Hospital data dictionary mapping method, device, computing equipment and storage medium

Through the pre-training mapping model, the mapping between the hospital's main data dictionary and the business system dictionary is automatically completed, which solves the problem of inconsistent dictionary standards in the hospital's information system, and realizes efficient data dictionary mapping, reduces manpower consumption and ensures data consistency .

CN118588313BActive Publication Date: 2024-11-15四川互慧软件有限公司 +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411067303.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-15
Estimated Expiration
2044-08-06

AI Technical Summary

Technical Problem

In the hospital information system, due to the self-built data dictionary of departments and departments, the dictionary standards are inconsistent, which affects the aggregation and statistical efficiency of business data, and may even cause safety accidents. The existing manual mapping methods are inefficient and consume a lot of manpower.

Method used

The pre-trained mapping model is used to automatically complete the mapping between the main data dictionary and the business system dictionary. By obtaining entity descriptions, calculating the correlation scores and establishing association relationships, the standardized mapping of the dictionary is realized.

Benefits of technology

It realizes automated dictionary mapping, reduces manpower consumption, ensures the unity and consistency of hospital data dictionaries, improves data processing efficiency, and avoids the inefficiency problem of manual mapping.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118588313B_ABST
    Figure CN118588313B_ABST
Patent Text Reader

Abstract

The present application provides a hospital data dictionary mapping method, apparatus, computing device and storage medium, wherein the hospital data dictionary mapping method comprises the following steps: S10, obtaining the hospital's master data dictionary and the dictionary of the business system to be mapped; S20, selecting any entity description contained in the dictionary of the business system to be mapped as the entity description to be processed; S30, determining the correlation score between the entity description contained in the master data dictionary and the entity description to be processed based on a pre-trained mapping model; S40 determining the entity description with a correlation score higher than a preset threshold as the target entity description corresponding to the entity description to be processed; S50, executing step S20 until the target entity description corresponding to all entity descriptions in the dictionary of the business system to be mapped is obtained to achieve the mapping between the dictionary of the business system to be mapped and the master data dictionary, thereby ensuring the consistency between the master data dictionary under the hospital standard and the dictionaries of each business system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a hospital data dictionary mapping method. The present application also relates to a hospital data dictionary mapping device, a computing device, and a computer-readable storage medium. Background Art

[0002] With the development of computer technology, information systems are being used more and more widely in various fields. Similarly, hospitals also have their own information systems, namely HIS systems. However, due to historical reasons, the construction of hospital information systems before the advent of the big data era was based on departments and divisions, each of which built its own data dictionary. This caused the problem of inconsistent dictionary standards. After the advent of the big data era, all business data will be aggregated to form a data lake. This inconsistent dictionary standard will cause various statistical-related problems, affect the operating efficiency of the hospital, and even cause safety accidents.

[0003] In the existing collection of hospital business data, each data dictionary needs to be manually mapped to achieve dictionary standardization. However, this method of mapping each business dictionary with the master data dictionary by manual coding is not only inefficient, but also wastes a lot of manpower. Summary of the invention

[0004] In view of this, the embodiment of the present application provides a hospital data dictionary mapping method to solve the technical defects existing in the prior art. The embodiment of the present application also provides a hospital data dictionary mapping device, a computing device, and a computer-readable storage medium.

[0005] According to a first aspect of an embodiment of the present application, a hospital data dictionary mapping method is provided, comprising the following steps:

[0006] S10, obtaining the master data dictionary of the hospital and the dictionary of the business system to be mapped;

[0007] S20, selecting any entity description included in the dictionary of the business system to be mapped as the entity description to be processed;

[0008] S30, determining, according to the pre-trained mapping model, a correlation score between the entity description contained in the master data dictionary and the entity description to be processed;

[0009] S40 determines the entity description whose relevance score is higher than a preset threshold as the target entity description corresponding to the entity description to be processed;

[0010] S50, executing step S20 until the target entity descriptions corresponding to all entity descriptions in the dictionary of the business system to be mapped are obtained, and the mapping between the master data dictionary and the dictionary of the business system to be mapped is completed.

[0011] Optionally, the step S30 includes:

[0012] S31, based on a preset vector conversion algorithm, converting the entity description to be processed into a vector to be processed;

[0013] S32, selecting any entity description included in the master data dictionary as a reference entity description;

[0014] S33, based on the vector conversion algorithm, converting the comparison entity description into a comparison vector;

[0015] S34, taking the vector to be processed and the reference vector as inputs of the mapping model, and outputting a correlation score between the entity description to be processed and the reference entity description;

[0016] S35, executing step S32 until the correlation scores of all entity descriptions in the master data dictionary and the entity description to be processed are determined.

[0017] Optionally, the training process of the mapping model in step S30 includes:

[0018] Constructing a sample data set according to the master data dictionary;

[0019] Inputting the sample data pairs contained in the sample data set into the initial mapping model to obtain a correlation prediction score;

[0020] Constructing a loss function according to the sample scores corresponding to the sample data and the correlation prediction scores contained in the sample data set;

[0021] The initial mapping model is adjusted by using the loss function until the training termination condition is met to obtain the pre-trained mapping model.

[0022] Optionally, constructing a sample data set according to the master data dictionary includes:

[0023] Select any two entity descriptions in the master data dictionary to construct the sample data pair;

[0024] Determining whether two entity descriptions included in the sample data pair correspond to the same entity;

[0025] If so, mark the sample score corresponding to the sample data as 1;

[0026] If not, mark the sample score corresponding to the sample data as 0;

[0027] The step of selecting any two entity descriptions in the main data dictionary and constructing the sample data pairs is performed until all entity descriptions in the main data dictionary are selected, and the sample data set is constructed based on all the obtained sample data pairs and the corresponding sample scores.

[0028] Optionally, the step of inputting the sample data pairs in the sample data set into an initial mapping model to obtain a correlation prediction score includes:

[0029] Performing vector conversion on the two entity descriptions in the sample data pair to obtain sample vectors corresponding to the two entity descriptions, and receiving the sample vectors through the input layer of the initial mapping model;

[0030] The sample vector is processed by the convolution layer of the initial mapping model, and the processing result of the sample vector is reduced in dimension by the pooling layer of the initial mapping model to obtain corresponding sample features;

[0031] The sample features are processed through the fully connected layer of the initial mapping model to obtain corresponding feature representations, and the feature representations are mapped through the output layer of the initial mapping model to obtain the association degree prediction score.

[0032] Optionally, the adjusting parameters of the initial mapping model by using the loss function until a training termination condition is met to obtain the pre-trained mapping model includes:

[0033] The parameters of the initial mapping model are adjusted through the loss function, and the learning rate of the initial mapping model is adjusted through a preset learning rate adjustment strategy. The complexity of the initial mapping model is controlled by combining L1 regularization and L2 regularization until the training termination condition is met, thereby obtaining the pre-trained mapping model.

[0034] Optionally, the adjusting parameters of the initial mapping model by using the loss function until a training termination condition is met to obtain the pre-trained mapping model includes:

[0035] Adjusting the parameters of the initial mapping model by using the loss function;

[0036] Calculate the accuracy, precision and recall of the initial mapping model after parameter adjustment;

[0037] Repeat the step of adjusting the parameters of the initial mapping model through the loss function until the accuracy, the precision and the recall rate meet the preset parameter requirements to obtain the mapping model.

[0038] According to a second aspect of an embodiment of the present application, a hospital data dictionary mapping device is provided, comprising:

[0039] An acquisition module is configured to acquire a master data dictionary of a hospital and a dictionary of a business system to be mapped;

[0040] A selection module is configured to select any entity description contained in the dictionary of the business system to be mapped as the entity description to be processed;

[0041] A scoring module is configured to determine a relevance score between the entity description contained in the master data dictionary and the entity description to be processed according to a pre-trained mapping model;

[0042] A determination module is configured to determine the entity description having a relevance score higher than a preset threshold as a target entity description corresponding to the entity description to be processed;

[0043] The mapping module is configured to call the selection module until the target entity descriptions corresponding to all entity descriptions in the dictionary of the business system to be mapped are obtained, and the mapping between the master data dictionary and the dictionary of the business system to be mapped is completed.

[0044] According to a third aspect of an embodiment of the present application, a computing device is provided, including:

[0045] Memory and processor;

[0046] The memory is used to store computer executable instructions, and the processor implements the steps of the hospital data dictionary mapping method when executing the computer executable instructions.

[0047] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the hospital data dictionary mapping method are implemented.

[0048] According to a fifth aspect of an embodiment of the present application, a chip is provided, which stores a computer program, and when the computer program is executed by the chip, the steps of the hospital data dictionary mapping method are implemented.

[0049] The hospital data dictionary mapping method provided in the present application obtains the hospital's master data dictionary and the dictionary of the business system to be mapped through S10; S20, selects any entity description contained in the business system dictionary to be mapped as the entity description to be processed; S30, determines the correlation score between the entity description contained in the master data dictionary and the entity description to be processed according to the pre-trained mapping model; S40 determines the entity description with the correlation score higher than the preset threshold as the target entity description corresponding to the entity description to be processed; S50, executes step S20 until the target entity description corresponding to all entity descriptions in the business system dictionary to be mapped is obtained, and the mapping between the master data dictionary and the business system dictionary to be mapped is completed. The mapping between the business system dictionary to be mapped and the master data dictionary is automatically completed without manual participation, reducing manpower consumption, and ensuring the consistency of the master data dictionary under the hospital standard and the dictionaries of each business system. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 It is a flowchart of a hospital data dictionary mapping method provided by an embodiment of the present application;

[0052] Figure 2 It is a structural schematic diagram of a hospital data dictionary mapping device provided by an embodiment of the present application;

[0053] Figure 3 It is a structural block diagram of a computing device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0054] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0055] The terms used in one or more embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of the present application. The singular forms of "a", "said" and "the" used in one or more embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of the present application refers to and includes any or all possible combinations of one or more associated listed items.

[0056] It should be understood that, although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present application, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0057] First, the terms involved in one or more embodiments of the present invention are explained.

[0058] Word2vec: is a group of related models used to generate word vectors. These models are shallow, two-layer neural networks that are trained to reconstruct linguistic word texts.

[0059] ReLU: Linear rectification function, also known as rectified linear unit, is an activation function commonly used in artificial neural networks, usually referring to nonlinear functions represented by ramp functions and their variants.

[0060] Sigmoid: is an activation function that maps the input vector to a real value between 0 and 1.

[0061] Embedding: Embedded features can be representations of text, images, audio, or other types of data. Their purpose is to convert raw data into a vector form so that machine learning models can better understand and process it.

[0062] In the present application, a hospital data dictionary mapping method is provided. The present application also relates to a hospital data dictionary mapping device, a computing device, and a computer-readable storage medium, which are described in detail one by one in the following embodiments.

[0063] Figure 1 A flowchart of a hospital data dictionary mapping method provided according to an embodiment of the present application is shown, which specifically includes the following steps:

[0064] S10, obtaining the master data dictionary of the hospital and the dictionary of the business system to be mapped;

[0065] S20, selecting any entity description included in the dictionary of the business system to be mapped as the entity description to be processed;

[0066] S30, determining, according to the pre-trained mapping model, a correlation score between the entity description contained in the master data dictionary and the entity description to be processed;

[0067] S40 determines the entity description whose relevance score is higher than a preset threshold as the target entity description corresponding to the entity description to be processed;

[0068] S50, executing step S20 until the target entity descriptions corresponding to all entity descriptions in the dictionary of the business system to be mapped are obtained, and the mapping between the master data dictionary and the dictionary of the business system to be mapped is completed.

[0069] In the actual information construction of medical systems, since hospital-related medical equipment comes from different manufacturers in different countries, different departments use different types of text for medical equipment from different sources, and have different description habits for the same entity. This leads to inconsistent dictionary standards for business systems built on this basis by various departments.

[0070] In addition, even if the relevant equipment is localized in Chinese by each department or the equipment itself has a Chinese version, the equipment from different sources will also have inconsistent business system dictionary standards due to different manufacturers' different description habits of the same entity. For departments that use less medical equipment and more manual records in the field work environment, the problem of inconsistent business system dictionary standards will be more serious when the manually recorded content is imported into storage devices for information processing due to the different recording habits of medical workers.

[0071] Based on this, obtain the master data dictionary as the standard and the dictionary of the business system that needs to be standardized, that is, the dictionary of the business system to be mapped. Select an entity description contained in the dictionary of the system to be mapped in turn. The currently selected entity description is called the entity description to be processed. It should be noted that the entity description includes information such as name, definition and attributes. For example, the name "male" corresponds to the definition of "a male human, as opposed to a female human, that is, a woman", and the corresponding attribute is "gender". The above is a schematic description of the information contained in the entity description. The specific name, definition, attribute and other information are determined by the actual usage scenario, and this embodiment does not limit it.

[0072] In addition, in actual usage scenarios, since some entity descriptions have similar characteristics, such as "current address" and "place of registered residence", these two types of entity descriptions correspond to information related to location coordinates. When processing these two types of entity descriptions, confusion is prone to occur. Therefore, attributes are introduced to distinguish different types of entity descriptions to prevent confusion.

[0073] Afterwards, the entity descriptions selected in sequence are mapped with the entity descriptions contained in the main data dictionary. The mapping process is carried out through a pre-trained mapping model. Through the mapping model, the association scores between the entity descriptions selected in sequence from the dictionary of the business system to be mapped and the entity descriptions contained in the main data dictionary can be obtained. For any entity description selected from the dictionary of the business system to be mapped, the association score between it and all the entity descriptions in the main data dictionary represents the probability that the entity description corresponds to the same entity with each entity description in the main data dictionary.

[0074] By establishing an association relationship between entity descriptions in the master data dictionary whose association scores are higher than a preset threshold and entity descriptions in the dictionary of the business system to be mapped, mapping between the business dictionary to be mapped and the master data dictionary is achieved after all entity descriptions contained in the dictionary of the business system to be mapped have established an association relationship with the entity descriptions in the master data dictionary.

[0075] It should be noted that in the traditional mapping process between data dictionaries, the entity descriptions contained in different data dictionaries are manually associated and the corresponding mapping tables are established. However, through the mapping model, after determining the association between the entity descriptions contained in the dictionary of the business system to be mapped and the entity descriptions contained in the main data dictionary, only manual judgment of the association results is required. This reduces the workload of personnel and realizes the construction of data dictionary mapping that meets user application needs.

[0076] Furthermore, in step S30, the process of determining the correlation score between the entity description contained in the master data dictionary and the entity description to be processed according to the pre-trained mapping model is specifically implemented as follows in this embodiment:

[0077] S31, based on a preset vector conversion algorithm, converting the entity description to be processed into a vector to be processed; S32, selecting any entity description contained in the master data dictionary as a reference entity description; S33, based on the vector conversion algorithm, converting the reference entity description into a reference vector; S34, using the vector to be processed and the reference vector as inputs of the mapping model, and outputting a correlation score between the entity description to be processed and the reference entity description; S35, executing step S32 until all entity descriptions in the master data dictionary and the correlation scores with the entity description to be processed are determined.

[0078] Among them, the vector conversion algorithm can be Word2Vec, Embidding, or a neural network model, etc. The specific model selection is determined by the actual usage scenario and is not limited in this embodiment.

[0079] Based on this, the description of the entity to be processed is converted into a vector to obtain the vector to be processed, and then the entity descriptions contained in the main data dictionary are selected in turn. The currently selected entity description is called the reference entity description, and the entity description selected in the main data dictionary is also converted into a vector. The vector to be processed and the vector of the entity description in the main data dictionary are input into the mapping model, and the mapping model outputs the corresponding association score.

[0080] Furthermore, the training process of the mapping model in step S30 is specifically implemented as follows in this embodiment:

[0081] According to the master data dictionary, a sample data set is constructed; the sample data pairs contained in the sample data set are input into the initial mapping model to obtain a correlation prediction score; a loss function is constructed according to the sample scores corresponding to the sample data contained in the sample data set and the correlation prediction score; the initial mapping model is adjusted by the loss function until the training termination condition is met, thereby obtaining the pre-trained mapping model.

[0082] Among them, since the mapping model is used to process the entity description of the data dictionary of the hospital system, its training sample should select the data dictionary of the hospital system, so the main data dictionary is used as the training sample for training.

[0083] Based on this, a sample data set is constructed through the master data dictionary. The sample data set contains multiple sample data pairs, and the sample data pairs are composed of two entity descriptions in the data dictionary. The sample data pairs are input into the initial mapping model to obtain the association prediction score that predicts the probability that the two entity descriptions in the sample data pair correspond to the same entity. The loss function is constructed through the predicted score and the sample score that characterizes whether the two entity descriptions in the sample data pair actually correspond to the same entity. The initial mapping model is adjusted through the loss function to obtain the mapping model.

[0084] Furthermore, the process of constructing a sample data set according to the master data dictionary is specifically implemented as follows in this embodiment:

[0085] Select any two entity descriptions in the master data dictionary to construct the sample data pair; determine whether the two entity descriptions included in the sample data pair correspond to the same entity; if so, mark the sample score corresponding to the sample data as 1; if not, mark the sample score corresponding to the sample data as 0; execute the step of selecting any two entity descriptions in the master data dictionary to construct the sample data pair until all entity descriptions in the master data dictionary are selected, and construct the sample data set based on all the obtained sample data pairs and the corresponding sample scores.

[0086] Among them, select any entity description in the master data dictionary as the source entity description, and then select another entity description as the target entity description. The source entity description and the target entity description constitute a sample data pair. It should be noted that the target entity description can be another version of the same entity description or a description of a different entity compared to the source entity description.

[0087] Based on this, when the source entity description and the target entity description belong to the same entity, the sample score is marked as 1, otherwise the sample score is marked as 0. Based on the mapping model accuracy required by the user, a sample data pair is constructed. When the user needs a mapping model with higher accuracy, the number of sample data pairs should be increased, and when the user does not need a mapping model with higher accuracy, the number of sample data pairs can be appropriately reduced. Finally, the constructed sample data pairs and the corresponding sample score sets are used to obtain a sample data set. It should be noted that the sample descriptions contained in the sample data set can be adjusted, such as padding or truncating the text to ensure consistent length, etc., to ensure that the initial mapping model better processes the entity descriptions in the sample data set.

[0088] Furthermore, the sample data pairs in the sample data set are input into the initial mapping model to obtain the correlation prediction score. In this embodiment, the specific implementation method is as follows:

[0089] The two entity descriptions in the sample data pair are vectorized to obtain sample vectors corresponding to the two entity descriptions, and the sample vectors are received through the input layer of the initial mapping model; the sample vectors are processed through the convolution layer of the initial mapping model, and the processing results of the sample vectors are reduced in dimension through the pooling layer of the initial mapping model to obtain corresponding sample features; the sample features are processed through the fully connected layer of the initial mapping model to obtain corresponding feature representations, and the feature representations are mapped through the output layer of the initial mapping model to obtain the association prediction score.

[0090] For any sample data pair in the sample data set, the source entity description and the target entity description contained therein are vector-converted to obtain a corresponding sample vector in vector form.

[0091] Specifically, for example, the source entity description "M" is "M is the abbreviation of male, which is translated into male in Chinese", and the target entity description "male" is "male refers to a male human, as opposed to a female human, that is, a woman"; the two entity descriptions are vectorized by the Word2Vec model to obtain corresponding vectors; in the process of vectorization, the words that make up the entity description need to be vectorized. If the word vector of "male" is [0.2, 0.3, -0.1, ...], and the word vector of "male" is [0.1, 0.4, -0.2,...], the corresponding sample vector can be obtained by calculating the average or weighted average of the vectors of all words contained in each entity description. The actual method of determining the sample vector is determined by the actual usage scenario, and this embodiment does not limit it.

[0092] Based on this, by inputting the sample vector into the input layer of the initial mapping model, the input layer performs preprocessing operations such as normalization and standardization on the received sample vector, and converts the sample data into a form that can be processed by the initial mapping model. It should be noted that the specific preprocessing operation is determined by the actual usage scenario and is not limited in this embodiment.

[0093] Afterwards, the convolution layer processes the preprocessed sample vector through multiple convolution kernels of different sizes, each of which is used to capture features of different sizes; each convolution kernel slides on the sample vector, extracts local features and retains the word order information of the sample vector, thereby converting local information into a more advanced feature representation. It should be noted that the number of convolution kernels is determined by actual usage requirements and is not limited in this embodiment.

[0094] In addition, the convolutional layer also introduces the learning of nonlinear association relationships through activation functions to enhance the expression and fitting capabilities of the initial mapping model. For example, the ReLU activation function is introduced. For the input sample vector X = [x1, x2, x3], the ReLU activation function is expressed by the formula , process the elements in the sample vector X, where The subscripts representing the elements in the sample vector are 1, 2, and 3, which means that if the element in vector X is greater than 0, the element itself is output, and if it is less than 0, 0 is input.

[0095] The sample vector processed by the convolution layer will be processed by the pooling layer, that is, the sample vector processed by the convolution layer is reduced in dimension through the maximum pooling layer, and the most significant features are extracted to obtain the sample features. The maximum pooling operation retains the most important features captured by each convolution kernel. Then, through the fully connected layer whose neurons are connected to all neurons in the previous layer, the sample features are mapped to a higher-dimensional space to obtain feature representation, so that the fully connected layer learns nonlinear combinations between features.

[0096] Finally, according to the node-based output layer, the feature representation is predicted for the binary classification task. For example, the feature representation is mapped to the [0, 1] interval through the Sigmoid activation function to obtain a correlation prediction score. The closer the correlation prediction score is to 1, the more likely the two entity descriptions in the sample data pair are to belong to the same entity. Conversely, if the correlation prediction score is closer to 0, the more likely the two entity descriptions in the sample data pair are to belong to different entities.

[0097] Specifically, the Sigmoid activation function formula is:

[0098] ,

[0099] Among them, y is each element in the feature representation.

[0100] Furthermore, the initial mapping model is adjusted by the loss function until the training termination condition is met to obtain the pre-trained mapping model. In this embodiment, the specific implementation method is as follows:

[0101] The parameters of the initial mapping model are adjusted through the loss function, and the learning rate of the initial mapping model is adjusted through a preset learning rate adjustment strategy. The complexity of the initial mapping model is controlled by combining L1 regularization and L2 regularization until the training termination condition is met, thereby obtaining the pre-trained mapping model.

[0102] Among them, in the process of training the initial mapping model, the learning rate is adjusted through methods such as learning rate decay and adaptive learning rate to improve the stability and convergence speed of model training; the batch size is adjusted to affect the training speed and generalization ability of the model. In actual usage scenarios, different batch sizes are usually tried and the model performance is observed; in addition, L1 regularization or L2 regularization terms are added to control the complexity of the model to prevent overfitting; data enhancement operations such as random rotation, translation, scaling, etc. are performed on the entity descriptions in the sample data set used for training to increase the diversity of training samples and improve the generalization ability of the model; model integration methods such as voting, averaging, stacking, etc. can also be used to combine the prediction results of multiple models to improve performance.

[0103] Furthermore, regarding the process of satisfying the training termination condition, in this embodiment, the specific implementation method is as follows:

[0104] The initial mapping model is adjusted through the loss function; the accuracy, precision and recall of the initial mapping model after parameter adjustment are calculated; and the step of adjusting the parameters of the initial mapping model through the loss function is repeated until the accuracy, precision and recall meet the preset parameter requirements to obtain the mapping model.

[0105] Among them, the calculation formula of accuracy A is:

[0106] ,

[0107] Among them, TP represents true positive examples, that is, the model correctly predicts positive class samples as positive classes, TN represents true negative examples, that is, the model correctly predicts negative class samples as negative classes, FP represents false positive examples, that is, the model incorrectly predicts negative class samples as positive classes, and FN represents false negative examples, that is, the model incorrectly predicts positive class samples as negative classes.

[0108] The precision rate P is the proportion of samples that are actually positive in the samples predicted by the model. The calculation formula of the precision rate P is:

[0109] ,

[0110] The recall rate R is the ratio of the number of positive class samples correctly identified by the model to the total number of actual positive class samples, also known as sensitivity or true rate. The calculation formula of the recall rate R is:

[0111] ,

[0112] In addition, the F1 value can be calculated to comprehensively consider the accuracy and coverage of the model. The calculation formula of the F1 value is:

[0113] ,

[0114] In the process of repeatedly adjusting the parameters of the initial mapping model, when the accuracy, precision, recall rate and F1 value of the initial mapping model all meet the preset parameter requirements, it can be considered that the initial mapping model can now more accurately predict whether two entity descriptions correspond to the same entity. At this time, the training of the initial mapping model is completed and the pre-trained mapping model is obtained.

[0115] Corresponding to the above method embodiment, the present application also provides a hospital data dictionary mapping device embodiment, Figure 2 FIG. 1 is a schematic diagram showing the structure of a hospital data dictionary mapping device provided by an embodiment of the present application. Figure 2 As shown, the device comprises:

[0116] The acquisition module 202 is configured to acquire the master data dictionary of the hospital and the dictionary of the business system to be mapped;

[0117] A selection module 204 is configured to select any entity description contained in the dictionary of the business system to be mapped as the entity description to be processed;

[0118] A scoring module 206 is configured to determine a correlation score between the entity description included in the master data dictionary and the entity description to be processed according to a pre-trained mapping model;

[0119] A determination module 208 is configured to determine the entity description having a relevance score higher than a preset threshold as a target entity description corresponding to the entity description to be processed;

[0120] The mapping module 210 is configured to call the selection module 204 until the target entity descriptions corresponding to all entity descriptions in the business system dictionary to be mapped are obtained, thereby completing the mapping between the master data dictionary and the business system dictionary to be mapped.

[0121] In an optional embodiment, the scoring module 206 is further configured to:

[0122] The first conversion submodule is configured to convert the entity description to be processed into a vector to be processed based on a preset vector conversion algorithm; the comparison submodule is configured to select any entity description contained in the main data dictionary as a comparison entity description; the second conversion submodule is configured to convert the comparison entity description into a comparison vector based on the vector conversion algorithm; the input submodule is configured to use the vector to be processed and the comparison vector as inputs of the mapping model, and output a correlation score between the entity description to be processed and the comparison entity description; the repetition submodule is configured to call the comparison submodule until the correlation scores of all entity descriptions in the main data dictionary and the entity description to be processed are determined.

[0123] In an optional embodiment, the hospital data dictionary mapping device further includes:

[0124] The pre-training module is configured to construct a sample data set according to the master data dictionary; input the sample data pairs contained in the sample data set into the initial mapping model to obtain a correlation prediction score; construct a loss function according to the sample scores corresponding to the sample data contained in the sample data set and the correlation prediction score; adjust the parameters of the initial mapping model through the loss function until the training termination condition is met, thereby obtaining the pre-trained mapping model.

[0125] In an optional embodiment, the pre-training module is further configured as follows:

[0126] Select any two entity descriptions in the master data dictionary to construct the sample data pair; determine whether the two entity descriptions included in the sample data pair correspond to the same entity; if so, mark the sample score corresponding to the sample data as 1; if not, mark the sample score corresponding to the sample data as 0; execute the step of selecting any two entity descriptions in the master data dictionary to construct the sample data pair until all entity descriptions in the master data dictionary are selected, and construct the sample data set based on all the obtained sample data pairs and the corresponding sample scores.

[0127] In an optional embodiment, the pre-training module is further configured as follows:

[0128] The two entity descriptions in the sample data pair are vectorized to obtain sample vectors corresponding to the two entity descriptions, and the sample vectors are received through the input layer of the initial mapping model; the sample vectors are processed through the convolution layer of the initial mapping model, and the processing results of the sample vectors are reduced in dimension through the pooling layer of the initial mapping model to obtain corresponding sample features; the sample features are processed through the fully connected layer of the initial mapping model to obtain corresponding feature representations, and the feature representations are mapped through the output layer of the initial mapping model to obtain the association prediction score.

[0129] In an optional embodiment, the pre-training module is further configured as follows:

[0130] The parameters of the initial mapping model are adjusted through the loss function, and the learning rate of the initial mapping model is adjusted through a preset learning rate adjustment strategy. The complexity of the initial mapping model is controlled by combining L1 regularization and L2 regularization until the training termination condition is met, thereby obtaining the pre-trained mapping model.

[0131] In an optional embodiment, the pre-training module is further configured as follows:

[0132] The initial mapping model is adjusted through the loss function; the accuracy, precision and recall of the initial mapping model after parameter adjustment are calculated; and the step of adjusting the parameters of the initial mapping model through the loss function is repeated until the accuracy, precision and recall meet the preset parameter requirements to obtain the mapping model.

[0133] The hospital data dictionary mapping device provided in the present application obtains the hospital's master data dictionary and the dictionary of the business system to be mapped through S10; S20, selects any entity description contained in the business system dictionary to be mapped as the entity description to be processed; S30, determines the correlation score between the entity description contained in the master data dictionary and the entity description to be processed according to the pre-trained mapping model; S40 determines the entity description with the correlation score higher than the preset threshold as the target entity description corresponding to the entity description to be processed; S50, executes step S20 until the target entity description corresponding to all entity descriptions in the business system dictionary to be mapped is obtained, and the mapping between the master data dictionary and the business system dictionary to be mapped is completed. The mapping between the business system dictionary to be mapped and the master data dictionary is automatically completed without manual participation, reducing manpower consumption, and ensuring the consistency of the master data dictionary under the hospital standard and the dictionaries of each business system.

[0134] The above is a schematic scheme of a hospital data dictionary mapping device of the present embodiment. It should be noted that the technical scheme of the hospital data dictionary mapping device and the technical scheme of the hospital data dictionary mapping method mentioned above belong to the same concept. For the details not described in detail in the technical scheme of the hospital data dictionary mapping device, please refer to the description of the technical scheme of the hospital data dictionary mapping method mentioned above. In addition, each component in the device embodiment should be understood as a functional module that must be established to implement each step of the program flow or each step of the method, and each functional module is not an actual functional division or separation definition. The device claim defined by such a group of functional modules should be understood as a functional module architecture that mainly implements the solution through the computer program recorded in the specification, and should not be understood as a physical device that mainly implements the solution through hardware.

[0135] Figure 3 The block diagram of a computing device 300 provided according to an embodiment of the present application is shown. The components of the computing device 300 include but are not limited to a memory 310 and a processor 320. The processor 320 is connected to the memory 310 via a bus 330, and the database 350 is used to store data.

[0136] The computing device 300 also includes an access device 340 that enables the computing device 300 to communicate via one or more networks 360. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 340 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a World Wide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0137] In one embodiment of the present application, the above components of the computing device 300 and Figure 3 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 3 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of the present application. Those skilled in the art may add or replace other components as needed.

[0138] The computing device 300 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or PC. The computing device 300 may also be a mobile or stationary server.

[0139] The processor 320 is used to execute computer executable instructions of each step of the hospital data dictionary mapping method.

[0140] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the hospital data dictionary mapping method described above are of the same concept, and the details not described in detail in the technical scheme of the computing device can be found in the description of the technical scheme of the hospital data dictionary mapping method described above.

[0141] An embodiment of the present application also provides a computer-readable storage medium storing computer instructions, which are used to execute the steps of the hospital data dictionary mapping method when executed by a processor.

[0142] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the hospital data dictionary mapping method mentioned above belong to the same concept, and the details not described in detail in the technical scheme of the storage medium can be found in the description of the technical scheme of the hospital data dictionary mapping method mentioned above.

[0143] An embodiment of the present application further provides a chip storing a computer program, which implements the steps of the hospital data dictionary mapping method when executed by the chip.

[0144] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0145] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0146] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0147] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0148] The preferred embodiments of the present application disclosed above are only used to help explain the present application. The optional embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the present application. The present application selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can understand and use the present application well. The present application is only limited by the claims and their full scope and equivalents.

Claims

1. A hospital data dictionary mapping method, characterized in that: The following steps are involved: S10, obtaining the master data dictionary of the hospital and the dictionary of the business system to be mapped; S20, selecting any entity description contained in the dictionary of the business system to be mapped as the entity description to be processed, wherein the entity description includes name information, definition information and attribute information; S30, according to the pre-trained mapping model, determine the correlation score between the entity description contained in the master data dictionary and the entity description to be processed. Specifically, S31, based on the preset vector conversion algorithm, convert the entity description to be processed into a vector to be processed; S32, select any one entity description contained in the master data dictionary as the reference entity description; S33, based on the vector conversion algorithm, convert the reference entity description into a reference vector; S34, use the reference vector to be processed and the reference vector as inputs of the mapping model, and output the correlation score between the entity description to be processed and the reference entity description; S35, execute step S32 until the correlation score between all entity descriptions in the master data dictionary and the entity description to be processed is determined; the training process of the mapping model is to select any two entity descriptions in the master data dictionary to construct a sample data pair; determine whether the two entity descriptions contained in the sample data pair correspond to the same entity; if so, mark the sample score corresponding to the sample data as 1; if not, mark the sample score corresponding to the sample data as 0; execute the selection any two entity descriptions in the master data dictionary, construct the sample data pair step, until all entity descriptions in the master data dictionary are selected, construct a sample data set based on all the obtained sample data pairs and the corresponding sample scores, perform vector conversion on the two entity descriptions in the sample data pair to obtain sample vectors corresponding to the two entity descriptions, receive the sample vectors through the input layer of the initial mapping model, process the sample vectors through the convolution layer of the initial mapping model, and perform dimension reduction on the processing results of the sample vectors through the pooling layer of the initial mapping model to obtain corresponding sample features, process the sample features through the fully connected layer of the initial mapping model to obtain corresponding feature representations, and map the feature representations through the output layer of the initial mapping model to obtain the association prediction score, construct a loss function based on the sample scores corresponding to the sample data contained in the sample data set and the association prediction score, adjust the parameters of the initial mapping model through the loss function until the training termination condition is met to obtain the pre-trained mapping model; S40, determining the entity description whose relevance score is higher than a preset threshold as the target entity description corresponding to the entity description to be processed; S50, executing step S20 until the target entity descriptions corresponding to all entity descriptions in the dictionary of the business system to be mapped are obtained, and the mapping between the master data dictionary and the dictionary of the business system to be mapped is completed.

2. The method according to claim 1, characterized in that: The initial mapping model is adjusted by the loss function until a training termination condition is met to obtain the pre-trained mapping model, including: The parameters of the initial mapping model are adjusted through the loss function, and the learning rate of the initial mapping model is adjusted through a preset learning rate adjustment strategy. The complexity of the initial mapping model is controlled by combining L1 regularization and L2 regularization until the training termination condition is met, thereby obtaining the pre-trained mapping model.

3. The method according to claim 1, characterized in that: The initial mapping model is adjusted by the loss function until a training termination condition is met to obtain the pre-trained mapping model, including: Adjusting the parameters of the initial mapping model by using the loss function; Calculate the accuracy, precision and recall of the initial mapping model after parameter adjustment; Repeat the step of adjusting the parameters of the initial mapping model through the loss function until the accuracy, the precision and the recall rate meet the preset parameter requirements to obtain the mapping model.

4. A hospital data dictionary mapping device, characterized in that: include: An acquisition module is configured to acquire a master data dictionary of a hospital and a dictionary of a business system to be mapped; A selection module is configured to select any entity description contained in the dictionary of the business system to be mapped as the entity description to be processed, wherein the entity description includes name information, definition information and attribute information; The scoring module is configured to determine the correlation score between the entity description contained in the main data dictionary and the entity description to be processed according to the pre-trained mapping model. Specifically, based on a preset vector conversion algorithm, the entity description to be processed is converted into a vector to be processed; any entity description contained in the main data dictionary is selected as a reference entity description; based on the vector conversion algorithm, the reference entity description is converted into a reference vector; the vector to be processed and the reference vector are used as inputs of the mapping model to output the correlation score between the entity description to be processed and the reference entity description; the scoring module is called cyclically until the correlation score between all entity descriptions in the main data dictionary and the entity description to be processed is determined; the training process of the mapping model is to select any two entity descriptions in the main data dictionary to construct a sample data pair; determine whether the two entity descriptions contained in the sample data pair correspond to the same entity; if so, mark the sample score corresponding to the sample data as 1; if not, mark the sample score corresponding to the sample data as 0; execute the selection of the main data dictionary any two entity descriptions in the dictionary, construct the sample data pair step, until all entity descriptions in the master data dictionary are selected, construct a sample data set based on all the obtained sample data pairs and the corresponding sample scores, perform vector conversion on the two entity descriptions in the sample data pair to obtain sample vectors corresponding to the two entity descriptions, receive the sample vectors through the input layer of the initial mapping model, process the sample vectors through the convolution layer of the initial mapping model, and reduce the dimension of the processing results of the sample vectors through the pooling layer of the initial mapping model to obtain corresponding sample features, process the sample features through the fully connected layer of the initial mapping model to obtain corresponding feature representations, and map the feature representations through the output layer of the initial mapping model to obtain the association prediction score, construct a loss function according to the sample scores corresponding to the sample data contained in the sample data set and the association prediction score, adjust the parameters of the initial mapping model through the loss function until the training termination condition is met, and obtain the pre-trained mapping model; A determination module is configured to determine the entity description having a relevance score higher than a preset threshold as a target entity description corresponding to the entity description to be processed; The mapping module is configured to call the selection module until the target entity descriptions corresponding to all entity descriptions in the dictionary of the business system to be mapped are obtained, and the mapping between the master data dictionary and the dictionary of the business system to be mapped is completed.

5. A computing device, characterized in that: include: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the steps of the hospital data dictionary mapping method according to any one of claims 1 to 3.

6. A computer-readable storage medium storing computer instructions, characterized in that: When the instruction is executed by the processor, the steps of the hospital data dictionary mapping method described in any one of claims 1 to 3 are implemented.