WO2020233256A1

WO2020233256A1 - General medical termbase-based multi-center medical terminology standardization system

Info

Publication number: WO2020233256A1
Application number: PCT/CN2020/083586
Authority: WO
Inventors: 李劲松; 田雨; 王执晓; 周天舒; 董凯奇
Original assignee: 之江实验室
Priority date: 2019-07-12
Filing date: 2020-04-07
Publication date: 2020-11-26
Also published as: CN110349639A; CN110349639B; JP7093593B2; JP2022508350A

Abstract

A general medical termbase-based multi-center medical terminology standardization system, comprising a source database, a database connection management module, a pre-analysis module, a terminology mapping unit, an incremental update module, an anomaly processing module and a multi-center interaction module. The system solves the problem of medical terminology standardization between multiple medical data centers, and maintains the consistency of medical terminology expressions among various medical data centers. A medical data center source database is automatically scanned and analyzed, and medical terminology having standard codes are automatically mapped on this basis. The inevitable complexities of medical terminology mapping are fully taken into account, and a spiral rising process from automated mapping to fuzzy match mapping to custom terminology mapping is implemented. An incremental update mechanism makes full use of previous mapping records, significantly reducing the pressure of subsequent work, and significantly increasing the degree of standardization of medical terminology mapping.

Description

A multi-center medical terminology standardization system based on general medical terminology database

Technical field

The invention belongs to the field of terminology standardization, and in particular relates to a multi-center medical terminology standardization system based on a general medical term database.

Background technique

With the rapid development of medical informatization, the type and scale of medical data are growing rapidly. The medical data of multiple medical data centers (referred to as "multi-center") is used for data analysis and mining, and then for clinical decision-making, medical management services, and scientific research. Providing support has become an inevitable trend. However, domestic medical terminology related standards are scarce, the system is not complete, and there are many medical information system vendors. Therefore, the heterogeneity of term names and codes between medical data centers and even medical data centers is very serious, and accompanied by a large number of semi-structured And unstructured data; while the internationally mature related term sets are limited in domestic application, and due to language barriers, the mapping relationship between existing international standard term sets is difficult to apply to the standardization of domestic medical terminology; the above reasons This makes it impossible to interoperate between medical information systems, and it is difficult to realize the standardization and sharing of medical data among multiple medical data centers.

The general medical term database is a set of "Medical Terminology Nomenclature-Clinical Terminology" (SNOMED-CT), "International Classification and Codes of Diseases" (ICD-10), Clinical Drug Standard Nomenclature (RxNorm), "Observation Index Identifier" Logical Naming Precoding System (LOINC) and other international general medical term sets as the core medical concept terminology standard library covering the entire medical process. After the multi-center medical data is mapped to a unified general medical term database, operations such as big data analysis can be conveniently performed. Before using multi-center medical data for data analysis, how to standardize and clean the medical data of different medical information systems has become a big problem.

Existing technical solutions [CN201510922676-Method and system for automatically constructing medical term mapping relationship based on word segmentation coding], [CN201710101827-A data standardization processing method and device for medical big data] and [CN201710152584-Method and device for determining medical synonyms] More from the perspective of Chinese word segmentation, medical terms are segmented based on string matching and equal word segmentation, and then the similarity between medical terms is calculated, so as to select the most similar medical term to establish a mapping relationship between the target term . This type of solution only focuses on solving the problem of matching Chinese medical terms, rather than solving the problem of terminology standardization between the entire medical information system, and only aims at the mapping between Chinese medical terms, and has not achieved standardization with foreign standard medical term sets. .

Existing technical solutions [CN201610173625-A method and system for automatic standardization of medical data dictionaries] patent documents are mainly to establish a cloud-based data dictionary standardization model at the logical level, and it is necessary to extract the term set of all medical data centers to the cloud For unified mapping processing, in reality, medical data centers such as hospitals basically will not allow the terminology set of medical data centers to be opened to the cloud outside the central local area network in order to ensure data security. Therefore, it does not meet the actual medical terminology standardization requirements.

At present, the more extensive processing method is that information technology personnel and doctors and other personnel with medical background knowledge determine the mapping relationship between data in the medical system and general medical terminology one by one, and then perform semi-automatic mapping by executing SQL scripts. Obtain standardized medical terminology; another operation of standardized medical terminology is to require medical staff to enter data in a standardized format according to a standard format when entering data. However, the current method has obvious disadvantages:

1. The prior art only focuses on the establishment of the mapping relationship between medical terms, rather than solving the standardization of medical terms in the entire medical information system.

2. The existing schemes are all aimed at a specific data model, not only lacking practicability and pertinence, but also limited to the mapping between Chinese medical terms, and cannot establish a mapping relationship with the international general medical term database.

3. For the medical data encoded in the international general medical term set, the existing mapping relationship between the existing term sets is not fully explored, and the medical terminology in the medical data center that does not use the standard medical term set is related to the medical data center. The defined medical terms are generally resolved by fuzzy matching, or discarded directly, and a complete mapping process and mechanism have not been established.

4. For terminology mapping that requires personnel with medical background knowledge to participate, a friendly interactive interface and standardized manual review and exception handling mechanisms are not provided.

5. Due to the messy data, the previous medical terminology standardization process did not combine terminology mapping with subsequent data cleaning, and it was impossible to use the mapping relationship between terms to actually complete the terminology standardization process, let alone guarantee the quality of the mapped data, which seriously affected Subsequent data analysis results.

6. For the medical data after mapping and cleaning, no detailed quality assessment mechanism has been established to ensure the accuracy of terminology mapping and data cleaning.

7. The processing mechanism after the update of the relevant international general medical terminology database and the subsequent incremental update mechanism are not considered.

Summary of the invention

In view of the shortcomings of the prior art, the present invention proposes a multi-center medical terminology standardization system based on a general medical term library, which solves the multi-center medical terminology standardization problem based on multiple standardized medical term sets, simplifies the operation of medical terminology mapping, and enriches The whole process of standardization of medical terminology.

The purpose of the present invention is achieved through the following technical solutions: a multi-center medical terminology standardization system based on general medical terminology database, which includes source database, database connection management module, pre-analysis module, term mapping unit, incremental update Module, exception handling module and multi-center interaction module;

The source database is distributed in the front-end server of each medical data center, and stores the business data of each medical data center;

The database connection management module: manages the information required to access the source database, and provides support for the terminology mapping tool to access and modify the source database;

The pre-analysis module: automatically scans the source database, counts the appearance frequency of each medical term in the original medical data, gives suggestions for rejection of terms whose appearance frequency is less than the set threshold, and sends the terms greater than or equal to the set threshold to the term mapping unit Perform subsequent term mapping;

The term mapping unit includes an automatic mapping module, a fuzzy matching module and a custom term module;

The automatic mapping module: supports automatic mapping of medical terminology, and realizes multi-directional mapping according to the existing mapping relationship between the standard codes of the international general medical terminology library for terms using the international general medical terminology library standard encoding;

The fuzzy matching module: For medical terms that cannot be directly mapped based on the mapping relationship between the internal standard codes of the existing medical terminology database, traverse and query in the general medical terminology database through fuzzy matching, and provide several sets of standards with the highest similarity Medical terms are available for selection as the target term of the term mapping;

The custom term module: For medical terms that cannot rely on the mapping relationship between the standard codes in the existing medical term database and cannot be fuzzy matching the target term in the existing general medical term database, after the user generates a custom term application, Send to the multi-center interactive module for review and feedback;

The multi-center interaction module: after receiving the custom term application from each medical data center sent by the custom term module, the custom term will be reviewed, and the reviewed custom term will be added to the general medical term database as a standard term , And send it to each medical data center to keep the common medical terminology database consistent in each medical data center;

The incremental update module: For the medical term standardization process in which the source database that has performed the standardized mapping of medical terminology generates incremental data due to business reasons, the historical mapping relationship record generated by the term mapping unit is retrieved to complete the terminology standardization mapping of the incremental data ；

The exception handling module: records the execution process of each of the above modules, generates an error log for the occurrence of errors, and can trace the entire process of medical term mapping according to the error log.

Further, the system also includes a data cleaning module, which is used to formulate cleaning rules, assign weights to each data element, and filter out severely missing data, including cleaning structure-level and instance-level dirty data.

Further, the database connection management module specifically includes: composing a JDBC module through classes and interfaces written in a programming language, providing a unified access interface for multiple types of databases, realizing the establishment of a connection with a database or other data sources, and sending SQL to the database Commands and functions to process the results returned by the database.

Further, after the database connection management module realizes the connection to the source database, the pre-analysis module automatically scans the structural information of all data in the source database and the statistical information of specific fields through the module to generate a statistical table, including two parts:

First, summarize the statistics of all tables in the source database, including the field names in each table, the value type, the maximum length of all values, the total number of rows in the table, and the proportion of empty values;

Secondly, make statistics on the detailed information and frequency of occurrence of specific terms in a specific table, and arrange them in descending order of frequency of occurrence, so that subsequent term mapping will give priority to terms with higher frequency of occurrence. Whether it is necessary for a term with a very low frequency to participate in the subsequent term mapping suggestions, if it is not defined, all terms participate in the mapping by default. The user can also adjust according to the specific situation to determine the minimum frequency threshold for not participating in the subsequent term mapping.

Further, the automatic mapping module: for terms with international general medical terminology standard codes in the source database, after determining the standard to which the codes belong, select the target term set to be mapped, if the term in the source database belongs to the standard There is already a reference mapping relationship between the term set encoding and the target term set encoding, then this part of the terminology can automatically generate mapping SQL statements to complete the automatic mapping and corresponding data loading of the terms in the source database.

Further, in the fuzzy matching module, the specific method of fuzzy matching is as follows:

(1) Terminology segmentation: segment all vocabularies in the general medical terminology database, and perform frequency statistics for each segmentation as the basic word frequency; segment the source medical term M that needs fuzzy matching before matching.

(2) Fuzzy matching: By comparing the probability difference between medical terms as the standard of similarity, the specific operation is as follows:

(2.1) Filter out all terms including word segmentation from the general medical term database, and perform word segmentation, and combine them into term set A;

(2.2) Use the following formula to calculate the matching degree to find the average weighted probability of all terms in term M and term set A; where n is the number of word segmentation for each term, and P1, P2, P3, P4...Pn is each The corresponding probability of word segmentation in basic word frequency:

(2.3) Make the difference between the average weighted probability of all standard terms in the term set A and the term M that needs fuzzy matching, and take the negative value as the matching degree. The greater the matching degree, the higher the similarity between the two. The formula is as follows:

S(M,A)=|D(M)-D(A)|

Further, the custom term module: define constraints in advance to avoid conflicts between custom terms and known standard terms; when adding custom terms, each medical data center needs to maintain the added custom standard terms To prevent repeated addition, while ensuring that multi-center medical data can be shared after terminology mapping is standardized. Before adding a custom term, you need to submit an application for adding a custom term to the multi-center interaction module. The application content includes: the custom term that needs to be added, the specific description of the custom term, the code of the custom term; pending multi-center interaction module After passing the review, the relevant operators confirm that there is no custom code of similar repeated medical terms, then a custom standard term code is generated, and then the automatic mapping module can be called to complete the term mapping and the loading of the covered data; if the review fails , Then return the existing custom term code for the medical data center to complete subsequent mapping or return the reason for the failure of custom term generation, generate an error document and prompt the user.

Further, the multi-center interaction module is responsible for the coordination and unification of the general medical terminology database and term coding of each medical data center, and the highest authority personnel of the multi-center interaction module review and coordinate the use of custom standard terms.

Further, the incremental update module is used in the subsequent medical term standardization process of the medical data center that has already operated the medical term mapping, and is mainly based on the previous terminology standardized mapping records generated by the term mapping unit to update the incremental data. For medical terms that still cannot be standardized, execute the custom term module repeatedly.

Further, the exception handling module is used to save all logs during system operation, and record whether each module is running normally; classified and save error logs, including: errors that occur during system operation, errors that occur when each module is called, Errors that occur when each module is running for a single term mapping; classified and saved terms that were not successfully mapped, including those that were ignored in the automatic analysis module and the terms that were ignored in the custom module, and the failed term document was generated; the exception handling module passed Set the time stamp on the database, support the database backtracking function, and support users to backtrack the matched database to the data on the specified date.

The beneficial effects of the present invention are: the present invention systematically solves the problem of medical terminology standardization in multiple medical data centers, and maintains the consistency of medical terminology expression in each medical data center; automatically realizes automatic scanning and analysis of the source database of the medical data center, and On this basis, the automatic mapping of medical terminology with standard coding is realized; the inevitable complexity of medical term mapping is fully considered, and a spiral process of automatic mapping to fuzzy matching mapping to custom term mapping is realized; incremental update The mechanism makes full use of past mapping records, greatly reduces the pressure of follow-up work, and greatly improves the standardization of medical terminology mapping.

Description of the drawings

Figure 1 is the system flow chart;

Figure 2 is a system data flow diagram;

Figure 3 is a schematic diagram of JDBC implementation of database connection management;

Figure 4 is a flowchart of standardized mapping of medical terms;

Figure 5 is a schematic diagram of multi-center interaction.

Detailed ways

The present invention will be further described in detail below with reference to the drawings and specific embodiments.

As shown in Figure 1, the present invention provides a multi-center medical terminology standardization system based on a general medical term database. The system includes a source database, a database connection management module, a pre-analysis module, a term mapping unit, an incremental update module, and an exception The processing module and the multi-center interaction module may also include a data cleaning module;

The source database is distributed in the front-end server of each medical data center, and stores business data of medical information systems such as HIS, LIS, PACS and EMR in each medical data center, including basic patient information, medical information, cost information, diagnosis information, medication information, Surgery information, inspection information, examination information, text medical record information and nursing vital signs information;

Database connection management module: manage (including loading, modifying, storing) the information needed to access the source database, and provide support for the terminology mapping tool to access and modify different types of source databases;

Pre-analysis module: Automatically scan the source database, count the frequency of occurrence of each medical term in the original medical data, give suggestions for rejection of terms whose frequency of occurrence is less than the set threshold, and send the terms greater than or equal to the set threshold to the term mapping unit for follow-up Term mapping

Term mapping unit includes automatic mapping module, fuzzy matching module and custom term module;

Automatic mapping module: supports automatic mapping of medical terminology. For terms that use international general medical terminology standard codes, multi-directional mapping is realized according to the mapping relationship between the existing general medical terminology standard codes, and only the quality of the mapping results is required Control

Fuzzy matching module: For medical terms that cannot be directly mapped based on the mapping relationship between the internal standard codes of the existing medical terminology database, the general medical terminology database can be traversed and searched through fuzzy matching to provide the most similar groups of standard medical care Term for selection as the target term of the term mapping;

Custom terminology module: For medical terms that cannot rely on the mapping relationship between standard codes in the existing medical terminology database and cannot fuzzy match the target term in the existing general medical terminology database, after the user generates a custom term application (can pass The joint decision of technicians and doctors), sent to the multi-center interactive module for review and feedback;

Multi-center interaction module: After receiving the custom term application from each medical data center sent by the custom term module, the custom term will be reviewed, and the reviewed custom term will be added to the general medical term database as a standard term, and Send to each medical data center to keep the common medical terminology database consistent in each medical data center;

Incremental update module: for the medical terminology standardization process in which the source database that has performed the standardized mapping of medical terminology generates incremental data due to business reasons, the historical mapping relationship records generated by the terminology mapping unit are retrieved to complete the terminology standardization mapping of the incremental data;

Exception handling module: Record the execution process of each of the above modules, especially generate error logs when errors occur, and ensure that the entire process of medical term mapping can be traced back based on the error logs.

Data cleaning module: Formulate cleaning rules, assign weight to each data element, filter out serious missing data, and improve data quality.

The specific implementation of each module is as follows:

1. Database connection management module

Manage (including loading, modifying, storing) the information needed to access the source database. The source database and the target database can be the same database system at the physical level. The realization method is mainly to compose the JDBC module through the existing classes and interfaces written in the java programming language, which provides a unified access interface for various types of databases. It has good cross-platform performance and mainly realizes the establishment of the database or other data sources. Connect, send SQL commands to the database and process the returned results of the database. The schematic diagram is shown in Figure 3.

2. Pre-analysis module: After the database connection management module realizes the connection to the source database, the module automatically scans the structure information of all data in the source database and the statistical information of specific fields to generate statistical table A, which consists of two parts:

First, the summary statistics of all tables in the source database, including the field names in each table, the value type, the maximum length of all values, the total number of rows in the table, and the proportion of empty values, are as follows:

AA	BB	CC	DD	EE	FF
表名Table Name	列名Column name	数值类型Numeric type	最大长度The maximum length	行数Rows	空行比例Blank line ratio
PATIENTPATIENT	患者标识Patient identification	NUMBERNUMBER	88	30003000	00
PATIENTPATIENT	姓名Name	VARCHAR2VARCHAR2	2020	30003000	00
PATIENTPATIENT	出生日期date of birth	DATEDATE	1010	30003000	00

Second, make statistics on the detailed information and frequency of occurrence of specific terms in a specific table, and arrange them in descending order of frequency of occurrence. The subsequent term mapping can give priority to terms with higher occurrence frequency for processing. The system will give Whether it is necessary for the terms with extremely low frequency to participate in the subsequent term mapping suggestions. If it is not defined, all terms participate in the mapping by default. Users can also adjust the parameters according to the specific situation to determine the minimum frequency threshold that does not participate in the subsequent term mapping. This can greatly simplify the subsequent term mapping process, reduce a certain amount of work and improve data quality.

AA	BB	CC
编码coding	性别gender	频次frequency
Z03.001Z03.001	男male	200200
Z03.002Z03.002	女Female	100100

For example: a term A is a non-standard term, its total amount is N2, and the total amount of data is N1, then the frequency of A is P=N1/N2. M is the set minimum frequency of participation in the mapping. If P≥M, then A is the object of participation in the mapping; P<M, then A is a non-standard term with a very low frequency and does not participate in subsequent term mapping, where M is the user's basis Threshold set by the actual situation.

The document information generated by the above modules can be exported in pdf, excel, CSV and other formats.

Three, automatic mapping module

For terms with international general medical terminology standard codes in the source database, after determining the standard to which the codes belong, select the target term set to be mapped, if the term in the source database belongs to the standard term set code and the target term set code. There is already a referenceable mapping relationship, then this part of terms can automatically generate mapping SQL statements to complete the automatic mapping of terms in the source database and the corresponding data loading.

Four, fuzzy matching module

Through fuzzy matching of the part of the medical terms with the standard terms in the general medical terminology database one by one, the recommended standard terms for mapping and the standard term set codes are given. Fuzzy matching generally recommends multiple standard terms as matching objects. Professionals with medical knowledge are required to manually determine the only matching object. After determining the mapping relationship, call the automatic mapping module to complete the mapping of this part of the medical term and the data it covers. load. The specific method of fuzzy matching is as follows:

(1) Term segmentation

Medical terms are mostly composed of multiple vocabularies in an orderly combination. Here, medical terms are subdivided into multiple vocabularies again according to specific rules.

(1.1) According to this method, all words in the general medical terminology database are segmented, and the frequency of each segmentation is counted as the basic word frequency.

(1.2) The source medical terms that need fuzzy matching are also segmented before matching. For example: after the term M is participle, you get [participle 1, participle 2, ... participle n].

(2) Fuzzy matching

The present invention compares the probability difference between medical terms as the standard of similarity, and the specific operations are as follows:

(2.1) Filter out all terms including word segmentation from the general medical term database, and perform word segmentation, and combine them into term set A{a,b,c,d,e,...};

(2.2) Use the following formula to calculate the matching degree and find the average weighted probability of all terms in term M and term set A. Among them, n is the number of word segmentation for each term, P1, P2, P3, P4...Pn is the probability of each word segmentation in the basic word frequency:

S(M,A)=|D(M)-D(A)|

Take the term "Ejiao Yishou Oral Liquid" as an example:

a) Perform word segmentation on the general medical terminology database and get the probability of each word segmentation;

b) The term "Ejiao Yishou Oral Liquid" is divided into "Ejiao\Yishou\Oral Liquid". Find the corresponding probability in the basic word frequency, get the frequency of "Donkey-hide gelatin" p1, "Yi Shou" p2, "Oral Liquid" p3, and find the average probability D(M) of each word segmentation;

c) Search all the terms containing "Ejiao", "YiShou" and "Oral Liquid" in the general medical term database, and perform word segmentation to obtain the term set A{["Ejiao", "Calcium", "Oral Liquid"] ,["Ejiao","granules"],["Ejiao","blood-enriching","oral liquid"]...}, get D(a), D(b), D(c)...;

d) Find the matching degree and sort

模糊匹配术语Fuzzy matching term	通用数据库术语General database terminology	匹配度suitability
阿胶益寿口服液Ejiao Yishou Oral Liquid	阿胶钙口服液Ejiao calcium oral solution	S(M,a)S(M,a)
To	阿胶补血口服液Ejiao Buxue Oral Liquid	S(M,c)S(M,c)
To	阿胶颗粒Donkey-hide gelatin particles	S(M,b)S(M,b)

Five, custom terminology module

In complex situations, especially for the actual situation where the domestic medical data center has redundant data and many medical terms related to Chinese medicine and traditional treatment methods, there is a situation that cannot match the international general medical term database. The custom term module will define necessary constraints in advance to avoid conflicts between custom terms and known standard terms. For example, in coding, the custom term is forced to use a limited coding range.

When adding custom terms, medical data centers must maintain the consistency of the added custom standard terms to prevent repeated additions, and at the same time ensure that multi-center medical data can be shared after terminology mapping is standardized. Therefore, when performing terminology standardization mapping on medical data in the medical data center, before adding custom terms, you need to submit a report adding custom terms to the multi-center interaction module. The report content includes: custom terms that need to be added, custom terms The specific description of the user-defined term code (automatically generated by the system). After the relevant operators of the center have passed the review and determined that there is no custom code of similar repeated medical terms, a custom standard term code will be generated, and then the automatic mapping module can be called to complete the term mapping and the loading of the covered data; if the review is not Pass, return the existing custom term code for the medical data center to complete subsequent mapping or return the reason for the failure of custom term generation, generate an error document and prompt the user. The operation diagram of the custom term module is shown in Figure 4.

Six, multi-center interactive module

To achieve data standardization and data sharing between the medical information systems of various medical data centers, all medical data centers are required to use a unified general medical term database and a unified medical term set coding. The invention adopts the method of uniformly adding after submission for review, preventing each medical data center from generating differences in terminology expression when self-defining standard terminology. In the process of submission, review, and authorization, there are interaction issues among multiple medical data centers. The multi-center interaction module is responsible for the coordination and unification of the general medical terminology database and term coding in each medical data center. The highest authority personnel of the multi-center interaction module will review and coordinate the use of custom standard terms, and the multi-center custom term interaction network As shown in Figure 5.

Seven, incremental update module

It is used in the subsequent medical term standardization process of the medical data center that has already operated the medical terminology mapping. It is mainly based on the previous terminology standardization mapping records generated by the terminology mapping unit to update the incremental data. For medical terms that still cannot be standardized , Repeat the custom term module.

Eight, exception handling module

It is used to save all the logs of the system during operation and record whether each module is operating normally; the error logs are stored in categories, including: errors during system operation, errors during each module call, and when each module is running for a single term mapping Errors occurred; classified and saved terms that were not successfully mapped, including those that were ignored in the automatic analysis module and the terms that were ignored in the custom module, and generated a failed term document. The exception handling module supports the database backtracking function by setting a time stamp on the database, and supports users to backtrack the matched database to the data on the specified date.

Nine, data cleaning module

After completing the standardized mapping of medical terminology, medical data cleaning is extremely necessary to improve the quality of medical data for subsequent data mining and analysis; here is a common data cleaning strategy, which mainly cleans the "dirty data" at the structure level and the instance level. Including data that violates the data model and integrity constraint requirements, such as data value out of range, property dependency damage, unique relationship damage, referential integrity damage, etc., as well as the value corresponding to the wrong attribute and the data of the dependency damage between the attributes, such as missing Values, duplicate records, contradictory records, reference errors, etc.; to maximize data integrity, uniqueness, authority, legality, and consistency, reduce data redundancy, and improve data quality.

1) Structure-level cleaning rules: unified data model (including data type) definition; unified integrity constraint definition; unified function dependency requirement definition.

2) Instance-level cleaning rules: Analyze dirty data, formulate cleaning rules, evaluate and verify, and record cleaning actions in a log for traceability.

The present invention is to realize the data sharing and sharing between multiple medical data centers (mainly hospitals) as the current data mining and analysis requirements for the quantity and quality of data continue to increase, and at the same time fully ensure the safety of the data of each medical data center A collaborative model designed based on nature, with the aim of optimizing the medical process, accelerating the progress of related scientific research and ultimately improving the quality of patients’ medical services by sharing medical data. The prerequisite for data sharing among multiple medical data centers is the standardization of medical data, which includes two parts, one is data structure standardization, and the second is medical terminology standardization. The above content is designed for the latter standardization. . The technical points of the present invention are summarized as follows:

1. Through the interaction between the various modules, the automatic analysis and scanning of the database in the medical information system is realized, and statistical information such as the frequency of occurrence of medical terms in the database is returned to provide a practical basis for subsequent medical term mapping and performance optimization.

2. For the data encoded by the international general medical term set, the mapping of the medical data covered by the part of the medical terminology is automatically realized according to the mapping relationship between the existing medical term set encodings.

3. For data in the medical data center that is not coded with the standard term set, custom medical terms, or unique domestic medical terms, such as Chinese medicine, etc., support information based on the frequency of occurrence of the medical term in the data in the medical data center. Support relevant personnel to visually perform reasonable and scientific fuzzy matching or directly add custom standard terms.

4. Complete the interaction requirements with multiple medical data centers on a regular basis to ensure that after each medical data center completes the standardization of medical terminology, the standards between the general medical databases of each family are maintained uniform, and data sharing can be realized.

5. Clean the data in accordance with the cleaning strategy to ensure data quality.

6. Record all errors and exceptions and write them in the log to facilitate the implementation of functions such as error troubleshooting and quality evaluation.

7. Make full use of the established mapping relationship between the medical terminology in the medical data center and the international standard medical terminology to realize the semi-automatic or even automated mapping and standardization of the terminology in the subsequent medical data center.

The above are only implementation examples of the present invention and are not used to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made without creative work within the spirit and principle of the present invention are all included in the protection scope of the present invention.

Claims

A multi-center medical terminology standardization system based on a general medical term database, which is characterized in that the system includes a source database, a database connection management module, a pre-analysis module, a term mapping unit, an incremental update module, an exception handling module, and multi-center interaction Module

The source database is distributed in the front-end server of each medical data center, and stores the business data of each medical data center;

The database connection management module: manages the information required to access the source database, and provides support for the terminology mapping tool to access and modify the source database;

The pre-analysis module: automatically scans the source database, counts the appearance frequency of each medical term in the original medical data, gives suggestions for rejection of terms whose appearance frequency is less than the set threshold, and sends the terms greater than or equal to the set threshold to the term mapping unit Perform subsequent term mapping;

The term mapping unit includes an automatic mapping module, a fuzzy matching module and a custom term module;

The automatic mapping module: supports automatic mapping of medical terminology, and realizes multi-directional mapping according to the existing mapping relationship between the standard codes of the international general medical terminology library for terms using the international general medical terminology library standard encoding;

The fuzzy matching module: For medical terms that cannot be directly mapped based on the mapping relationship between the internal standard codes of the existing medical terminology database, traverse and query in the general medical terminology database through fuzzy matching, and provide several sets of standards with the highest similarity Medical terms are available for selection as the target term of the term mapping;

The custom term module: For medical terms that cannot rely on the mapping relationship between the standard codes in the existing medical term database and cannot be fuzzy matching the target term in the existing general medical term database, after the user generates a custom term application, Send to the multi-center interactive module for review and feedback;

The multi-center interaction module: after receiving the custom term application from each medical data center sent by the custom term module, the custom term will be reviewed, and the reviewed custom term will be added to the general medical term database as a standard term , And send it to each medical data center to keep the common medical terminology database consistent in each medical data center;

The incremental update module: For the medical term standardization process in which the source database that has performed the standardized mapping of medical terminology generates incremental data due to business reasons, the historical mapping relationship record generated by the term mapping unit is retrieved to complete the terminology standardization mapping of the incremental data ；

The exception handling module: records the execution process of each of the above modules, generates an error log for the occurrence of errors, and can trace the entire process of medical term mapping according to the error log.
A multi-center medical terminology standardization system based on a general medical term database according to claim 1, characterized in that the system further includes a data cleaning module for formulating cleaning rules, assigning weights to each data element, and avoiding missing Severe data filtering, including cleaning dirty data at the structure level and instance level.
The multi-center medical terminology standardization system based on a general medical term database according to claim 1, wherein the database connection management module specifically includes: classes and interfaces written in a programming language form a JDBC module, which is a variety of The type database provides a unified access interface to realize the functions of establishing a connection with a database or other data sources, sending SQL commands to the database, and processing the results returned by the database.
The multi-center medical terminology standardization system based on a general medical term database according to claim 1, wherein the pre-analysis module automatically scans the source after the database connection management module realizes the connection to the source database. The structure information of all data in the database and the statistical information of specific fields are generated to generate a statistical table, including two parts:

First, summarize the statistics of all tables in the source database, including the field names in each table, the value type, the maximum length of all values, the total number of rows in the table, and the proportion of empty values;

Secondly, make statistics on the detailed information and frequency of occurrence of specific terms in a specific table, and arrange them in descending order of frequency of occurrence, so that subsequent term mapping will give priority to terms with higher frequency of occurrence. Whether it is necessary for a term with a very low frequency to participate in the subsequent term mapping suggestions, if it is not defined, all terms participate in the mapping by default. The user can also adjust according to the specific situation to determine the minimum frequency threshold for not participating in the subsequent term mapping.
The multi-center medical terminology standardization system based on a general medical terminology database according to claim 1, wherein the automatic mapping module: for terms in the source database that have international general medical terminology standard codes, determine After the code belongs to the standard, select the target term set to be mapped. If there is a reference mapping relationship between the standard term set code of the term in the source database and the target term set code, then this part of the term can automatically generate a mapping SQL statement , To complete the automatic mapping of terms in the source database and the corresponding data loading.
The multi-center medical terminology standardization system based on a general medical term database according to claim 1, wherein the specific method of fuzzy matching in the fuzzy matching module is as follows:

(1) Terminology segmentation: segment all vocabularies in the general medical terminology database, and perform frequency statistics for each segmentation as the basic word frequency; segment the source medical term M that needs fuzzy matching before matching.

(2) Fuzzy matching: By comparing the probability difference between medical terms as the standard of similarity, the specific operation is as follows:

(2.1) Filter out all terms including word segmentation from the general medical term database, and perform word segmentation, and combine them into term set A;

(2.2) Use the following formula to calculate the matching degree to find the average weighted probability of all terms in term M and term set A; where n is the number of word segmentation for each term, and P1, P2, P3, P4...Pn is each The corresponding probability of word segmentation in basic word frequency:

(2.3) Make the difference between the average weighted probability of all standard terms in the term set A and the term M that needs fuzzy matching, and take the negative value as the matching degree. The greater the matching degree, the higher the similarity between the two. The formula is as follows:

S(M,A)=|D(M)-D(A)|
The multi-center medical terminology standardization system based on a general medical term database according to claim 1, wherein the custom term module: defines constraints in advance to avoid conflicts between custom terms and known standard terms ; When adding custom terminology, medical data centers must maintain the consistency of the added custom standard terminology to prevent repeated additions, while ensuring that multi-center medical data can be shared after terminology mapping is standardized. Before adding a custom term, you need to submit an application for adding a custom term to the multi-center interaction module. The application content includes: the custom term that needs to be added, the specific description of the custom term, the code of the custom term; pending multi-center interaction module After passing the review, the relevant operators confirm that there is no custom code of similar repeated medical terms, then a custom standard term code is generated, and then the automatic mapping module can be called to complete the term mapping and the loading of the covered data; if the review fails , Then return the existing custom term code for the medical data center to complete subsequent mapping or return the reason for the failure of custom term generation, generate an error document and prompt the user.
The multi-center medical terminology standardization system based on a general medical term library according to claim 1, wherein the multi-center interaction module is responsible for the coordination and unification of the general medical terminology database and its term coding of each medical data center , The highest authority personnel of the multi-center interaction module will review and coordinate the use of custom standard terms.
The multi-center medical terminology standardization system based on a general medical term database according to claim 1, wherein the incremental update module is used in the subsequent medical terminology standardization process of a medical data center that has already operated medical terminology mapping , It is mainly based on the previous terminology standardized mapping records generated by the term mapping unit to update incremental data. For medical terms that still cannot be standardized, the custom term module is repeatedly executed.
The multi-center medical terminology standardization system based on a general medical term database according to claim 1, wherein the exception handling module is used to save all logs during system operation and record whether each module is operating normally; Category save error logs, including: errors that occur when the system is running, errors that occur when each module is called, and errors that occur when each module is running for a single term mapping; category and save terms that are not successfully mapped, including in the automatic analysis module Ignored and ignored terms in the custom module, generate a failed term document; the exception handling module supports the database backtracking function by setting a time stamp on the database, and supports users to backtrack the matched database to the data on the specified date.