Matching method and device based on medical terms
Through the matching method based on medical terms, the text content input by the doctor is combined in segments and recursively, the Cartesian product is calculated and similarity matched, which solves the problems of low efficiency and poor portability of existing medical information matching methods. It realizes efficient matching and sharing of medical information.
Patent Information
- Application Number
- CN202411592173.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-08
- Publication Date
- 2025-02-14
AI Technical Summary
The existing medical information matching methods have problems such as mutual exclusion of rules, low maintenance efficiency and low matching rate, and it is difficult to achieve efficient sharing of medical information and portability in multiple scenarios.
Using a matching method based on medical terms, the text content input by the doctor is segmented, the element set is combined using a recursive method, the Cartesian product is calculated and the similarity match is performed, and the generation of structured data is achieved.
The ability of semantic recognition in medical documents has been improved, the generalization ability and portability of matching algorithms has been improved, the intervention of human resources has been reduced, and the accuracy and matching rate of 95% has been achieved.
Smart Images
Figure CN119443078A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of medical term matching, and in particular relates to a matching method and device based on medical terminology. Background Art
[0002] Electronic medical records, as the basic information carrier for recording patients' medical treatment in hospitals, record a large amount of meaningful data. However, in the current medical environment, these case information is full of a large amount of distributed heterogeneous data, which brings many obstacles to the expression, storage, exchange, sharing and system collaboration of medical information. If we want to realize the digitalization and informatization of the medical industry and then realize efficient sharing of medical resources, we undoubtedly need a set of electronic medical records mapped to a standardized data structure based on medical terminology to achieve the purpose of expressing, exchanging and sharing medical information.
[0003] At present, traditional methods are based on simple text matching, search engines, etc., plus manually established rule bases for matching, which have problems such as rule mutual exclusion, low rule maintenance efficiency, and low matching rate. In addition, the portability between different scenarios is poor, requiring a large number of rule establishment and system modification, and cannot meet the support of later decision-making models.
[0004] Although, with the development of artificial intelligence in recent years, many researchers have proposed the use of big data, deep learning and other technologies to conduct text matching research. However, due to the large difference between medical terms and general terms, it is difficult to use the annotation and recognition of general corpus information for medical texts. This is because the judgment of medical terms requires professionals, which greatly increases the cost of medical text recognition and matching. Summary of the invention
[0005] The purpose of the present invention is to overcome the above problems and provide a matching method and device based on medical terminology. In this technology, doctors can freely input a text, and after pressing Enter, the system automatically matches it into standardized structured data.
[0006] In order to achieve the above purpose, the technical solution adopted by the present invention is:
[0007] A matching method based on medical terms comprises the following steps:
[0008] A corresponding matching decomposition method is used for the sequentially segmented text content, date and / or natural language in a numerical type format to obtain a redundant decomposed element set; wherein the matching method used for the natural language of the text content includes: dividing the text content into a plurality of text segments; matching the number of characters of each text segment + 1 with an element in a medical term set; wherein, if a match is found, recording the starting offset and ending offset of the element in the text content;
[0009] Using a recursive method, based on a start offset and an end offset, the element set is combined into a plurality of concatenated sets;
[0010] Traversing the concatenated set, and extracting a dictionary value set associated with elements in the concatenated set;
[0011] Based on the dictionary value set, a Cartesian product is calculated to obtain a plurality of multisets, wherein the multiset is composed of a plurality of subsets;
[0012] Each subset is taken from the multiple sets for similarity matching, and the matching results are arranged in reverse order to complete automatic matching.
[0013] As a further improvement, matching the number of characters of each text segment + 1 with an element in the medical term set comprises the following steps:
[0014] If no match is found, the number of characters + 2 is taken again to match the elements in the set. If no match is found again, the process is repeated recursively until a match is found.
[0015] If a match is found, the element and type are also recorded.
[0016] As a further improvement, the recursive method combines the element sets into multiple concatenated sets based on the start offset bit and the stop offset bit, comprising the following steps:
[0017] Setting a threshold value for the number of elements in the element set and a threshold value for the number of elements within the element set;
[0018] Based on the set quantity threshold and the element quantity threshold, taking the number of sets within the threshold range from the element set to form a basic set;
[0019] The basic sets are concatenated and inverted to obtain the concatenated set.
[0020] As a further improvement, the step of calculating the Cartesian product based on the dictionary value set to obtain multiple multisets includes the following steps:
[0021] Calculate the Cartesian product of each layer of the dictionary value set that is arranged in reverse order, and count the total amount of the Cartesian product; wherein, when the total amount is greater than 500, the calculation of the Cartesian product of the set is interrupted; the formula for calculating the Cartesian product of the multiset is A=A i *A i+1 , where A represents a multi-set, A i Represents a collection of dictionary values;
[0022] Obtain a set X of sets, wherein the set X includes multiple sets x, and the set x includes multiple elements a i , ai ∈A.
[0023] As a further improvement, taking each subset from the multiple sets for similarity matching comprises the following steps:
[0024] Take each set x in the set X, and match the set Y of all term combinations written by Director Min; if no matching set Y is found, an empty set is returned; wherein the set Y includes multiple term combination elements y i , the term combination element y i Contains a unique matching degree, which is the basic matching score.
[0025] As a further improvement, taking each subset from the multiple sets for similarity matching further comprises the following steps:
[0026] The element a i Combine the elements y one by one with the term i The data elements in match;
[0027] Among them, the matched element a i Assign the value code and value name attributes to the value to obtain the basic elements of the case.
[0028] As a further improvement, the process of taking each subset from the multiple sets for similarity matching further includes the following steps:
[0029] In the matching process, when the matching fails, the matching score is reduced according to the set weight on the basis of the basic matching score; wherein, each time the matching fails, the set weight value is reduced once as the matching score;
[0030] Among them, for the term combination element y with a fixed preset value i And the element a i If they are not consistent, the matching score is equal to the basic matching score - 1.
[0031] As a further improvement, the process of taking each subset from the multiple sets for similarity matching further includes the following steps:
[0032] If the natural language and the term combination element y i If they are equal, the matching score is the basic matching score + 10.
[0033] As a further improvement, the matching results are arranged in reverse order to complete the automatic matching, including the following steps:
[0034] Based on the order of the matching scores from high to low, the set Y obtained from the concatenated set in the element set is merged and arranged for the second time.
[0035] On the other hand, the present invention also provides a matching system based on medical terminology, the system comprising at least one processor and a memory storing instructions, which implement the method in the above technical solution when the instructions are executed by at least one processor.
[0036] A computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method described in the steps of the above technical solution.
[0037] A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method described in the steps of the above technical solution is implemented.
[0038] Beneficial effects of the present invention:
[0039] Through the method and device, the ability of semantic recognition in medical documents is realized. The rule engine that is not based on semantics greatly improves the generalization ability of the matching algorithm and can be used in multiple application scenarios. It is not only highly portable but also reduces the intervention of human resources. When matching medical documents, the accuracy matching rate can be stabilized at 95%, meeting the requirements of use. When the text content is too long, the number of combinations may be large, so only sets with less than or equal to 8 internal elements are considered, and then the elements in the set are concatenated and reversed, and the content closest to the user input is placed at the front, reducing the use of computing resources. Medical terminology introduces the concept of data elements, which improves the accuracy of text matching medical terminology. The present invention can structure the decomposition and storage of medical text content, which is conducive to scientific research analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The present invention has the following accompanying drawings:
[0041] Figure 1 is a flow chart of a matching method based on medical terms according to some specific embodiments of the present invention;
[0042] Figure 2 is a schematic diagram of a structure of a matching device based on medical terminology in some specific embodiments of the present invention;
[0043] Figure 3 A schematic diagram of a matching system based on medical terminology of the present invention. DETAILED DESCRIPTION
[0044] The following examples further illustrate the content of the present invention, but should not be construed as limiting the present invention. Without departing from the spirit and substance of the present invention, modifications or substitutions made to the methods, steps or conditions of the present invention all fall within the scope of the present invention.
[0045] Embodiment 1:
[0046] like Figure 1 As shown, the present invention provides a matching method based on medical terms, comprising the following steps:
[0047] Step S101: adopting a corresponding matching decomposition method to the sequentially segmented text content, date and / or natural language in a numerical type format to obtain a redundant decomposed element set; wherein the matching method adopted for the natural language of the text content includes: dividing the text content into a plurality of text segments; matching the number of characters of each text segment + 1 with an element in a medical term set; wherein, if a match is found, recording the start offset and end offset of the element in the text content;
[0048] For example, continuous natural language text segments input by a client doctor are received and recognized; the type format of the natural language text segments is determined, and corresponding matching decomposition methods are used for natural language text segments in text content, date and numerical type formats to obtain a redundant set of decomposed elements.
[0049] In some embodiments of the present invention, after the doctor inputs the content and presses Enter to confirm, the front end divides the content input by the user according to general commas, periods or exclamation marks, and each text segment is taken as a whole, and the back end interface is called for decomposition (in parallel); after the back end receives the content, it segments the text content from back to front in sequence, taking one more character each time and matching it with the elements in the set. If no match is found, one more character is taken for matching. If a match is found, the element is recorded, and the starting offset and ending offset of the element in the text content are recorded, and the record type is 1; the date format of yyyy-MM-dd entered by the user (regular match) is also recorded as an element, and the type is 2; the numerical type (regular match) entered by the user is directly skipped; if the decomposition fails and the obtained set is empty, an empty set is directly returned, and the front end directly processes it as a text data element; after completing all the decompositions, a large number of redundant decomposed element sets M will be obtained.
[0050] Step S102: using a recursive method, based on the start offset and the end offset, combining the element set into a plurality of splicing sets;
[0051] For example, a recursive method is used, combined with the start offset and end offset of the record. The sets are combined into multiple reasonable concatenation sets (m). When the text content is too long, the number of combinations may be large. At this time, considering that the actual elements of the term combination are not many, only 1,000 are taken first, and only sets with less than or equal to 8 internal elements are considered (there are almost no term combinations that can match more than 8). Then, the elements in the set m are concatenated and reversed, which means that the content closest to the user input is placed at the front. If there is no such set, an empty set is returned directly, and the front end directly processes it as a text data element.
[0052] In some embodiments of the present invention, in order to avoid a large pressure on the server caused by first obtaining 1000 sets, only the first 8 sets are still retained for subsequent processing.
[0053] Step S103: traverse the concatenated set and extract the dictionary value set associated with the elements in the concatenated set;
[0054] For example, for each concatenated set m, we traverse it and extract the dictionary value set A associated with each element. i If it is an element of date type, query the database to obtain all date type data elements that can be used for writing purposes as a set.
[0055] Step S104: Calculate the Cartesian product based on the dictionary value set to obtain a plurality of multisets, wherein the multiset is composed of a plurality of subsets;
[0056] For example, for a sequence of multiple sets, find the Cartesian product (A = A i *A i+1 , where A represents a multi-set, A i Represents a collection of dictionary values);
[0057] In some embodiments of the present invention, in order to prevent the number of explosions, the total amount of Cartesian products calculated at each level will be recorded. When the total amount of Cartesian products exceeds 500, the Cartesian product of the set is interrupted and the matching is no longer continued. Finally, a set of sets X is obtained, and the elements inside X are also a set x (x∈X), where the number of elements inside x is n, x={a1+a2+a3+...+a i +...+a n}, where a i ∈A.
[0058] Step S105: taking each subset from the multiple sets for similarity matching, and arranging the matching results in reverse order to complete automatic matching.
[0059] For example, using the Cartesian product to solve all sets X, take each x (x = {a1 + a2 + a3 + ... + a i +...+a n}), match the set Y of all term combinations written by Director Min (the numerical data elements in the term combinations will also be ignored during the query), and sort them in reverse order by similarity.
[0060] If no matching term combination is found, an empty set is returned directly and the processing of this set ends.
[0061] For the queried set Y, y∈Y, y i is a term combination, and y i A matching degree is recorded internally as the basic matching score (maximum 100). i The data elements in the x set are sorted one by one according to the elements a of the x set. i Match (selective, Boolean, and more standardized date data sources participate in the matching, others do not participate), for the matching y i , a i The value of is filled into the value code and value name attributes to form the basic elements of the medical record.
[0062] During the matching process, the score will be reduced according to the matching situation. The experience value 10 / n is taken as the increase or decrease weight avgWeight. Every time a match fails, avgWeight is reduced. For y i There is already a preset value and a i If inconsistent, the score is reduced by 1.
[0063] The step S101 of matching the number of characters of each text segment + 1 with an element in a medical term set includes the following steps:
[0064] Step S1011: If no match is found, then the character number + 2 is taken again to match the elements in the set. If no match is found again, the process is repeated recursively until a match is found.
[0065] Step S1012: If a match is found, the element and type are also recorded.
[0066] Step S102 includes the following steps:
[0067] Step S1021: setting a threshold value for the number of elements in the element set and a threshold value for the number of elements within the element set;
[0068] Step S1022: based on the set quantity threshold and the element quantity threshold, taking the number of sets within the threshold range from the element set to form a basic set;
[0069] Step S1023: concatenate and invert the basic set to obtain the concatenated set.
[0070] Step S104 includes the following steps:
[0071] Step S1041: Calculate the Cartesian product of each layer of the dictionary value set that has been sorted in reverse order, and count the total amount of the Cartesian product; wherein, when the total amount is greater than 500, the calculation of the Cartesian product of the set is interrupted; the formula for calculating the Cartesian product of the multi-set is A=A i *A i+1 , where A represents a multi-set, A i Represents a collection of dictionary values;
[0072] Step S1042: Obtain a set X of sets, wherein the set X includes multiple sets x, and the set x includes multiple elements a i , a i ∈A.
[0073] In step S105, taking each subset from the multiple sets for similarity matching includes the following steps:
[0074] Step S1051: Take each set x in the set X and match it to the set Y of all term combinations written by Director Min; if no matching set Y is found, an empty set is returned; wherein the set Y includes multiple term combination elements y i , the term combination element y i Contains a unique matching degree, which is the basic matching score.
[0075] In step S105, taking each subset from the multiple sets for similarity matching further includes the following steps:
[0076] Step S1052: The element a i Combine the elements y one by one with the term i The data element in is matched; wherein the matched element a i Assign the value code and value name attributes to the value to obtain the basic elements of the case.
[0077] In step S105, taking each subset from the multiple sets for similarity matching further includes the following steps:
[0078] Step S1053: During the matching process, when the matching fails, the matching score is reduced according to the set weight based on the basic matching score; wherein, each time the matching fails, the set weight value is reduced once as the matching score; wherein, for the term combination element y with a fixed preset value iAnd the element a i If they are not consistent, the matching score is equal to the basic matching score - 1.
[0079] In step S105, taking each subset from the multiple sets for similarity matching further includes the following steps:
[0080] Step S1054: If the natural language and the term combination element y i If they are equal, the matching score is the basic matching score + 10.
[0081] In step S105, the matching results are arranged in reverse order to complete the automatic matching, including the following steps:
[0082] Step S1055: Based on the order of the matching scores from high to low, the set Y obtained from the concatenated set in the element set is merged and arranged for the second time.
[0083] For example, when a complete term combination (with user-entered values, natural language) is finally assembled, if the text concatenation value of the combination is exactly the same as the value used for input (ignoring numerical data), add 10 to the original score.
[0084] Finally, the set Y obtained by each m in the M set is merged again and sorted in reverse order according to the score, with the one with the highest score at the front and returned to the front end.
[0085] Embodiment 2:
[0086] like Figure 2 As shown, the present invention provides a matching device based on medical terms, comprising:
[0087] A decomposition module is used to adopt a corresponding matching decomposition method for the sequentially segmented text content, date and / or natural language in a numerical type format to obtain a redundant decomposed element set; wherein the matching method adopted for the natural language of the text content includes: dividing the text content into a plurality of text segments; matching the number of characters of each text segment + 1 with an element in a medical term set; wherein, if a match is found, recording the starting offset and ending offset of the element in the text content;
[0088] For example, receiving and identifying continuous natural language text segments input by a client doctor; determining the type format of the natural language text segments, using corresponding matching decomposition methods for natural language text segments in text content, date and numerical type formats, and obtaining a redundant decomposed element set;
[0089] A splicing module, used for combining the element sets into a plurality of splicing sets based on a start offset and a stop offset by using a recursive method;
[0090] An extraction module, used for traversing the concatenated set and extracting a dictionary value set associated with elements in the concatenated set;
[0091] A calculation module, configured to calculate a Cartesian product based on the dictionary value set to obtain a plurality of multisets, wherein the multiset is composed of a plurality of subsets;
[0092] The matching module is used to take each subset from the multiple sets for similarity matching, and arrange the matching results in reverse order to complete automatic matching.
[0093] The decomposition module matches the number of characters of each text segment + 1 with an element in the medical term set, including:
[0094] The first judgment module is used to re-take the number of characters + 2 to match the elements in the set if no match is found, and if no match is found again, recursively perform the operation until a match is found; and is also used to record the element and type if a match is found.
[0095] Splicing module, including:
[0096] A setting module, used to set a threshold value of the number of elements in the element set and a threshold value of the number of elements in the element set;
[0097] A composition module, configured to take a number of sets within a threshold range from the element set to form a basic set based on the set number threshold and the element number threshold;
[0098] The first concatenation module is used to concatenate and invert the basic set to obtain the concatenated set.
[0099] Extraction modules, including:
[0100] The first calculation module is used to calculate the Cartesian product of each layer of the dictionary value set arranged in reverse order, and count the total amount of the Cartesian product; wherein, when the total amount is greater than 500, the calculation of the Cartesian product of the set is interrupted; the formula for calculating the Cartesian product of the multi-set is A=A i *A i+1 , where A represents a multiset and Ai represents a dictionary value set;
[0101] The first result module is used to obtain a set X of sets, wherein the set X includes multiple sets x, and the set x includes multiple elements a i , a i ∈A.
[0102] In the matching module, taking each subset from the multiple sets for similarity matching includes:
[0103] The first extraction module is used to take each set x in the set X and match the set Y of all term combinations written by Director Min; wherein, if no matching set Y is found, an empty set is returned; wherein, the set Y includes multiple term combination elements y i , the term combination element y i Contains a unique matching degree, which is the basic matching score.
[0104] In the matching module, taking each subset from the multiple sets for similarity matching further includes:
[0105] The first matching module is used to match the element a i Combine the elements y one by one with the term i The data element in is matched; wherein the matched element a i Assign the value code and value name attributes to the value to obtain the basic elements of the case.
[0106] In the matching module, taking each subset from the multiple sets for similarity matching further includes:
[0107] The second matching module is used to reduce the matching score according to the set weight on the basis of the basic matching score when the matching fails during the matching process; wherein, each time the matching fails, the set weight value is reduced once as the matching score; wherein, for the term combination element y with a fixed preset value i And the element a i If they are not consistent, the matching score is equal to the basic matching score - 1.
[0108] In the matching module, taking each subset from the multiple sets for similarity matching further includes:
[0109] The third matching module is used to match the natural language and the term combination element y i If they are equal, the matching score is the basic matching score + 10.
[0110] In the matching module, the matching results are arranged in reverse order to complete the automatic matching, including:
[0111] The permutation and combination module is used to perform a second merging and permutation on the set Y obtained from the concatenated set in the element set based on the order of the matching scores from high to low.
[0112] In some embodiments of the present invention, continuous natural language text segments input by a client doctor are received and recognized; the type format of the natural language text segment is determined, and the natural language text segment in the text content, date and numerical type format is subjected to corresponding matching decomposition methods to obtain a redundant decomposed element set; the element set is combined into a plurality of concatenated sets using a recursive method in combination with the start offset and end offset of the record; the concatenated sets are traversed to extract the dictionary value set associated with the elements in the concatenated sets; based on the dictionary value set, the Cartesian product is calculated to obtain a plurality of multi-sets, wherein the multi-set is composed of a plurality of subsets; each subset is taken from the multi-set for similarity matching, and the matching results are arranged in reverse order, and the combinations arranged in reverse order are sent to the client.
[0113] In this technology, text structuring relies on a preset medical term combination, which is composed of multiple data elements, which are also the basic components of medical records. For example, "fever for 3 days" can be decomposed into a combination of three data elements: symptoms, duration, and time unit. The symptom data element is a selective data element, which is associated with a symptom dictionary (also known as medical terminology, and is still referred to as a dictionary below to avoid confusion). The duration is a numerical data element, and the time unit is a selective data element, which is associated with a time unit dictionary. The term combination is represented in the program as a combination of a drop-down option, a number box, and a text box.
[0114] A data element is the basic unit of representation and storage of data. A data element stores one type of data. Each data element has an independent identifier and a bound data type.
[0115] Arrange the data elements in order from left to right according to the subject classification code, major category code, minor category code, sequence code, and additional code.
[0116] 1. Subject classification code: represented by two uppercase English letters. The code for the health information field is uniformly set as "DE".
[0117] 2. Major category code: represented by two digits, the size of the digits has no meaning.
[0118] 3. Sub-category code: It is represented by two digits, and the size of the digits has no meaning. If there is no sub-category, the sub-category code is 00. A “.” is added between the sub-category and the major category code to distinguish them.
[0119] 4. Sequence code: It is represented by 3 digits, representing the sequence number of the data element under a certain subcategory. The size of the digits has no meaning. The sequence code starts from 001. A “.” is added between the sequence code and the subcategory code to distinguish them.
[0120] 5. Additional code: It is a two-digit number representing the serial relationship of a group of data elements. It is coded sequentially starting from 01, and a “.” is added between the additional code and the sequence number to distinguish them. The additional code of a data element without serial relationship is “00”.
[0121]
[0122] Data Types
[0123]
[0124] Description of format characters
[0125]
[0126]
[0127] Description of format characters
[0128]
[0129] dictionary
[0130] The sources of the dictionary are mainly national standards (GB) and health standards (WS), and also include medical terms written by Director Min (please refer to the data in the system). It is an item that can be selected. For example, gender dictionary (GB), administrative division (GB), identity document category code table (health), ICD-10 diagnosis code (health), etc. The dictionary items, or dictionary values (or medical term values), are composed of codes and names. The code should theoretically be able to fully express the meaning, range, level and other information of the dictionary value. The name can be regarded as an explanation of the meaning. This part belongs to the category of coding and will not be explained too much here. Some dictionaries may have some specific extended attributes inside the program for specific purposes, such as marking the parent-child relationship on the administrative division dictionary value to express the three-level association of province, city and district, which is not part of the standard content.
[0131] Dictionaries cannot be used directly for writing and must be bound to selective data elements. The same dictionary can be bound to different data elements to express different meanings. For example, an administrative division dictionary can be bound to a current address data element or to a work address data element to express different purposes.
[0132] Structuring process
[0133] text fever 3 sky Data element symptom Duration of symptoms Time Unit
[0134] The task of text structuring is to decompose the continuous natural language input by doctors into term combinations. When writing medical records, doctors do not directly write the names of data elements, but only the dictionary values behind the data elements.
[0135] Through the method and device, the ability of semantic recognition in medical documents is realized. The rule engine that is not based on semantics greatly improves the generalization ability of the matching algorithm and can be used in multiple application scenarios. It is not only highly portable but also reduces the intervention of human resources. When matching medical documents, the accuracy matching rate can be stabilized at 95%, meeting the requirements of use. When the text content is too long, the number of combinations may be large, so only sets with less than or equal to 8 internal elements are considered, and then the elements in the set are concatenated and reversed, and the content closest to the user input is placed at the front, reducing the use of computing resources. Medical terminology introduces the concept of data elements, which improves the accuracy of text matching medical terminology. The present invention can structure the decomposition and storage of medical text content, which is conducive to scientific research analysis.
[0136] like Figure 3 As shown, a matching system based on medical terminology includes at least one processor and a memory storing instructions to implement any one of the matching methods based on medical terminology in the above technical solutions.
[0137] In some specific embodiments, a computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps in the above technical solution.
[0138] In some specific embodiments, a computer program product is provided, the computer program product includes a computer program, and when the computer program is executed by a processor, the method of implementing the steps in the above technical solution is implemented.
[0139] The embodiments and functional operations of the subject matter described in this specification may be implemented in: digital electronic circuitry, tangibly implemented computer software or firmware, computer hardware, including the structures disclosed in this specification and their structural equivalents, or a combination of more than one of the above. The embodiments of the subject matter described in this specification may be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded on one or more tangible non-transitory program carriers, for execution by a data processing device or to control the operation of a data processing device.
[0140] Alternatively or additionally, the program instructions may be encoded on an artificially generated propagation signal, such as a machine-generated electrical, optical, or electromagnetic signal, which is generated to encode information for transmission to an appropriate receiver device for execution by a data processing device. The computer storage medium may be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or a combination of one or more of the above devices.
[0141] A computer program (which may also be referred to or described as a program, software, software application, module, software module, script or code) may be written in any form of programming language, including compiled or interpreted languages or declarative or procedural languages, and the computer program may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment. A computer program may, but need not, correspond to a file in a file system. A program may be stored in a portion of a file that stores other programs or data, for example, one or more scripts stored in the following: in a markup language document; in a single file dedicated to a related program; or in multiple collaborative files, for example, files storing one or more modules, subroutines or code portions. A computer program may be deployed to execute on one or more computers, the computers being located in one place or distributed to multiple locations and interconnected by a communication network.
[0142] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.
[0143] Although the present invention has been described in detail above by general description, specific implementation methods and experiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.
Claims
1. A matching method based on medical terminology, characterized in that: The following steps are involved: A corresponding matching decomposition method is used for the sequentially segmented text content, date and / or natural language in a numerical type format to obtain a redundant decomposed element set; wherein the matching method used for the natural language of the text content includes: dividing the text content into a plurality of text segments; matching the number of characters of each text segment + 1 with an element in a medical term set; wherein, if a match is found, recording the starting offset and ending offset of the element in the text content; Using a recursive method, based on a start offset and an end offset, the element set is combined into a plurality of concatenated sets; Traversing the concatenated set, and extracting a dictionary value set associated with elements in the concatenated set; Based on the dictionary value set, a Cartesian product is calculated to obtain a plurality of multisets, wherein the multiset is composed of a plurality of subsets; Each subset is taken from the multiple sets for similarity matching, and the matching results are arranged in reverse order to complete automatic matching.
2. A matching method based on medical terminology according to claim 1, characterized in that: The step of matching the number of characters of each text segment plus 1 with an element in a medical term set comprises the following steps: If no match is found, the number of characters + 2 is taken again to match the elements in the set. If no match is found again, the process is repeated recursively until a match is found. If a match is found, the element and type are also recorded.
3. A matching method based on medical terminology according to claim 1, characterized in that: The recursive method combines the element sets into a plurality of concatenated sets based on the start offset and the stop offset, comprising the following steps: Setting a threshold value for the number of elements in the element set and a threshold value for the number of elements within the element set; Based on the set quantity threshold and the element quantity threshold, taking the number of sets within the threshold range from the element set to form a basic set; The basic sets are concatenated and inverted to obtain the concatenated set.
4. A matching method based on medical terminology according to claim 1, characterized in that: The step of calculating the Cartesian product based on the dictionary value set to obtain multiple multisets includes the following steps: Calculate the Cartesian product of each layer of the dictionary value set that is arranged in reverse order, and count the total amount of the Cartesian product; wherein, when the total amount is greater than 500, the calculation of the Cartesian product of the set is interrupted; the formula for calculating the Cartesian product of the multiset is A=A i *A i+1 , where A represents a multi-set, A i Represents a collection of dictionary values; Obtain a set X of sets, wherein the set X includes multiple sets x, and the set x includes multiple elements a i , a i ∈A.
5. A matching method based on medical terminology according to claim 4, characterized in that: The taking each subset from the multiple sets for similarity matching comprises the following steps: Take each set x in the set X, and match the set Y of all term combinations written by Director Min; if no matching set Y is found, an empty set is returned; wherein the set Y includes multiple term combination elements y i , the term combination element y i Contains a unique matching degree, which is the basic matching score.
6. A matching method based on medical terminology according to claim 5, characterized in that: The taking each subset from the multiple sets for similarity matching further comprises the following steps: The element a i Combine elements y one by one with the term i The data elements in match; Among them, the matched element a i Assign the value code and value name attributes to the value to obtain the basic elements of the case.
7. A matching method based on medical terminology according to claim 5, characterized in that: The taking each subset from the multiple sets for similarity matching further comprises the following steps: In the matching process, when the matching fails, the matching score is reduced according to the set weight on the basis of the basic matching score; wherein, each time the matching fails, the set weight value is reduced once as the matching score; Among them, for the term combination element y with a fixed preset value i And the element a i If they are not consistent, the matching score is equal to the basic matching score - 1.
8. The medical terminology matching method according to claim 5, characterized in that: The taking each subset from the multiple sets for similarity matching further comprises the following steps: If the natural language and the term combination element y i If they are equal, the matching score is the basic matching score + 10.
9. A matching method based on medical terminology according to claim 5, characterized in that: The step of arranging the matching results in reverse order to complete the automatic matching includes the following steps: Based on the order of the matching scores from high to low, the set Y obtained from the concatenated set in the element set is merged and arranged for the second time.
10. A matching device based on medical terminology, the device comprising at least one processor and a memory storing instructions, and when the instructions are executed by at least one processor, a matching method based on medical terminology as described in any one of claims 1 to 9 is implemented.