JP2015106361A - Data retrieval system and data retrieval method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve comprehensiveness of retrieval processing for data including an orthographical variant.SOLUTION: A data retrieval system including processing in which a character string forming each notation included in master information of each event is divided into words, a partial character string of each notation in each event is identified and between the partial character strings, a synonymous relation is determined in accordance with the positions in each notation relating to the same event. The partial character strings for which the presence of the synonymous relation is determined are made correspondent to generate a partial orthographical variant dictionary 102, the partial character strings forming the character string of the notation are collated with the partial orthographical variant dictionary 102, and another partial character string corresponding to the partial character string is identified. A combination pattern relating to each partial character string of the notation and the other partial character string obtained for the partial character string is generated, and the character string indicated by each pattern is made correspondent to the relevant event indicated by the notation, thereby generating an orthographical variant dictionary 103. The data retrieval system also includes an arithmetic unit 14 which identifies a pattern registered for a correspondent event by collating a retrieval word with the orthographical variant dictionary 103, retrieves a predetermined database 105 with a character string indicated by the identified pattern, and outputs a retrieval result.

Description

  The present invention relates to a data search system and a data search method, and more specifically, to a technique that can improve the comprehensiveness of search processing for data including notation fluctuation.

  With the progress of computerization of medical information, there is an increasing trend to use the accumulated medical information to improve medical quality and improve management. On the other hand, when registering such medical information in a prescribed management system, registration processes with different notation for the same event, so-called notation fluctuations, frequently occur due to various factors such as the individual's personal recognition, thoughts, and skills. Yes. Such notation fluctuation is an obstacle when searching for the corresponding medical information, and makes it difficult to efficiently use the medical information.

  As a coping method when using medical information corresponding to the above-mentioned notation fluctuation, there is a method of using a synonym dictionary that defines a correspondence relationship between each vocabulary as synonyms for each event in which the notation fluctuation occurs. . For example, the following techniques have been proposed as conventional techniques related to such a synonym dictionary. That is, this is a technique for extracting a character string pattern common to a plurality of existing synonyms as a synonym rule and automatically generating a synonym using the rule and character string information in a database (see Patent Document 1).

JP 2009-128968 A

  In the prior art, a common pattern existing between synonyms existing in the synonym dictionary is extracted as a synonym rule, and a synonym with high accuracy is generated. However, since the synonym generation is based on the extracted rule, the synonym rule extraction target is used for a character string that cannot have a common pattern itself, such as a character string that appears only once in the synonym dictionary. I can't even do that. That is, even if such a character string is appropriate as a synonym for the vocabulary of a certain event, the medical information tied to the character string is excluded from the search target.

  On the other hand, when performing data analysis in the medical field, such as drug efficacy assessment and side effect investigation, comprehensively extracting relevant medical information from the database without omission is particularly important in terms of maintaining the accuracy of the analysis results. It becomes important. On the other hand, as described above, the accuracy of the analysis result is inevitably lowered if the medical information with the shaking cannot be extracted and cannot be included in the evaluation target.

  Accordingly, the present invention has been invented in view of the above problems, and it is an object of the present invention to provide a technique capable of improving the comprehensiveness of search processing for data including notation fluctuation.

  The data search system of the present invention that solves the above problems includes a storage device that stores, for each event, master information in which a plurality of notations indicating the same event are associated, and reads the master information of each event from the storage device. The character string constituting each notation included in the information is divided into words, the process of specifying the partial character string of each notation in each event, and between each of the specified partial character strings, The synonym relation is determined according to the position, the partial character strings determined to have the synonymous relation are stored in the storage device in association with each other and the partial notation fluctuation dictionary is generated, and the character string of the notation is configured Each partial character string is collated with the partial notation fluctuation dictionary, the other partial character string associated with the corresponding partial character string is specified, and each partial character string obtained with respect to each partial character string of the notation and the corresponding partial character string A combination pattern is generated, a character string indicated by each pattern is stored in a storage device in association with a corresponding event indicated by the notation, a notation fluctuation dictionary is generated, and a search request is accepted from the input device or a predetermined terminal The search word indicated by the search request is collated with the notation fluctuation dictionary, the pattern registered for the event corresponding to the search word is specified, and the predetermined database is searched with the character string indicated by the specified pattern. And an arithmetic unit that executes processing for outputting the information extracted in the search to an output device or the predetermined terminal.

  Further, according to the data search method of the present invention, an information processing device including a storage device that stores, for each event, master information in which a plurality of notations indicating the same event are associated is stored in the storage device. Reading, dividing the character string constituting each notation included in the master information into words, specifying the partial character string of each notation in each event, and between each of the specified partial character strings, A process of determining a synonym relationship according to a position in the notation, associating partial character strings that are determined to have a synonym relationship by the determination, storing them in a storage device, and generating a partial notation fluctuation dictionary; and a character string of the notation Are compared with the partial notation fluctuation dictionary, the other partial character strings associated with the corresponding partial character strings are specified, and each partial character string of the notation and each of the obtained partial character strings are obtained. Other parts A combination pattern relating to a character string is generated, a character string indicated by each pattern is stored in a storage device in association with a corresponding event indicated by the notation, and a notation fluctuation dictionary is generated, and a search request is made from an input device or a predetermined terminal The search word indicated by the search request is collated with the notation fluctuation dictionary, the registered pattern for the event corresponding to the search word is specified, and the predetermined database is searched with the character string indicated by the specified pattern. And the process which outputs the information extracted by the said search to an output device thru | or the said predetermined terminal is performed, It is characterized by the above-mentioned.

  According to the present invention, the comprehensiveness of search processing for data including notation fluctuation is improved.

It is a network block diagram including the data search system of this embodiment. It is a figure which shows the hardware structural example of the data search system in this embodiment. It is a figure which shows the example of a data structure of the master information table of this embodiment. It is a figure which shows the hardware structural example of the user terminal in this embodiment. It is a flowchart which shows process sequence example 1 of the data search method in this embodiment. It is a figure which shows the example of a data structure of the partial notation fluctuation dictionary of this embodiment. It is a figure which shows the example of a data structure of the notation fluctuation dictionary of this embodiment. It is a figure which shows the pattern production example 1 of the notation fluctuation dictionary in this embodiment. It is a figure which shows the pattern production example 2 of the notation fluctuation dictionary in this embodiment. It is a figure which shows the pattern generation example 3 of the notation fluctuation dictionary in this embodiment. It is a flowchart which shows process sequence example 2 of the data search method in this embodiment.

--- System configuration ---
Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a network configuration diagram including a data search system 100 of the present embodiment. A data search system 100 shown in FIG. 1 is a computer system that can improve the completeness of search processing for data including notation fluctuation. The data search system 100 is connected to the user terminal 200 and the like via the network 20, receives a search request including the search word 107 transmitted from the user terminal 200, and extracts from the clinical data 105 accordingly. The information processing apparatus returns the retrieved data as the search result 106.

  The data search system 100 according to the present embodiment is a search target, for example, clinical data 105 accumulated in a medical institution. However, in addition to such clinical data, medical information in general (eg, various examination data related to medical care, experimental data at medical research institutions, statistical data aggregated by public institutions, etc.), etc., are enormous and the same event Various data groups that are likely to cause notation fluctuation (eg, cancer, cancer, cancer) (eg, disease name, examination name, drug name, etc.) are also search targets in the search system 100 of the present embodiment.

  By the way, in medical institutions, there is a standard master of terms indicating each event such as examinations and diseases, and it is referred to when recording in electronic medical records and subsequent use of information. In many cases, a table or the like defines the correspondence between the term to be used, that is, the notation fluctuation and the standard master. Therefore, in the present embodiment, in the above-described correspondence table created independently by the standard master used by medical institutions or the like or by the medical institution, information on the notation of each term defined as having a synonym relationship with each other is used. Therefore, it is assumed that the notation fluctuation pattern is efficiently expanded and the comprehensiveness in the search processing of the corresponding clinical data 105 is improved.

  The configuration of such a data search system 100 will be described. FIG. 2 is a diagram illustrating a hardware configuration example of the data search system 100 according to the present embodiment. First, the hardware configuration of the data search system 100 is as follows. The data search system 100 reads out to the memory 13 a storage device 11 composed of an appropriate nonvolatile storage device such as a hard disk drive, a memory 13 composed of a volatile storage device such as RAM, and a program 12 held in the storage device 11. And the like, and a control device 14 such as a CPU for performing various determinations, calculations and control processes, and a communication device 17 connected to the network 20 and responsible for communication processing with other devices.

  In the storage device 11, a master 12 that stores, for each event, master information that associates a program 12 for implementing functions necessary for the data search system 100 of the present embodiment and a plurality of notations indicating the same event. The information table 101 and at least a normalization rule 104, which is a conversion rule for converting characters constituting a character string into a specified character type, are stored. However, after the partial notation fluctuation dictionary 102 and the notation fluctuation dictionary 103 described later are generated by the data search system 100, the partial notation fluctuation dictionary 102 and the notation fluctuation dictionary 103 are stored in the storage device 11. .

  Further, the clinical data 105 to be searched may be held in the storage device 11 by the data search system 100 (indicated by a broken line in FIG. 2 in a non-essential meaning), as illustrated in FIG. It is also possible to hold the data searchable in other devices connected to the network 20 via the network 20. In addition, as the master information table 101 in the present embodiment, master information that prescribes examination names corresponding to medical examination codes is assumed, but the master information table 101 is not limited to this, and any master information such as an injury and illness master or a medicine master. Can be assumed.

  FIG. 3 shows an example of such a master information table 101. The master information table 101 includes a plurality of records, and one record stores information related to one inspection event. Each record stores one inspection code and at least one inspection name indicating the inspection of the inspection code. When a plurality of examination names exist, that is, when a plurality of synonyms exist for one examination code, a plurality of examination names are stored in one record corresponding to the examination code. For example, the record 201 in the master information table 101 in FIG. 3 stores two notations “β-TG” and “b-thromboglobulin” as examination names corresponding to the examination code “1000”. Note that each record in the master information table 101 may include not only the above-described items (inspection code, inspection name) but also items according to other arbitrary attributes (those that cause notation fluctuation).

  The data search system 100 does not receive a search request for the data search system 100 from the user terminal 200 via the network 20 and the communication device 17, but the data search system 100 directly receives a search request from the user. It is good. In this case, the data search system 100 includes an input device 15 that accepts key input and voice input from the user, and an output device 16 such as a display that displays processing data (indicated by a broken line in FIG. 2 in a non-essential sense). ing).

  On the other hand, as illustrated in FIG. 4, the user terminal 200 has a hardware configuration as a general information processing device, as in the data search system 100, and is configured by an appropriate nonvolatile storage device such as a hard disk drive. Storage device 21, a memory 23 composed of a volatile storage device such as a RAM, and a program 22 held in the storage device 21 is read and executed to the memory 23 to execute overall control of the device itself, and various determinations and calculations And an arithmetic device 24 such as a CPU that performs control processing, an input device 25 that receives key input and voice input from a user, and an output device 26 such as a display that displays processing data. The user terminal 200 is a terminal operated by a user who desires to retrieve desired data from the clinical data 105 and use it.

  Next, functions provided in the data search system 100 of this embodiment will be described. As described above, it can be said that the functions described below are implemented by executing the program 12 included in the data search system 100, for example. In the example of FIG. 2, the master information reading unit 111, the character string normalization unit 112, the character string division unit 113, the partial notation fluctuation dictionary generation unit 114, the notation fluctuation dictionary generation unit 115, The notation fluctuation pattern generation unit 116 and the data extraction unit 117 are mounted.

  The data search system 100 reads the master information of each event from the master information table 101 of the storage device 11 (function of the master information reading unit 111), divides the character string constituting each notation included in the master information into words, A function (function of the character string dividing unit 113) for specifying a partial character string of each notation in each event is provided.

  In addition, the data search system 100 determines the synonym relation between the partial character strings specified above according to the position in each notation related to the same event, and the partial character strings that are determined to have the synonymous relation by the determination. Are stored in the storage device 11 in association with each other and a partial notation fluctuation dictionary 102 is generated (function of the partial notation fluctuation dictionary generation unit 114).

  Further, the data search system 100 collates each partial character string constituting the above-described character string with the partial notation fluctuation dictionary 102, specifies the other partial character string associated with the corresponding partial character string, and A combination pattern regarding each partial character string of the notation and each other partial character string obtained with respect to the corresponding partial character string is generated, and the character string indicated by each pattern is stored in the storage device 11 in association with the corresponding event indicated by the above notation. And a function for generating the notation fluctuation dictionary 103 (function of the notation fluctuation dictionary generation unit 115).

  Further, the data search system 100 receives a search request from the input device 15 or the user terminal 200 (predetermined terminal), collates the search word indicated by the search request with the notation fluctuation dictionary 103, and the event corresponding to the search word. A function (data extraction) that specifies a registered pattern, searches clinical data (predetermined database) 105 with a character string indicated by the specified pattern, and outputs information extracted by the search to the output device 16 or the user terminal 200 The function of the unit 117).

  Further, in the process of specifying the partial character string described above, the data search system 100 reads the master information of each event from the master information table 101 of the storage device 11, and sets a predetermined character string constituting each notation included in the master information. Applying the normalization rule 104 that is a conversion rule, a normalization process is performed to convert the characters forming the corresponding character string into a specified character type suitable for the subsequent processing (function of the character string normalization unit 112), and the normalization process It has a function to divide a subsequent character string into words and specify a partial character string of each notation in each event.

  In addition, when the data search system 100 generates the notation fluctuation dictionary 103, the partial character string and the other partial character string, the partial character string and the partial character string, the other partial character string and the other part indicated by the pattern generated above. For any pair of character strings, the accuracy level to be a reliability index of the corresponding pattern is determined according to the registration status of the corresponding character string in the master information table 101 or the commonality between the patterns, and the accuracy level Is registered in the notation fluctuation dictionary 103 in association with the corresponding pattern (function of the notation fluctuation dictionary generation unit 115).

  Further, in the process of searching the clinical data 105 in response to the search request and outputting information, the data search system 100 divides the search word indicated by the search request into words, and specifies the partial character string. The partial character string specified here is collated with the partial notation fluctuation dictionary 102, the other partial character string associated with the corresponding partial character string is specified, and each partial character string of the search word and the corresponding partial character string are related. A combination pattern related to each of the other partial character strings is generated (function of the notation fluctuation pattern generation unit 116), the clinical data 105 is searched with the character string indicated by each pattern, and the information extracted by the search is output. 15 to a function of outputting to the user terminal 200.

--- Example of processing procedure 1 ---
Hereinafter, the actual procedure of the data search method in the present embodiment will be described with reference to the drawings. Various operations corresponding to the data search method described below are realized by a program that the data search system 100 reads into a memory or the like and executes. And this program is comprised from the code | cord | chord for performing the various operation | movement demonstrated below.

  FIG. 5 is a flowchart showing a processing procedure example 1 of the data search method in the present embodiment. Here, first, a process in which the data search system 100 generates the partial notation fluctuation dictionary 102 and the notation fluctuation dictionary 103 will be described. In this case, the master information reading unit 111 in the data search system 100 reads each record of the master information table 101 in the storage device 11 into the memory 13 (s100), and extracts each value of the inspection code and the inspection name included in each record. (S101). In the example of FIG. 2, for example, for test code “1000”, each value of test name 1 “β-TG” and test name 2 “b-thromboglobulin” is extracted, and test code “1001” is tested. Extraction processing such as extraction of each value of name 1 “CK iso” and test name 2 “CK isozyme” is sequentially executed for each record.

  Subsequently, the character string normalization unit 112 in the data search system 100 executes character conversion based on the normalization rule 104 on the value of the examination name extracted from each record in step 100 described above, that is, the character string (s102). ). As an example of the normalization rule 104, a rule is assumed that unifies character types, such as converting lowercase letters to uppercase letters, half-width characters to full-width characters, and "" to ().

  Next, the character string dividing unit 113 in the data search system 100 converts each character string indicating the examination name corresponding to the examination code to the partial character after the character conversion is performed in accordance with the normalization rule 104 in step s102 described above. The process is divided into elements (hereinafter referred to as elements) that are columns (s103). In the present embodiment, a character string is divided into elements by using a difference in character type or an existing morpheme analysis technique using a predetermined dictionary.

  Subsequently, the partial notation fluctuation dictionary generation unit 133 in the data search system 100 uses a plurality of examination names associated with one examination event as synonyms in the record to be processed in the above steps. The specified examination name is selected (s104). Further, the partial notation fluctuation dictionary generation unit 113 compares the elements obtained in step s103 with respect to the corresponding examination name among the plurality of examination names selected in step s104 described above, and determines whether there is a matching element (s105). ).

  For example, when examination name 1 “β-TG” and examination name 2 “b-thromboglobulin” are selected in step s104, the partial notation fluctuation dictionary generation unit 113 includes elements “β”, “−” included in each examination name. ”,“ TG ”,“ b ”,“ − ”,“ thromboglobulin ”(these are obtained in step s103), and the presence of“ − ”is identified.

  As a result of such determination, when it is found that there is a matching element between a plurality of examination names (s105: Y), the partial notation fluctuation dictionary generation unit 113 matches between examination names in the character string of the corresponding examination name. An element set in front of the element and an element set behind the matching element are specified (s106). In the above example, the partial notation fluctuation dictionary generation unit 113 determines the examination name from the element “-” common to the two examination names of the examination name 1 “β-TG” and the examination name 2 “b-thromboglobulin”. For 1 “β-TG”, the character string “β” as the front element is specified, and for the examination name 2 “b-thromboglobulin”, the character string “b” as the front element is specified. Similarly, for the test name 1 “β-TG”, the rear character string “TG” is specified, and for the test name 2 “b-thromboglobulin”, the rear character string “thromboglobulin” is specified. When the number of elements (at the front or rear of the common elements) in at least one examination name is 0, the examination name is excluded from the subsequent processing targets.

  Subsequently, the partial notation fluctuation dictionary generation unit 113 compares the number of elements for each examination name for which the element set has been specified in step s106 described above, and if the number of elements is not equal between the examination names (s107: N), (Partial character string) is integrated to reduce the number of elements in the element set having a large number of elements (element set in front of or behind the above-mentioned common element), thereby reducing the number of elements between the above examination names. Unify (s108). The partial notation fluctuation dictionary generation unit 113 determines which elements are to be integrated based on whether each element or a character string combining the elements exists in the partial notation fluctuation dictionary 102 or the like. For example, when three elements “A, B, C” are converted into two elements by integration processing, each element “A”, “B”, “C” exists in the partial notation fluctuation dictionary 102, It is determined whether or not a character string, such as “AB” or “BC”, in which the elements are connected to each other and the continuity as a character string is not impaired, is present in the partial notation fluctuation dictionary 102.

  When an element (partial character string) or a character string in which elements are integrated exists in the partial notation fluctuation dictionary 102, the partial notation fluctuation dictionary generation unit 113 gives priority to the notation existing in the partial notation fluctuation dictionary 102. Perform the integration process. For example, when the notation “AB” is present in the partial notation fluctuation dictionary 102, the elements “A” and “B” of the elements “A”, “B”, and “C” are integrated into “AB”. And this element “AB” and the remaining element “C” are defined as two elements. On the other hand, if the element or the character string that combines the elements does not exist in the partial notation fluctuation dictionary 102, the partial notation fluctuation dictionary generation unit 113 does not impair continuity as a character string when the elements are connected. Generate multiple integration patterns. For example, when “A”, “B”, “C”, “AB”, “BC”, and “ABC” do not exist in the partial notation fluctuation dictionary 102, the partial notation fluctuation dictionary generation unit 113 selects “AB, Two patterns of “C”, “A, BC” are generated, and the subsequent processing is performed.

  Next, the partial notation fluctuation dictionary generation unit 113 obtains the positional relationship of the elements in each examination name obtained up to step s108 described above, that is, the positional relation of the elements between the examination names with the same number of elements selected as synonyms. Is used to determine the partial notation fluctuation, and the elements that have a synonymous relationship by the determination (partial character strings in front of or behind the common element) are stored in the storage device 11 in association with each other. Is generated (s109). Before describing the step s109 in detail, the partial notation fluctuation dictionary 102 will be described.

  FIG. 6 shows an example of the partial notation fluctuation dictionary 102. The partial notation fluctuation dictionary 102 is a partial character string of a partial character string corresponding to the above-described element, an ID of the partial character string, and another partial character string (synonymous partial character string) determined to be a synonym of the partial character string. ID is stored. At least one partial character string ID of the synonymous partial character string is stored in one record. For example, the record 301 indicates that “b” (partial character string ID “0002” and “beta” (partial character string ID “0015”) is determined as a synonym for the partial character string “β”. It should be noted that the partial notation fluctuation dictionary 102 may include partial notation fluctuation information obtained as a result of executing the data search method of the present embodiment using other master information.

  The process in step s109 will be described by taking as an example the case where two element sets “AB, C” and “D, E” are obtained in step s108. In this case, the partial notation fluctuation dictionary generation unit 113 calculates “AB” and “D” from the positional relationship of each element in the element set (for example, in the element set, that is, between the leading elements in the character string or between the trailing elements) It is determined that “C” and “E” are synonyms, and a character string corresponding to the corresponding element is added to the partial notation fluctuation dictionary 102.

  At this time, the partial notation fluctuation dictionary generation unit 113 determines whether the partial character strings “AB” and “D” as elements are registered in the partial notation fluctuation dictionary 102. When the partial character string is not registered in the partial notation fluctuation dictionary 102, the partial notation fluctuation dictionary generation unit 113 registers the character string in the partial notation fluctuation dictionary 102. On the other hand, if the partial character string is registered in the partial notation fluctuation dictionary 102, the partial notation fluctuation dictionary generation unit 113 determines whether the partial character string that is an element determined to have a synonym relationship is registered as a synonym. Determine and perform the registration process. For example, in the partial notation fluctuation dictionary 102, it is determined whether or not the partial character string ID of the character string “D” is registered as a synonym ID of the character string “AB”. The unit 113 adds the partial character string ID of the character string “D” to the synonym ID of the partial character string “AB”. At the same time, it is determined whether the partial character string ID of the character string “AB” is registered in the partial notation fluctuation dictionary 102 as a synonym ID of the character string “D”. 113 adds the partial character string ID of the character string “AB” to the synonym ID of the character string “D”. The partial notation fluctuation dictionary generation unit 113 performs the above processing for the character strings “C” and “E” determined to be synonymous.

  The partial notation fluctuation dictionary generation unit 113 performs the processes of s104 to s109 described above for all combinations of synonyms. For example, when there are three or more inspection names having the same inspection code, the above-described processing of s104 to s109 is executed for all combinations that are established by selecting two from the set of inspection names. For example, when three examination names having the same meaning of “TG”, “triglyceride”, and “neutral fat” are obtained from one record, the partial notation fluctuation dictionary generation unit 113 selects “TG”, “triglyceride”, The processing of s104 to s109 is performed on three combinations of “triglyceride” and “neutral fat”, “TG” and “neutral fat”.

  Next, generation processing of the notation fluctuation dictionary 103 by the notation fluctuation dictionary generation unit 115 of the data search system 100 will be described. The notation fluctuation dictionary generation unit 115 collates each element registered in the partial notation fluctuation dictionary 102, that is, each partial character string (element) constituting the character string of the examination name, with the partial notation fluctuation dictionary 102, and the corresponding partial character string. Identifies the element corresponding to the other partial character string, that is, the synonymous partial character string ID, and generates a combination pattern relating to each partial character string of the examination name and each other partial character string obtained with respect to the corresponding partial character string Then, the character string indicated by each pattern is stored in association with the inspection code corresponding to the corresponding inspection name, and the notation fluctuation dictionary 103 is generated.

  Here, as an example, an example will be described in which pattern generation is performed for fluctuations in notation of “β-TG” and “b-thromboglobulin” (hereinafter referred to as target terms) that are synonymous in the master information table 101. Shall. The notation fluctuation dictionary generation unit 115 collates the ID of the partial character string that constitutes the target term described above with the partial notation fluctuation dictionary 102, and synonymous partial characters that are associated with the ID of the partial character string that constitutes the target term. A column ID is acquired (s110). For example, focusing on the partial character string “β” of the target term “β-TG”, the ID acquired in the partial notation fluctuation dictionary 102 is “0001”, which is the partial character string ID of “β”, and this “β "," "0002" and "0015" which are synonymous partial character string IDs associated with "". Similarly, IDs obtained in the partial notation fluctuation dictionary 102 by paying attention to the partial character string “TG” are “0003”, “0004”, and “0039”.

  Next, the notation fluctuation dictionary generation unit 115 generates a notation fluctuation pattern to be registered in the notation fluctuation dictionary 103 (s111). Specifically, all the notation fluctuation patterns are created by combining the partial character string ID and the synonymous partial character string ID obtained in step s110 with respect to the target term as much as possible. This notation fluctuation pattern is an examination name in which character strings corresponding to the combined partial character string ID and synonymous partial character string ID are connected. That is, the data search system 100 generates a test name that may be derived from an existing test name registered in the master information table 101.

  Therefore, the above-described notation fluctuation pattern is created by the product of the number of partial character string IDs and synonymous partial character string IDs acquired from the partial notation fluctuation dictionary 102 based on each partial character string of the target term. For example, when there are three IDs acquired in the partial notation fluctuation dictionary 102 based on the partial character string “β” and the partial character string “TG” that constitute the target term, the notation fluctuation pattern created is As shown in Table 601 in FIG. 8, “3 × 3” is nine. When similar processing is performed on “b-thromboglobulin” which is synonymous with the target term “β-TG”, six notation fluctuation patterns are created as shown in Table 602 of FIG.

  Subsequently, the notation fluctuation dictionary generation unit 115 determines the accuracy level of the notation fluctuation pattern obtained in the above-described step 111 and registers it in the notation fluctuation dictionary 103 (s112). In this case, the notation fluctuation dictionary generation unit 115 indicates the partial character string that forms the target term, the synonymous partial character string, the partial character strings that form the target term, the target, indicated by each notation fluctuation pattern obtained in step s111 described above. For each set of synonymous partial character strings related to a partial character string that constitutes a term, the accuracy level that is a reliability index of the corresponding pattern is determined according to the registration status in the master information table 101 or the commonality between the notation fluctuation patterns. Then, the accuracy level value is registered in the notation fluctuation dictionary 103 in association with the corresponding pattern.

  For example, a notation fluctuation dictionary is generated for “β-TG” corresponding to “0001, 0003” in Table 601 and “b-thromboglobulin” corresponding to “0002, 0004” in Table 602 among the notation fluctuation patterns. The unit 115 recognizes that all are names included in the record of the master information table 101, and sets a value of the accuracy level “1” in the record of the corresponding notation fluctuation pattern in the notation fluctuation dictionary 103. FIG. 10 exemplifies a table 603 that summarizes the fluctuation pattern of the accuracy level “1”.

  In addition, the notation fluctuation dictionary generation unit 115, for the notation fluctuation pattern that exists in any of the notation fluctuation patterns created based on the above-described target terms, the accuracy level “ A value of “2” is set. For example, the character strings that exist in both the written fluctuation pattern created from “β-TG” and the written fluctuation pattern created from “b-thromboglobulin”, ie, “β-thromboglobulin”, “b-TG” The value of the accuracy level “2” is registered in the notation fluctuation dictionary 103. FIG. 10 exemplifies a table 604 that summarizes the fluctuation pattern of the accuracy level “2”.

  In addition, the notation fluctuation dictionary generation unit 115 stores a notation fluctuation pattern that exists in only one of the notation fluctuation patterns created based on the above-described target terms in a record of the corresponding notation fluctuation pattern in the notation fluctuation dictionary 103. Set the value of accuracy level “3”. For example, among the notation fluctuation patterns created from the subject terms “β-TG” and “b-thromboglobulin”, as shown in Tables 601 and 602, “β-triglyceride”, “b-triglyceride”, “beta-TG” ”And the like are present only in the written fluctuation pattern for only one of the target terms, so that they have an accuracy level of“ 3 ”. FIG. 10 exemplifies a table 605 that summarizes the fluctuation pattern of the accuracy level “3”.

  A specific example of the notation fluctuation dictionary 103 generated in this way is shown in FIG. The notation fluctuation dictionary 103 of this embodiment illustrated in FIG. 7 is an aggregate of records including three attribute values of an inspection code, an inspection name, and an accuracy level. Each record stores an inspection code, an inspection name, and an accuracy level one by one. The accuracy level described above is an index indicating the likelihood of the written fluctuation pattern created as described above. In this embodiment, as the accuracy, the corresponding examination name is included in the record of the master information table 101 (accuracy level 1), and the notation fluctuation created from each target term that is a synonym in the master information table 101 Three accuracy levels: a notation fluctuation pattern (accuracy level 2) that exists in any of the patterns, and a notation fluctuation pattern (accuracy level 3) that exists only in one of the notation fluctuation patterns created from the target terms that are synonyms Is set. The notation fluctuation dictionary 103 may include notation fluctuation information obtained as a result of executing the data search method of this embodiment using other master information. Moreover, you may include the synonym information acquired using means other than the data search method of this embodiment, for example, the means of patent document 1. FIG.

--- Processing procedure example 2 ---
The process up to the generation of the partial notation fluctuation dictionary 102 and the notation fluctuation dictionary 103 has been described above. Thereafter, the partial notation fluctuation dictionary 102 and the notation fluctuation dictionary 103 are used as appropriate, and a search process for performing an exhaustive search in response to a search request from the user, that is, only data that matches the search word from the clinical data 105. In addition, a description will be given of a process for outputting data that matches the term synonymous with the search word, that is, the notation fluctuation. FIG. 11 is a flowchart showing a processing procedure example 2 of the data search method according to this embodiment. Here, a search problem of extracting information on patients who have been examined for “β-TG” from clinical data 105 will be described as an example.

  In this case, the character string normalization unit 112 in the data search system 100 acquires the search word 107 from, for example, a search request received from the user terminal 200, applies the normalization rule 104 to the search word 107, and The character conversion process similar to that in step s102 is performed (s200). Details of processing similar to the processing procedure example 1 described above are omitted to avoid duplication of explanation (the same applies hereinafter). In addition, the character string dividing unit 113 in the data search system 100 performs the character string dividing process similar to that in step s103 described above on the search word normalized by the character string normalizing unit 112 in step s200 described above. (S201).

  Further, the notation fluctuation pattern generation unit 116 in the data search system 100 collates the partial character string identified by dividing the search word 107 with the partial notation fluctuation dictionary 102, and obtains the synonymous partial character string associated with the corresponding partial character string. The combination pattern regarding each partial character string of the search word 107 and the synonymous partial character string obtained regarding the corresponding partial character string is generated (s202). In this processing flow, each process of the character string dividing unit 113 and the notation fluctuation pattern generation unit 116 is a case where the search word normalized by the character string normalization part 112 is a term included in the notation fluctuation dictionary 103. Can be omitted.

  Subsequently, the data extraction unit 117 in the data search system 100 synonyms (search word) indicated by the search word normalized in step s200 described above and the notation fluctuation pattern generated by the notation fluctuation pattern generation unit 116 in step s202. 107, a search is performed on the clinical data 105 to extract data, and the corresponding data is output to the user terminal 200 as a search result (s203). For example, when the search word 107 is “β-TG”, “β-TG” itself and the notation fluctuation dictionary 103 are searched using “β-TG” as a key and the same inspection code as the “β-TG”. The clinical data 105 is searched using another examination name having “1000” and the term indicated by the notation fluctuation pattern obtained in step s202 described above as a search key, and the search result is output.

  The above-described data extraction unit 117, when executing step s203, lists the search word 107 and its synonyms (obtained from the notation fluctuation dictionary 103 and obtained from the search word 107) in descending order of accuracy level. It is good also as returning to the user terminal 200 as a search candidate, and making the searcher who browsed this select the term actually extracted. In this case, the display order and display range of the search word 107 and its synonyms may be preset on the data search system 100 side, or set by the searcher accessing the data search system 100 via the user terminal 200. You may do that. For example, by setting the display range of search candidates so that only the terms registered in the master information table 101, only the terms registered in the master information table 101 and the high-precision notation fluctuation, the searcher's wishes are met. Efficient search can be performed.

  Further, when the data search system 100 accepts a combination of a plurality of search words with a space symbol as the search word 107, the data search system 100 performs an AND search on the plurality of search words. For example, when the search word 107 “triglyceride LDL cholesterol” is received from the user terminal 200, the above steps s200 to s203 are executed for each word of “triglyceride” and “LDL cholesterol”, and data including both words Is returned to the user terminal 200 as a search result.

  Although the best mode for carrying out the present invention has been specifically described above, the present invention is not limited to this, and various modifications can be made without departing from the scope of the invention.

  According to the present embodiment, it is possible to efficiently construct an exhaustive synonym dictionary that is effective in accurately extracting information to be analyzed from medical information including notation fluctuation. It is also possible to present to the user the accuracy of the synonyms registered in the synonym dictionary (probability and validity as synonyms). Since such a synonym dictionary can be used at the time of search processing, a search based on a synonym based on a search word designated by the user is executed, and the completeness of the search result is improved. Note that even if the above search is performed with synonyms that are registered in the synonym dictionary but do not actually exist, they are not extracted because they are not included in the search target database. , Does not affect the correctness of search results.

  Therefore, according to the present embodiment, the completeness of search processing for data including notation fluctuation is improved.

  At least the following will be clarified by the description of the present specification. That is, in the process of specifying the partial character string, the arithmetic device reads the master information of each event from the storage device, applies a predetermined conversion rule to the character string constituting each notation included in the master information, Executes a normalization process that converts the characters that make up the character string into a standard character type suitable for the subsequent processing, divides the character string after the normalization process into words, and identifies the partial character string of each notation in each event It is good also as what to do.

  According to this, in the character string included in the master information, the character type fluctuations such as uppercase and lowercase letters, half-width and full-width are unified in advance, that is, normalized, and the subsequent partial character string identification, partial notation fluctuation dictionary and notation fluctuation dictionary It is possible to improve the accuracy and efficiency of each process such as generation.

  Further, in the data search system of the present embodiment, the arithmetic unit, in the process of generating the fluctuation dictionary, shows a partial character string and another partial character string, a partial character string and a partial character string, etc. indicated by the generated pattern. For any pair of partial character string and other partial character string, the accuracy level that is a reliability index of the corresponding pattern is determined according to the registration status in the master information or the commonality between the patterns, and the accuracy It is also possible to further execute a process of associating the level value with the corresponding pattern and registering it in the notation fluctuation dictionary.

  According to this, it is possible to clearly indicate to the user that there is a difference in reliability among the notation fluctuations, and it is possible to present the certainty about the search results at the time of each search.

  Further, in the data search system of the present embodiment, in the process of searching the predetermined database in response to the search request and outputting information, the arithmetic device divides the search word indicated by the search request into words, A partial character string is identified, the identified partial character string is checked against the partial notation fluctuation dictionary, another partial character string associated with the corresponding partial character string is identified, and each partial character string of the search word corresponds to A combination pattern related to each other partial character string obtained with respect to the partial character string is generated, a predetermined database is searched with the character string indicated by each pattern, and information extracted by the search is output to the output device or the predetermined terminal. The process may be executed.

  According to this, it is possible to generate a synonym for the search word in consideration of the fluctuation of the notation, and to execute the search not only for the search word but also for the synonym, and the search accuracy is further improved. Even if the synonym obtained from the search word does not exist in the database to be searched, if the database is searched with such a synonym, the data is extracted because it is not included in the database in the first place. Will not affect the correctness of the search results.

DESCRIPTION OF SYMBOLS 11 Storage device 12 Program 13 Memory 14 Arithmetic device 17 Communication device 20 Network 100 Data retrieval system 101 Master information table 102 Partial notation fluctuation dictionary 103 Notation fluctuation dictionary 104 Normalization rule 105 Clinical data 111 Master information reading part 112 Character string normalization part 113 Character string division unit 114 Partial notation fluctuation dictionary generation part 115 Notation fluctuation dictionary generation part 116 Notation fluctuation pattern generation part 117 Data extraction part 200 User terminal 21 Storage device 22 Program 23 Memory 24 Arithmetic device 25 Input device 26 Output device 27 Communication device

Claims (5)

A storage device that stores, for each event, master information in which a plurality of notations indicating the same event are associated,
Reading the master information of each event from the storage device, dividing a character string constituting each notation included in the master information into words, and specifying a partial character string of each notation in each event;
For each of the identified partial character strings, a synonym relationship is determined according to the position in each notation relating to the same event, and the partial character strings that are determined to have a synonym relationship are associated with each other and stored in the storage device. Processing to generate a partial notation fluctuation dictionary;
Each partial character string constituting the character string of the notation is collated with the partial notation fluctuation dictionary, the other partial character string associated with the corresponding partial character string is specified, and each partial character string and the corresponding partial character of the notation are specified. Generating a combination pattern related to each other partial character string obtained with respect to the column, storing the character string indicated by each pattern in a storage device in association with the corresponding event indicated by the notation, and generating a notation fluctuation dictionary;
A search request is received from an input device or a predetermined terminal, a search word indicated by the search request is collated with the notation fluctuation dictionary, a pattern registered for an event corresponding to the search word is specified, and the specified pattern is An arithmetic device that executes a process of searching a predetermined database with a character string indicating, and outputting information extracted by the search to an output device or the predetermined terminal;
A data retrieval system comprising: The arithmetic unit is:
In the process of specifying the partial character string,
The master information of each event is read from the storage device, and a predetermined conversion rule is applied to a character string constituting each notation included in the master information, and characters constituting the corresponding character string are converted into a prescribed character type suitable for subsequent processing. The normalization process is executed, the character string after the normalization process is divided into words, and the partial character string of each notation in each event is specified.
The data search system according to claim 1. The arithmetic unit is:
In the process of generating the notation fluctuation dictionary,
The registration status or pattern in the master information for any combination of partial character string and other partial character string, partial character string and partial character string, other partial character string and other partial character string, indicated by the generated pattern In accordance with the commonality between them, the accuracy level that becomes the reliability index of the corresponding pattern is determined, and the processing of registering the accuracy level value in the notation fluctuation dictionary in association with the corresponding pattern is further executed. ,
The data search system according to claim 2, The arithmetic unit is:
In the process of searching the predetermined database in response to the search request and outputting information,
The search word indicated by the search request is divided into words, a partial character string is specified, the specified partial character string is checked against the partial notation fluctuation dictionary, and another partial character string associated with the corresponding partial character string And generating a combination pattern relating to each partial character string of the search word and each other partial character string obtained with respect to the corresponding partial character string, searching a predetermined database with the character string indicated by each pattern, and performing the search The process of outputting the information extracted in the output device or the predetermined terminal is executed.
The data search system according to claim 3. An information processing apparatus provided with a storage device that stores, for each event, master information in which a plurality of notations indicating the same event are associated,
Reading the master information of each event from the storage device, dividing a character string constituting each notation included in the master information into words, and specifying a partial character string of each notation in each event;
For each of the identified partial character strings, a synonym relationship is determined according to the position in each notation relating to the same event, and the partial character strings that are determined to have a synonym relationship are associated with each other and stored in the storage device. Processing to generate a partial notation fluctuation dictionary;
Each partial character string constituting the character string of the notation is collated with the partial notation fluctuation dictionary, the other partial character string associated with the corresponding partial character string is specified, and each partial character string and the corresponding partial character of the notation are specified. Generating a combination pattern related to each other partial character string obtained with respect to the column, storing the character string indicated by each pattern in a storage device in association with the corresponding event indicated by the notation, and generating a notation fluctuation dictionary;
A search request is received from an input device or a predetermined terminal, a search word indicated by the search request is collated with the notation fluctuation dictionary, a pattern registered for an event corresponding to the search word is specified, and the specified pattern is A process of searching a predetermined database with a character string indicating, and outputting information extracted by the search to an output device or the predetermined terminal;
The data search method characterized by performing.

Images