regulation relationship.

[0090] In an embodiment of the present invention, by counting the number of negatively regulated relationships between nodes in the signal pathway, the specific entity relationship between the disease entity and the target entity, i.e., a positive regulatory relationship or a negative regulatory relationship, can be determined based on the number of relationships, thereby providing a basis for subsequently obtaining the specific entity relationship between the drug entity and the disease entity.

[0091] Based on any of the above embodiments, the actual association relationship and potential association relationship between the disease entity and the target entity can also be mined through the disease relationship map. The specific steps of determining the relationship between the disease entity and the target entity include:

[0092] If the signal pathway exists in the disease signal pathway map, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined to be an actual association relationship;

[0093] Otherwise, the relationship between the disease entity and the target entity corresponding to the signaling pathway is determined to be a potential association relationship.

[0094] Specifically, the signal pathway is an information pathway used to connect the disease entity and the target entity in the disease relationship map. When the disease entity and the target entity exist in the same signal pathway, it can be determined that there is an association relationship between the disease entity and the target entity. In order to further explore the potential association relationship between the disease and the target, it is necessary to further identify and judge the signal pathway to determine whether the relationship between the disease entity and the target entity is an actual association relationship or a potential association relationship.

[0095] Since the disease relationship map is obtained by associating the disease signal pathway maps of various diseases based on the same molecular entity, new signal pathways may be added to the disease relationship map compared with the original disease signal pathway maps of various diseases. Based on the newly added signal pathways, the relationship between the disease entity and the target entity can be determined as a potential association relationship.

[0096] In order to identify each signal pathway in the disease relationship map, after obtaining all the signal pathways in the disease relationship map, these signal pathways can be used as target signal pathways, and each target signal pathway can be compared with the signal pathway in the disease signal pathway map of each disease in turn. When the target signal pathway is the same as the signal pathway in the disease signal pathway map, it indicates that the target signal pathway exists in the disease signal pathway map, that is, the target signal pathway is not a newly added signal pathway, indicating that the disease entity corresponding to the target signal pathway and the target entity originally have an association relationship, therefore, it can be determined that the relationship between the disease entity corresponding to the target signal pathway and the target entity is an actual association relationship; when the target signal pathway is different from the signal pathways in the disease signal pathway map, it indicates that the target signal pathway does not exist in the disease signal pathway map, that is, the target signal pathway is a newly added signal pathway, therefore, it can be determined that the relationship between the disease entity corresponding to the target signal pathway and the target entity is a potential association relationship.

[0097] In an embodiment of the present invention, by further identifying and judging each signal pathway in the disease relationship map based on the disease relationship map and the disease signal pathway map of each disease, the potential correlation between the disease and the target can be excavated, thereby facilitating the subsequent excavation of the potential correlation between the drug and the disease based on the predetermined relationship between the drug and the target. In this way, a drug-disease relationship map can be constructed, thereby helping users to efficiently and accurately discover new indications for drugs.

[0098] Based on any of the above embodiments, before executing step 110, medical research documents may be obtained from a document retrieval database in advance. The specific document acquisition steps include:

[0099] Get the initial document collection;

[0100] Based on the publication type of each initial document in the initial document set, the secondary processed documents are screened out from the initial document set to obtain a candidate document set;

[0101] The candidate document set is screened to obtain medical research documents.

[0102] Specifically, the initial document set refers to all document sets in the document retrieval database. The initial document set can be obtained from document retrieval databases from different sources, such as PubMed, Web of Science, MedPeer, etc., which is not specifically limited in the embodiment of the present invention.

[0103] Secondary processed literature is the product of processing and organizing primary literature. Its content is not as comprehensive as primary literature. If entity and entity relationship mining is performed based on secondary processed literature, it may easily lead to the constructed disease signal pathway map being incomplete. In order to improve the comprehensiveness and accuracy of entity and entity relationship mining, as well as improve the efficiency of mining, the initial literature collection can be screened based on publication type, and secondary processed literature can be screened out to reduce the number of literature that needs to be mined, thereby reducing the time and cost required for information mining.

[0104] In one embodiment, the initial document set is preliminarily screened based on the publication type, which can be achieved by rule matching. Each initial document in the initial document set corresponds to at least one publication type. Therefore, the specific rule matching method can be: if the publication type of the initial document includes at least one of the first publication types and does not include any of the second publication types, the initial document is considered not to be a secondary processed document; otherwise, the initial document is considered to be a secondary processed document and is deleted from the initial document set. The first publication type here can include "Journal Article", "Letter"; the second publication type can include "Systematic Review", "Meta-Analysis", "Practice Guideline", "Guideline".

[0105] In another embodiment, the initial document set is preliminarily screened based on the publication type, which can also be achieved through a text binary classification model. The text binary classification model is a commonly used text classification model, which can divide all the initial documents in the initial document set into two categories according to the publication type, namely, secondary processed documents and non-secondary processed documents. Each initial document in the initial document set is input into the text binary classification model, and the text binary classification model outputs whether the initial document is a secondary processed document or not. If the initial document is a secondary processed document, it will be deleted from the initial document set.

[0106] In another embodiment, the initial document collection is initially screened based on the publication type, which can also be achieved by combining rule matching with a text binary classification model, which will not be described in detail here.

[0107] Since the publication types provided in the literature retrieval database have a certain time lag, in order to ensure accurate screening of the latest published medical research literature, after the initial literature set is preliminarily screened to obtain a candidate literature set, the candidate literature set can be accurately screened to avoid incorrect document types obtained due to the lag in publication types, thereby ensuring that the obtained documents are all medical research documents, which is conducive to the subsequent efficient and accurate mining of entities and entity relationships from various medical research documents, and improving the efficiency and accuracy of disease signal pathway map construction.

[0108] When accurately screening the candidate document set, the screening can be performed based on a pre-trained model plus a fine-tuned document classifier. First, the text data of each candidate document in the candidate document set is preprocessed and converted into a form that can be input into the model. The converted text data is input into the document classifier, and the document classifier is used for classification to determine whether the candidate document belongs to a secondary processed document. If the candidate document belongs to a secondary processed document, it will be deleted from the candidate document set, so that medical research documents can be obtained based on the final candidate document set.

[0109] For example, when the initial document set is obtained from PubMed, the pre-trained model can be a PubMedBert pre-trained model, which is a BERT (Bidirectional Encoder Representation from Transformers) model trained in the medical field, and its pre-training data comes from the abstracts and full-text articles of PubMed, so the model can understand the professional terms and language in the medical field. When obtaining the pre-trained model, the PubMedBert pre-trained model can be downloaded and imported from the model library of the Hugging Face website.

[0110] After obtaining the PubMedBert pre-trained model, the pre-trained model can be used as a basis for fine-tuning on the target dataset. The target dataset here is the training data automatically constructed based on the above rule matching method. The literature data that meets the above rules is added as positive examples, and those that do not meet the rules are added as negative examples. There is no need to manually annotate the training data, thereby saving a lot of time and resources. In order to enable the model to classify the training samples and output whether the training samples belong to secondary processed documents, a fully connected output layer can be added on top of the pre-trained layer of the model to classify the training samples and determine whether the training samples belong to secondary processed documents. The training samples here refer to the sample documents in the training data. During the fine-tuning training, the data in the target dataset is first pre-processed and converted into a form that can be input into the model to obtain the converted text data, and then the text data is input into the pre-trained model. The text data is processed by the input layer and Transformer encoder of the pre-trained model, and features are extracted on the pre-trained layer, and then the extracted features are input into the fully connected layer for classification and judgment. The weights and biases of the fully connected layer are updated according to the training data to minimize the loss function on the target dataset, thereby obtaining a document classifier based on the pre-trained model plus fine-tuning.

[0111] It should be noted that for documents whose publication type provided by PubMed is only "Journal Article" and whose publication time is less than 12 months, it is generally believed that the publication type of the document is still to be updated. Therefore, in order to avoid screening errors in this type of documents, you can directly use the document classifier to screen this type of documents to ensure that the documents finally screened are all medical research documents, thereby ensuring that the entities and entity relationships subsequently mined from various medical research documents are more accurate and comprehensive.

[0112] The embodiment of the present invention first screens the initial document set based on the publication type to obtain a candidate document set, and then accurately screens the candidate document set through a document classifier based on a pre-trained model and fine-tuning. This not only ensures that the obtained documents are all medical research documents, but also reduces the number of documents that need to be mined, which is conducive to the subsequent efficient, comprehensive and accurate mining of entities and entity relationships from various medical research documents, thereby improving the efficiency and accuracy of disease signal pathway map construction.

[0113] Based on any of the above embodiments, Figure 4 is a schematic diagram of the process of step 110 in the method for constructing a map provided by the present invention, such as Figure 4 As shown, step 110 specifically includes:

[0114] Step 111, performing entity recognition on sentences in the medical research document to obtain entity pairs contained in the medical research document, wherein the entity pairs include entities and entity relationships between entities;

[0115] Specifically, the statements here refer to text information in basic research literature. Entities are the most basic elements in the disease signal pathway map. There are different relationships between different entities, namely entity relationships. When there is a relationship between two entities, the two entities and the entity relationship between them constitute an entity pair.

[0116] In order to efficiently, comprehensively and accurately mine biological entities related to disease pathogenesis and entity relationships between entities from medical research literature, entities can be defined in advance. For example, entities can include disease entities, molecular entities, and pathophysiological mechanism entities. For each type of entity defined above, a standard dictionary can be pre-constructed for entity extraction and normalization. When pre-constructing a standard dictionary, it can be constructed based on MeSH (Medical Subject Headings), HGNC (HUGO Gene Nomenclature Committee, Human Gene Nomenclature Database) and internal data accumulated by the enterprise independently, and the embodiment of the present invention does not specifically limit this. For example, when obtaining relevant entities of the standard dictionary from MeSH, appropriate inclusion and exclusion rules can be preset based on target requirements, and filtered from the MeSH tree structure. Taking the disease entity in the pre-constructed standard dictionary as an example, MeSH contains the subject vocabulary of various diseases. When obtaining disease-related entities from MeSH, inclusion and exclusion rules suitable for the disease entity can be preset, and the disease vocabulary in the MeSH tree structure can be filtered, thereby constructing the disease entity in the standard dictionary.

[0117] Based on the numerous entities in the pre-built standard dictionary, the sentences in the medical research literature can be entity recognized to obtain all the entities contained in the medical research literature. After obtaining a large number of entities, it is also necessary to identify the relationship between the entities and associate these entities to obtain the entity relationship between the entities. In order to accurately identify the relationship between entities and improve the reliability of the entity relationship finally obtained, the entity relationship can be defined in advance based on expert experience, and the entity relationship includes but is not limited to: positive (upregulation, activation, promotion) regulatory relationship, negative (downregulation, inactivation, inhibition) regulatory relationship, etc. For example, the entity relationship between molecule A and disease B can be positive regulation or negative regulation; the relationship between molecule A and pathophysiological mechanism C can be activation regulation, etc.

[0118] After extracting all entities contained in the medical research literature, the relationships between different entities can be judged based on pre-defined entity relationships and combined with the sentences in the medical research literature to obtain the entity relationships between entities, thereby obtaining several entity pairs.

[0119] Step 112, based on entities and entity relationships, construct a disease signal pathway map with diseases as units.

[0120] Specifically, after obtaining entities and entity relationships based on various medical research documents, a disease signaling pathway map can be constructed based on entities and entity relationships. Each entity constitutes a node in the disease signaling pathway map, and each entity relationship constitutes an edge connection between entities in the disease signaling pathway map. The specific relationship between entities (i.e., the specific content of the entity relationship) constitutes the attribute of the edge. In order to make the constructed disease signaling pathway map clearer, more comprehensive, and more accurate, a disease signaling pathway map can be constructed based on the heat of entity pairs on the basis of entity-entity relationships. For example, the heat of the entity pair can be represented by the thickness of the edge, making the information display more intuitive, thereby facilitating users to quickly obtain relevant knowledge information from the map.

[0121] It should be understood that the popularity of an entity pair can be determined based on the literature citation information of medical research literature. Here, literature citation information refers to the relevant information and data cited in medical research literature, which can be used to evaluate the influence and usefulness of medical research literature. Literature citation information includes, but is not limited to: impact factor, JCR (Journal Citation Reports) division, Chinese Academy of Sciences division, number of citations, monthly average number of citations, etc. When determining the popularity of an entity pair, each medical research literature can be scored based on the literature citation information to obtain the influence score of each medical research literature, and then the popularity of the entity pair can be obtained based on the influence scores of all medical research literature where the entity pair appears.

[0122] Here, when obtaining the influence score of each medical research document, the literature citation information of each medical research document can be obtained first, that is, the impact factor, JCR classification, Chinese Academy of Sciences classification, citation number and monthly average citation number and other indicators, and then the medical research document can be scored after comprehensive consideration of all the above indicators. The score can be set to 1 to 5 points, and the score obtained is the influence score of the medical research document. It is also possible to score each medical research document based on the above indicators to obtain the score corresponding to each indicator. For example, when scoring medical research documents based on the impact factor, the impact factor can be divided into five levels of 0 to 20%, 21% to 40%, 41% to 60%, 61% to 80%, and 81% to 100%, and each level corresponds to 1 point, 2 points, 3 points, 4 points, and 5 points respectively. When the impact factor of the medical research document is 57%, the medical research document can be scored 3 points. After obtaining the score corresponding to each indicator, the sum of the scores corresponding to all indicators is used as the influence score of the medical research document. The embodiment of the present invention does not specifically limit the method for obtaining the influence score of the medical research literature.

[0123] After obtaining the influence score of each medical research document, a secondary weighted score can be performed on all medical research documents where the entity pairs appear to obtain the popularity of the entity pairs, thereby constructing a disease signaling pathway map based on the popularity of the entity pairs.

[0124] In an embodiment of the present invention, entity recognition is performed on sentences in various medical research documents to comprehensively and accurately mine entities and entity relationships between entities, and a disease signal pathway map is constructed based on entities, entity relationships, and heat. Thus, information obtained from massive biomedical data can be comprehensively and accurately extracted into the disease signal pathway map, thereby improving the efficiency and reliability of information mining and extraction, reducing the time and cost required for information mining, and marking the heat of entity relationships in the disease signal pathway map to make information display more intuitive.

[0125] Based on the above embodiments, Figure 5 is a schematic diagram of the process of step 111 in the method for constructing a map provided by the present invention, such as Figure 5 As shown, step 111 specifically includes:

[0126] Step 1111, obtaining the title text and abstract text of the medical research document;

[0127] Specifically, when mining entities and entity relationships from the text of each medical research document, since the overall text of each medical research document is very long and most of it is redundant information, in order to improve the efficiency of entity recognition extraction, the text extraction range can be set to the title text and abstract text. After obtaining the text information of each medical research document, the title information and abstract information can be extracted based on the text information to obtain the title text and abstract text.

[0128] In order to further improve the efficiency of entity mining and extraction, the following steps 1112 and 1113 can be executed to classify and judge the title text and the summary text, distinguish the main information from the title text and the summary text, retain the sentences containing the main information in the title text and the summary text, and discard the sentences containing the secondary information in the title text and the summary text, thereby performing entity recognition on the sentences in the retained title text and the summary text, which can further improve the efficiency of entity extraction.

[0129] Step 1112, input the title text and the summary text into a sentence classifier, and obtain the sentence type of each sentence in the title text and the summary text output by the sentence classifier;

[0130] Specifically, the sentence classifier is obtained by training the pre-trained model based on manually annotated training data, and is used to determine the sentence type of each sentence in the title text and the summary text. The sentence types here include sentences to be identified and sentences that do not need to be identified, wherein the sentences to be identified refer to the sentences containing the main information in the title text and the summary text, and the sentences that do not need to be identified refer to the sentences that do not contain the main information in the title text and the summary text, and the above-mentioned main information refers to the text information related to the entities and entity relationships to be extracted.

[0131] The obtained title text and abstract text of the medical research document are input into the sentence classifier, and classified by the sentence classifier to determine the sentence type of each sentence in the title text and abstract text of the medical research document. If the sentence contains the main information, the sentence is retained and the sentence type of the sentence is output as the sentence to be identified; if the sentence does not contain the main information, the sentence is discarded.

[0132] Before executing step 1112, a sentence classifier can be trained. Specifically, the sentence classifier can be trained in the following manner: obtain a pre-trained model, such as a PubMedBert pre-trained model, collect a large number of sample title texts and abstract texts, manually annotate and identify the sentence type of each sentence in the sample title texts and abstract texts, and train the pre-trained model based on the sample title texts and abstract texts and the manually annotated sentence types to obtain a sentence classifier.

[0133] Step 1113, perform entity recognition on the sentence type to be recognized to obtain entity pairs contained in the medical research document.

[0134] Specifically, the sentences to be identified refer to the sentences containing the main information in the title text and the abstract text that are retained after classification and judgment by the sentence classifier. When performing entity recognition on this type of sentence, firstly, based on the pre-built standard dictionary, the rule matching scheme can be used to extract entities, and all entities that have appeared in the pre-built standard dictionary can be extracted to ensure the accuracy of the entity extraction based on the standard dictionary; secondly, since there are still many entities or entity aliases that may not be included in the pre-built standard dictionary, the named entity recognition model can be used to further perform entity recognition extraction on the sentences to be identified, and the semantics in the sentences can be parsed by the named entity recognition model, so as to extract the entities required for constructing the disease signaling pathway map, and ensure that the entities finally extracted are more comprehensive and accurate.

[0135] It should be noted that the named entity recognition model here is a basic information extraction task model in the field of natural language processing, which is used to identify named entities from a given text and classify the entities. Since there are many types of entities to be extracted, including disease entities, molecular entities, pathophysiological mechanism entities, etc., in order to achieve the extraction of multiple entities, a multi-task architecture can be used on the named entity recognition model. The input of the named entity recognition model is the sentences to be recognized in the title text and the abstract text, and its output is the result of multiple tasks, each task is the extraction of an entity, such as disease entity extraction task, molecular entity extraction task, etc.

[0136] The embodiment of the present invention obtains the title text and abstract text of each medical research document and sets the text extraction scope to the title text and abstract text, thereby improving the efficiency of entity recognition and entity relationship extraction. By classifying and judging the title text and abstract text through a sentence classifier, the scope of text extraction can be further narrowed, thereby improving the efficiency of entity and entity relationship extraction, and further improving the efficiency of disease signal pathway map construction.

[0137] Based on the above embodiment, since the title text and abstract text contain a large number of pronouns, these pronouns may also represent certain entities. In order to further ensure the comprehensiveness and accuracy of entity and entity relationship recognition and extraction, a new task can be added to the named entity recognition model in the multi-task architecture, that is, extracting pronouns in the title text and abstract text. To this end, entity recognition is performed in step 111 or step 1113 to obtain entity pairs contained in the medical research literature, including:

[0138] Perform entity recognition on the sentence to obtain the entities and pronouns in the sentence;

[0139] Input the entity and the pronoun into the reference relationship classifier, and obtain the reference relationship between the entity and the pronoun output by the reference relationship classifier;

[0140] When the reference relationship is yes, the pronouns in the sentence are replaced with the corresponding entities to obtain the optimized sentence;

[0141] The optimized statement and entities in the optimized statement are input into the entity relationship classifier to obtain the entity relationship between the entities output by the entity relationship classifier.

[0142] Specifically, when performing entity recognition on sentences in the title text and abstract text, entities can be extracted based on a pre-built standard dictionary using rule matching; then, a named entity recognition model with a multi-task architecture can be used for extraction to obtain all entities and pronouns contained in the sentence. After obtaining a large number of entities and related pronouns, it is necessary to associate entities with entities and entities with pronouns. First, for all pronouns and entities, a reference relationship classifier can be used to determine whether each pronoun and each entity have a reference relationship. The identified entities and pronouns as well as the title text and abstract text are input into the reference relationship classifier. Here, the reference relationship classifier uses a binary classification method to classify the relationship between pronouns and entities, and determines whether the pronoun and entity have a reference relationship. If the reference relationship is yes, it indicates that the pronoun refers to the entity, that is, the pronoun is equivalent to the entity, and the pronoun can be replaced by the entity; if the reference relationship is no, it indicates that the pronoun is not equivalent to the entity, and the pronoun can be deleted to obtain an optimized sentence.

[0143] Secondly, after associating the entity with the pronoun, it is also necessary to associate the entity with the entity to obtain the entity relationship between the entities. Since there may be many different representations of the same entity in the same medical research document, such as the full name and abbreviation of an entity, after obtaining the optimized statement, the relationship between different entities in the same entity category can be judged based on the entity relationship classifier. For example, the optimized statement and entity A and entity B in the optimized statement are input into the entity relationship classifier, and the entity relationship between entity A and entity B output by the entity relationship classifier can be obtained. The entity relationship here can be any one of the following entity relationships: entity A is equal to entity B, entity A contains entity B, entity B contains entity A, and entity A and entity B are unrelated. Since there are many types of relationships between different entities in the same entity category, an entity relationship classifier based on a multi-classification model can be used to classify and judge this type of entity relationship.

[0144] Finally, after associating entities with pronouns and entities with different representations in the same category, it is necessary to discriminate the relationships between all entities to obtain the entity relationships between entities. Here, the relationships between all entities are discriminated, including the entity relationship discrimination between entities of the same entity category, such as disease entity 1 and disease entity 2; it also includes the entity relationship discrimination between entities of different entity categories, such as an entity in the disease and an entity in the molecule. After obtaining the optimized statement, the optimized statement and the two entities of different entity categories in the optimized statement are input into the entity relationship classifier, and the entity relationship between the two entities output by the entity relationship classifier can be obtained. In order to judge the entity relationships between all entities, an entity relationship classifier based on the token-classification model can be used for implementation.

[0145] In an embodiment of the present invention, by first processing the pronouns in the sentence, and when the reference relationship between the pronouns and the entities is yes, replacing the pronouns in the sentence with corresponding entities, the semantics of the obtained optimized sentence can be made clearer, and it is convenient to subsequently identify the entity relationship between entities based on the optimized sentence, so that the entity relationship can be identified and mined more efficiently, comprehensively and accurately.

[0146] Based on the above embodiment, entity recognition is performed in step 111 or step 1113 to obtain entity pairs contained in the medical research document, which also includes:

[0147] Fill the optimized statement and the entities in the optimized statement into the relational query template to obtain the relational query statement;

[0148] The relational query statement is input into the question-answering language model to obtain the entity relationship output by the question-answering language model.

[0149] Specifically, when performing entity recognition and mining to obtain entities and entity relationships, the question-answering language model can also be used to predict. Here, the question-answering language model can be an LLM (Large Language Model) language model. After obtaining the optimized statement, the optimized statement and the entities extracted from the optimized statement are filled into the relational query template, that is, according to the optimized statement and the entities extracted from the optimized statement, the relevant content in the relational query template is replaced accordingly to obtain a relational query statement, and the relational query statement is used as the input of the question-answering language model, and is parsed by the question-answering language model based on a given format to obtain the entity relationship output by the question-answering language model.

[0150] Exemplarily, the relationship query template may be:

[0151] "Below is a biomedical section and the entities it contains. Please determine the possible binary relationships between entities based on the text semantics. Input [text content] input [entity content], possible entity relationships: [entity relationship description]. Please output the results line by line in the following format "(entity 1, entity 2, relationship)"."

[0152] In the above relational query template, the content in the brackets needs to be replaced accordingly according to the optimized statement and the extracted entity to obtain the relational query statement.

[0153] In order to further obtain more comprehensive and accurate entity relationships, the entity relationships output by the question-answering language model may be merged with the entity relationships obtained based on the entity relationship classifier in the above embodiment.

[0154] Based on any of the above embodiments, an embodiment of the present invention provides a map construction method, which generates a disease signal pathway map based on diseases by efficiently, comprehensively and accurately combing the disease pathogenesis based on medical research literature, and associating the disease signal pathway maps of each disease based on the same molecular entities in the disease signal pathway maps of each disease to obtain a disease relationship map, and based on the disease relationship map, the potential correlation between diseases and targets is mined, and combined with the predetermined relationship between drugs and targets, the potential correlation between drugs and diseases is further mined, thereby constructing a drug-disease relationship map to help users efficiently and accurately find new indication information for drugs.

[0155] Specifically, Figure 6 It is a structural schematic diagram of the disease signal pathway map provided by the present invention, and the disease signal pathway map can be constructed in the following manner: first obtain medical research literature, which is obtained after excluding secondary processing literature such as reviews and guidelines from the literature in the PubMed literature retrieval database; based on the medical research literature, obtain target entities and entity relationships between target entities, the target entities may include disease entities, molecular entities, pathophysiological mechanism entities, and the entity relationships may include positive (upregulation, activation, promotion) regulatory relationships, negative (downregulation, inactivation, inhibition) regulatory relationships, etc.; based on the literature citation information of the medical research literature, such as impact factor, JCR division, Chinese Academy of Sciences division, citation number, monthly average citation number, etc., the heat of the entity relationship between the target entities can be obtained; thereby, based on the target entity, entity relationship and heat, a disease signal pathway map with disease as the unit can be constructed.

[0156] like Figure 6As shown, in the embodiment of the present invention, disease signal pathway maps of disease A, disease B and disease C are respectively constructed. In the disease signal pathway map, each target entity constitutes a node, and the entity name can be annotated on the node, such as molecule 1, pathophysiological mechanism 1, molecule 2, disease A, etc.; the entity relationship between entities constitutes an edge connection, and the direction of the arrow or the style of the arrow can be used to indicate whether the entity relationship between the entities is positive regulation or negative regulation. Figure 6 It is shown in the arrow style, that is, a solid arrow indicates that the entity relationship between entities is positively regulated, and a hollow arrow indicates that the entity relationship between entities is negatively regulated; the specific relationship between entities constitutes the attributes of the edge, such as Figure 6 The entity relationship between molecule 1 and pathophysiological mechanism 1 shown in the figure is a promotion relationship, so "promotion" constitutes the attribute of the connecting edge between the two nodes. In addition, the number of medical research documents involving the entity can be represented by the size of the node. For example, the node of disease A is larger, indicating that the number of medical research documents involving disease A is larger; the heat of the entity relationship can also be represented by the thickness of the edge. The thicker the edge, the higher the heat of the entity relationship.

[0157] Furthermore, the molecular entity may be a target entity and / or a biomarker entity, and thus, the target and / or biomarker may be marked at the molecular node in the disease signaling pathway map of each disease, such as Figure 6 As shown, molecules 1 and 3 are both labeled as “targets”, and molecules 2 and 4 are both labeled as “biomarkers”.

[0158] Figure 7 It is a schematic diagram of the structure of the disease relationship map provided by the present invention. After constructing the disease signal pathway map of each disease, the disease signal pathway map can be associated based on the same molecular entity to generate a disease relationship map. Figure 6 As shown, when associating the disease signal pathway maps of each disease, the disease signal pathway map corresponding to disease A can be selected as the target disease signal pathway map. Based on the target disease signal pathway map, all molecular entities can be obtained as: molecule 1, molecule 2, molecule 3, and molecule 4. Each of the obtained molecular entities is compared with the disease signal pathway maps corresponding to disease B and disease C respectively to determine whether the disease signal pathway maps corresponding to disease B and disease C contain the molecular entity. Since the disease signal pathway map corresponding to disease C contains molecule 2, disease C can be associated with the target disease signal pathway map based on molecule 2; since the disease signal pathway map corresponding to disease B contains molecule 3, disease B can be associated with the target disease signal pathway map based on molecule 3. After the association is completed, the following can be generated. Figure 7 The disease relationship graph shown in the figure has the same heat as the entity relationship in the disease signal pathway graph.

[0159] Based on the disease relationship map, the potential association between diseases and targets can be mined. Specifically, the mining can be performed in the following way: based on the disease relationship map, a molecular entity connecting at least two disease entities is obtained, and the molecular entity is only marked as a biomarker (such as molecule 2). Then, among all disease signal pathways connected by the molecular entity (such as the signal pathways of disease A and disease C), if one of the disease signal pathways contains a target entity (such as the signal pathway of disease A contains molecule 1), and the target entity was originally unrelated to the disease entities of other disease signal pathways (such as molecule 1 and disease C are not associated in the disease signal pathway map), then it can be determined that the relationship between the target entity and the disease entity is a potential association relationship (such as the relationship between molecule 1 and disease C is a potential association relationship).

[0160] like Figure 7 As shown in the figure, based on the disease relationship map, the potential association between diseases and targets can also be mined in the following ways: based on the disease relationship map, the signal pathways used to connect each disease entity with each target entity are obtained, such as molecule 1-pathophysiological mechanism 1-molecule 2-disease A, molecule 1-pathophysiological mechanism 1-molecule 2-disease C, molecule 3-disease A, molecule 3-disease B, and each signal pathway is sequentially connected to Figure 6 The signal pathways in the disease signal pathway maps of each disease shown in the figure are compared to determine whether the signal pathway exists in the disease signal pathway map. Since the signal pathway molecule 1-pathophysiological mechanism 1-molecule 2-disease A, molecule 3-disease A, and molecule 3-disease B all exist in Figure 6 In the disease signal pathway map shown, it can be determined that the relationship between the target entity and the disease entity corresponding to the above signal pathway is an actual association relationship, that is, the relationship between molecule 1 and disease A, molecule 3 and disease A, and molecule 3 and disease B are all actual association relationships; since the signal pathway molecule 1-pathophysiological mechanism 1-molecule 2-disease C does not exist in Figure 6 In the disease signal pathway map shown, it can be determined that the relationship between the target entity corresponding to the signal pathway and the disease entity is a potential association relationship, that is, the relationship between molecule 1 and disease C is a potential association relationship.

[0161] After the potential relationship between the disease and the target is mined, it can be highlighted on the disease relationship map, for example Figure 7 In the example, the nodes of molecule 1 and disease C can be highlighted to make the information display more intuitive.

[0162] Figure 8 is a schematic diagram of the structure of the target-disease relationship map provided by the present invention. Based on the disease relationship map, a target-disease relationship map can be constructed, such as Figure 8 As shown in Figure 1, the target-disease relationship map includes: target entity, disease entity, and entity relationship between target entity and disease entity. Based on the disease relationship map, the target-disease relationship map can be generated in the following way: Figure 7 As shown, the signal pathway used to connect each disease entity and each target entity is first obtained, such as molecule 1-pathophysiological mechanism 1-molecule 2-disease A, molecule 1-pathophysiological mechanism 1-molecule 2-disease C, molecule 3-disease A, molecule 3-disease B. For each signal pathway, the entity relationship between each node is obtained in turn, and then the number of negatively regulated relationships of all entity relationships obtained in the signal pathway is counted. When the number of relationships is an even number, it can be determined that the relationship between the disease entity and the target entity corresponding to the signal pathway is a positively regulated relationship; when the number of relationships is an odd number, it can be determined that the relationship between the disease entity and the target entity corresponding to the signal pathway is a negatively regulated relationship.

[0163] For example, for the signal pathway molecule 1-pathophysiological mechanism 1-molecule 2-disease A, the entity relationships between the nodes obtained include: promotion, inhibition, and positive regulation. Since promotion belongs to a positive regulation relationship and inhibition belongs to a negative regulation relationship, the number of negative regulation relationships between the nodes in the signal pathway is 1, so it can be determined that the relationship between molecule 1 and disease A is a negative regulation relationship. Similarly, it can be determined that the relationship between molecule 1 and disease C is a negative regulation relationship, the relationship between molecule 3 and disease A is a positive regulation relationship, and the relationship between molecule 3 and disease B is a negative regulation relationship.

[0164] Since the entities in the target-disease relationship map include target entities and disease entities, in order to more clearly display the target entities, "molecule 1" can be replaced by "target 1" and "molecule 3" can be replaced by "target 2" for display. In addition, the relationship between target 1 and disease C is a potential association relationship mined based on the disease relationship map. When associating, the potential association relationship and the actual association relationship can be distinguished by changing the line style (such as using dotted lines to represent potential association relationships and solid lines to represent actual association relationships), so as to obtain the following: Figure 8 Target-disease relationship map shown.

[0165] Fig. 9 is a schematic diagram of the structure of the drug-target association relationship and drug-disease relationship map provided by the present invention, wherein: Fig. 9The drug-target association relationship shown in (a) is a relationship between a drug and a target that is predetermined based on a known drug target, such as drug 1-target 1, drug 2-target 2. After constructing the target-disease relationship map, the target-disease relationship map and the predetermined relationship between a drug and a target can be associated based on the same target entity, thereby constructing a drug-disease relationship map. The specific association method is as follows:

[0166] like Figure 8 and Fig. 9 As shown in (a), since drug 1 is associated with target 1, target 1 is negatively regulated and actually associated with disease A, and target 1 is negatively regulated and potentially associated with disease C, based on target 1, we can get the relationship between drug 1 and disease A to be negatively regulated and actually associated, and the relationship between drug 1 and disease C to be negatively regulated and potentially associated. Similarly, we can determine that the relationship between drug 2 and disease A is positively regulated and actually associated, and the relationship between drug 2 and disease B is negatively regulated and actually associated, so we can construct the following Fig. 9 (b) shows the drug-disease relationship map.

[0167] The relationship between drugs and diseases is the same as the relationship between targets and diseases. Therefore, the form of the drug-disease relationship map is the same as that of the target-disease relationship map. After the potential correlation between drugs and diseases is discovered, the potential correlation and the actual correlation can be distinguished by different line styles, or displayed in a prominent manner such as highlighting, so that users can quickly learn about the potential correlation between drugs and diseases.

[0168] Based on any of the above embodiments, Fig.10 It is a schematic diagram of the process of the retrieval method provided by the present invention, such as Fig.10 As shown, the method includes:

[0169] Step 1010, obtaining the target entity to be queried;

[0170] Step 1020, based on the drug-disease relationship map, determine the relevant knowledge information of the target entity, the drug-disease relationship map is determined based on any of the map construction methods described above.

[0171] Specifically, for medical research literature from various sources, the method provided in the above embodiment can be used to efficiently, comprehensively and accurately sort out the disease pathogenesis to generate a disease signal pathway map, and by associating the disease signal pathway maps of various diseases, the relationship between the disease and the target can be mined, so that based on the association between the drug and the target, a drug-disease relationship map can be constructed. On this basis, an information retrieval system can be constructed so that users can quickly query and obtain relevant knowledge information about diseases, targets or drugs.

[0172] The user can input the target entity to be queried through a user terminal in the form of a mobile phone, computer, tablet computer, etc., and send the target entity to be queried to the server of the information retrieval system. Here, the target entity to be queried can be any one or a combination of entity information such as disease, target, drug, etc.

[0173] In the process of constructing the drug-disease relationship map, a disease signal pathway map, a disease relationship map, and a target-disease relationship map will also be generated. Therefore, after receiving the target entity to be queried, the server can search based on the disease signal pathway map, the disease relationship map, the target-disease relationship map, and the drug-target relationship map, locate the node corresponding to the target entity and other nodes or sub-nodes at all levels that are connected to the node, and intercept the local map containing the node of the target entity and its connection relationship from the above maps, and return it to the user terminal as the relevant knowledge information of the target entity for the user to view. For example, when a user searches for a target, a disease signal pathway map and / or a disease relationship map and / or a target-disease relationship map containing the target target can be output.

[0174] The retrieval method provided in the embodiment of the present invention can help users efficiently and accurately obtain new drug indication information by realizing rapid retrieval and query of relevant knowledge information such as diseases, drugs, targets, etc. based on various types of graphs constructed, so that users can optimize product structure, increase market share, increase corporate profits, and achieve sustainable development goals.

[0175] Based on any of the above embodiments, Fig.11 is a schematic diagram of the structure of the atlas construction device provided by the present invention, such as Fig.11 As shown, the device comprises:

[0176] An initial map construction unit 1110 is used to construct a disease signal pathway map based on diseases based on medical research literature;

[0177] A map association unit 1120 is used to associate the disease signal pathway maps of each disease based on the same molecular entities in the disease signal pathway maps of each disease to obtain a relationship map between diseases, wherein the molecular entities are biomarker entities and / or target entities;

[0178] The target map construction unit 1130 is used to construct a drug-disease relationship map based on the relationship between disease entities and target entities in the disease relationship map, and the relationship between the predetermined drugs and targets.

[0179] The map construction device provided in the embodiment of the present invention, by constructing a disease signal pathway map based on diseases based on medical research literature, can enable the information in massive medical research literature to be comprehensively and accurately extracted into the disease signal pathway map of each disease, thereby improving the efficiency of information mining and extraction. By associating the disease signal pathway maps of each disease, the potential correlation between the disease and the target can be mined, and then based on the predetermined relationship between the drug and the target, the potential correlation between the drug and the disease can be further mined. The drug-disease relationship map thus constructed can help users efficiently and accurately mine new indications for drugs, thereby improving the efficiency, reliability and accuracy of drug information mining.

[0180] Based on any of the above embodiments, the target graph construction unit 1130 includes a first relationship determination unit, and the first relationship determination unit includes:

[0181] A signal pathway acquisition subunit, used to acquire the signal pathways used to connect each disease entity with each target entity in the disease relationship map;

[0182] The relationship determination subunit is used to determine the relationship between the disease entity and the target entity corresponding to the signal pathway based on the relationship between the nodes in the signal pathway.

[0183] Based on any of the above embodiments, the relationship determination subunit is specifically used for:

[0184] The number of negatively regulated relationships between nodes in the signaling pathway was counted;

[0185] Based on the number of relationships, the relationship between the disease entity and the target entity corresponding to the signaling pathway is determined to be a positive regulatory relationship or a negative regulatory relationship.

[0186] Based on any of the above embodiments, the target map construction unit 1130 further includes a second relationship determination unit, which is used to:

[0187] If the signal pathway exists in the disease signal pathway map, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined to be an actual association relationship;

[0188] Otherwise, the relationship between the disease entity and the target entity corresponding to the signaling pathway is determined as a potential association relationship.

[0189] Based on any of the above embodiments, the device further includes a document acquisition unit, which is used to:

[0190] Get the initial document collection;

[0191] Based on the publication type of each initial document in the initial document set, the secondary processed documents are screened out from the initial document set to obtain a candidate document set;

[0192] The candidate document set is screened to obtain medical research documents.

[0193] Based on any of the above embodiments, the initial graph construction unit 1110 includes:

[0194] An entity acquisition unit, used to perform entity recognition on sentences in medical research documents to obtain entity pairs contained in the medical research documents, wherein the entity pairs include entities and entity relationships between entities;

[0195] The graph construction unit is used to construct a disease signal pathway graph based on diseases based on entities and entity relationships.

[0196] Based on any of the above embodiments, the entity acquisition unit includes:

[0197] A text acquisition subunit is used to acquire the title text and abstract text of medical research documents;

[0198] A sentence classification subunit, used for inputting the title text and the abstract text into a sentence classifier, and obtaining the sentence type of each sentence in the title text and the abstract text output by the sentence classifier;

[0199] The entity recognition subunit is used to perform entity recognition on the sentence type to be recognized, and obtain the entity pairs contained in the medical research document.

[0200] Based on any of the above embodiments, the entity acquisition unit or the entity identification subunit is used to:

[0201] Perform entity recognition on the sentence to obtain the entities and pronouns in the sentence;

[0202] Input the entity and the pronoun into the reference relationship classifier, and obtain the reference relationship between the entity and the pronoun output by the reference relationship classifier;

[0203] When the reference relationship is yes, the pronouns in the sentence are replaced with the corresponding entities to obtain the optimized sentence;

[0204] The optimized statement and entities in the optimized statement are input into the entity relationship classifier to obtain the entity relationship between the entities output by the entity relationship classifier.

[0205] Based on any of the above embodiments, the entity acquisition unit or the entity identification subunit is further used for:

[0206] Fill the optimized statement and the entities in the optimized statement into the relational query template to obtain the relational query statement;

[0207] The relational query statement is input into the question-answering language model to obtain the entity relationship output by the question-answering language model.

[0208] Based on any of the above embodiments, Fig.12 is a schematic diagram of the structure of the retrieval device provided by the present invention, such as Fig.12 As shown, the device comprises:

[0209] An acquisition unit 1210 is used to acquire a target entity to be queried;

[0210] The retrieval unit 1220 is used to determine the relevant knowledge information of the target entity based on the drug-disease relationship map, and the drug-disease relationship map is determined based on any of the map construction methods described above.

[0211] The retrieval device provided in the embodiment of the present invention can help users efficiently and accurately obtain new drug indication information by realizing rapid retrieval and query of relevant knowledge information such as diseases, drugs, targets, etc. based on various types of graphs constructed, so that users can optimize product structure, increase market share, increase corporate profits, and achieve sustainable development goals.

[0212] Fig.13 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.13 As shown, the electronic device may include: a processor 1310, a communication interface 1320, a memory 1330 and a communication bus 1340, wherein the processor 1310, the communication interface 1320 and the memory 1330 communicate with each other through the communication bus 1340. The processor 1310 may call the logic instructions in the memory 1330 to execute the map construction method, which includes: constructing a disease signal pathway map based on diseases based on medical research literature; correlating the disease signal pathway maps of each disease based on the same molecular entities in the disease signal pathway maps of each disease to obtain a disease relationship map, wherein the molecular entities are biomarker entities and / or target entities; constructing a drug-disease relationship map based on the relationship between the disease entity and the target entity in the disease relationship map, and the relationship between the predetermined drug and the target.

[0213] In addition, the processor 1010 can call the logic instructions in the memory 1030 to execute the retrieval method, which includes: obtaining the target entity to be queried; determining the relevant knowledge information of the target entity based on the drug-disease relationship map, and the drug-disease relationship map is determined based on the above-mentioned map construction method.

[0214] In addition, the logic instructions in the above-mentioned memory 1330 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0215] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the map construction method provided by the above methods, which method includes: based on medical research literature, constructing a disease signal pathway map based on diseases; based on the same molecular entities in the disease signal pathway maps of each disease, associating the disease signal pathway maps of each disease to obtain a disease relationship map, where the molecular entities are biomarker entities and / or target entities; based on the relationship between disease entities and target entities in the disease relationship map, and the relationship between predetermined drugs and targets, constructing a drug-disease relationship map.

[0216] In addition, the computer can also execute the retrieval method provided by the above methods, which includes: obtaining the target entity to be queried; determining the relevant knowledge information of the target entity based on the drug-disease relationship map, and the drug-disease relationship map is determined based on the above map construction method.

[0217] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the map construction method provided by the above-mentioned methods, the method comprising: constructing a disease signal pathway map based on diseases based on medical research literature; associating the disease signal pathway maps of each disease based on the same molecular entities in the disease signal pathway maps of each disease to obtain a disease relationship map, wherein the molecular entities are biomarker entities and / or target entities; constructing a drug-disease relationship map based on the relationship between disease entities and target entities in the disease relationship map, and the relationship between predetermined drugs and targets.

[0218] When the computer program is executed by the processor, it is implemented to execute the retrieval method provided by the above methods, which includes: obtaining the target entity to be queried; determining the relevant knowledge information of the target entity based on the drug-disease relationship map, and the drug-disease relationship map is determined based on the above map construction method.

[0219] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0220] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0221] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a graph, characterized in that: include: Based on medical research literature, construct a disease signaling pathway map based on disease; Based on the same molecular entities in the disease signal pathway maps of each disease, the disease signal pathway maps of each disease are associated to obtain a disease relationship map, wherein the molecular entities are biomarker entities and / or target entities; Constructing a drug-disease relationship map based on the relationship between the disease entity and the target entity in the disease relationship map, and the predetermined relationship between the drug and the target; The step of determining the relationship between the disease entity and the target entity in the disease relationship map includes: Obtaining a signal pathway for connecting each disease entity with each target entity in the disease relationship map; Based on the relationship between the nodes in the signal pathway, determining the relationship between the disease entity and the target entity corresponding to the signal pathway; Determining the relationship between the disease entity and the target entity corresponding to the signal pathway based on the relationship between the nodes in the signal pathway includes: Counting the number of negatively regulated relationships between nodes in the signaling pathway; Based on the number of relationships, determining whether the relationship between the disease entity and the target entity corresponding to the signaling pathway is a positive regulation relationship or a negative regulation relationship; The step of determining the relationship between the disease entity and the target entity in the disease relationship map further includes: If the signal pathway exists in the disease signal pathway map, determining that the relationship between the disease entity and the target entity corresponding to the signal pathway is an actual association relationship; Otherwise, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined to be a potential association relationship.

2. The method for constructing a map according to claim 1, characterized in that: The disease signaling pathway map based on disease is constructed based on medical research literature, including: Performing entity recognition on sentences in the medical research document to obtain entity pairs contained in the medical research document, wherein the entity pairs include entities and entity relationships between the entities; Based on the entities and the entity relationships, the disease signal pathway map based on diseases is constructed.

3. The method for constructing a map according to claim 2, characterized in that: The performing entity recognition on the sentences in the medical research document to obtain entity pairs contained in the medical research document includes: Obtaining the title text and abstract text of the medical research document; Inputting the title text and the summary text into a sentence classifier to obtain the sentence type of each sentence in the title text and the summary text output by the sentence classifier; Entity recognition is performed on the sentence type to be recognized to obtain entity pairs contained in the medical research document.

4. The method for constructing a map according to claim 2 or 3, characterized in that: The entity recognition is performed to obtain entity pairs contained in the medical research document, including: Performing entity recognition on the sentence to obtain entities and pronouns in the sentence; Inputting the entity and the pronoun into a reference relationship classifier, and obtaining a reference relationship between the entity and the pronoun output by the reference relationship classifier; In the case where the reference relationship is yes, the pronouns in the sentence are replaced with corresponding entities to obtain an optimized sentence; The optimization statement and entities in the optimization statement are input into an entity relationship classifier, and entity relationships between the entities output by the entity relationship classifier are obtained.

5. A search method, characterized in that: include: Get the target entity to be queried; Based on a drug-disease relationship map, relevant knowledge information of the target entity is determined, and the drug-disease relationship map is determined based on the map construction method according to any one of claims 1 to 4.

6. A graph construction device, characterized in that: include: The initial map construction unit is used to construct a disease signaling pathway map based on disease based on medical research literature; A map association unit, for associating the disease signal pathway maps of each disease based on the same molecular entities in the disease signal pathway maps of each disease to obtain a disease relationship map, wherein the molecular entities are biomarker entities and / or target entities; A target map construction unit, used to construct a drug-disease relationship map based on the relationship between the disease entity and the target entity in the disease relationship map, and the predetermined relationship between the drug and the target; The target graph construction unit includes a first relationship determination unit, and the first relationship determination unit includes: A signal pathway acquisition subunit, used to acquire the signal pathways used to connect each disease entity with each target entity in the disease relationship map; A relationship determination subunit, used to determine the relationship between the disease entity and the target entity corresponding to the signal pathway based on the relationship between the nodes in the signal pathway; The relationship determination subunit is specifically used for: The number of negatively regulated relationships between nodes in the signaling pathway was counted; Based on the number of relationships, determine whether the relationship between the disease entity and the target entity corresponding to the signaling pathway is a positive regulation relationship or a negative regulation relationship; The target graph construction unit further includes a second relationship determination unit, which is used to: If the signal pathway exists in the disease signal pathway map, the relationship between the disease entity and the target entity corresponding to the signal pathway is determined to be an actual association relationship; Otherwise, the relationship between the disease entity and the target entity corresponding to the signaling pathway is determined to be a potential association relationship.

7. A search device, characterized in that: include: An acquisition unit, used for acquiring a target entity to be queried; A retrieval unit is used to determine the relevant knowledge information of the target entity based on a drug-disease relationship map, wherein the drug-disease relationship map is determined based on the map construction method according to any one of claims 1 to 4.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the graph construction method according to any one of claims 1 to 4 or the retrieval method according to claim 5 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the graph construction method according to any one of claims 1 to 4 or the retrieval method according to claim 5 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the graph construction method according to any one of claims 1 to 4 or the retrieval method according to claim 5 is implemented.

Images