CN114330267A - Structural report template design method based on semantic association - Google Patents
Structural report template design method based on semantic association Download PDFInfo
- Publication number
- CN114330267A CN114330267A CN202111461852.7A CN202111461852A CN114330267A CN 114330267 A CN114330267 A CN 114330267A CN 202111461852 A CN202111461852 A CN 202111461852A CN 114330267 A CN114330267 A CN 114330267A
- Authority
- CN
- China
- Prior art keywords
- entity
- diagnosis
- treatment
- disease
- semantic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a structural report template design method based on semantic association, which comprises the following steps: s1, constructing based on the big data semantics of the historical medical record to obtain a first diagnosis and treatment knowledge map; s2, constructing a second diagnosis and treatment knowledge graph based on big data of the disease diagnosis and treatment guide; step S3, performing entity fusion on the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to obtain a diagnosis and treatment structural knowledge graph fusing and representing diagnosis and treatment practice experience and diagnosis and treatment expert experience, and constructing a structural report template for disease categories based on the diagnosis and treatment structural knowledge graph; and step S4, performing unified diagnosis and treatment on the disease categories according to the structured report template so as to improve the diagnosis and treatment normative. The invention ensures that the disease diagnosis and treatment according to the structured report template accords with the diagnosis and treatment practical experience of doctors and the diagnosis and treatment expert experience, realizes the consideration of diagnosis and treatment standard and maneuverability, and breaks through the limitation that a computer-aided diagnosis method is only driven by a diagnosis and treatment guide.
Description
Technical Field
The invention relates to the technical field of medical reports, in particular to a structural report template design method based on semantic association.
Background
Compared with the traditional report, the structured report is structured in content, the traditional report is mostly written according to personal habits of doctors, the content logic is complex, words are diversified, value information in the report is difficult to effectively extract, and informationized management and utilization cannot be performed. Not only wastes a large amount of precious medical record information, but also easily causes omission and errors of writing reports. When the times of big data and artificial intelligence come, structured medical record information is the most basic data. The popularity and use of structured reports has therefore been pressing.
CN202110892281.6 discloses a structured report design method for prostate MR cancer, which comprises the following steps of logging in a user interface and providing a plurality of data options on the user interface; filling the PI _ RADS score of the prostate and inserting the PACS image in a structured report template, and then uploading the result to a database; the database receives the uploaded data and classifies and archives the key information; the database stores the received information; inputting a data query request on a user interface and sending the data query request to a database; the database receives the request, and the data information in the request is matched and screened with the existing information in the database; the database returns the data that the user so inquired about and automatically generates a structured report form.
Although the above-mentioned prior art can generate a structured report, the physician user is still required to select the index of the examination item, so that the generated structured report is different from person to person and has poor normativity, and the generated structured report depends on the subjectivity of the physician user, thereby resulting in low objectivity and low credibility of the structured report.
Disclosure of Invention
The invention aims to provide a structural report template design method based on semantic association, which aims to solve the technical problems that in the prior art, a doctor user is still required to select an inspection item index, so that a generated structural report is different from person to person and has poor normative, and the generated structural report depends on the subjectivity of the doctor user, so that the objectivity of the structural report is low.
In order to solve the technical problems, the invention specifically provides the following technical scheme:
a structural report template design method based on semantic association comprises the following steps:
step S1, extracting a first disease category and a first examination item respectively as a first entity and a second entity based on historical disease case big data semantics, extracting a relation attribute of the first disease category and the first examination item semantically as a first entity relation, and performing knowledge graph construction on the first entity, the second entity and the first entity relation to obtain a first diagnosis and treatment knowledge graph, wherein the first relation attribute is characterized by a deterministic relation and a topological relation between the first disease category and the first examination item which are determined by diagnosis and treatment practice experience;
step S2, extracting a second disease category and a second inspection item respectively as a third entity and a fourth entity based on the big data semantics of the disease diagnosis and treatment guide, extracting the relation attribute of the second disease category and the second inspection item as a third entity relation semantically, and performing knowledge map construction on the third entity, the fourth entity and the third entity relation to obtain a second diagnosis and treatment knowledge map, wherein the second relation attribute is characterized by the deterministic relation and the topological relation of the second disease category and the second inspection item determined by the experience of a diagnosis and treatment expert;
step S3, performing entity fusion on the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to obtain a diagnosis and treatment structural knowledge graph fusing and representing diagnosis and treatment practice experience and diagnosis and treatment expert experience, and constructing a structural report template for disease categories based on the diagnosis and treatment structural knowledge graph;
step S4, according to the structured report template, the disease category is treated uniformly to improve the treatment normative, wherein,
if the disease category does not have the corresponding structured report template, returning to the step S1 and sequentially executing the steps S1 to S3 to update the diagnosis and treatment structured knowledge graph;
and if the disease category has the corresponding structured report template, directly using the corresponding structured report template to diagnose and treat.
As a preferred aspect of the present invention, the semantic extracting a first disease category and a first examination item based on historical disease big data as a first entity and a second entity, respectively, includes:
randomly selecting a group of historical medical records from historical medical record big data, and respectively extracting semantic texts representing a first medical record category and a first inspection item from the historical medical records as a first entity and a second entity, wherein the historical medical records comprise the semantic texts representing the first medical record category and the first inspection item and the semantic texts representing the relationship attributes of the first medical record category and the first inspection item;
quantizing the historical medical records, the first entity and the second entity from a text form to a vector form to obtain semantic vectors, first entity vectors and second entity vectors of the historical medical records, and correspondingly calibrating the semantic vectors, the first entity vectors and the second entity vectors of the historical medical records into a single first entity sample and a single second entity sample, wherein the single first entity sample is characterized as [ the semantic vectors and the first entity vectors of the historical medical records ], and the single second entity sample is characterized as [ the semantic vectors and the second entity vectors of the historical medical records ];
randomly selecting 70% of the total sample number from all the first entity samples as a first training set, using the remaining 30% as a first test set, and using the first training set and the first test set for training a crf model to obtain a first entity extraction model;
randomly selecting 70% of the total sample number from all second entity samples as a second training set, using the remaining 30% as a second testing set, and using the second training set and the second testing set for training a crf model to obtain a second entity extraction model;
the first entity extraction model and the second entity extraction model are used for carrying out semantic extraction on entities on historical medical record big data to obtain a first entity and a second entity;
preferably, the removal of sample redundancy is required before the cdf model training using the first physical sample and the second physical sample, wherein,
circularly calculating the similarity between any two first entity samples/second entity samples, and randomly removing one first entity sample/second entity sample from the two first entity samples/second entity samples with the similarity exceeding a set threshold until the similarity between any two first entity samples/second entity samples does not exceed the set threshold;
the calculation formula of the similarity is as follows:
in which I is characterized by a similarity value, Ai、AjRespectively characterized as the ith and jth first/second physical samples,i, j are the metering constants.
As a preferable aspect of the present invention, the semantically extracting relationship attributes of the first disease category and the first examination item as a first entity relationship, includes:
extracting semantic texts representing relation attributes of a first disease category and a first inspection item from historical disease cases corresponding to the first entity sample/the second entity sample after sample redundancy removal to serve as a first entity relation;
quantizing the historical medical records and the first entity relations into a vector form from a text form to obtain semantic vectors of the historical medical records and semantic vectors of the first entity relations, and correspondingly calibrating the semantic vectors of the historical medical records and the semantic vectors of the first entity relations into a single first entity relation sample, wherein the single first entity relation sample is characterized as [ the semantic vectors of the historical medical records, the first entity relation sample ];
randomly selecting 70% of the total sample number from all the first entity relationship samples as a first relationship training set, taking the rest 30% as a first relationship test set, and using the first relationship training set and the first relationship test set for training a BP neural network to obtain a first entity relationship extraction model;
and the first entity relationship extraction model is used for semantic extraction of relationship attributes on historical case big data to obtain a first entity relationship.
As a preferred embodiment of the present invention, the constructing a knowledge graph of the relationship among the first entity, the second entity, and the first entity to obtain a first diagnosis knowledge graph includes:
carrying out graph connection on a first entity, a second entity and a first entity relationship extracted from the same historical case to form a sub-graph of the first entity, the first entity relationship and the second entity, and sequentially setting the merging priority of the sub-graphs as the first entity, the first entity relationship and the second entity;
carrying out sub-graph merging on all sub-graphs obtained from the big data of the historical case according to the priority of sub-graph merging in sequence, wherein,
carrying out node combination on the subgraphs with the same first entity at the first entity, then carrying out node combination on graph structures with the same first entity relation in the combined subgraphs at the first entity relation, and finally combining the nodes with the same second entity in the combined graph structures so as to realize that all the subgraphs are combined to generate the first diagnosis and treatment knowledge graph;
preferably, determining that the first entities are identical comprises:
summarizing all semantic names of disease categories represented by the first entity to form a disease standard name query table, converting the semantic names of the disease categories represented by the first entity into standard semantic names based on the disease standard name query table, and judging all first entities capable of being converted into the same standard semantic names as the same first entities;
preferably, determining that the second entities are identical comprises:
and summarizing all semantic names of the inspection items represented by the second entities to form an item standard name query table, converting all the semantic names of the inspection items represented by the second entities into standard semantic names based on the item standard name query table, and judging all third entities capable of being converted into the same standard semantic names as the second entities.
As a preferred aspect of the present invention, the extracting a second disease category and a second examination item based on the big data semantic of the disease diagnosis and treatment guideline as a third entity and a fourth entity respectively includes:
randomly selecting a group of disease diagnosis and treatment guidelines from the disease diagnosis and treatment guideline big data, and respectively extracting semantic texts representing a second disease category and a second inspection item from the disease diagnosis and treatment guidelines to be used as a third entity and a fourth entity, wherein the disease diagnosis and treatment guidelines comprise the semantic texts representing the second disease category and the second inspection item and the semantic texts representing the relationship attributes of the second disease category and the second inspection item;
quantizing a disease diagnosis and treatment guide, a third entity and a fourth entity from a text form to a vector form to obtain a semantic vector, a third entity vector and a fourth entity vector of the disease diagnosis and treatment guide, and correspondingly calibrating the semantic vector, the third entity vector and the fourth entity vector of the disease diagnosis and treatment guide into a single third entity sample and a single fourth entity sample, wherein the single third entity sample is characterized as [ the semantic vector and the third entity vector of the disease diagnosis and treatment guide ], and the single fourth entity sample is characterized as [ the semantic vector and the fourth entity vector of the disease diagnosis and treatment guide ];
randomly selecting 70% of the total sample number from all third entity samples as a third training set, using the remaining 30% as a third testing set, and using the third training set and the third testing set for training a crf model to obtain a third entity extraction model;
randomly selecting 70% of the total sample number from all the fourth entity samples as a fourth training set, using the remaining 30% as a fourth test set, and using the fourth training set and the fourth test set for training a crf model to obtain a fourth entity extraction model;
and the third entity extraction model are used for carrying out semantic extraction on entities in the disease diagnosis and treatment guide big data to obtain a third entity and a third entity.
As a preferable aspect of the present invention, the semantic extracts a relationship attribute of the second disorder category and the second examination item as a third entity relationship,
extracting semantic texts representing relationship attributes of a second disease category and a second examination item from a group of disease diagnosis and treatment guidelines to serve as second entity relationships;
quantizing the disease diagnosis and treatment guide and the second entity relationship from a text form to a vector form to obtain a semantic vector of the disease diagnosis and treatment guide and a semantic vector of the second entity relationship, and correspondingly marking the semantic vector of the disease diagnosis and treatment guide and the semantic vector of the second entity relationship as a single second entity relationship sample, wherein the single second entity relationship sample is characterized as [ the semantic vector of the disease diagnosis and treatment guide, the second entity relationship sample ];
randomly selecting 70% of the total sample number from all the second entity relationship samples as a second relationship training set, using the remaining 30% as a second relationship test set, and using the second relationship training set and the second relationship test set for training a BP neural network to obtain a second entity relationship extraction model;
and the second entity relationship extraction model is used for carrying out semantic extraction on relationship attributes on big data of the disease diagnosis and treatment guide to obtain a second entity relationship.
As a preferred embodiment of the present invention, the constructing a knowledge graph of the relationship among the third entity, the fourth entity, and the third entity to obtain a second diagnosis knowledge graph includes:
carrying out graph connection on the third entity, the fourth entity and the second entity relationship extracted from the same disease diagnosis and treatment guide to form a subgraph of the third entity-the second entity relationship-the fourth entity, and sequentially setting the merging priority of the subgraphs as the third entity, the second entity relationship and the fourth entity;
and all sub-graphs obtained from big data of the disease diagnosis and treatment guide are sequentially merged according to the priority of sub-graph merging, wherein,
and finally, combining the nodes identical to the fourth entity in the combined graph structure so as to combine all the sub-graphs to generate the second diagnosis and treatment knowledge graph.
As a preferred embodiment of the present invention, the obtaining of the diagnosis and treatment structured knowledge graph fusing the diagnosis and treatment actual practice experience and the diagnosis and treatment expert experience by performing entity fusion on the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph includes:
setting the merging priority of all graph structures in the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph as a first entity/a third entity, a first entity relation/a second entity relation and a second entity/a fourth entity in sequence, merging the graph structures according to the merging priority, wherein,
and finally, combining the second entity and the fourth same node in the combined graph structure to realize the combination of all graph structures in the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to generate the diagnosis and treatment structured knowledge graph, so that the structural fusion of diagnosis and treatment practical experience and diagnosis and treatment expert experience is realized, and the standard property and the mobility are considered in the disease category structured diagnosis and treatment process.
As a preferred embodiment of the present invention, the constructing a structured report template for disease category based on the diagnosis and treatment structured knowledge graph includes:
outputting all examination items corresponding to the disease categories by inquiring the diagnosis and treatment structured knowledge graph according to the disease categories judged by the doctor, and integrating all the examination items in the same report template to generate a structured report template used as a diagnosis and treatment standard guide for the doctor according to the disease categories.
As a preferred aspect of the present invention, the updating of the diagnosis and treatment structured knowledge graph includes:
when the disease category does not have the corresponding structured report template, a doctor formulates a medical record of the disease category, and executes step S1 based on the medical record to construct a sub-graph which is in the form of a first entity-first entity relation and a second entity, and the sub-graph is added into the first diagnosis and treatment knowledge graph to merge the graph structure so as to update the first diagnosis and treatment knowledge graph;
and step S3 is executed, the updated first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph are subjected to entity fusion to update the diagnosis and treatment structured knowledge graph, and amplification of the diagnosis and treatment structured knowledge graph is achieved.
Compared with the prior art, the invention has the following beneficial effects:
the diagnosis and treatment structured knowledge graph representing the diagnosis and treatment practical operation experience and the diagnosis and treatment expert experience is constructed based on the historical medical record big data and the disease diagnosis and treatment guide big data in a fusion mode, and the structured report template is constructed for disease categories based on the diagnosis and treatment structured knowledge graph, so that disease diagnosis and treatment are carried out according to the structured report template, namely the diagnosis and treatment practical operation experience of a doctor is met, the diagnosis and treatment expert experience is also met, the diagnosis and treatment standard and mobility are considered, and the limitation that a computer-aided diagnosis method is only driven by the diagnosis and treatment guide is broken through.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.
FIG. 1 is a flowchart of a method for designing a structured report template based on semantic association according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in FIG. 1, the invention provides a structural report template design method based on semantic association, which comprises the following steps:
step S1, extracting a first disease category and a first examination item respectively as a first entity and a second entity based on historical disease case big data semantics, extracting a relation attribute of the first disease category and the first examination item semantically as a first entity relation, and performing knowledge graph construction on the first entity, the second entity and the first entity relation to obtain a first diagnosis and treatment knowledge graph, wherein the first relation attribute is characterized by a deterministic relation and a topological relation between the first disease category and the first examination item which are determined by diagnosis and treatment practice experience;
the examination items required to be done by each doctor for the disease have the judgment of self-practice experience, the examination items required to be done by the disease are judged through the self-practice experience, so that the patient can check according to the check item list given by the doctor, and after the check is finished, the data of each item is recorded into a medical record by the doctor, the doctor can master the actual operation experience of the doctor on the medical record, the clinical experience accumulation of the doctor is reflected, can quickly and accurately carry out initial diagnosis and necessary examination items on the disease of the patient so as to realize quick control on the disease condition of the disease, therefore, the first diagnosis and treatment knowledge graph representing the practical practice experience is constructed on the basis of the medical record big data, so as to combine the experience knowledge of practical diagnosis and treatment when patients are diagnosed, embody the mobility of doctors in disease diagnosis, namely, the examination items are selected according to the degree of urgency of diseases so as to realize medical assistance for patient diagnosis.
The method for extracting the first disease category and the first check item based on the historical disease case big data semantics as a first entity and a second entity respectively comprises the following steps:
randomly selecting a group of historical medical records from historical medical record big data, and respectively extracting semantic texts representing a first medical record category and a first inspection item from the historical medical records as a first entity and a second entity, wherein the historical medical records comprise the semantic texts representing the first medical record category and the first inspection item and the semantic texts representing the relationship attributes of the first medical record category and the first inspection item;
quantizing the historical medical records, the first entity and the second entity from a text form to a vector form to obtain semantic vectors, first entity vectors and second entity vectors of the historical medical records, and correspondingly calibrating the semantic vectors, the first entity vectors and the second entity vectors of the historical medical records into a single first entity sample and a single second entity sample, wherein the single first entity sample is characterized as [ the semantic vectors and the first entity vectors of the historical medical records ], and the single second entity sample is characterized as [ the semantic vectors and the second entity vectors of the historical medical records ];
randomly selecting 70% of the total sample number from all the first entity samples as a first training set, using the remaining 30% as a first test set, and using the first training set and the first test set for training a crf model to obtain a first entity extraction model;
randomly selecting 70% of the total sample number from all second entity samples as a second training set, using the remaining 30% as a second testing set, and using the second training set and the second testing set for training a crf model to obtain a second entity extraction model;
all 70% and 30% in the embodiment are adjustable data, and a user can perform custom modification according to actual needs.
The first entity extraction model and the second entity extraction model are used for carrying out semantic extraction on entities on historical medical record big data to obtain a first entity and a second entity;
preferably, before the cdf model training is performed by using the first entity sample and the second entity sample, the sample redundancy needs to be removed, so as to avoid the occupation of the redundant data on the operation resources of the model training and improve the efficiency of the model training, wherein,
circularly calculating the similarity between any two first entity samples/second entity samples, and randomly removing one first entity sample/second entity sample from the two first entity samples/second entity samples with the similarity exceeding a set threshold until the similarity between any two first entity samples/second entity samples does not exceed the set threshold;
the calculation formula of the similarity is as follows:
in which I is characterized by a similarity value, Ai、AjRespectively characterized as the ith and jth first/second entity samples, i, j being a metering constant.
The higher the similarity between the two first entity samples/the second entity samples is, the more similar the two first entity samples/the second entity samples are, namely, one of the two first entity samples/the second entity samples can be randomly used for representing, so that the entity sample type is reserved, redundant items in the entity sample type are removed, one-variable data simplification is realized, and finally, the sample redundancy is removed while the sample diversity is reserved.
The semantic extracting a relationship attribute of a first condition category and a first examination item as a first entity relationship, including:
extracting semantic texts representing relation attributes of a first disease category and a first inspection item from historical disease cases corresponding to the first entity sample/the second entity sample after sample redundancy removal to serve as a first entity relation;
quantizing the historical medical records and the first entity relations into a vector form from a text form to obtain semantic vectors of the historical medical records and semantic vectors of the first entity relations, and correspondingly calibrating the semantic vectors of the historical medical records and the semantic vectors of the first entity relations into a single first entity relation sample, wherein the single first entity relation sample is characterized as [ the semantic vectors of the historical medical records, the first entity relation sample ];
randomly selecting 70% of the total sample number from all the first entity relationship samples as a first relationship training set, taking the rest 30% as a first relationship test set, and using the first relationship training set and the first relationship test set for training a BP neural network to obtain a first entity relationship extraction model;
and the first entity relationship extraction model is used for semantic extraction of relationship attributes on historical case big data to obtain a first entity relationship.
The knowledge graph construction of the relationship among the first entity, the second entity and the first entity to obtain the first diagnosis and treatment knowledge graph comprises the following steps:
carrying out graph connection on a first entity, a second entity and a first entity relationship extracted from the same historical case to form a sub-graph of the first entity, the first entity relationship and the second entity, and sequentially setting the merging priority of the sub-graphs as the first entity, the first entity relationship and the second entity;
carrying out sub-graph merging on all sub-graphs obtained from the big data of the historical case according to the priority of sub-graph merging in sequence, wherein,
carrying out node combination on the subgraphs with the same first entity at the first entity, then carrying out node combination on graph structures with the same first entity relation in the combined subgraphs at the first entity relation, and finally combining the nodes with the same second entity in the combined graph structures so as to realize that all the subgraphs are combined to generate the first diagnosis and treatment knowledge graph;
preferably, determining that the first entities are identical comprises:
summarizing all semantic names of disease categories represented by the first entity to form a disease standard name query table, converting the semantic names of the disease categories represented by the first entity into standard semantic names based on the disease standard name query table, and judging all first entities capable of being converted into the same standard semantic names as the same first entities;
preferably, determining that the second entities are identical comprises:
and summarizing all semantic names of the inspection items represented by the second entities to form an item standard name query table, converting all the semantic names of the inspection items represented by the second entities into standard semantic names based on the item standard name query table, and judging all third entities capable of being converted into the same standard semantic names as the second entities.
The reason why the judgment of the identity of the entities is needed is that the names of disease categories may be written into various alias forms instead of standard semantic names (academic names) due to different habits of doctors, so the identity judgment can be carried out to identify the entities representing the same disease categories, and the accuracy of establishing the knowledge graph is improved.
The embodiment provides an example of constructing a first medical knowledge graph, for example: there are four sub-graphs (for convenience of description here, the number of sub-graphs is set to four, and in fact, the number of sub-graphs is far more than four), a first entity A-a first entity relationship 1-a second entity a, a first entity B-a first entity relationship 2-a second entity C, a first entity C-a first entity relationship 1-a second entity B, and a first entity A-a first entity relationship 2-a second entity B, wherein sub-graphs with the same first entity are firstly determined, the first entity and the second entity of all sub-graphs are converted into standard semantic names, and the first entity A-the first entity relationship 1-the second entity a, and the first entity B-the first entityThe entity relationship 2-second entity c, the first entity A-first entity relationship 1-second entity b, the first entity A-first entity relationship 2-second entity c, three same first entities are found, and the first entities are combined to obtain: first entity
(or
) (ii) a Finding two identical first entity relations in the merged graph structure, and merging the first entity relations to obtain: first entity
(or
) (ii) a There is no third entity that is the same, and therefore the first medical knowledge-graph that is finally obtained is:
(or
)。
The entity relationships 1 and 2 are respectively characterized as a deterministic relationship and a topological relationship, the deterministic relationship refers to the examination items that the disease category must do, and the topological relationship refers to the examination items that the disease category can choose to do, such as for determining whether a complication exists.
Step S2, extracting a second disease category and a second inspection item respectively as a third entity and a fourth entity based on the big data semantics of the disease diagnosis and treatment guide, extracting the relation attribute of the second disease category and the second inspection item as a third entity relation semantically, and performing knowledge map construction on the third entity, the fourth entity and the third entity relation to obtain a second diagnosis and treatment knowledge map, wherein the second relation attribute is characterized by the deterministic relation and the topological relation of the second disease category and the second inspection item determined by the experience of a diagnosis and treatment expert;
the diagnosis and treatment practical experience of the doctor has one sidedness to a certain extent, and the disease diagnosis and treatment guideline is a guidance suggestion formulated by medical experts of all boundaries and used for guiding the clinical diagnosis of the doctor, so that the second diagnosis and treatment knowledge map for representing the diagnosis and treatment expert experience is constructed based on the disease diagnosis and treatment guideline big data, the experience knowledge of the diagnosis and treatment expert is combined when the patient is diagnosed, the one sidedness of the diagnosis and treatment experience of the doctor is compensated, the patient can be diagnosed more comprehensively and standard, namely, the standard property and the comprehensive property of diagnosis are improved, and the medical assistance of the patient diagnosis is realized.
The method for extracting a second disease category and a second inspection item based on disease diagnosis and treatment guide big data semantics to serve as a third entity and a fourth entity respectively comprises the following steps:
randomly selecting a group of disease diagnosis and treatment guidelines from the disease diagnosis and treatment guideline big data, and respectively extracting semantic texts representing a second disease category and a second inspection item from the disease diagnosis and treatment guidelines to be used as a third entity and a fourth entity, wherein the disease diagnosis and treatment guidelines comprise the semantic texts representing the second disease category and the second inspection item and the semantic texts representing the relationship attributes of the second disease category and the second inspection item;
quantizing a disease diagnosis and treatment guide, a third entity and a fourth entity from a text form to a vector form to obtain a semantic vector, a third entity vector and a fourth entity vector of the disease diagnosis and treatment guide, and correspondingly calibrating the semantic vector, the third entity vector and the fourth entity vector of the disease diagnosis and treatment guide into a single third entity sample and a single fourth entity sample, wherein the single third entity sample is characterized as [ the semantic vector and the third entity vector of the disease diagnosis and treatment guide ], and the single fourth entity sample is characterized as [ the semantic vector and the fourth entity vector of the disease diagnosis and treatment guide ];
randomly selecting 70% of the total sample number from all third entity samples as a third training set, using the remaining 30% as a third testing set, and using the third training set and the third testing set for training a crf model to obtain a third entity extraction model;
randomly selecting 70% of the total sample number from all the fourth entity samples as a fourth training set, using the remaining 30% as a fourth test set, and using the fourth training set and the fourth test set for training a crf model to obtain a fourth entity extraction model;
and the third entity extraction model are used for carrying out semantic extraction on entities in the disease diagnosis and treatment guide big data to obtain a third entity and a third entity.
The semantics extracts a relationship attribute of the second condition category and the second examination item as a third entity relationship,
extracting semantic texts representing relationship attributes of a second disease category and a second examination item from a group of disease diagnosis and treatment guidelines to serve as second entity relationships;
quantizing the disease diagnosis and treatment guide and the second entity relationship from a text form to a vector form to obtain a semantic vector of the disease diagnosis and treatment guide and a semantic vector of the second entity relationship, and correspondingly marking the semantic vector of the disease diagnosis and treatment guide and the semantic vector of the second entity relationship as a single second entity relationship sample, wherein the single second entity relationship sample is characterized as [ the semantic vector of the disease diagnosis and treatment guide, the second entity relationship sample ];
randomly selecting 70% of the total sample number from all the second entity relationship samples as a second relationship training set, using the remaining 30% as a second relationship test set, and using the second relationship training set and the second relationship test set for training a BP neural network to obtain a second entity relationship extraction model;
and the second entity relationship extraction model is used for carrying out semantic extraction on relationship attributes on big data of the disease diagnosis and treatment guide to obtain a second entity relationship.
The knowledge graph construction of the relationship among the third entity, the fourth entity and the third entity to obtain a second diagnosis and treatment knowledge graph comprises the following steps:
carrying out graph connection on the third entity, the fourth entity and the second entity relationship extracted from the same disease diagnosis and treatment guide to form a subgraph of the third entity-the second entity relationship-the fourth entity, and sequentially setting the merging priority of the subgraphs as the third entity, the second entity relationship and the fourth entity;
and all sub-graphs obtained from big data of the disease diagnosis and treatment guide are sequentially merged according to the priority of sub-graph merging, wherein,
and finally, combining the nodes identical to the fourth entity in the combined graph structure so as to combine all the sub-graphs to generate the second diagnosis and treatment knowledge graph.
The second diagnosis and treatment knowledge map has similarity with the first diagnosis and treatment knowledge map in the construction process, but the second diagnosis and treatment knowledge map does not need to carry out the removal of sample redundancy and the judgment of the same entity, because the redundancy is removed in the compiling process of the disease diagnosis and treatment guide, and the third entity and the fourth entity are presented by standard semantic names (academic names).
Step S3, performing entity fusion on the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to obtain a diagnosis and treatment structural knowledge graph fusing and representing diagnosis and treatment practice experience and diagnosis and treatment expert experience, and constructing a structural report template for disease categories based on the diagnosis and treatment structural knowledge graph;
the method for carrying out entity fusion on the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to obtain the diagnosis and treatment structured knowledge graph fusing and representing the diagnosis and treatment practical exercise experience and the diagnosis and treatment expert experience comprises the following steps:
setting the merging priority of all graph structures in the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph as a first entity/a third entity, a first entity relation/a second entity relation and a second entity/a fourth entity in sequence, merging the graph structures according to the merging priority, wherein,
and finally, combining the second entity and the fourth same node in the combined graph structure to realize the combination of all graph structures in the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to generate the diagnosis and treatment structured knowledge graph, so that the structural fusion of diagnosis and treatment practical experience and diagnosis and treatment expert experience is realized, and the standard property and the mobility are considered in the disease category structured diagnosis and treatment process.
The embodiment provides a construction example of a diagnosis and treatment structured knowledge graph, wherein the first diagnosis and treatment knowledge graph comprises the following components:
or
The second diagnosis and treatment knowledge graph comprises:
or
And carrying out node combination on the graph structures of the first entity and the third entity which are the same at the first entity to obtain:
and in the merged graph structure, node merging is carried out on the graph structure with the same first entity relation and second entity relation at the first entity relation, so that:
and combining the second entity and the fourth same node in the combined graph structure to obtain a diagnosis and treatment structured knowledge graph, wherein the diagnosis and treatment structured knowledge graph comprises the following steps:
step S4, according to the structured report template, the disease category is treated uniformly to improve the treatment normative, wherein,
if the disease category does not have the corresponding structured report template, returning to the step S1 and sequentially executing the steps S1 to S3 to update the diagnosis and treatment structured knowledge graph;
if the disease category has a corresponding structured report template, directly using the corresponding structured report template to diagnose and treat, wherein,
if the doctor judges that the disease category of the patient is A, inputting A in a computer terminal in which a diagnosis and treatment structured knowledge map is stored, generating all examination items of the disease category A, and generating a corresponding structured report template as follows:
the patient directly uses the corresponding structured report template to carry out disease examination so as to obtain various index data of the disease condition.
The method for constructing the structured report template for the disease category based on the diagnosis and treatment structured knowledge graph comprises the following steps:
outputting all examination items corresponding to the disease categories by inquiring the diagnosis and treatment structured knowledge graph according to the disease categories judged by the doctor, and integrating all the examination items in the same report template to generate a structured report template used as a diagnosis and treatment standard guide for the doctor according to the disease categories.
The updating of the diagnosis and treatment structured knowledge graph comprises the following steps:
when the disease category does not have the corresponding structured report template, a doctor formulates a medical record of the disease category, and executes step S1 based on the medical record to construct a sub-graph which is in the form of a first entity-first entity relation and a second entity, and the sub-graph is added into the first diagnosis and treatment knowledge graph to merge the graph structure so as to update the first diagnosis and treatment knowledge graph;
and step S3 is executed, the updated first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph are subjected to entity fusion to update the diagnosis and treatment structured knowledge graph, and amplification of the diagnosis and treatment structured knowledge graph is achieved.
The diagnosis and treatment structured knowledge graph representing the diagnosis and treatment practical operation experience and the diagnosis and treatment expert experience is constructed based on the historical medical record big data and the disease diagnosis and treatment guide big data in a fusion mode, and the structured report template is constructed for disease categories based on the diagnosis and treatment structured knowledge graph, so that disease diagnosis and treatment are carried out according to the structured report template, namely the diagnosis and treatment practical operation experience of a doctor is met, the diagnosis and treatment expert experience is also met, the diagnosis and treatment standard and mobility are considered, and the limitation that a computer-aided diagnosis method is only driven by the diagnosis and treatment guide is broken through.
The above embodiments are only exemplary embodiments of the present application, and are not intended to limit the present application, and the protection scope of the present application is defined by the claims. Various modifications and equivalents may be made by those skilled in the art within the spirit and scope of the present application and such modifications and equivalents should also be considered to be within the scope of the present application.
Claims (10)
1. A structural report template design method based on semantic association is characterized in that: the method comprises the following steps:
step S1, extracting a first disease category and a first examination item respectively as a first entity and a second entity based on historical disease case big data semantics, extracting a relation attribute of the first disease category and the first examination item semantically as a first entity relation, and performing knowledge graph construction on the first entity, the second entity and the first entity relation to obtain a first diagnosis and treatment knowledge graph, wherein the first relation attribute is characterized by a deterministic relation and a topological relation between the first disease category and the first examination item which are determined by diagnosis and treatment practice experience;
step S2, extracting a second disease category and a second inspection item respectively as a third entity and a fourth entity based on the big data semantics of the disease diagnosis and treatment guide, extracting the relation attribute of the second disease category and the second inspection item as a third entity relation semantically, and performing knowledge map construction on the third entity, the fourth entity and the third entity relation to obtain a second diagnosis and treatment knowledge map, wherein the second relation attribute is characterized by the deterministic relation and the topological relation of the second disease category and the second inspection item determined by the experience of a diagnosis and treatment expert;
step S3, performing entity fusion on the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to obtain a diagnosis and treatment structural knowledge graph fusing and representing diagnosis and treatment practice experience and diagnosis and treatment expert experience, and constructing a structural report template for disease categories based on the diagnosis and treatment structural knowledge graph;
step S4, according to the structured report template, the disease category is treated uniformly to improve the treatment normative, wherein,
if the disease category does not have the corresponding structured report template, returning to the step S1 and sequentially executing the steps S1 to S3 to update the diagnosis and treatment structured knowledge graph;
and if the disease category has the corresponding structured report template, directly using the corresponding structured report template to diagnose and treat.
2. The method for designing the structural report template based on semantic association according to claim 1, wherein: the method for extracting the first disease category and the first check item based on the historical disease case big data semantics as a first entity and a second entity respectively comprises the following steps:
randomly selecting a group of historical medical records from historical medical record big data, and respectively extracting semantic texts representing a first medical record category and a first inspection item from the historical medical records as a first entity and a second entity, wherein the historical medical records comprise the semantic texts representing the first medical record category and the first inspection item and the semantic texts representing the relationship attributes of the first medical record category and the first inspection item;
quantizing the historical medical records, the first entity and the second entity from a text form to a vector form to obtain semantic vectors, first entity vectors and second entity vectors of the historical medical records, and correspondingly calibrating the semantic vectors, the first entity vectors and the second entity vectors of the historical medical records into a single first entity sample and a single second entity sample, wherein the single first entity sample is characterized as [ the semantic vectors and the first entity vectors of the historical medical records ], and the single second entity sample is characterized as [ the semantic vectors and the second entity vectors of the historical medical records ];
randomly selecting 70% of the total sample number from all the first entity samples as a first training set, using the remaining 30% as a first test set, and using the first training set and the first test set for training a crf model to obtain a first entity extraction model;
randomly selecting 70% of the total sample number from all second entity samples as a second training set, using the remaining 30% as a second testing set, and using the second training set and the second testing set for training a crf model to obtain a second entity extraction model;
the first entity extraction model and the second entity extraction model are used for carrying out semantic extraction on entities on historical medical record big data to obtain a first entity and a second entity;
preferably, the removal of sample redundancy is required before the cdf model training using the first physical sample and the second physical sample, wherein,
circularly calculating the similarity between any two first entity samples/second entity samples, and randomly removing one first entity sample/second entity sample from the two first entity samples/second entity samples with the similarity exceeding a set threshold until the similarity between any two first entity samples/second entity samples does not exceed the set threshold;
the calculation formula of the similarity is as follows:
in which I is characterized by a similarity value, Ai、AjRespectively characterized as the ith and jth first/second entity samples, i, j being a metering constant.
3. The method for designing the structural report template based on semantic association according to claim 2, wherein: the semantic extracting a relationship attribute of a first condition category and a first examination item as a first entity relationship, including:
extracting semantic texts representing relation attributes of a first disease category and a first inspection item from historical disease cases corresponding to the first entity sample/the second entity sample after sample redundancy removal to serve as a first entity relation;
quantizing the historical medical records and the first entity relations into a vector form from a text form to obtain semantic vectors of the historical medical records and semantic vectors of the first entity relations, and correspondingly calibrating the semantic vectors of the historical medical records and the semantic vectors of the first entity relations into a single first entity relation sample, wherein the single first entity relation sample is characterized as [ the semantic vectors of the historical medical records, the first entity relation sample ];
randomly selecting 70% of the total sample number from all the first entity relationship samples as a first relationship training set, taking the rest 30% as a first relationship test set, and using the first relationship training set and the first relationship test set for training a BP neural network to obtain a first entity relationship extraction model;
and the first entity relationship extraction model is used for semantic extraction of relationship attributes on historical case big data to obtain a first entity relationship.
4. The method for designing the structural report template based on semantic association according to claim 3, wherein: the knowledge graph construction of the relationship among the first entity, the second entity and the first entity to obtain the first diagnosis and treatment knowledge graph comprises the following steps:
carrying out graph connection on a first entity, a second entity and a first entity relationship extracted from the same historical case to form a sub-graph of the first entity, the first entity relationship and the second entity, and sequentially setting the merging priority of the sub-graphs as the first entity, the first entity relationship and the second entity;
carrying out sub-graph merging on all sub-graphs obtained from the big data of the historical case according to the priority of sub-graph merging in sequence, wherein,
carrying out node combination on the subgraphs with the same first entity at the first entity, then carrying out node combination on graph structures with the same first entity relation in the combined subgraphs at the first entity relation, and finally combining the nodes with the same second entity in the combined graph structures so as to realize that all the subgraphs are combined to generate the first diagnosis and treatment knowledge graph;
preferably, determining that the first entities are identical comprises:
summarizing all semantic names of disease categories represented by the first entity to form a disease standard name query table, converting the semantic names of the disease categories represented by the first entity into standard semantic names based on the disease standard name query table, and judging all first entities capable of being converted into the same standard semantic names as the same first entities;
preferably, determining that the second entities are identical comprises:
and summarizing all semantic names of the inspection items represented by the second entities to form an item standard name query table, converting all the semantic names of the inspection items represented by the second entities into standard semantic names based on the item standard name query table, and judging all third entities capable of being converted into the same standard semantic names as the second entities.
5. The method for designing the structural report template based on semantic association according to claim 4, wherein: the method for extracting a second disease category and a second inspection item based on disease diagnosis and treatment guide big data semantics to serve as a third entity and a fourth entity respectively comprises the following steps:
randomly selecting a group of disease diagnosis and treatment guidelines from the disease diagnosis and treatment guideline big data, and respectively extracting semantic texts representing a second disease category and a second inspection item from the disease diagnosis and treatment guidelines to be used as a third entity and a fourth entity, wherein the disease diagnosis and treatment guidelines comprise the semantic texts representing the second disease category and the second inspection item and the semantic texts representing the relationship attributes of the second disease category and the second inspection item;
quantizing a disease diagnosis and treatment guide, a third entity and a fourth entity from a text form to a vector form to obtain a semantic vector, a third entity vector and a fourth entity vector of the disease diagnosis and treatment guide, and correspondingly calibrating the semantic vector, the third entity vector and the fourth entity vector of the disease diagnosis and treatment guide into a single third entity sample and a single fourth entity sample, wherein the single third entity sample is characterized as [ the semantic vector and the third entity vector of the disease diagnosis and treatment guide ], and the single fourth entity sample is characterized as [ the semantic vector and the fourth entity vector of the disease diagnosis and treatment guide ];
randomly selecting 70% of the total sample number from all third entity samples as a third training set, using the remaining 30% as a third testing set, and using the third training set and the third testing set for training a crf model to obtain a third entity extraction model;
randomly selecting 70% of the total sample number from all the fourth entity samples as a fourth training set, using the remaining 30% as a fourth test set, and using the fourth training set and the fourth test set for training a crf model to obtain a fourth entity extraction model;
and the third entity extraction model are used for carrying out semantic extraction on entities in the disease diagnosis and treatment guide big data to obtain a third entity and a third entity.
6. The method for designing the structural report template based on semantic association according to claim 5, wherein: the semantics extracts a relationship attribute of the second condition category and the second examination item as a third entity relationship,
extracting semantic texts representing relationship attributes of a second disease category and a second examination item from a group of disease diagnosis and treatment guidelines to serve as second entity relationships;
quantizing the disease diagnosis and treatment guide and the second entity relationship from a text form to a vector form to obtain a semantic vector of the disease diagnosis and treatment guide and a semantic vector of the second entity relationship, and correspondingly marking the semantic vector of the disease diagnosis and treatment guide and the semantic vector of the second entity relationship as a single second entity relationship sample, wherein the single second entity relationship sample is characterized as [ the semantic vector of the disease diagnosis and treatment guide, the second entity relationship sample ];
randomly selecting 70% of the total sample number from all the second entity relationship samples as a second relationship training set, using the remaining 30% as a second relationship test set, and using the second relationship training set and the second relationship test set for training a BP neural network to obtain a second entity relationship extraction model;
and the second entity relationship extraction model is used for carrying out semantic extraction on relationship attributes on big data of the disease diagnosis and treatment guide to obtain a second entity relationship.
7. The method for designing the structural report template based on semantic association according to claim 6, wherein the knowledge-graph construction of the relationship among the third entity, the fourth entity and the third entity to obtain the second diagnosis knowledge-graph comprises:
carrying out graph connection on the third entity, the fourth entity and the second entity relationship extracted from the same disease diagnosis and treatment guide to form a subgraph of the third entity-the second entity relationship-the fourth entity, and sequentially setting the merging priority of the subgraphs as the third entity, the second entity relationship and the fourth entity;
and all sub-graphs obtained from big data of the disease diagnosis and treatment guide are sequentially merged according to the priority of sub-graph merging, wherein,
and finally, combining the nodes identical to the fourth entity in the combined graph structure so as to combine all the sub-graphs to generate the second diagnosis and treatment knowledge graph.
8. The method according to claim 7, wherein the step of performing entity fusion on the first medical knowledge graph and the second medical knowledge graph to obtain the medical structured knowledge graph fusing diagnosis practical experience and diagnosis expert experience comprises:
setting the merging priority of all graph structures in the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph as a first entity/a third entity, a first entity relation/a second entity relation and a second entity/a fourth entity in sequence, merging the graph structures according to the merging priority, wherein,
and finally, combining the second entity and the fourth same node in the combined graph structure to realize the combination of all graph structures in the first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph to generate the diagnosis and treatment structured knowledge graph, so that the structural fusion of diagnosis and treatment practical experience and diagnosis and treatment expert experience is realized, and the standard property and the mobility are considered in the disease category structured diagnosis and treatment process.
9. The method according to claim 8, wherein the constructing a structured report template for the disease category based on the diagnosis and treatment structured knowledge graph comprises:
outputting all examination items corresponding to the disease categories by inquiring the diagnosis and treatment structured knowledge graph according to the disease categories judged by the doctor, and integrating all the examination items in the same report template to generate a structured report template used as a diagnosis and treatment standard guide for the doctor according to the disease categories.
10. The method for designing the structural report template based on semantic association according to claim 9, wherein the updating of the diagnosis and treatment structural knowledge graph comprises:
when the disease category does not have the corresponding structured report template, a doctor formulates a medical record of the disease category, and executes step S1 based on the medical record to construct a sub-graph which is in the form of a first entity-first entity relation and a second entity, and the sub-graph is added into the first diagnosis and treatment knowledge graph to merge the graph structure so as to update the first diagnosis and treatment knowledge graph;
and step S3 is executed, the updated first diagnosis and treatment knowledge graph and the second diagnosis and treatment knowledge graph are subjected to entity fusion to update the diagnosis and treatment structured knowledge graph, and amplification of the diagnosis and treatment structured knowledge graph is achieved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111461852.7A CN114330267A (en) | 2021-12-02 | 2021-12-02 | Structural report template design method based on semantic association |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111461852.7A CN114330267A (en) | 2021-12-02 | 2021-12-02 | Structural report template design method based on semantic association |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114330267A true CN114330267A (en) | 2022-04-12 |
Family
ID=81048975
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111461852.7A Pending CN114330267A (en) | 2021-12-02 | 2021-12-02 | Structural report template design method based on semantic association |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114330267A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115062120A (en) * | 2022-08-18 | 2022-09-16 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Reading knowledge graph construction method and device, processor and report generation method |
| CN115270779A (en) * | 2022-06-30 | 2022-11-01 | 山东大学齐鲁医院 | Method and system for generating ulcerative colitis structured report |
| CN116469542A (en) * | 2023-04-20 | 2023-07-21 | 智远汇壹(苏州)健康医疗科技有限公司 | Personalized medical image diagnosis path generation system and method |
-
2021
- 2021-12-02 CN CN202111461852.7A patent/CN114330267A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115270779A (en) * | 2022-06-30 | 2022-11-01 | 山东大学齐鲁医院 | Method and system for generating ulcerative colitis structured report |
| CN115270779B (en) * | 2022-06-30 | 2024-04-12 | 山东大学齐鲁医院 | Method and system for generating ulcerative colitis structured report |
| CN115062120A (en) * | 2022-08-18 | 2022-09-16 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Reading knowledge graph construction method and device, processor and report generation method |
| CN116469542A (en) * | 2023-04-20 | 2023-07-21 | 智远汇壹(苏州)健康医疗科技有限公司 | Personalized medical image diagnosis path generation system and method |
| CN116469542B (en) * | 2023-04-20 | 2024-01-02 | 智远汇壹(苏州)健康医疗科技有限公司 | Personalized medical image diagnosis path generation system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ding et al. | Knowledge-based approaches in software documentation: A systematic literature review | |
| CN114330267A (en) | Structural report template design method based on semantic association | |
| CN111708874A (en) | Man-machine interaction question-answering method and system based on intelligent complex intention recognition | |
| CN112015917A (en) | Data processing method and device based on knowledge graph and computer equipment | |
| CN107145511A (en) | Structured medical data library generating method and system based on medical science text message | |
| CN110675944A (en) | Triage method and device, computer equipment and medium | |
| Hu et al. | Automatic construction of chinese herbal prescriptions from tongue images using CNNs and auxiliary latent therapy topics | |
| CN107111608A (en) | N members and conceptual relation are automatically generated from language in-put data | |
| CN109670727A (en) | A kind of participle mark quality evaluation system and appraisal procedure based on crowdsourcing | |
| WO2022267678A1 (en) | Video consultation method and apparatus, device and storage medium | |
| CN111048167A (en) | Hierarchical case structuring method and system | |
| Pavlov et al. | Reference medical datasets (MosMedData) for independent external evaluation of algorithms based on artificial intelligence in diagnostics | |
| CN112541086A (en) | Knowledge graph construction method for cerebral apoplexy | |
| CN114023462A (en) | Computerized clinical guideline construction method and device based on graphical representation | |
| CN112071431B (en) | Clinical path automatic generation method and system based on deep learning and knowledge graph | |
| Ross et al. | A case-based reasoning system for conflict resolution: design and implementation | |
| CN116721699A (en) | Intelligent recommendation method based on tumor gene detection result | |
| Basharat et al. | Semantically enriched task and workflow automation in crowdsourcing for linked data management | |
| CN113257439B (en) | Function customization method and system for remote medical consultation platform | |
| Azzi et al. | Competency questions for biomedical ontology reuse | |
| Li et al. | Mapping client messages to a unified data model with mixture feature embedding convolutional neural network | |
| CN112579790A (en) | Method and device for constructing severe disease knowledge base, storage medium and electronic equipment | |
| CN110289065A (en) | A kind of auxiliary generates the control method and device of medical electronic report | |
| Pankhurst et al. | Introduction of Systematized Nomenclature of Medicine–Clinical Terms Coding Into an Electronic Health Record and Evaluation of its Impact: Qualitative and Quantitative Study | |
| WO2009151093A1 (en) | Database system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |