Methods, training methods, devices, equipment and media for determining medical entity relationships

Abstract

The present disclosure provides a method for determining medical entity relationships, a training method, an apparatus, an electronic device, and a medium, and relates to the field of artificial intelligence technology, particularly to the field of natural language processing, knowledge graphs, and deep learning technology. A specific implementation scheme is as follows: using a diagnostic item labeling model to process target diagnostic item data of a target diagnostic item to obtain a first object set, wherein the first object set represents an object set corresponding to the target diagnostic item; using a treatment item labeling model to process target treatment item data of a target treatment item to obtain a second object set, wherein the second object set represents an object set corresponding to the target treatment item; and determining the relationship between the target diagnostic item and the target treatment item based on the first object set and the second object set.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, particularly natural language processing, knowledge graphs, and deep learning technologies. Specifically, it relates to a method, training method, apparatus, electronic device, and storage medium for determining medical entity relationships. Background Art

[0002] In the healthcare field, the relationships between different medical entities are complex. For example, the relationship between diagnosis and treatment. Determining the relationship between diagnosis and treatment is crucial to standardizing medical resource utilization and rationalizing medical billing. Summary of the Invention

[0003] The present disclosure provides a method for determining a medical entity relationship, a training method, an apparatus, an electronic device, and a storage medium.

[0004] According to one aspect of the present disclosure, a method for determining medical entity relationships is provided, comprising: processing target diagnosis item data of a target diagnosis item using a diagnosis item label model to obtain a first object set, wherein the first object set represents an object set corresponding to the target diagnosis item; processing target treatment item data of a target treatment item using a treatment item label model to obtain a second object set, wherein the second object set represents an object set corresponding to the target treatment item; and determining the relationship between the target diagnosis item and the target treatment item based on the first object set and the second object set.

[0005] According to another aspect of the present disclosure, a model training method is provided, comprising: training a predetermined model using sample diagnostic item data and first object label data to obtain a diagnostic item label model; and training the above-mentioned diagnostic item label model using sample treatment item data and second object label data to obtain a treatment item label model.

[0006] According to another aspect of the present disclosure, a device for determining a medical entity relationship is provided, comprising: a first acquisition module for processing target diagnosis item data of a target diagnosis item using a diagnosis item label model to obtain a first object set, wherein the first object set represents an object set corresponding to the target diagnosis item; a second acquisition module for processing target treatment item data of a target treatment item using a treatment item label model to obtain a second object set, wherein the second object set represents an object set corresponding to the target treatment item; and a first determination module for determining the relationship between the target diagnosis item and the target treatment item based on the first object set and the second object set.

[0007] According to another aspect of the present disclosure, a model training device is provided, including: a third acquisition module, used to train a predetermined model using sample diagnostic item data and first object label data to obtain a diagnostic item label model; and a fourth acquisition module, used to train the above-mentioned diagnostic item label model using sample treatment item data and second object label data to obtain a treatment item label model.

[0008] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described above.

[0009] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described above.

[0010] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the method described above when executed by a processor.

[0011] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0013] Figure 1 Schematically illustrates an exemplary system architecture to which the medical entity relationship determination method, model training method, and apparatus according to an embodiment of the present disclosure can be applied;

[0014] Figure 2 A flowchart schematically illustrates a method for determining medical entity relationships according to an embodiment of the present disclosure;

[0015] Figure 3 Schematically illustrates an example of determining the relationship between a target diagnosis item and a target treatment item based on a relationship graph and an entity attribute graph in conjunction with a label model according to an embodiment of the present disclosure;

[0016] Figure 4 The following schematically shows a flow chart of a model training method according to an embodiment of the present disclosure;

[0017] Figure 5An example diagram of a model training process according to an embodiment of the present disclosure is schematically shown;

[0018] Figure 6 A block diagram schematically illustrates a device for determining medical entity relationships according to an embodiment of the present disclosure;

[0019] Figure 7 A block diagram schematically illustrates a model training device according to an embodiment of the present disclosure; and

[0020] Figure 8 A block diagram of an electronic device suitable for implementing a method for determining medical entity relationships and a model training method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0021] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0022] The medical record homepage is a highly condensed version of the medical record information, which can reflect the comprehensive information of the medical record. The main diagnosis item is the core of the medical record homepage. The selection of the main diagnosis item needs to follow relevant principles. For example, the diagnosis item that poses a great threat to the user's health, consumes a lot of medical resources and has a long hospitalization time should be selected. The main diagnosis item can be determined by using Diagnosis Related Groups (DRGs). Therefore, the correct selection of the main diagnosis item is related to the standardization of medical resource utilization and the rationalization of charges. There is an association between diagnosis items and treatment items. Therefore, for example, at least one diagnosis item can be sorted according to the medical resources consumed by the treatment items associated with the diagnosis item, thereby determining the main diagnosis item from at least one diagnosis item. This shows that the determination of the relationship between diagnosis items and treatment items is related to the standardization of medical resource utilization and the rationalization of charges.

[0023] Furthermore, hospital performance can be evaluated based on primary diagnosis items, making the correct selection of primary diagnosis items even more crucial. Assisting doctors in selecting the correct primary diagnosis items for the medical record homepage is a pressing issue.

[0024] With the continuous development of artificial intelligence technology, it can be used to assist doctors in correctly selecting the primary diagnosis item on the medical record homepage. For example, if a diagnosis item is associated with a treatment cost item, at least one diagnosis item can be sorted based on the cost information of the treatment item associated with the diagnosis item, thereby determining the primary diagnosis item from the at least one diagnosis item. This improves the standardization of medical resource utilization and the rationalization of charges.

[0025] To determine the primary diagnosis using the aforementioned method based on the association between diagnosis and treatment, it is necessary to be able to reasonably predict the relationship between diagnosis and treatment. This can be achieved using a relationship graph. Knowledge graphs can be constructed by extracting the association between diagnosis and treatment using medical texts and medical records.

[0026] There is an intricate relationship between diagnostic items and treatment items, but the data that medical books and medical records can provide is relatively limited. Therefore, it is difficult to extract a large number of association relationships using limited medical books and medical records data, and it is difficult to achieve a high-precision output for the rationality detection of the relationship between diagnostic items and treatment items.

[0027] To this end, the embodiment of the present disclosure proposes an entity relationship determination scheme. That is, the target diagnosis item data of the target diagnosis item is processed using a diagnosis item label model to obtain a first object set corresponding to the target diagnosis item. The target treatment item data of the target treatment item is processed using a treatment item label model to obtain a second object set corresponding to the target treatment item. Based on the first object set and the second object set, the relationship between the target diagnosis item and the target treatment item is determined. The prediction of the association relationship between any diagnosis item and any treatment item is achieved, thereby realizing the rationality detection of the relationship between the diagnosis item and the treatment item. Figure 1 An exemplary system architecture to which the medical entity relationship determination method, model training method, and apparatus according to an embodiment of the present disclosure can be applied is schematically illustrated.

[0028] It should be noted that Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present disclosure may be applied, to help those skilled in the art understand the technical content of the present disclosure, but do not imply that the embodiments of the present disclosure may not be applied to other devices, systems, environments, or scenarios. For example, in another embodiment, an exemplary system architecture to which the method and apparatus for determining medical entity relationships may be applied may include a terminal device, but the terminal device may implement the method for determining medical entity relationships, the model training method, and the apparatus provided in the embodiments of the present disclosure without interacting with a server.

[0029] like Figure 1As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links.

[0030] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software (for example only).

[0031] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0032] The server 105 may be any type of server that provides various services, such as a background management server (for example only) that supports the content browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process received data such as user requests, and feed back processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0033] Server 105 can be a cloud server, also known as a cloud computing server or cloud host. It is a host product in the cloud computing service system. It solves the management difficulties and poor business scalability of traditional physical hosts and VPS services. Server 105 can also be a server in a distributed system or a server integrated with blockchain.

[0034] It should be noted that the method for determining and training medical entity relationships provided in the embodiments of the present disclosure can generally be executed by the server 105. Accordingly, the apparatus for determining and training medical entity relationships provided in the embodiments of the present disclosure can generally be set in the server 105. The method for determining and training medical entity relationships provided in the embodiments of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the apparatus for determining and training medical entity relationships provided in the embodiments of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0035] Alternatively, the medical entity relationship determination method and training method provided in the embodiments of the present disclosure may also be generally executed by the terminal device 101, 102, or 103. Accordingly, the medical entity relationship determination apparatus and training method provided in the embodiments of the present disclosure may also be provided in the terminal device 101, 102, or 103.

[0036] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0037] Figure 2 The flowchart of the method for determining medical entity relationships according to an embodiment of the present disclosure is schematically shown.

[0038] like Figure 2 As shown, the method 200 includes operations S210 to S230.

[0039] In operation S210, target diagnostic item data of a target diagnostic item is processed using a diagnostic item label model to obtain a first object set. The first object set represents a set of objects corresponding to the target diagnostic item.

[0040] In operation S220, the target treatment item data of the target treatment item is processed using the treatment item label model to obtain a second object set. The second object set represents a set of objects corresponding to the target treatment item.

[0041] Operation S230 : determining a relationship between a target diagnosis item and a target treatment item according to the first object set and the second object set.

[0042] According to an embodiment of the present disclosure, a diagnostic item label model can be used to predict the label corresponding to the diagnostic item. The diagnostic item label model can be used to characterize the correspondence between the diagnostic item and the label. The treatment item label model can be used to predict the label corresponding to the treatment item. The diagnostic item label model can be obtained by training a deep learning model using the first sample data. The treatment item label model can be obtained by training a deep learning model using the second sample data. In addition, the treatment item label model can also be obtained by training the diagnostic item label model using the second sample data.

[0043] According to an embodiment of the present disclosure, a deep learning model may include a pre-training model. For example, the pre-training model may include at least one of the following: an autoregressive pre-trained language model for unidirectional feature representation, an autoencoding pre-trained language model for bidirectional feature representation, and an autoregressive pre-trained language model for bidirectional feature representation. The autoregressive pre-trained language model for unidirectional feature representation may include at least one of the following: an ELMo (Embedding from Language Models) model and a GPT (Generative Pre-Training) model. The autoencoding pre-trained language model for bidirectional feature representation may include an ENRIE (Enhanced Language Representation with Informative Entities) model, a BERT (Bidirectional Encoder Representations from Transformers) model, a MASS (Masked Sequence to Sequence Pre-training for Language Generation) model, and a UniLM (Unified Language Model) model. The autoregressive pre-trained language model for bidirectional feature representation may include an XLNet model.

[0044] According to an embodiment of the present disclosure, both diagnostic items and medical items are medical entities. A diagnostic item may refer to a diagnosis result of a disease. For example, diagnostic items may include fractures, arthritis, type 2 diabetes, rheumatic heart disease, coronary heart disease, and viral colds. A treatment item (i.e., a treatment consumable item) may refer to an operation used to treat a disease. Treatment items may include non-universal treatment items and universal treatment items. For example, non-universal treatment items may include bone traction, plaster fixation, radiotherapy, chemotherapy, and surgical operations. Universal treatment items may include at least one of the following: non-invasive electrocardiogram monitoring, dynamic blood pressure monitoring, finger pulse oximetry monitoring, and finger pulse oxygen saturation monitoring.

[0045] According to embodiments of the present disclosure, diagnostic items may be associated with treatment consumable items. For example, the diagnostic item "fracture" may be associated with both the treatment items "bone traction" and "cast fixation." The association between diagnostic items and treatment items can be one-to-one or one-to-many, meaning that one diagnostic item may correspond to one or more treatment items.

[0046] According to an embodiment of the present disclosure, a label may refer to an object corresponding to a diagnostic item and a treatment item. The object may include at least one of the following: a system, a primary site, and a secondary site. A secondary site may refer to a secondary site corresponding to a primary site. A system may include at least one of the following: a musculoskeletal system, an immune system, a lymphatic system, a digestive system, a circulatory system, a nervous system, an endocrine system, a sensory system, and a respiratory system. In addition, a system may also include other systems in addition to the above. A primary site may include at least one of the following: a non-systemic site and a systemic site. A systemic site may refer to a site throughout the entire body. The primary sites and secondary sites may refer to the following Table 1. It should be noted that Table 1 is only an exemplary description. In addition to the primary sites and secondary sites shown in Table 1, other primary sites and secondary sites corresponding to the primary sites may also be included. In addition, there may also be a tertiary site corresponding to the secondary site. For example, the secondary site "joint" may include at least one of the following: a wrist joint, a finger joint, and a knee joint.

[0047]

[0048] Table 1

[0049] According to an embodiment of the present disclosure, the association relationship between diagnosis items and objects (i.e., labels) can be found in Table 2. The association relationship between treatment items and objects (i.e., labels) can be found in Table 3. It should be noted that Tables 1 and 2 are only exemplary.

[0050] Diagnostic items Object premature ventricular contractions Circulatory system, heart, chest Thyroid tumors Endocrine system, thyroid, neck Gastric cancer Digestive system, stomach Dislocation of the hand joint musculoskeletal system, upper limbs, joints, hands Acute pharyngitis Respiratory system, throat …… ……

[0051] Table 2

[0052] Treatment Object Manual reduction of old fractures musculoskeletal system, bones Gastric tube placement Digestive system, stomach Tooth defect filling Teeth Bone marrow aspiration musculoskeletal system, bones Wrist traction locomotor system, wrist, joints Skin traction musculoskeletal system, bones, and joints Massage treatment for cervical spondylosis Neck and cervical spine …… ……

[0053] Table 3

[0054] According to an embodiment of the present disclosure, the first object set may represent an object set corresponding to the target diagnosis item. The first object set may include one or more objects. For example, if the target diagnosis item is "supracondylar fracture of femur", the first object set corresponding to "supracondylar fracture of femur" may include "lower limb" and "knee joint".

[0055] According to an embodiment of the present disclosure, the second object set may represent an object set corresponding to the target treatment item. The second object set may include one or more objects. For example, if the target treatment item is "massage treatment for dislocated joints of the limbs," the second object set corresponding to "massage treatment for dislocated joints of the limbs" may include "joints." If the target treatment item is "cast fixation of lower limbs," the second object set corresponding to "cast fixation of lower limbs" may include "joints."

[0056] According to an embodiment of the present disclosure, the target diagnostic item data can be input into the diagnostic item label model to obtain a first object set. The target treatment item data can be input into the treatment item label model to obtain a second object set. After obtaining the first object set and the second object set, the relationship between the target diagnostic item and the target treatment item can be determined based on the first object set and the second object set. The relationship between the target diagnostic item and the target treatment item can include having an association relationship and not having an association relationship. For example, it can be determined whether the first object set and the second object set have an intersection, and based on the determination result, the relationship between the target diagnostic item and the target treatment is determined.

[0057] According to an embodiment of the present disclosure, a diagnostic item label model is used to process the target diagnostic item data of a target diagnostic item to obtain a first set of objects corresponding to the target diagnostic item. A therapeutic item label model is used to process the target therapeutic item data of a target therapeutic item to obtain a second set of objects corresponding to the target therapeutic item. Based on the first object set and the second object set, the relationship between the target diagnostic item and the target therapeutic item is determined. This enables the prediction of the association relationship between any diagnostic item and any therapeutic item, thereby enabling the rationality detection of the relationship between the diagnostic item and the therapeutic item.

[0058] According to an embodiment of the present disclosure, operation S230 may include the following operations.

[0059] If it is determined that the first object set and the second object set have an intersection, it is determined that the target diagnosis item and the target treatment item have an association relationship. If it is determined that the first object set and the second object set do not have an intersection, it is determined that the target diagnosis item and the target treatment item do not have an association relationship.

[0060] According to an embodiment of the present disclosure, it is determined whether there is an intersection between the first object set and the second object set, that is, whether the first object set and the second object set have the same objects. If it is determined that the first object set and the second object set have the same objects, it can be determined that there is an intersection between the first object set and the second object set, thereby determining that there is an association relationship between the target diagnosis item and the target treatment item. If it is determined that the first object set and the second object set do not have the same objects, it can be determined that there is no intersection between the first object set and the second object set, thereby determining that there is no association relationship between the target diagnosis item and the target treatment item.

[0061] For example, the target diagnosis item is "supracondylar fracture of the femur." The target treatment item is "cast fixation of the lower limb." The first object set corresponding to "supracondylar fracture of the femur" includes "lower limb" and "knee joint." The second object set corresponding to "cast fixation of the lower limb" includes "lower limb." Because both the first object set and the second object set include "lower limb," it can be determined that there is an intersection between the first object set and the second object set. Therefore, it can be determined that there is an association relationship between "supracondylar fracture of the femur" and "cast fixation of the lower limb."

[0062] According to an embodiment of the present disclosure, in order to further improve the accuracy of rationality detection, with respect to determining whether the target diagnostic item and the target treatment item have an association relationship, it can be determined based on the first ratio, the second ratio and the first predetermined ratio threshold, based on whether the same objects exist in the first object set and the second object set, that is, whether the first object set and the second object set have an intersection. The first predetermined ratio threshold can be configured according to actual business needs and is not limited here. The first ratio can represent the ratio of the number of objects included in the intersection to the number included in the first object set. The second ratio can represent the ratio of the number of objects included in the intersection to the number included in the second object set. If both the first ratio and the second ratio are greater than the first predetermined ratio threshold, it can be determined that the target diagnostic item and the target treatment item have an association relationship. If at least one of the first ratio and the second ratio is less than the first predetermined ratio threshold, it can be determined that there is no association relationship between the target diagnostic item and the target treatment item.

[0063] According to an embodiment of the present disclosure, operation S210 may include the following operations.

[0064] When it is determined based on the relationship map and the entity attribute map that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data of the target diagnosis item is processed using the diagnosis item label model to obtain a first object set.

[0065] According to an embodiment of the present disclosure, operation S210 may include the following operations.

[0066] When it is determined based on the entity attribute map that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data of the target diagnosis item is processed using the diagnosis item label model to obtain a first object set.

[0067] According to an embodiment of the present disclosure, operation S210 may include the following operations.

[0068] When it is determined based on the relationship graph that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data of the target diagnosis item is processed using the diagnosis item label model to obtain a first object set.

[0069] According to embodiments of the present disclosure, a knowledge graph may include a relationship graph, an entity attribute graph, or both. A knowledge graph may be constructed based on the associations between diagnosis items and treatment items extracted from medical books and medical records. A relationship graph may be used to characterize the associations between diagnosis items and treatment items.

[0070] Before using the diagnosis item labeling model and the treatment item labeling model for processing, a knowledge graph can be used to determine whether the target diagnosis item and the target treatment item are associated. If the knowledge graph determines that the target diagnosis item and the target treatment item are not associated, the target diagnosis item data and the target treatment item data can be processed using the diagnosis item labeling model and the treatment item labeling model, respectively.

[0071] For example, if the knowledge graph includes a relationship graph, whether a target diagnosis and a target treatment have an association relationship can be determined based on the relationship graph. If the relationship graph determines that the target diagnosis and the target treatment do not have an association relationship, the target diagnosis data and the target treatment data can be processed separately based on the diagnosis label model and the treatment label model. Determining whether a target diagnosis and the target treatment have an association relationship based on the relationship graph can include: determining, based on the relationship graph and the target diagnosis data, a second set of associated diagnosis items corresponding to the target diagnosis. Determining, based on the relationship graph and the second set of associated diagnosis item data corresponding to the second set of associated treatment items. Determining a third similarity between the target treatment data and each of the multiple associated treatment item data included in the second set of associated treatment item data in the second set of associated treatment items to obtain multiple third similarities. If all of the multiple third similarities are determined to be less than a predetermined similarity threshold, it is determined that there is no association relationship between the target diagnosis and the target treatment. If at least one third similarity is determined to be less than the predetermined similarity threshold, it is determined that there is an association relationship between the target diagnosis and the target treatment. Determining, based on the relationship graph and according to the target diagnostic item data, a set of second associated diagnostic items corresponding to the target diagnostic item may include: determining a fourth similarity between the target diagnostic item data and each of a plurality of second candidate diagnostic item data included in the relationship graph to obtain a plurality of fourth similarities; and determining, based on the plurality of fourth similarities, a set of second associated diagnostic items corresponding to the target diagnostic item from a plurality of second candidate diagnostic items corresponding to the plurality of second candidate diagnostic item data.

[0072] For example, if the knowledge graph includes an entity attribute graph, the entity attribute graph can be used to determine whether the target diagnosis item and the target treatment item have an association relationship. If the entity attribute graph determines that the target diagnosis item and the target treatment item do not have an association relationship, the target diagnosis item data and the target treatment item data can be processed separately based on the diagnosis item labeling model and the treatment item labeling model.

[0073] For example, if the knowledge graph includes a relationship graph and an entity attribute graph, then the relationship graph can be used to determine whether the target diagnosis item and the target treatment item have an association relationship. If the relationship graph determines that the target diagnosis item and the target treatment item do not have an association relationship, then the entity attribute graph can be used to determine whether the target diagnosis item and the target treatment item have an association relationship. If the entity attribute graph determines that the target diagnosis item and the target treatment item do not have an association relationship, then the target diagnosis item data and the target treatment item data can be processed separately based on the diagnosis item label model and the treatment item label model.

[0074] According to an embodiment of the present disclosure, the entity attribute graph may include a first entity attribute graph and a second entity attribute graph.

[0075] According to an embodiment of the present disclosure, a first entity attribute map may be constructed based on the association between diagnosis items and medical conditions extracted from medical books and medical records. A second entity attribute map may be constructed based on the association between treatment items and medical records extracted from medical books and medical records. The first entity attribute map may represent the association between diagnosis items and medical conditions. The second entity attribute map may represent the association between treatment properties and medical conditions.

[0076] According to an embodiment of the present disclosure, the above-mentioned method for determining medical entity relationships may further include the following operations.

[0077] Based on the first entity attribute map and the target diagnosis item data, a first set of associated diagnosis items corresponding to the target diagnosis item is determined. Based on the first entity attribute map and the first set of associated diagnosis item data of the first associated diagnosis item set, a first set of disease conditions corresponding to the first set of associated diagnosis items is determined. Based on the second entity attribute map and the target treatment item data, a first set of associated treatment items corresponding to the target treatment item is determined. Based on the second entity attribute map and the first set of associated treatment item data of the first associated treatment item set, a second set of disease conditions corresponding to the first set of associated treatment items is determined. If it is determined that the first set of disease conditions and the second set of disease conditions do not intersect, it is determined that there is no association relationship between the target diagnosis item and the target treatment item.

[0078] According to an embodiment of the present disclosure, a first associated diagnostic item set may include one or more first associated diagnostic items. A first associated diagnostic item may refer to a diagnostic item that has an associated relationship with a target diagnostic item. A first condition set may include one or more first conditions. A first condition may refer to a condition that has an associated relationship with a target diagnostic item. A first associated treatment item set may include one or more first associated treatment items. A first associated treatment item may refer to a treatment item that has an associated relationship with a target treatment item. A second condition set may include one or more second conditions. A second condition may refer to a condition that has an associated relationship with a target treatment item. The associated relationship may be characterized by similarity.

[0079] According to an embodiment of the present disclosure, a first associated diagnostic item data set corresponding to the target diagnostic item data can be first determined from the first entity attribute map, and the diagnostic items corresponding to the first associated diagnostic item data set can be determined as the first associated diagnostic item set. Then, a first medical condition data set corresponding to the first associated diagnostic item data set can be determined from the first entity attribute map, and the medical conditions corresponding to the first medical condition data set can be determined as the first medical condition set.

[0080] According to an embodiment of the present disclosure, a first associated treatment item data set corresponding to the target treatment item data can be first determined from the second entity attribute graph, and the treatment items corresponding to the first associated treatment item data set can be determined as the first associated treatment item set. Then, a second condition data set corresponding to the first associated treatment item data set can be determined from the second entity attribute graph, and the conditions corresponding to the second condition data set can be determined as the second condition set.

[0081] According to an embodiment of the present disclosure, after determining a first condition set and a second condition set, it can be determined whether the first condition set and the second condition set have an intersection, that is, whether the first condition set and the second condition set have the same condition. If it is determined that the first condition set and the second condition set have the same condition, it can be determined that the first condition set and the second condition set have an intersection, thereby determining that there is an association relationship between the target diagnosis item and the target treatment item. If it is determined that the first condition set and the second condition set do not have the same condition, it can be determined that there is no intersection between the first condition set and the second condition set, thereby determining that there is no association relationship between the target diagnosis item and the target treatment item.

[0082] For example, the target diagnosis item is "supracondylar fracture of the femur". The target treatment item is "bone traction". The first condition set corresponding to "supracondylar fracture of the femur" includes "lower limb tension", "lower limb pain", "knee joint swelling", "lower limb edema", "limited knee joint movement", "joint dysfunction" and "knee flexion deformity". The second condition set corresponding to "bone traction" includes "lower limb pain", "lower limb tension" and "joint dysfunction". Since the first condition set and the second condition set both include "lower limb pain", "lower limb tension" and "joint dysfunction", it can be determined that there is an intersection between the first condition set and the second condition set, and thus, it can be determined that there is a correlation between "supracondylar fracture of the femur" and "bone traction".

[0083] According to an embodiment of the present disclosure, in order to further improve the accuracy of rationality detection, in order to determine whether the target diagnosis item and the target treatment item have an association relationship, it can be determined based on the third ratio, the fourth ratio and the second predetermined ratio threshold on the basis of whether the first condition set and the second condition set have the same condition, that is, whether the first condition set and the second condition set have an intersection. The second predetermined ratio threshold can be configured according to actual business needs and is not limited here. The third ratio can represent the ratio of the number of conditions included in the intersection to the number included in the first condition set. The fourth ratio can represent the ratio of the number of conditions included in the intersection to the number included in the second condition set. If both the third ratio and the fourth ratio are greater than the second predetermined ratio threshold, it can be determined that the target diagnosis item and the target treatment item have an association relationship. If at least one of the third ratio and the fourth ratio is less than the second predetermined ratio threshold, it can be determined that there is no association relationship between the target diagnosis item and the target treatment item.

[0084] According to an embodiment of the present disclosure, based on the first entity attribute graph and according to the target diagnostic item data, determining a first associated diagnostic item set corresponding to the target diagnostic item may include the following operations.

[0085] Determine a first similarity between the target diagnostic item data and each of the plurality of first candidate diagnostic item data included in the first entity attribute graph to obtain a plurality of first similarities. Based on the plurality of first similarities, determine a first associated diagnostic item set corresponding to the target diagnostic item from the plurality of first candidate diagnostic items corresponding to the plurality of first candidate diagnostic item data.

[0086] According to embodiments of the present disclosure, similarity can represent the degree of similarity between two objects. Similarity can be configured based on actual business needs and is not limited here. For example, similarity can include set similarity, cosine similarity, Pearson correlation coefficient, Euclidean distance, or Jaccard distance. Set similarity can be represented by Dice distance.

[0087] According to an embodiment of the present disclosure, the first similarity may represent the degree of similarity between the target diagnostic item corresponding to the target diagnostic item data and the first candidate diagnostic item corresponding to the first candidate diagnostic item data. For example, a larger value of the first similarity may represent a greater degree of similarity between the target diagnostic item and the first candidate diagnostic item. Conversely, a smaller value may represent a smaller degree of similarity between the target diagnostic item and the first candidate diagnostic item.

[0088] According to an embodiment of the present disclosure, determining a set of first associated diagnostic items corresponding to a target diagnostic item from a plurality of first candidate diagnostic items corresponding to a plurality of first candidate diagnostic item data based on a plurality of first similarities may include: sorting the plurality of first similarities to obtain a first sorting result. Based on the first sorting result, determining a set of first associated diagnostic items corresponding to a target diagnostic item from a plurality of first candidate diagnostic items. The sorting may be in ascending order of the first similarities or in descending order of the first similarities. For example, if the sorting is in ascending order of the first similarities, the first candidate diagnostic items corresponding to a predetermined number of first similarities at the end of the sorting may all be determined as first associated diagnostic items.

[0089] Alternatively, a set of first associated diagnostic items corresponding to the target diagnostic item can be determined from multiple first candidate diagnostic items based on multiple first similarities and a first predetermined similarity threshold. The first predetermined similarity threshold can be configured according to actual business needs and is not limited here. For example, the first predetermined similarity threshold is 0.9. For example, for each first candidate diagnostic item in the multiple first candidate diagnostic items, if it is determined that the first similarity corresponding to the first candidate diagnostic item is greater than or equal to the first predetermined similarity threshold, the first candidate diagnostic item can be determined as the first associated diagnostic item.

[0090] According to an embodiment of the present disclosure, based on the second entity attribute map and according to the target treatment item data, determining a first associated treatment item set corresponding to the target treatment item may include the following operations.

[0091] Determine a second similarity between the target treatment item data and each of the plurality of candidate treatment item data included in the second entity attribute graph to obtain a plurality of second similarities. Determine a first associated treatment item set corresponding to the target treatment item from the plurality of candidate treatment items corresponding to the plurality of candidate treatment item data based on the plurality of second similarities.

[0092] According to an embodiment of the present disclosure, the second similarity may represent the degree of similarity between the target treatment item corresponding to the target treatment item data and the candidate treatment item corresponding to the candidate treatment item data. For example, a larger value of the second similarity may represent a greater degree of similarity between the target treatment item and the candidate treatment item. Conversely, a smaller value may represent a smaller degree of similarity between the target treatment item and the candidate treatment item.

[0093] According to an embodiment of the present disclosure, determining a first set of associated treatment items corresponding to a target treatment item from a plurality of candidate treatment items corresponding to a plurality of candidate treatment item data based on a plurality of second similarities may include: sorting the plurality of second similarities to obtain a second sorting result. Based on the second sorting result, determining a first set of associated treatment items corresponding to the target treatment item from a plurality of candidate treatment items. Alternatively, a first set of associated treatment items corresponding to the target treatment item may be determined from a plurality of candidate treatment items based on a plurality of second similarities and a second predetermined similarity threshold. The second predetermined similarity threshold may be configured according to actual business needs and is not limited here. For example, the second predetermined similarity threshold is 0.9. For example, for each candidate treatment item among a plurality of candidate treatment items, when it is determined that the second similarity corresponding to the candidate treatment item is greater than or equal to the second predetermined similarity threshold, the candidate treatment item may be determined as the first associated treatment item.

[0094] According to an embodiment of the present disclosure, based on the second entity attribute map and according to the target treatment item data, determining a first associated treatment item set corresponding to the target treatment item may include the following operations.

[0095] Determine a second similarity between the target treatment item data and each of the plurality of candidate treatment item data included in the second entity attribute graph to obtain a plurality of second similarities. Determine a first associated treatment item set corresponding to the target treatment item from the plurality of candidate treatment items corresponding to the plurality of candidate treatment item data based on the plurality of second similarities.

[0096] According to an embodiment of the present disclosure, determining a first associated diagnostic item set corresponding to a target diagnostic item based on the first entity attribute graph and according to target diagnostic item data may include the following operations.

[0097] When it is determined based on the relationship map that there is no association relationship between the target diagnosis item and the target treatment item, a first associated diagnosis item set corresponding to the target diagnosis item is determined based on the first entity attribute map and the target diagnosis item data.

[0098] According to an embodiment of the present disclosure, before processing using the first entity attribute map and the second entity attribute map, it is possible to first determine whether there is an association relationship between the target diagnosis item and the target treatment item based on the relationship map. If it is determined based on the relationship map that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data and the target treatment item data can be processed based on the first entity attribute map and the second entity attribute map.

[0099] According to an embodiment of the present disclosure, the above-mentioned method for determining medical entity relationships may further include the following operations.

[0100] Based on the relationship graph, a second set of associated diagnostic items corresponding to the target diagnostic item is determined based on the target diagnostic item data. Based on the relationship graph, a second set of associated treatment items corresponding to the second associated diagnostic item set is determined based on the second associated diagnostic item data set of the second associated diagnostic item set. A third similarity is determined between the target treatment item data and each of the plurality of associated treatment item data included in the second associated treatment item data set of the second associated treatment item set, to obtain a plurality of third similarities. If it is determined that the plurality of third similarities are all less than a predetermined similarity threshold, it is determined that there is no association relationship between the target diagnostic item and the target treatment item.

[0101] According to an embodiment of the present disclosure, the second associated diagnostic item set may include one or more second associated diagnostic items. A second associated diagnostic item may refer to a diagnostic item associated with a target diagnostic item. The second associated treatment item set may include one or more second associated treatment items. A second associated treatment item may refer to a treatment item associated with a target treatment item. The association relationship may be characterized by similarity.

[0102] According to an embodiment of the present disclosure, a second associated diagnostic item data set corresponding to the target diagnostic item data can be first determined from the relationship graph, and the diagnostic items corresponding to the second associated diagnostic item data set can be determined as the second associated diagnostic item set. A second associated treatment item data set corresponding to the second associated diagnostic item data set can then be determined from the relationship graph, and the treatment items corresponding to the second associated treatment item data set can be determined as the second associated treatment item set.

[0103] According to an embodiment of the present disclosure, after determining the second associated treatment item set, the third similarity between the target treatment item data and each of the multiple second associated treatment item data included in the second associated treatment item data set can be determined to obtain multiple third similarities. If it is determined that the multiple third similarities are all less than the third predetermined similarity threshold, it can be determined that there is no association relationship between the target diagnosis item and the target treatment item. If it is determined that there is at least one third similarity that is greater than or equal to the third predetermined similarity threshold, it can be determined that there is an association relationship between the target diagnosis item and the target treatment item. The third predetermined similarity threshold can be configured according to actual business needs and is not limited here. For example, the third predetermined similarity threshold is 0.9.

[0104] According to an embodiment of the present disclosure, determining a second associated diagnostic item set corresponding to a target diagnostic item based on a relationship graph and target diagnostic item data may include the following operations.

[0105] Determine a fourth similarity between the target diagnostic item data and each of the plurality of second candidate diagnostic item data included in the relationship graph to obtain a plurality of fourth similarities. Determine a second associated diagnostic item set corresponding to the target diagnostic item from the plurality of second candidate diagnostic items corresponding to the plurality of second candidate diagnostic item data based on the plurality of fourth similarities.

[0106] According to an embodiment of the present disclosure, the fourth similarity may represent the degree of similarity between the target diagnostic item corresponding to the target diagnostic item data and the second candidate diagnostic item corresponding to the second candidate diagnostic item data. For example, a larger value of the fourth similarity may represent a greater degree of similarity between the target diagnostic item and the second candidate diagnostic item. Conversely, a smaller value may represent a smaller degree of similarity between the target diagnostic item and the second candidate diagnostic item.

[0107] According to an embodiment of the present disclosure, determining a set of second associated diagnostic items corresponding to a target diagnostic item from a plurality of second candidate diagnostic items corresponding to a plurality of second candidate diagnostic item data based on a plurality of fourth similarities may include: sorting the plurality of fourth similarities to obtain a third sorting result. Based on the third sorting result, determining a set of second associated diagnostic items corresponding to a target diagnostic item from a plurality of second candidate diagnostic items. Alternatively, a set of second associated diagnostic items corresponding to a target diagnostic item may be determined from a plurality of second candidate diagnostic items based on a plurality of fourth similarities and a fourth predetermined similarity threshold. The fourth predetermined similarity threshold may be configured according to actual business needs and is not limited here. For example, the fourth predetermined similarity threshold is 0.9. For example, for each second candidate diagnostic item in a plurality of second candidate diagnostic items, when it is determined that the fourth similarity corresponding to the second candidate diagnostic item is greater than or equal to the fourth predetermined similarity threshold, the second candidate diagnostic item may be determined as a second associated diagnostic item.

[0108] According to an embodiment of the present disclosure, the above-mentioned method for determining medical entity relationships may further include the following operations.

[0109] When it is determined that the target treatment item is a universal treatment item, it is determined that there is an association relationship between the target diagnosis item and the target treatment item.

[0110] According to an embodiment of the present disclosure, before determining whether a target diagnosis item and a target treatment item have an association relationship based on a relationship map, it is possible to determine whether the target treatment item is a universal treatment item. If the target treatment item is determined to be a universal treatment item, it is possible to determine whether the target diagnosis item and the target treatment item have an association relationship. If the target treatment item is determined not to be a universal treatment item, the relationship between the target diagnosis item and the target treatment item can be determined based on the relationship map.

[0111] According to an embodiment of the present disclosure, the first similarity and the second similarity may both be set similarities.

[0112] According to an embodiment of the present disclosure, the third similarity and the fourth similarity may both be set similarities.

[0113] According to the embodiments of the present disclosure, set similarity can achieve character-level similarity determination, thereby improving the accuracy of predicting the association relationship between diagnosis items and treatment items.

[0114] According to an embodiment of the present disclosure, when it is determined that a target diagnosis item and a target treatment item are associated, a primary diagnosis item is determined from the target diagnosis item and other diagnosis items corresponding to the other treatment items based on the medical resources consumed by each of the target treatment item and the other treatment items. The target diagnosis item and the other diagnosis items are diagnosis items for the same user.

[0115] According to the embodiments of the present disclosure, the medical entity relationship determination solution provided by the embodiments of the present disclosure can be used to determine the relationship between other diagnosis items and other treatment items, which will not be described in detail here. Medical resources may include medical supplies and medical personnel.

[0116] According to an embodiment of the present disclosure, the financial consumption values ​​of the target treatment item and the other treatment items can be determined based on the medical resources consumed by each of the target treatment item and the other treatment items. The target diagnosis items corresponding to the target treatment item and the other diagnosis items corresponding to the other treatment items are sorted based on their respective financial consumption values ​​to obtain a fourth sorting result. Based on the fourth sorting result, a primary diagnosis item is determined from the target diagnosis item and the other diagnosis items.

[0117] According to the embodiments of the present disclosure, the above can assist doctors in correctly selecting the main diagnosis items on the medical record homepage.

[0118] Based on the above, the entity relationship determination scheme described in the embodiments of the present disclosure may include a scheme for determining the relationship between the target diagnosis item and the target treatment item based on a label model, or a scheme for determining the relationship between the target diagnosis item and the target treatment item based on a combination of a knowledge graph and a label model. The knowledge graph may include at least one of the following: a relationship graph and an entity attribute graph, wherein the entity attribute graph includes a first entity attribute graph and a second entity attribute graph.

[0119] The scheme for determining the relationship between the target diagnosis item and the target treatment item based on the knowledge graph and the label model may include the following: a scheme for determining the relationship between the target diagnosis item and the target treatment item based on the relationship graph and the label model, a scheme for determining the relationship between the target diagnosis item and the target treatment item based on the entity attribute graph (i.e., including the first entity attribute graph and the second entity attribute graph) and the label model, and a scheme for determining the relationship between the target diagnosis item and the target treatment item based on the relationship graph and the entity attribute graph and the label model.

[0120] Reference below Figure 3, the method for determining the medical entity relationship according to the embodiment of the present disclosure is further explained in combination with specific embodiments.

[0121] Figure 3 An example diagram of determining the relationship between a target diagnosis item and a target treatment item based on a relationship graph and an entity attribute graph in conjunction with a label model according to an embodiment of the present disclosure is schematically shown.

[0122] like Figure 3 As shown in FIG300 , the relationship between the target diagnosis item and the target treatment item is determined based on the relationship graph 300-1. That is, a fourth similarity 304 is determined between the target diagnosis item data 301 and each of the plurality of second candidate diagnosis item data 302 included in the relationship graph, resulting in a plurality of fourth similarities 304. Based on the plurality of fourth similarities 304, a second associated diagnosis item set corresponding to the target diagnosis item is determined from the plurality of second candidate diagnosis items corresponding to the second candidate diagnosis item data set 305.

[0123] A third similarity between the target treatment item data 306 and each of the plurality of associated treatment item data included in the second associated treatment item data set 305 is determined to obtain a plurality of third similarities.

[0124] Are all the third similarities less than a predetermined similarity threshold 307? If not, it is determined that the target diagnosis item and the target treatment item have an association relationship 308. If so, the relationship between the target diagnosis item and the target treatment item is determined based on the first entity attribute map and the second entity attribute map.

[0125] The relationship between the target diagnosis item and the target treatment item is determined based on the entity attribute map (i.e., including the first entity attribute map and the second entity attribute map) 300-2. That is, the first similarity 310 between the target diagnosis item data 301 and each of the multiple first candidate diagnosis item data 309 included in the first entity attribute map is determined, and multiple first similarities 310 are obtained. Based on the multiple first similarities 310, a first associated diagnosis item set corresponding to the target diagnosis item is determined from the multiple first candidate diagnosis items corresponding to the multiple first candidate diagnosis item data 309. Based on the first entity attribute map, a first condition set 312 corresponding to the first associated diagnosis item set is determined based on the first associated diagnosis item data set 311 of the first associated diagnosis item set.

[0126] A second similarity 314 is determined between the target treatment item data 306 and each of the plurality of candidate treatment item data 313 included in the second entity attribute graph, resulting in a plurality of second similarities 314. Based on the plurality of second similarities 314, a first set of associated treatment items corresponding to the target treatment item is determined from the plurality of candidate treatment items corresponding to the plurality of candidate treatment item data 313. Based on the second entity attribute graph, a second set of conditions 316 corresponding to the first set of associated treatment items is determined based on the first set of associated treatment item data 315.

[0127] Do the first condition set 312 and the second condition set 316 intersect 317? If so, determine that the target diagnosis item and the target treatment item have an association relationship 308. If not, determine the relationship between the target diagnosis item and the target treatment item based on the label model.

[0128] The relationship between the target diagnosis item and the target treatment item is determined based on the label model 300-3. That is, the target diagnosis item data 301 is input into the diagnosis item label model 318 to obtain a first object set 319. The target treatment item data 306 is input into the treatment item label model 320 to obtain a second object set 321.

[0129] Do the first object set 319 and the second object set 321 intersect 322? If so, it is determined that there is an association relationship between the target diagnosis item and the target treatment item 308. If not, it is determined that there is no association relationship between the target diagnosis item and the target treatment item 323.

[0130] When evaluating the entity relationship determination scheme described in the embodiment of the present disclosure, the accuracy, recall rate and F1 score can all reach above 0.91.

[0131] Figure 4 The flowchart of the model training method according to an embodiment of the present disclosure is schematically shown.

[0132] like Figure 4 As shown, the method 400 includes operations S410 to S420.

[0133] In operation S410, a predetermined model is trained using sample diagnosis item data and first object label data to obtain a diagnosis item label model.

[0134] In operation S420 , a diagnosis item label model is trained using the sample treatment item data and the second object label data to obtain a treatment item label model.

[0135] According to an embodiment of the present disclosure, the first object tag and the second object tag may each include at least one of the following: a system tag, a primary site tag, and a secondary site tag. The secondary site tag may refer to a secondary site tag corresponding to the primary site tag. The system tag may include at least one of the following: a musculoskeletal system tag, an immune system tag, a lymphatic system tag, a digestive system tag, a circulatory system tag, a nervous system tag, an endocrine system tag, a sensory system tag, and a respiratory system tag. In addition, the system tag may also include other system tags in addition to the above. The primary site tag may include at least one of the following: a non-whole body site tag and a whole body site tag. The whole body site tag may refer to a site tag throughout the whole body. The primary site tags and the secondary site tags can be found in Table 1 shown above and will not be repeated here.

[0136] According to an embodiment of the present disclosure, the predetermined model may be a deep learning model. For example, the predetermined model may include a pre-trained language module, a GRU (i.e., a gated recurrent unit), a fully connected layer, and an activation function layer.

[0137] According to an embodiment of the present disclosure, the sample diagnosis item data has corresponding first object label data. The association relationship between the sample diagnosis item data and the first object label data can be established based on system anatomy and the opinions of medical experts. Similarly, the sample treatment item data has corresponding second object label data. The association relationship between the sample treatment item data and the second object label data can also be established based on system anatomy and the opinions of medical experts.

[0138] According to an embodiment of the present disclosure, sample diagnostic item data can be input into a predetermined model to obtain a first object prediction result. The first object prediction result and first sample label data are input into a first loss function to obtain a first output value. Model parameters of the predetermined model are adjusted based on the first output value until a predetermined condition is satisfied. The predetermined model obtained when the predetermined condition is satisfied is determined as the diagnostic item label model.

[0139] According to an embodiment of the present disclosure, sample treatment item data can be input into a diagnosis item label model to obtain a second object prediction result. The second object prediction result and the second sample label data are input into a loss function to obtain a second output value. The model parameters of the diagnosis item label model are adjusted based on the second output value to obtain a treatment item label model.

[0140] Based on the above content, a diagnosis item label model and a treatment item label model can be obtained.

[0141] According to the embodiments of the present disclosure, the diagnostic item label model and the treatment item label model obtained based on the above-mentioned method can realize the prediction of the association relationship between any diagnostic item and any treatment item. The accuracy and recall rate of the diagnostic item label model and the treatment item label model can both reach above 0.95. The coverage rate of the diagnostic item label model for diagnostic items can reach above 0.98. The coverage rate of the treatment item label model for treatment items can reach above 0.9. Based on the above-mentioned implementation, the association prediction of the relationship map or entity attribute map between the diagnostic item and the treatment item is not established, thereby further realizing the rationality detection of the relationship between the diagnostic item and the treatment item.

[0142] According to an embodiment of the present disclosure, the first object label data includes system label data and multi-level part label data.

[0143] According to an embodiment of the present disclosure, operation S410 may include the following operations.

[0144] The sample diagnostic item data is encoded to obtain a sample diagnostic item vector. The sample diagnostic item vector is fully connected to obtain a first sample diagnostic item vector and a second sample diagnostic item vector. Based on the sample diagnostic item vector, the first sample diagnostic item vector, and the second sample diagnostic item vector, a system prediction result and a multi-level part prediction result are obtained. The system label data, the system prediction result, the multi-level part label data, and the multi-level part prediction result are used to train a predetermined model to obtain a diagnostic item label model.

[0145] According to an embodiment of the present disclosure, system label data and system prediction results can be input into a first loss function to obtain a first output value. Multi-level part label data and multi-level part prediction results can be input into the first loss function to obtain a second output value. Model parameters of a predetermined model are adjusted based on the first and second output values ​​until predetermined conditions are met. The predetermined model obtained when the predetermined conditions are met is determined as the diagnostic item label model.

[0146] According to an embodiment of the present disclosure, the multi-level part label data includes first-level part label data and second-level part label data. The first part label data may include non-whole-body part label data and whole-body part label data.

[0147] According to an embodiment of the present disclosure, the multi-level prediction results include a first-level part prediction result and a second-level part prediction result. The first-level part prediction result may include a non-whole body part prediction result and a whole body part prediction result.

[0148] According to an embodiment of the present disclosure, obtaining a system prediction result and a multi-level part prediction result based on a sample diagnosis item vector, a first sample diagnosis item vector, and a second sample diagnosis item vector may include the following operations.

[0149] Based on the sample diagnostic item vector and the first sample diagnostic item vector, a third sample diagnostic item vector is obtained. Based on the sample diagnostic item vector, the first sample diagnostic item vector, and the second sample diagnostic item vector, a fourth sample diagnostic item vector is obtained. Based on the first sample diagnostic item vector, a non-whole body part prediction result is obtained. Based on the second sample diagnostic item vector, a whole body part prediction result is obtained. Based on the third sample diagnostic item vector, a secondary part prediction result is obtained. Based on the fourth sample diagnostic item vector, a system prediction result is obtained.

[0150] According to an embodiment of the present disclosure, obtaining a system prediction result based on the fourth sample diagnosis item vector may include: performing a full connection process on the fourth sample diagnosis item vector to obtain a fifth sample diagnosis item vector; and performing an activation process on the fifth sample diagnosis item vector to obtain a system part prediction result.

[0151] According to an embodiment of the present disclosure, obtaining a non-whole-body part prediction result based on the first sample diagnostic item vector may include: performing a full-connection process on the first sample diagnostic item vector to obtain a sixth sample diagnostic item vector; and performing an activation process on the sixth sample diagnostic item vector to obtain a non-whole-body part prediction result.

[0152] According to an embodiment of the present disclosure, obtaining a whole-body part prediction result based on the second sample diagnostic item vector may include: performing a full-connection process on the second sample diagnostic item vector to obtain a seventh sample diagnostic item vector; and performing an activation process on the seventh sample diagnostic item vector to obtain a whole-body part prediction result.

[0153] According to an embodiment of the present disclosure, obtaining a secondary part prediction result based on the third sample diagnostic item vector may include: performing a full connection process on the third sample diagnostic item vector to obtain an eighth sample diagnostic item vector; and performing an activation process on the eighth sample diagnostic item vector to obtain a secondary part prediction result.

[0154] Reference below Figure 5 , the model training method according to the embodiment of the present disclosure is further explained in combination with specific embodiments.

[0155] Figure 5 An example diagram of a model training process according to an embodiment of the present disclosure is schematically shown.

[0156] like Figure 5 As shown in 500, the predetermined model includes a pre-trained language module 502, a gated recurrent unit 503, multiple fully connected layers, and multiple activation layers. The multiple fully connected layers include fully connected layers 505, 510, 514, 518, and 522. The multiple activation layers include activation layers 512, 516, 520, and 524.

[0157] The sample diagnosis item data 501 is processed using the pre-trained language module 502 and the gated recurrent unit 503 to obtain a sample diagnosis item vector 504. The sample diagnosis item vector 504 is input into the fully connected layer 505 to obtain a first sample diagnosis item vector 506 and a second sample diagnosis item vector 507.

[0158] According to the sample diagnosis item vector 504 and the first sample diagnosis item vector 506, a third sample diagnosis item vector 508 is obtained. According to the sample diagnosis item vector 504, the first sample diagnosis item vector 506 and the second sample diagnosis item vector 507, a fourth sample diagnosis item vector 509 is obtained.

[0159] The fourth sample diagnosis item vector 509 is input into the fully connected layer 510 to obtain the fifth sample diagnosis item vector 511. The fifth sample diagnosis item vector 511 is input into the activation layer 512 to obtain the system part prediction result 513.

[0160] The first sample diagnosis item vector 506 is input into the fully connected layer 514 to obtain the sixth sample diagnosis item vector 515. The sixth sample diagnosis item vector 515 is input into the activation layer 516 to obtain the non-whole body part prediction result 517.

[0161] The second sample diagnosis item vector 507 is input into the fully connected layer 518 to obtain the seventh sample diagnosis item vector 519. The seventh sample diagnosis item vector 519 is input into the activation layer 520 to obtain the whole body part prediction result 521.

[0162] The third sample diagnosis item vector 508 is input into the fully connected layer 522 to obtain the eighth sample diagnosis item vector 523. The eighth sample diagnosis item vector 523 is input into the activation layer 524 to obtain the secondary part prediction result 525.

[0163] The system label data and the system prediction result 513 are input into the first loss function to obtain a first output value. The non-whole body part label data and the non-whole body part prediction result 517 are input into the first loss function to obtain a third output value. The whole body part label data and the whole body part prediction result 521 are input into the first loss function to obtain a fourth output value. The secondary part label data and the secondary part prediction result 523 are input into the first loss function to obtain a fifth output value. The model parameters of the predetermined model are adjusted according to the first output value and the second output value until the predetermined conditions are met. The predetermined model obtained when the predetermined conditions are met is determined as the diagnostic item label model. The second output value includes a third output value, a fourth output value and a fifth output value.

[0164] The above are merely exemplary embodiments, but are not limited thereto. Other methods for determining medical entity relationships and model training methods known in the art may also be included, as long as reasonable predictions of the relationships between diagnostic items and treatment items can be achieved.

[0165] It should be noted that in the technical solutions disclosed herein, the acquisition, collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0166] Figure 6 A block diagram schematically illustrates a device for determining medical entity relationships according to an embodiment of the present disclosure.

[0167] like Figure 6 As shown, the medical entity relationship determination device 600 may include a first obtaining module 610 , a second obtaining module 620 and a first determining module 630 .

[0168] The first obtaining module 610 is configured to process the target diagnostic item data of the target diagnostic item using the diagnostic item label model to obtain a first object set. The first object set represents a set of objects corresponding to the target diagnostic item.

[0169] The second obtaining module 620 is configured to process the target treatment item data of the target treatment item using the treatment item label model to obtain a second object set. The second object set represents a set of objects corresponding to the target treatment item.

[0170] The first determination module 630 is configured to determine the relationship between the target diagnosis item and the target treatment item according to the first object set and the second object set.

[0171] According to an embodiment of the present disclosure, the first determining module 630 may include a first determining submodule and a second determining submodule.

[0172] The first determining submodule is configured to determine that the target diagnosis item and the target treatment item have an association relationship when it is determined that the first object set and the second object set have an intersection.

[0173] The second determining submodule is configured to determine that the target diagnosis item and the target treatment item do not have an association relationship when it is determined that the first object set and the second object set do not have an intersection.

[0174] According to an embodiment of the present disclosure, the first obtaining module 610 may include a first obtaining sub-module.

[0175] The first acquisition submodule is used to process the target diagnosis item data of the target diagnosis item using the diagnosis item label model to obtain a first object set when it is determined based on the relationship graph and the entity attribute graph that there is no association relationship between the target diagnosis item and the target treatment item.

[0176] According to an embodiment of the present disclosure, the first obtaining module 610 may include a second obtaining submodule.

[0177] The second acquisition submodule is used to process the target diagnosis item data of the target diagnosis item using the diagnosis item label model to obtain the first object set when it is determined based on the entity attribute map that there is no association relationship between the target diagnosis item and the target treatment item.

[0178] According to an embodiment of the present disclosure, the first obtaining module 610 may include a third obtaining submodule.

[0179] The third acquisition submodule is used to process the target diagnosis item data of the target diagnosis item using the diagnosis item label model to obtain the first object set when it is determined based on the relationship graph that there is no association relationship between the target diagnosis item and the target treatment item.

[0180] According to an embodiment of the present disclosure, the entity attribute graph includes a first entity attribute graph and a second entity attribute graph.

[0181] According to an embodiment of the present disclosure, the medical entity relationship determination device 600 may further include a second determination module, a third determination module, a fourth determination module, a fifth determination module, and a sixth determination module.

[0182] The second determination module is used to determine a first associated diagnostic item set corresponding to the target diagnostic item based on the first entity attribute map and the target diagnostic item data.

[0183] The third determining module is configured to determine a first disease condition set corresponding to the first associated diagnosis item set based on the first entity attribute map and the first associated diagnosis item data set of the first associated diagnosis item set.

[0184] The fourth determination module is used to determine a first associated treatment item set corresponding to the target treatment item based on the second entity attribute map and the target treatment item data.

[0185] The fifth determination module is used to determine the second condition set corresponding to the first associated treatment item set based on the second entity attribute map and the first associated treatment item data set of the first associated treatment item set.

[0186] The sixth determination module is used to determine that there is no association relationship between the target diagnosis item and the target treatment item when it is determined that there is no intersection between the first condition set and the second condition set.

[0187] According to an embodiment of the present disclosure, the second determining module may include a third determining submodule.

[0188] The third determining submodule is configured to determine a first similarity between the target diagnosis item data and each of the plurality of first candidate diagnosis item data included in the first entity attribute graph, and obtain a plurality of first similarities.

[0189] The fourth determining submodule is configured to determine, based on the plurality of first similarities, a first associated diagnostic item set corresponding to the target diagnostic item from the plurality of first candidate diagnostic items corresponding to the plurality of first candidate diagnostic item data.

[0190] According to an embodiment of the present disclosure, the fourth determining module may include a fifth determining submodule and a sixth determining submodule.

[0191] The fifth determining submodule is configured to determine a second similarity between the target treatment item data and each of the plurality of candidate treatment item data included in the second entity attribute graph, and obtain a plurality of second similarities.

[0192] The sixth determining submodule is configured to determine, based on the plurality of second similarities, a first associated treatment item set corresponding to the target treatment item from the plurality of candidate treatment items corresponding to the plurality of candidate treatment item data.

[0193] According to an embodiment of the present disclosure, both the first similarity and the second similarity are set similarities.

[0194] According to an embodiment of the present disclosure, the second determining module may include a seventh determining submodule.

[0195] The seventh determination submodule is used to determine the first associated diagnosis item set corresponding to the target diagnosis item based on the first entity attribute map and the target diagnosis item data when it is determined based on the relationship map that there is no association relationship between the target diagnosis item and the target treatment item.

[0196] According to an embodiment of the present disclosure, the above-mentioned method for determining medical entity relationships may further include a seventh determination module, an eighth determination module, a ninth determination module, and a tenth determination module.

[0197] The seventh determination module is used to determine a second associated diagnostic item set corresponding to the target diagnostic item based on the relationship graph and the target diagnostic item data.

[0198] An eighth determining module is configured to determine, based on the relationship graph and according to the second associated diagnosis item data set of the second associated diagnosis item set, a second associated treatment item set corresponding to the second associated diagnosis item set.

[0199] The ninth determining module is configured to determine a third similarity between the target treatment item data and each of the plurality of associated treatment item data included in the second associated treatment item data set of the second associated treatment item set, to obtain a plurality of third similarities.

[0200] The tenth determining module is configured to determine that there is no association between the target diagnosis item and the target treatment item when it is determined that the plurality of third similarities are all less than a predetermined similarity threshold.

[0201] According to an embodiment of the present disclosure, the seventh determining module may include an eighth determining submodule and a ninth determining submodule.

[0202] The eighth determining submodule is configured to determine a fourth similarity between the target diagnosis item data and each of the plurality of second candidate diagnosis item data included in the relationship graph, and obtain a plurality of fourth similarities.

[0203] The ninth determining submodule is configured to determine, based on the plurality of fourth similarities, a second associated diagnostic item set corresponding to the target diagnostic item from the plurality of second candidate diagnostic items corresponding to the plurality of second candidate diagnostic item data.

[0204] According to an embodiment of the present disclosure, the above-mentioned medical entity relationship determination device 600 may further include an eleventh determination module.

[0205] The eleventh determination module is configured to, when determining that the target diagnosis item and the target treatment item are associated, determine a primary diagnosis item from the target diagnosis item and other diagnosis items corresponding to the other treatment items based on the medical resources consumed by each of the target treatment item and the other treatment items. The target diagnosis item and other diagnosis items are diagnosis items for the same user.

[0206] Figure 7 A block diagram of a model training device according to an embodiment of the present disclosure is schematically shown.

[0207] like Figure 7 As shown, the model training device 700 may include a second obtaining module 710 and a third obtaining module 720 .

[0208] The third obtaining module 710 is configured to train a predetermined model using the sample diagnosis item data and the first object label data to obtain a diagnosis item label model.

[0209] The fourth obtaining module 720 is used to train the diagnosis item label model using the sample treatment item data and the second object label data to obtain the treatment item label model.

[0210] According to an embodiment of the present disclosure, the first object label data includes system label data and multi-level part label data.

[0211] According to an embodiment of the present disclosure, the third obtaining module 710 may include a second obtaining sub-module, a third obtaining sub-module, a fourth obtaining sub-module, and a fifth obtaining sub-module.

[0212] The second obtaining submodule is used to encode the sample diagnosis item data to obtain a sample diagnosis item vector.

[0213] The third obtaining submodule is configured to perform full connection processing on the sample diagnosis item vector to obtain a first sample diagnosis item vector and a second sample diagnosis item vector.

[0214] The fourth obtaining submodule is used to obtain a system prediction result and a multi-level part prediction result according to the sample diagnosis item vector, the first sample diagnosis item vector and the second sample diagnosis item vector.

[0215] The fifth acquisition submodule is used to train a predetermined model using the system label data, the system prediction results, the multi-level part label data and the multi-level part prediction results to obtain a diagnosis item label model.

[0216] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0217] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method described above.

[0218] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method described above.

[0219] According to an embodiment of the present disclosure, a computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the method described above.

[0220] Figure 8 A block diagram of an electronic device suitable for implementing a method for determining a medical entity relationship and a model training method according to an embodiment of the present disclosure is schematically shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0221] like Figure 8As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the electronic device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0222] Multiple components in the electronic device 800 are connected to the I / O interface 805, including an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a magnetic disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the electronic device 800 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0223] The computing unit 801 can be various general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as the method for determining medical entity relationships or the model training method. For example, in some embodiments, the method for determining medical entity relationships or the model training method can be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as the storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the method for determining medical entity relationships or the model training method described above can be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the medical entity relationship determination method or the model training method in any other appropriate manner (eg, by means of firmware).

[0224] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0225] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0226] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0227] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0228] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0229] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0230] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0231] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for determining a medical entity relationship, comprising: Processing target diagnostic item data of a target diagnostic item using a diagnostic item label model to obtain a first object set, wherein the first object set represents a set of objects corresponding to the target diagnostic item and includes one or more objects; Processing target treatment item data of a target treatment item using a treatment item label model to obtain a second object set, wherein the second object set represents a set of objects corresponding to the target treatment item and includes one or more objects; and Determine whether there is an association relationship between the target diagnosis item and the target treatment item based on whether there is an intersection between the first object set and the second object set, In which, when it is determined that there is an intersection between the first object set and the second object set, whether there is an association relationship between the target diagnostic item and the target treatment item is determined based on a first ratio, a second ratio and a first predetermined ratio threshold, the first ratio is the ratio of the number of objects included in the intersection to the number included in the first object set, and the second ratio is the ratio of the number of objects included in the intersection to the number included in the second object set. When the first ratio and the second ratio are both greater than the first predetermined ratio threshold, it is determined that there is an association relationship between the target diagnostic item and the target treatment item.

2. The method according to claim 1, wherein The determining whether there is an association relationship between the target diagnosis item and the target treatment item based on whether there is an intersection between the first object set and the second object set further includes: When it is determined that there is no intersection between the first object set and the second object set, it is determined that the target diagnosis item and the target treatment item have no association relationship.

3. The method according to claim 1 or 2, wherein: The step of processing the target diagnostic item data of the target diagnostic item using the diagnostic item label model to obtain a first object set includes: When it is determined based on the relationship map and the entity attribute map that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data of the target diagnosis item is processed using the diagnosis item label model to obtain the first object set.

4. The method according to claim 1 or 2, wherein: The step of processing the target diagnostic item data of the target diagnostic item using the diagnostic item label model to obtain a first object set includes: When it is determined based on the entity attribute map that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data of the target diagnosis item is processed using the diagnosis item label model to obtain the first object set.

5. The method according to claim 1 or 2, wherein: The step of processing the target diagnostic item data of the target diagnostic item using the diagnostic item label model to obtain a first object set includes: When it is determined based on the relationship graph that there is no association relationship between the target diagnosis item and the target treatment item, the target diagnosis item data of the target diagnosis item is processed using the diagnosis item label model to obtain the first object set.

6. The method according to claim 3, wherein: The entity attribute graph includes a first entity attribute graph and a second entity attribute graph; The method further comprises: Based on the first entity attribute map and according to the target diagnosis item data, determining a first associated diagnosis item set corresponding to the target diagnosis item; Based on the first entity attribute map, and according to the first associated diagnosis item data set of the first associated diagnosis item set, determining a first disease condition set corresponding to the first associated diagnosis item set; Based on the second entity attribute graph and according to the target treatment item data, determining a first associated treatment item set corresponding to the target treatment item; Based on the second entity attribute graph, determining a second set of medical conditions corresponding to the first set of associated treatment items according to the first set of associated treatment items data; and When it is determined that there is no intersection between the first condition set and the second condition set, it is determined that there is no association relationship between the target diagnosis item and the target treatment item.

7. The method according to claim 6, wherein: The determining, based on the first entity attribute graph and according to the target diagnosis item data, a first associated diagnosis item set corresponding to the target diagnosis item includes: Determining a first similarity between the target diagnosis item data and each of a plurality of first candidate diagnosis item data included in the first entity attribute graph to obtain a plurality of first similarities; and According to the plurality of first similarities, a first associated diagnostic item set corresponding to the target diagnostic item is determined from a plurality of first candidate diagnostic items corresponding to the plurality of first candidate diagnostic item data.

8. The method according to claim 7, wherein: The determining, based on the second entity attribute graph and according to the target treatment item data, a first associated treatment item set corresponding to the target treatment item includes: determining a second similarity between the target treatment item data and each of the plurality of candidate treatment item data included in the second entity attribute graph to obtain a plurality of second similarities; and According to the plurality of second similarities, a first associated treatment item set corresponding to the target treatment item is determined from a plurality of candidate treatment items corresponding to the plurality of candidate treatment item data.

9. The method according to claim 8, wherein Both the first similarity and the second similarity are set similarities.

10. The method according to any one of claims 6 to 9, wherein The determining, based on the first entity attribute graph and according to the target diagnosis item data, a first associated diagnosis item set corresponding to the target diagnosis item includes: When it is determined based on the relationship map that there is no association relationship between the target diagnostic item and the target treatment item, a first associated diagnostic item set corresponding to the target diagnostic item is determined based on the first entity attribute map and according to the target diagnostic item data.

11. The method according to claim 5, further comprising: Based on the relationship graph and according to the target diagnosis item data, determining a second associated diagnosis item set corresponding to the target diagnosis item; Based on the relationship graph, determining a second associated treatment item set corresponding to the second associated diagnosis item set according to the second associated diagnosis item data set of the second associated diagnosis item set; determining a third similarity between the target treatment item data and each of the plurality of associated treatment item data included in the second associated treatment item data set of the second associated treatment item set, to obtain a plurality of third similarities; as well as When it is determined that the plurality of third similarities are all smaller than a predetermined similarity threshold, it is determined that there is no association relationship between the target diagnosis item and the target treatment item.

12. The method according to claim 11, wherein The determining, based on the relationship graph and according to the target diagnosis item data, a second associated diagnosis item set corresponding to the target diagnosis item includes: determining a fourth similarity between the target diagnosis item data and each of the plurality of second candidate diagnosis item data included in the relationship graph to obtain a plurality of fourth similarities; and According to the plurality of fourth similarities, a second associated diagnostic item set corresponding to the target diagnostic item is determined from a plurality of second candidate diagnostic items corresponding to the plurality of second candidate diagnostic item data.

13. The method according to claim 1, further comprising: When it is determined that the target diagnostic item and the target treatment item are associated with each other, the main diagnostic item is determined from the target diagnostic item and other diagnostic items corresponding to other treatment items based on the medical resources consumed by each of the target treatment item and other treatment items, wherein the target diagnostic item and the other diagnostic items are diagnostic items for the same user.

14. A device for determining a medical entity relationship, comprising: a first obtaining module, configured to process target diagnostic item data of a target diagnostic item using a diagnostic item label model to obtain a first object set, wherein the first object set represents a set of objects corresponding to the target diagnostic item and includes one or more objects; a second obtaining module, configured to process target treatment item data of a target treatment item using a treatment item label model to obtain a second object set, wherein the second object set represents a set of objects corresponding to the target treatment item and includes one or more objects; and A first determining module is configured to determine whether there is an association relationship between the target diagnosis item and the target treatment item based on whether there is an intersection between the first object set and the second object set; In which, when it is determined that there is an intersection between the first object set and the second object set, whether there is an association relationship between the target diagnostic item and the target treatment item is determined based on a first ratio, a second ratio and a first predetermined ratio threshold, the first ratio is the ratio of the number of objects included in the intersection to the number included in the first object set, and the second ratio is the ratio of the number of objects included in the intersection to the number included in the second object set. When the first ratio and the second ratio are both greater than the first predetermined ratio threshold, it is determined that there is an association relationship between the target diagnostic item and the target treatment item.

15. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 13.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable the computer to execute the method according to any one of claims 1 to 13.

17. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method according to any one of claims 1 to 13.

Images