JP5050504B2 - Receipt extraction method and receipt inspection support system using reexamination results - Google Patents

Description

  The present invention relates to a receipt extraction method and a receipt inspection support system for improving the efficiency of inspection of a receipt (medical fee description) requested by a medical institution in a medical insurer.

  In the current medical insurance system, the person who received the medical care and the medical insurer (hereinafter referred to as the insurer) pay the cost related to the medical treatment performed at the medical institution. As a flow for an insurer to pay, first, a medical institution issues a receipt (medical remuneration statement) that records points regarding the medical practice, its contents, and its cost, and sends it to the examination and payment institution. The examination and payment institution checks the contents and sends a receipt to the insurer, and the insurer processes the medical expenses for the receipt through the examination and payment institution to the medical institution.

  In order to confirm that there is no error in the contents of the receipt, the medical institution that issues the receipt, the examination and payment institution that performs the contents examination, and the insurer that pays the medical expenses are each inspected and examined for the contents of the receipt. Currently, such examinations and inspections are often carried out by specialists (inspectors) who conduct inspections, but the number of issued issuances is enormous, and the medical remuneration score table has been established. Due to the enormous and complicated rules, the burden of examination and inspection at each organization is very large.

  Therefore, several methods have been proposed for supporting the inspection and examination of the receipt for such a situation. For example, in Patent Document 1, for a check at a medical institution, a master that describes the check rules is created, the last use date is recorded in the master item, and only the items used within a predetermined period are checked. A method to reduce the number of master items to be checked is proposed. Also, in Patent Document 2, for the examination by the examination / payment institution, the result of the reexamination and the upper limit of the number of claims for the medical treatment are managed as the actual results, and if the requested score exceeds the upper limit, the receipt has an error. Proposes a method to determine that there is.

  As another technique, a method of applying a machine learning technique such as correlation rule analysis or decision tree in Non-Patent Document 1 to this field can be considered. In this case, in order to create a rule for whether or not there is an error in the new receipt, the keyword is extracted from the receipt that has been judged to have an error as a result of past inspections and examinations, and the receipt that has been judged to have no error. By applying a machine learning method such as a tree, a rule is created that distinguishes an erroneous receipt from an erroneous receipt. Then, a method of determining whether there is an error in the new receipt using the created rule can be considered.

JP2005-108138 JP2005-222176 Artificial Intelligence Encyclopedia (Kyoritsu Publishing): pp303-305, pp673-674

  As mentioned above, the check of the receipt basically follows a standard such as a medical fee score table. For example, the rules stipulate that certain drugs can only be prescribed for certain illnesses, and that certain drugs can be used for a limited period of time. In the reception inspection, the inspector who performs the inspection work performs an operation of finding an error in the description on the reception that does not comply with this standard. However, since there are an enormous number of types of medicines and the rules are updated as needed, a system that supports receipt inspection needs to cope with the increase and update of such standards. In addition, it is said that the contents of medical care performed by doctors vary and are not uniformly defined. In addition, in the case of inspections by insurers, since insurers have accumulated records of insured persons from the past, general inspections for finding errors based on the relationship with the past receptions and out-of-hospital prescription medical institutions It is important to have a dispensing check that checks the receipt from the dispensing pharmacy. In particular, in general inspection, it is necessary to search for errors from the previous month and the month before the previous month, but to what extent it is necessary to go back, and what values are collected for the previous date. It is necessary to decide. However, in the above-described conventional example, a method for satisfying all of these has not been considered. For example, although the method of Patent Document 1 can reduce the number of rules to be used by using only the rules used within a certain period, it is necessary to prepare an inspection rule as a master in advance, and rules can be added or updated. If it is necessary, it is considered that work to update the master of the rule will occur, and the method of Patent Document 2 can automatically create a rule regarding the score to some extent based on the result of the reexamination. However, it was not considered how to deal with the general inspection that goes back in the past, especially how to specify the period that goes back through the general inspection.

  Further, when the machine learning method disclosed in Non-Patent Document 1 is applied, it is considered that rules can be added / updated by creating a rule from past receipt data. However, if a rule is to be created for a general inspection, it is necessary to specify how far the receipt to be learned needs to be traced, and what items in it should be aggregated and learned. If this is not done, it is assumed that the number of types of items will increase and learning will become difficult. In addition, when machine learning is simply performed, rules that are not meaningful to the inspection staff may appear, which may cause a problem that such rules are difficult to use for actual inspection support. There is sex.

In order to solve the above-described problem, the method for extracting a receptor of the present invention is a method for extracting a receptor corresponding to the above-described receptor extraction rule from a plurality of receptors using a receptor extraction rule having information on the characteristics of the receptor. ,
Rule creation means for creating the receipt extraction rule,
Re-examination history data including the claim text that is the claim when the insurer makes a request for re-examination of the receipt to the examination / payment institution, and the name of the wound, the name of the medicine, the name of the medical practice name written on the receipt A receipt code keyword for managing a keyword comprising at least a part of at least one of the names, a wording keyword that is a keyword that appears in the claim word and is not included in the receipt code keyword, and the receipt code keyword and the wording keyword Using a database having a combination pattern that defines how to create a receipt extraction rule for the combination,
First, search the receipt code keyword and the wording keyword from the claim wording,
Second, confirm the match between the combination of the receipt code keyword and the word keyword retrieved from the claim word and the combination pattern;
Third, create a receipt extraction rule using a receipt code keyword according to the receipt extraction rule creation method defined in the combination pattern,
The inspection necessity determination unit extracts a corresponding receipt from a receipt stored in a database in accordance with the receipt extraction rule created by the rule creation unit.

  Furthermore, in the receipt extraction method of the present invention, when the wording keyword stored in the database is a keyword indicating a numerical value and a unit of a period, the numerical value is converted to the month for the wording keyword. When the combination pattern includes a keyword indicating the numerical value and the unit of the period, the rule creating means converts the unit into a month when creating the receipt extraction rule in the third step. A receipt extraction rule is created in which a period including the numerical value is a period for searching for a receipt retroactively.

  Further, in the receipt extracting method of the present invention, the rule creating means, when the combination pattern matched in the second step includes the word keyword indicating the numerical value and the unit of the period, and the receipt code keyword, Step of summing up the value of the quantity of medicines or medical treatments indicated by the receipt code keyword that exists during the period of searching backwards for the receipt, and the aggregated value is the unit of the term keyword A search condition for the receipt extraction rule is to exceed the value obtained by converting the value into days.

  Furthermore, in the receipt extraction method of the present invention, the reexamination record data stored in the database is either “accepted” where the request was approved as a result of the reexamination, or “as per the original referee” where the request was not accepted. And the rule creation means, in the second step, includes a combination of the receipt code keyword and the wording keyword and the combination pattern. When confirming a match, the number of accepted cases and the number of accepted points are totaled for each combination pattern and the receipt code keyword, and when a receipt extraction rule is created in the third step, the re-registration is performed for each combination pattern and the receipt code keyword. The acceptance ratio indicating the percentage of acceptance with respect to the number of cases of the examination results and the average acceptance score per acceptance are given in the receipt. When the receipt necessity rule is extracted from the plurality of receipts stored in the database using the receipt extraction rule, the inspection necessity determination unit searches for the receipt extraction rule from the plurality of receipts. A value obtained by multiplying the number of cases by the acceptance ratio, and a value obtained by multiplying the number of cases retrieved by the receipt extraction rule by the acceptance ratio and the average acceptance point are calculated.

Furthermore, in the receipt extraction method of the present invention, the database includes receipt data related to the reexamination record data and receipt data that has not been requested for reexamination in the past, and the rule creation means includes:
Fourth, from the receipt data related to the reexamination record data, obtain the receipt data related to the reexamination record data that is the basis of the receipt extraction rule and accepted as a result of the reexamination,
Fifth, from the receipt that has not been requested for reexamination in the past, obtain the receipt data using the receipt extraction rule,
Sixth, from the receipt data that is the basis of the receipt extraction rule and the receipt data that is not requested for reexamination obtained using the receipt extraction rule, the difference between the items described is determined as the acceptance ratio or the average acceptance. Learning so that the score is higher than the receipt extraction rule,
Seventh, the item obtained as a result of the learning is corrected to be added to the receipt extraction rule.

  Furthermore, the method for extracting a receipt according to the present invention provides, for each receipt extraction rule managed by the database, an acceptance ratio indicating a ratio of the number of acceptances to the number of receipts extracted by the receipt extraction rule, An average acceptance score indicating the number of points is calculated, and when the acceptance ratio and the average acceptance score are less than a predetermined value, the receipt extraction rule is deleted.

In addition, in order to solve the above-described problem, the receipt inspection support system according to the present invention uses a receipt extraction rule having information on the characteristics of a receipt to be inspected preferentially, and receives a receipt corresponding to the receipt extraction rule from a plurality of receipts. Is a receipt inspection support system for supporting receipt inspection,
The receipt inspection support system has a receipt inspection support terminal and a database,
The database is
Re-examination record data management means for managing the re-examination record including the claim text that is the claim when the insurer makes a request for re-examination of the receipt to the examination and payment institution;
A receipt code keyword management means for managing a keyword comprising at least a part of at least one of a name of a disease, a medicine name, and a name of a medical practice displayed on a receipt;
Wording keyword management means for managing wording keywords that appear in the claim wording and are not included in the receipt code keyword;
A combination pattern managing means for managing a combination pattern defining a method for creating a receipt extraction rule for the combination of the receipt code keyword and the wording keyword;
Receipt data management means for accumulating past receipt data;
A receipt extraction rule managing means for managing the receipt extraction rule;
The receipt inspection support terminal is
An input unit for receiving a receipt to be inspected;
A receipt extractor that prioritizes inspections for the inspection target; and
An output unit for outputting the prioritized results;
The receipt extraction unit includes:
An inspection necessity determination unit that determines prioritization of inspection of the inspection target receipt using the receipt extraction rule managed by the receipt extraction rule management unit;
Comprising rule creation means for creating the receipt extraction rule,
The rule creation means includes:
A claim word search means for searching the claim word managed by the reexamination record data management means with a keyword managed by the receipt code keyword management means and a keyword managed by the word keyword management means;
A combination pattern conformity determination means for determining a match between the receipt code keyword, the appearance pattern of the wording keyword, and the combination pattern managed by the combination pattern management means;
And a receipt extraction rule generating means for generating a receipt extraction rule using the receipt code keyword in accordance with a receipt extraction rule creation method defined by the combination pattern.

  Furthermore, the receipt inspection support system according to the present invention provides a method in which the wording keyword management means converts the unit of the period to the month for the wording keyword when the wording keyword is a keyword indicating a numerical value and a unit of the period. And the receipt extraction rule generating means, when the combination pattern includes a keyword indicating the unit of the numerical value and the period, a period including the numerical value obtained by converting the unit into a month, It is characterized in that a receipt extraction rule is created.

  Furthermore, in the receipt inspection support system of the present invention, when the combination pattern determined by the combination pattern conformity determination unit includes the word keyword indicating the numerical value and the unit of the period and the receipt code keyword, the receipt extraction rule generation unit includes , The value of the quantity of medicines or medical treatments indicated by the receipt code keyword existing during the period of searching for the receipt retroactively is counted, and the counted value is the numerical value of the wording keyword in units of the period. A receipt inspection support system characterized in that exceeding the converted value is used as a retrieval condition for the receipt extraction rule.

  Further, in the receipt inspection support system according to the present invention, the reexamination result data management means is information indicating either “acceptance” in which the request is approved as a result of the reexamination or “as per the original trial” in which the request is not approved. When the result is accepted, the reduced acceptance score is managed, and when the combination pattern matching determination means confirms a match between the combination of the receipt code keyword and the wording keyword and the combination pattern, The number of acceptances and acceptance scores for each combination pattern and receipt code keyword are tabulated, and the acceptance percentage indicating the percentage of acceptance for the number of re-examination results for each combination pattern and receipt code keyword and the average acceptance score per acceptance. The rule index calculating means for calculating, and the receipt extraction rule generating means is configured to receive the receipt extraction rule. And the acceptance ratio calculated by the rule index calculation means and the average acceptance score are combined and stored in the receipt extraction rule management means, and the inspection necessity determination means performs the input receipt extraction when the input receipt is extracted. Using the value obtained by multiplying the number of cases searched by the receipt extraction rule from the above-mentioned acceptance ratio, and the value obtained by multiplying the number of receipts searched by the receipt extraction rule by the acceptance ratio and the average acceptance score, It is characterized by prioritizing receipt inspection.

  Furthermore, in the receipt inspection support system of the present invention, the receipt data management means manages the receipt data related to the reexamination result data and the receipt data that has not been requested for reexamination in the past, and the basis of the receipt extraction rule. A rule generation source receipt confirmation means for acquiring from the receipt data management means the receipt data related to the reexamination result data that has been approved and as a result of the reexamination, and the receipt data not requested for reexamination. The received receipt search means for obtaining the receipt data corresponding to the receipt extraction rule from the receipt data management means, the receipt data based on the receipt extraction rule, and the reacquisition obtained using the receipt extraction rule. From the receipt data that was not requested for examination, Additional rule analysis means for learning such that the acceptance ratio or the average acceptance score is higher than the receipt extraction rule, and a receipt extraction rule for performing correction to add the item obtained as a result of the learning to the receipt extraction rule Correction means.

  Furthermore, in the receipt inspection support system according to the present invention, the output unit includes a value obtained by multiplying the number of cases to be inspected by the receipt extraction rule and the receipt extraction rule by an acceptance ratio and an inspection by the receipt extraction rule. A receipt inspection support system which displays a value obtained by multiplying the number of target receipts by an average acceptable score and allows selection of a receipt inspection rule to be inspected.

  Furthermore, in the receipt inspection support system of the present invention, the output unit displays on the screen a receipt of a person having a receipt that conforms to the receipt extraction rule, and the receipt code included in the receipt extraction rule obtained by searching for the receipt. A receipt inspection support system that highlights a name of a wound or a medicine or a name of a medical practice indicated by a keyword.

  Furthermore, in the receipt inspection support system of the present invention, the receipt displayed on the screen is a plurality of receipts, and the output unit includes a time series display unit of the plurality of receipts, and the time series display unit A receipt inspection support system, wherein the receipt selected by the selection unit displayed on the display is displayed.

  Further, in the receipt inspection support system according to the present invention, the output unit includes an extraction result display unit that displays the extracted receipt result in a pie chart on the axis of the acceptance ratio and the average acceptance score. A receipt inspection support system, wherein the size of the circle of the pie chart is changed and displayed based on the number of applicable rules or the total number of acceptable points.

  Furthermore, in the receipt inspection support system of the present invention, the output unit further includes an inspection effect prediction display unit that displays the ratio of the extracted receptor to the inspection target receptor for each number of cases and / or points. An inspection support system, wherein a ratio in a circle selected from the circles of the pie chart of the extraction result display unit is displayed on the inspection effect prediction display unit.

  Further, in the receipt inspection support system according to the present invention, the output unit is configured such that the prioritization performed by the inspection necessity determination means is a priority based on an expected acceptance score or a corresponding rule number, and the expected acceptance score or An inspection support system that displays the results of receipt extraction in the order of corresponding rules.

  Furthermore, in the receipt inspection support system according to the present invention, the output unit includes a request word input unit for inputting a request word to the receipt extraction result.

  Furthermore, in the receipt inspection support system of the present invention, the output unit includes a template selection unit that selects a word to be input to the billing word input unit, and the billing word input unit charges the selected template. An inspection support system characterized by letting you input words.

  Furthermore, the receipt inspection support system of the present invention uses the reexamination record managed by the reexamination record data management unit to receive the receipt extracted by the receipt extraction rule for each of the receipt extraction rules managed by the receipt extraction rule management unit. An acceptance ratio indicating the ratio of the number of acceptances to the number of cases, an average acceptance score indicating the score per receipt extracted by the receipt extraction rule, and a rule in which the acceptance ratio and the average acceptance score do not reach a predetermined value are calculated. It has the receipt extraction rule evaluation means to delete.

In order to solve the above-mentioned problem, the rule creation method of the present invention includes the data consisting of a plurality of records having a plurality of items, the wording of the query sentence when the query of the record of the data is performed, and the query A rule creation method for creating a rule representing a relationship between a data item and a value of the query result by using a value of a query result in which a judgment of a query destination is recorded as a result of a query using a sentence. ,
Manage data keyword information that is a keyword that appears in both the data item and the query sentence, word keyword information that is a keyword that appears only in the sentence of the query sentence, and a combination of the data keyword and the word keyword in the query sentence Using a database having a combination pattern, the rule creation means
First, search for data keywords and wording keywords from the query,
Second, obtain the data keyword for each combination pattern from the search result of the first step and the combination pattern,
Third, in relation to the query statement corresponding to the combination pattern and data keyword, the query result obtains data having a value related to rule creation,
Fourth, from the data where the query result has a value different from the value related to rule creation, obtain the data of the search result using the data keyword obtained in the second step,
Fifth, the data obtained in the third step and the different items of the data obtained in the fourth step,
Sixth, the rule is created by combining the data keyword acquired in the third step and the value of the item acquired in the fifth step.

  The receipt inspection support method and the receipt inspection support system of the present invention manage a receipt code keyword, a wording keyword, and a combination pattern thereof, and create a receipt extraction rule from a request wording of reexamination results. Since the claim word points out the points that were noticed at the time of inspection, there is an effect that it is possible to create an easy-to-understand receipt extraction rule that is meaningful for the inspector who grasped the inspection points. Furthermore, because the invoices are the contents that the inspector inputs in daily receipt inspection work, regular rule creation processing can be performed to maintain the rules in response to the increase or update of the number of rules. This is effective in reducing the load of rule maintenance.

  In addition, the receipt inspection method and the receipt inspection support system of the present invention have an effect that can manage a general inspection by managing a keyword related to a period as a wording keyword and setting it as a retroactive period of the receipt extraction rule. is there. In addition, in the case of a combination pattern of word keywords related to periods and receipt code keywords, items to be checked as a general inspection by aggregating the quantity of medicines or medical treatments indicated by the retrospective receipt code keywords determined from the word keywords related to periods And the effect of determining the quantity.

  In addition, the receipt inspection support method and the receipt inspection support system of the present invention calculate an index such as an acceptance ratio and an average acceptance score when a reexamination request is made for a receipt extraction rule, and when extracting a receipt for priority inspection, Using these indices, the number of cases corresponding to the receipt inspection rule and the predicted acceptance score are calculated and displayed. By this method, there is an effect that it is possible to select and receive a priority by checking a receipt that has a high effect of suppressing the inspection efficiency and expenditure due to errors.

  Further, the receipt inspection support method and the receipt inspection support system of the present invention are free from errors searched by the receipt extraction rule and the receipt extraction rule for the receipt extraction rule created from the claim wording. By learning about the differences in the receipts and making corrections to add the feature items on the resulting receipt to the receipt extraction rule, it is possible to reflect the characteristics of the error receipt not listed in the claim text in the receipt extraction rule. is there.

  Furthermore, it is possible to increase the efficiency of the receipt inspection by learning the features of the receipt that increase the index such as the acceptance ratio and the average acceptance score and adding it to the receipt extraction rule. In particular, compared to simple machine learning of features such as keywords and quantities obtained from receipts, it is based on rules based on billing wordings, so it is possible to create meaningful rules. In addition, there is an effect that more efficient receipt inspection can be supported by learning and adding so that the index for the receipt extraction rule becomes higher.

  In addition, the receipt inspection support system of the present invention displays an indication such as the receipt extraction rule, and the corresponding number of cases extracted by the receipt extraction rule and the acceptance score on the output unit, so that the inspector has a high acceptance score, It is possible to select a highly efficient receptor, and to increase the efficiency and effectiveness of the receipt inspection.

  In addition, the receipt inspection support system of the present invention displays a receipt conforming to the receipt extraction rule on the output unit, and highlights the name of the wound, the medicine, and the medical treatment indicated by the receipt code keyword included in the receipt extraction rule on the receipt. By doing so, the inspector can easily know where to focus and can increase the efficiency of the inspection work.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The insurer determines whether the receipt should be reconsidered by checking the receipt of a newly received receipt by an inspection specialist. The receipt inspection support system of the present invention is a system that receives input of new receipt data and determines whether or not inspection by an inspector is better in order to improve the efficiency of receipt inspection.

  First, an explanation will be given of an example of a reception process and a hardware configuration of the receipt inspection support system according to the present invention. FIG. 16 is a diagram showing an example of the flow of the reception process and the hardware configuration of the reception inspection support system of the present invention. In the current flow of receipts, medical institutions such as hospitals 1601, clinics 1602, and dispensing pharmacies 1603 receive receipts that describe the contents of medical practices, medicines, etc. that are paid by the insurer. Remuneration statement) is sent to the examination and payment institution 1604 once a month (1621). The data to be sent may be sent via a medium such as a floppy disk, CD, or DVD, or may be sent via a network. The examination / payment institution 1604 examines the receipt for one month and, if it determines that there is no error, sends the receipt to the insurer 1605 (1622). The insurer inspects this receipt, and if it is determined that there is an error in the description of the receipt, the insurer makes a request for reexamination to the examination / payment institution (1623). The examination and payment institution performs a re-examination and notifies the insurer of the result (1624). The result of the notification is notified as “accepted” if the request for reexamination is judged to be in error, and “according to the original case” if there is no error and the first examination has been made. The insurer 1605 has a receipt inspection support system 1606. The insurer's receipt is extracted from the receipt 1622 sent from the examination / payment institution and it is better to file a reexamination request. Support the work of inspectors. The receipt inspection support system 1606 includes a receipt inspection support terminal 1611 and a database device 1612 which are computer devices. The receipt inspection support terminal 1611 has an output device such as a display screen 1613 and an input device such as a keyboard 1614 and a mouse 1615. In addition, the receipt inspection support terminal 1611 of the receipt inspection support system 1606 receives the receipt data 1622 sent from the examination / payment institution 1604. Therefore, when receiving the receipt data 1622, the receipt inspection support terminal 1611 is sent via a medium reader or network. If you want to capture data via the network interface. The staff who conducts the receipt inspection extracts the receipts to be inspected using the receipt inspection support terminal 1611 and conducts the inspection. If it is judged that the reexamination is necessary, prepare the receipt data and the data for requesting the reexamination. To the examination / payment institution 1604 (1623). Then, a re-examination is performed by the examination / payment institution 1604 and the result is received by the receipt inspection support terminal 1611 (1624). In this manner, the receipt inspection support system 1606 extracts from the receipt 1622 from the examination / payment institution 1604 a receipt that is highly likely to be accepted as a result of the reexamination request 1623, and the inspector 1605 conducts the receipt inspection. Support business.

  FIG. 1 is a diagram showing a configuration example of a receipt inspection support system according to the present invention. The receipt inspection support system includes a receipt inspection support terminal 130 and a database 104 that manages receipt data, re-examination results, and the like. The reception inspection support terminal 130 includes an input unit 101 for inputting the reception data to be inspected into the system, a reception extraction unit 102 for extracting a reception to be inspected from the reception data to be inspected received by the input unit 101, a reception As a determination result of the extraction unit 102, the receipt is composed of an output unit 103 that outputs a result of determining whether or not a priority inspection by an inspector should be performed. The database 104 includes a receipt code keyword management means 121 for managing keywords extracted from the code information of the receipt, a word keyword management means 122 for managing keywords appearing in the request wording at the time of reexamination, Re-examination result data management means 123 for managing results, combination pattern management means 124 for managing combinations of keywords managed by the receipt code keyword management means 121 and keywords managed by the wording keyword management means 122, and past receipt data is accumulated A receipt data management means 125 and a receipt extraction rule management means 126 for managing a rule for extracting a receipt are included. The receipt extraction unit 102 has an inspection necessity determination unit 105 that determines whether or not an inspection person needs to check the receipt received from the input unit 101, and a receipt that the inspection necessity determination unit 105 uses to determine whether or not the inspection is necessary. It comprises a rule creation means 106 for creating an extraction rule. The rule creation means 106 includes a claim text search means 107 for searching for keywords managed by the receipt code keyword management means 121 and the text keyword management means 122 for the claim text managed by the reexamination result data management means 123, A combination pattern matching determination unit 108 that determines matching with a combination pattern managed by the combination pattern management unit 124 with respect to a search result of the search unit 107, and a combination of keywords of a combination pattern corresponding to the combination pattern matching determination unit 108 Based on the rule index calculation means 109 for calculating an index such as the acceptance ratio and the average acceptance score, the keyword combination suitable for the combination pattern conformity determination means 108, and the index obtained by the rule index calculation means 109 Receipt extraction rule generation means 1 for creating a receipt extraction rule for use in the inspection necessity determination means 105 With ten.

  This system is configured by using a computer device, and each means is stored in a storage device on the computer, and is read out and executed by a computer memory and CPU as necessary. The input unit 101 is a means for receiving receipt data recorded on a medium such as a floppy disk or a CD-ROM, or receipt data sent via a network. The output unit is an output device such as a computer screen or a printer, and outputs the determination result by screen display or printing. The inspection necessity determination unit 105 and the rule creation unit 106 are operated by input from an input device such as a keyboard or a mouse (not shown).

  FIG. 2 is a diagram showing an example of receipt data received by the input unit 101, and shows an example of medical receipt data. There are various types of receipts such as medical, dental, dispensing, DPC, etc. In this embodiment, explanation will be made using two medical receipts and dispensing receipts. The receipt data is composed of a basic information table 201, a disease name table 202, and a summary column table 203. The basic information table 201 holds data such as a receipt ID 211, a medical treatment date 212, medical institution information 213 that issued the receipt, an insured person ID 214, and a claim score 215. The wound name table 202 holds a receipt ID 221, a wound name column 222, a medical treatment start date column 223, etc. that match the basic ID of the receipt ID 211 so that a plurality of wound names can be managed for one receipt. To do. The summary column table 203 also manages data such as a receipt ID 231, a summary column 232, and a quantity column 233 that match the receipt ID 211 of the basic information so that a plurality of data can be managed for one receipt.

  FIG. 3 is a diagram showing an example of receipt data received by the input unit 101, and shows dispensing receipt data. The receipt data includes a basic information table 301 and a dispensing content table 302. The basic information table 301 manages data such as a receipt ID 311, a pharmacy name 313 that issued a receipt, a medical institution 314 that issued a prescription, an insured person ID 315, and a claim score 316. In addition, the Dispensing Contents Table can manage multiple dispensing contents for one receipt. The Receipt ID 321 that matches the Receipt ID 311, the prescription date 322, the dispensing date 323, the drug name 324, and the unit It manages drug charges 325, quantity 326, etc.

  FIG. 4 is a diagram showing an example of reexamination request actual data managed by the reexamination actual data management means 123. Number field 401, Receipt ID 402 matching with the receipt, Request text 403, which is the text to be entered as the content of the re-examination upon request for re-examination, and “Acceptance” in which an error was recognized as requested as a result of the re-examination “1” The result field 404 indicates “0” as “original referee” as in the original original examination, and the result score 405 indicates the number of points reduced from the original receipt when “accepted”. . The wording in the billing wording 403 describes the text that points out / inquires about the part of the receipt that the insurer has determined to be in error. In this example, the claims No. 1, 3, and 11 have a restriction that a maximum of 20 days can be given for the drug “Iueo Tablets”, whereas the receipt 0001 specified by the receipt ID 402 It is a content that points out that it has been administered for the above days. In addition, in the claim language of No. 2 and 21, it is determined that the drug “Kakikukeko Tablet” or “Tatsutsutoto” can be administered for a certain wound, whereas in the case of the receipt of Recep ID 0002 or 0104, the name of the wound There is a statement that inquires about the fact that the drug is being administered even though there is no.

  FIG. 5 is a diagram showing an example of receipt code keyword data managed by the receipt code keyword management means 121. As shown in FIG. Data such as a code ID 501, a keyword 502, a code name 503, a code type 505, and a receipt type 504 indicating the type of receipt in which the receipt code appears are managed. Here, the receipt code is information published by the Ministry of Health, Labor and Welfare (http://202.214.127.149/) and is code information in the receipt computer processing. Masters such as injury and illness names, medicines, medical practices, dispensing practices, and specific equipment are published. The code ID 501 is a code in this master, and the code name 503 is a name for a code defined by the master. The keyword 502 is a part of the code name 503. For example, when the code name 503 is “Iueo Tablets 1 mg”, the keyword 502 is “1 mg” related to the unit amount or a word indicating a general medicine form. Except for the word “lock”, only “Iueo”, which is a part of the proper name, is used as a keyword. This is because when the inspector enters the request wording at the time of reexamination request, it may be written in an abbreviated form instead of an official name, and the form of tablets, capsules, liquids, etc. may be different even with the same medicine, etc. Therefore, a keyword 502 is created by extracting the unique name from the code name 503. Further, the code type 505 is the above-described master type (such as injury / disease name, medicine, medical practice, dispensing action, specific equipment, etc.).

  FIG. 6 is a diagram illustrating an example of data for managing keywords appearing in the billing wording managed by the wording keyword management unit 122. It has a keyword ID 601, a keyword 602 that appears in the billing wording, a search period 604, and a condition value 606. A search period 604 and a condition value 606 indicate a method of converting information related to a wording keyword period into a search condition when the information is used as a search condition for a receipt extraction rule. Since the receipt is issued on a monthly basis, in the search period 604, n is converted to the number of months. Since “n” in “n days” is the number of days, the number of months obtained by adding 1 divided by 30 (rounded down after the decimal point) is the period going back. For example, if it is 40 days, it goes back two months ago, and the search for three months can be performed together with this month, so that the interval of 40 days can be within the search period. Further, since the value of the quantity of the medicine is written in the number of days, the condition value 606 converts n into the number of days. For example, when the wording keyword is “n months”, it is calculated as a product of 30. This provides a method for managing the keywords appearing in the billing wording and converting them into search periods and condition values when the wording keyword is information relating to the period.

  FIG. 7 is a diagram illustrating an example of data for managing a combination of a receipt code keyword and a wording keyword managed by the combination pattern management unit 124. Pattern ID 701, pattern name 702, pattern 703 for specifying a combination of a receipt code keyword and a wording keyword appearing in the billing wording 403, a search type 704 for specifying the presence / absence of a keyword or a total as a search extraction type, and a general inspection Search period 705 that specifies the retroactive period, aggregation field 706 that specifies the items of the receipt to be aggregated in the case of the aggregation condition, condition field 707 that indicates the condition for extracting the receipt, the type of the receipt code keyword is the drug, the name of the disease, and the medical practice For example, a code keyword type 708 field is used. In the pattern 703, the receipt code keyword is indicated by “codeKW” in the drawing. And the pattern which appears in a claim wording in combination with ID of the wording keyword shown to 601 of FIG. 6 is shown. In the code keyword type 708 column, the code type indicated by the code type 505 in FIG. 5 is designated as the type of “codeKW”. Further, “L” in the search period indicates a value calculated from the numerical value “n” of the wording keyword in the search period 604 in FIG. 6, and “V” in the condition 707 is the condition value 606 in FIG. This indicates a value calculated from the quantity “n” of the wording keyword.

  For example, the pattern ID (701) 101 is a pattern in which a wording keyword 0001 and a receipt code keyword related to a medicine appear simultaneously in the claim wording pattern. Indicates that this is the case. As a method for creating the priority inspection receipt extraction rule in the case of this pattern, since the search type 704 is “keyword presence / absence” and the condition 707 is “Yes”, a rule for searching for a receipt including the keyword indicated by “codeKW” is used. Indicates to create. Further, since the search period 705 is “0”, it indicates that a rule for performing a search in the receipt of the month is created. The pattern ID (701) 102 is a pattern in which “n days” or “n weeks” or “n months”, “upper limit”, and a medicine receipt code keyword (“codeKW”) appear in the claim text. In this case, a method for creating a receipt extraction rule is shown. In this case, as the search period 705, the quantity column relating to the receipt code specified by “codeKW” is totaled by going back the receipt of the period calculated from “n” in the wording shown in the search period 604 of FIG. This indicates that a rule for extracting a receipt corresponding to the value of “V” or more obtained from “n” indicated by the condition value 606 is created. Similarly, 110 and 111 of the pattern ID (701) also manage the method of creating a receipt extraction rule from the receipt code keyword and the wording keyword.

  In this way, by managing the receipt extraction rule creation method for the keyword combination pattern in the billing statement, it is possible to create a receipt extraction rule from the reexamination results. In particular, it is possible to cope with a change in medical practice by managing separately the keyword of the word relating to the contents of the inspection to be re-examined and the receipt code keyword and managing the combination as a combination pattern.

  FIG. 8 is a diagram showing an example of a receipt extraction rule table managed by the receipt extraction rule management means 126. As shown in FIG. 7, a pattern ID 802 that matches the combination pattern in FIG. 7, an And connection field 803 that indicates the number of the And connection destination between rules, a condition type 804 that indicates whether a keyword is extracted or not, and a receipt type as a condition for receiving a receipt 805, a period column 806 indicating a search period, a keyword column 807 indicating a receipt code keyword of a search condition indicated by the extraction rule, an aggregate amount column 808 for specifying data to be aggregated, a condition column 809 indicating a search condition, a rule As an index, an acceptance ratio 810 indicating the ratio of acceptance to the request for reexamination and an average acceptance score 811 indicating the average reduction score per approval are managed.

  Type 804 is information from combination pattern search type 704, receipt type 805 is information from receipt code keyword receipt type 504, period 806 is information from combination pattern search period 705, and keyword 807 is combination pattern 703 “ The receipt code keyword corresponding to “codeKW”, the total amount 808 is information from the combination pattern total column 706, and the condition 809 is information from the combination pattern condition column 809. Further, the acceptance ratio 810 and the average acceptance score 811 are values calculated from re-examination results that match the combination pattern.

  For example, the No. 1 receipt extraction rule is a rule made from the pattern ID 101 of the combination pattern of FIG. 7 and includes “Kakikakeko” as a keyword from the receipt for one month of medical / dispensing (period 806 is 0). It shows that the receipt is extracted. In addition, the receipt extracted by this receipt extraction rule shows that 50% is accepted and 300 points are reduced per case when re-examination is requested from the past results.

In addition, the No. 11 receipt extraction rule is a rule created from the combination pattern ID 111 of FIG. 7, and the receipt of the keyword “Aiueo” has a total of 20 or more compared to the receipt dated one month before medical / dispensing. It shows that the receipt is extracted. Furthermore, the receipt extraction rule of No. 22 is used in conjunction with the rule of No. 21 at And connection 803, and it is a combination with the keywords “Hahifuheho” in the medical or dispensing receipt and “Prescription” in the medical reception. Indicates to search. In addition, No. 32 is ANDed with No. 31 to indicate that the XX treatment and XX instruction and management fee extract the receipts that appear in the medical receipt for two months that date back one month.

  Here, the receipt extraction rule is a search for a receipt related to a series of medical treatments. Receipts for a series of medical treatments refer to receipts issued by the same medical institution and dispensing pharmacies that dispense prescriptions issued by the medical institution for the same insured person Is a series of medical care. The receipt extraction rule performs a search based on a receipt to be inspected. Basically, receipts are issued once a month by medical institutions, so insurers check monthly. The receipt extraction rule can easily specify a search condition related to a series of medical treatments based on the receipt of the month subject to inspection.

  FIG. 9 is a diagram showing an example of a medical receipt of past receipt data managed by the receipt data management means 125. Similar to the input data of FIG. 2, it is composed of a basic information table 901, a sickness name table 902 and a summary column table 903. The basic information table 901 holds data such as a receipt ID 911, a medical treatment date 912, medical institution information 913, an insured person ID 914, a claim score 915, and a decision score 916. The request score 915 is a score requested from a medical institution, and the decision score 916 is a score paid as a result of examination and inspection. When the claim score 915 and the decision score 916 are the same, there is no error in the receipt, and when the decision score 916 is lower than the claim score 915, it indicates that the receipt is pointed out and the score is reduced. The wound name table 902 holds a receipt ID 921, a wound name column 922, a medical treatment start date field 923, etc. that match the basic ID of the receipt ID 911 so that a plurality of wound names can be managed for one receipt. To do. The summary column table 903 also manages data such as a receipt ID 931, a summary column 932, and a quantity column 933 that match the basic information receipt ID 911 so that a plurality of data can be managed for one receipt.

  FIG. 10 is a diagram showing an example of a dispensing receipt of past receipt data managed by the receipt data management means 125. As shown in FIG. The receipt data includes a basic information table 1001 and a dispensing content table 1002 in the same manner as the input data in FIG. The basic information table 1001 manages data such as a receipt ID 1011, a pharmacy name 1013, a medical institution 1014, an insured person ID 1015, a claim score 1016, and a decision score 1017. The requested score 1016 is the number requested from the medical institution, and the decision score 1017 is the score paid as a result of examination and inspection. When the number of requested points 1016 and the number of determined points 1016 are the same, there is no error in the receipt, and when the number of determined points 1017 is lower than the number of requested points 1016, it indicates that the receipt has been pointed out and the score has been reduced. In addition, the dispensing content table 1002 can manage a plurality of dispensing contents for one receipt, a receipt ID 1021 that matches the receipt ID 1011, a prescription date 1022, a dispensing date 1023, a drug name 1024, It manages unit drug fee 1025, quantity 1026, etc.

  FIG. 11 is a diagram showing an example of a receipt extraction rule list screen displayed on the output unit 103 by the inspection necessity determination unit 105. On the screen, as details of the rules, pattern display column 1101, keyword display column 1102, receipt type display column 1103, retroactive month display column 1104, condition display column 1105, acceptance ratio display column 1106, average acceptable score display column 1107 Is displayed. Further, an extraction button 1111 for extracting a receipt to be inspected using this rule is displayed. By displaying the name of the pattern in the pattern display field, the inspector can understand what type of error the following rule is to extract. Also, by referring to the acceptance ratio 1106, the average acceptance score 1107, and the like, it is possible to determine which rule is effective as a receipt extraction rule.

  FIG. 12 is a diagram showing an example of a priority inspection receipt extraction result list screen displayed on the output unit 103 by the inspection necessity determination unit 105. Display the details of the rule and the receipt information extracted by that rule. As in FIG. 11, the pattern display field 1201, the keyword display field 1202, the receipt type display field 1203, the retroactive month display field 1204, and the condition field 1205 are displayed, and the acceptance obtained from the number of rules corresponding to this rule and the acceptance ratio. An expected hit number 1206 indicating the number of received receipts, an expected hit number 1207 obtained from the expected hit number 1206, and the average acceptable score are displayed. Further, an inspection receipt selection field 1211 and an inspection start button 1212 are displayed. By this screen, the inspector can confirm the number of cases for the receipt extraction rule and the number of accepted points as the result of the receipt extraction, and can inspect from the rule with efficient inspection.

  FIG. 13 is a diagram illustrating an example of a receipt inspection screen displayed on the output unit 103 by the inspection necessity determination unit 105. Highlights 1303 and 1304 are displayed for the keywords of the rule from which the receipt is extracted in the two fields 1301 and 1302 for displaying the receipt. In the time-series receipt display field 1305 for the same person, two displays 1306 and 1307 currently displayed and a display 1308 indicating a range corresponding to the receipt extraction rule are displayed. In addition, although the case where there are two columns is shown here, the number is not limited to two, and one column may be displayed or a plurality of columns may be displayed according to the case. In addition, a field 1309 for inputting a request text for a reexamination request, a reexamination request button 1310 for requesting a reexamination, and a payment for inputting that there is no error as a result of the inspection and no reexamination request is made. Button 1311 is displayed. The receipt time series display 1305 shows a single circle (one month's receipt), the horizontal axis is the month, the vertical axis is the medical institution, and the presence or absence of a person who falls under the receipt corresponding to the receipt extraction rule. Displayed in time series. Here, it is displayed that there is a dispensing receipt of the A pharmacy corresponding to the rule and a medical receipt of the hospital B that issued a prescription for the dispensing. In addition, the receipts displayed in the receipt display fields 1301 and 1302 are displayed in different colors such as 1306 and 1307. By this screen, it is possible to easily inspect a plurality of receipts such as a general inspection and a dispensing check. Further, since the part corresponding to the keyword of the search condition is highlighted on the receipt, the inspector can easily find the part to be noted. In addition, the receipt time series display 1305 can refer to the receipts of the persons corresponding to the rules in time series, and the inspector can list the presence of the receipts to be checked.

  Next, an example of a specific processing flow will be described with reference to a flowchart and a sequence diagram regarding the configuration, data, and screen of the receipt inspection support system described so far. First, a process for creating a receipt extraction rule will be described, and then a flow of receipt inspection support using the created receipt extraction rule will be described.

  FIG. 14 is a flowchart illustrating an example of a flow of receipt extraction rule creation processing performed by the rule creation means 106. FIG. 17 is a diagram showing a sequence between the receipt inspection support terminal and the database in the receipt extraction rule creation process. Here, past receipts are accumulated in the receipt data management means, past reexamination results are managed by the reexamination result data management means, and receipt code keyword data (Fig. 5), wording keyword data (Fig. 6) The combination pattern (Fig. 7) shall be prepared in advance. First, the claim text search means 107 of the rule creation means 106 is configured to execute the reexamination result data management means 123 of the database 104 from the reexamination result data shown in FIG. 4, and the receipt code keyword management means 121 to the receipt code keyword data and text shown in FIG. The wording keyword data shown in FIG. 6 is acquired from the keyword management means 122 (1401). As a sequence, the re-examination result data is requested from the receipt inspection support terminal to the database (1701), and the database returns the re-examination result data (1702). Then, a receipt code keyword and a wording keyword are requested to the database from the receipt inspection support terminal (1703), and the database returns data of the receipt code keyword and the wording keyword (1704).

  Next, the claim word search means 107 searches the claim word 403 of the reexamination record data for the presence of the keyword 502 of the receipt code keyword and the keyword 602 of the claim word (1402). When you search, you can combine the keywords that appear for each re-examination record data. For example, for the re-examination result with the number (401) of 1, the keywords “Iueo”, “upper limit”, “20 days” are applicable, and data indicating that the code ID 121212, the keyword ID 0010, and the keyword ID 0004 have appeared. .

  Next, the combination pattern match determination unit 108 acquires the combination pattern data shown in FIG. 7 from the combination pattern management unit 124 of the database 104 (1403). In the sequence, the receipt inspection support terminal requests a combination pattern from the database (1705), and the database returns combination pattern data (1706). Then, a search is made as to which of the combination patterns the combination of keywords in the above-mentioned billing statement matches (1404). For example, in the reexamination result with the number (401) of 1, since the code ID 121212, the keyword ID 0010, and the keyword ID 0004 appear simultaneously as described above, it corresponds to the combination pattern ID (701) 102, and the drug use period has been exceeded. It is classified as a claim text that conforms to the pattern. This is performed for all reexamination record data.

  Next, the rule index calculation means 109 calculates an index from the result (404) and result score (405) of the reexamination result data (FIG. 4) for the keyword combination that matches the combination pattern (1405). For example, in the reexamination result data, numbers (401) 1, 3, and 11 indicate the same error and are the same keyword combination. Since 2 of the 3 cases have been approved as a result of the request for reexamination, the acceptance ratio indicator is 66%. In addition, two cases that were accepted with the same keyword combination are the result points of 500 points and 300 points, so 400 points obtained by dividing 500 + 300 by 2 cases are used as an index of the average acceptable point. Similarly, other re-examination results are also counted.

  Next, the receipt extraction rule generation means 110 creates a receipt extraction rule from the keyword combination that matches the combination pattern (1406). Here, a rule with the number (801) in FIG. Since the combination pattern ID 802 is the pattern ID (701) is 102, the code keyword is “Iueo”, and the wording keyword is “upper limit” and “n days”, the type 804 is searched for “total” from 704 The receipt type 805 is medical or dispensing than the receipt type 504, the period 806 is specified to go back one month before 20/30 + 1 from the search period 604, the keyword 807 is “Aiueo”, the total amount 808 is the total column 706 “Quantity”, Condition 809 is “≧ 20” from Conditions 606,707, Acceptance Ratio 810 is 66% from the aggregation result of Step 1406, Average Acceptance Score 811 is also 400 points from the aggregation result of Step 1406, and the 11th rule is created To do. Similarly, create other rules.

  Then, the receipt extraction rule generation unit 110 stores the created rules in the receipt extraction rule management unit 126 of the database 104 in the format shown in FIG. 8 (1407). In the sequence, the created receipt extraction rule is sent to request storage (1707), and when the database completes storing the receipt extraction rule, a completion notice is sent to the receipt inspection support terminal (1708). The receipt extraction rule creation process is thus completed.

  Next, a description will be given of the flow of processing for extracting a receipt that requires priority inspection. FIG. 15 is a diagram illustrating an example of a flow of processing when a reception to be inspected with priority is extracted from a reception performed by the inspection necessity determination unit 105 and an inspector inspects the reception. FIG. 18 is a diagram showing a sequence between the receipt inspection support terminal and the database at this time.

  When the receipt inspection necessity determination is started, first, the inspection necessity determination means 105 acquires the receipt extraction rule of FIG. 8 from the receipt extraction rule management means 126 of the database 104 (1501). In the sequence, the receipt inspection support terminal requests a receipt extraction rule from the database (1801), and the database returns the receipt extraction rule (1802). Here, the receipt extraction rule is displayed on the screen, and the inspector confirms the rule content. The inspection necessity determination unit 105 displays, on the output unit 103, a keyword and a condition for the keyword as shown in FIG. 11, and a rule index such as an acceptance ratio and an acceptance score. When the inspector presses the receipt extraction button 1111, the process proceeds to the next process.

  When the receipt extraction button 1111 is pressed, first, the receipt data of FIG. 2 and FIG. 3 is acquired from the input unit (1502). Here, one month's receipts are collectively sent from the examination and payment institution, and they are judged collectively. Here, the receipt inspection support terminal sends the receipt data acquired from the input unit in response to the receipt data storage request to the database (1803) and stores it in the receipt data management means 125 of the database. When the database completes storing the new receipt, the database sends the notification to the receipt inspection support terminal (1804). Next, the inspection necessity determination means 105 searches for a receipt from the new receipt by the receipt extraction rule (1503). For example, in the case of Rule No. (801) No. 1, “Kyokkeko Tablets” such as Receipt Receipt ID 2002 of the dispensing receipt are administered as a keyword containing “Kakyokekoko” as a keyword from among medical or dispensing receipts. Get a receipt. In the case of rule No. (801) 11, since the receipts of dispensing / medical departments containing “Aiweo” in the keyword are counted retroactively for one month, the receipts containing “Aiweo” are searched from the receipts of FIG. 2 and FIG. , Get the receipt with receipt ID (311, 321) 2001. Next, since the insured ID (315) of the receipt ID 2001 is 1111111 and the medical treatment date is May 2006, the same insured ID 1111111 is obtained from the receipt data of FIG. 9 and FIG. 10 managed by the receipt data management means 125. Search for a receipt containing the keyword “Aiueo” in a receipt in April 2006, one month before, and obtain a receipt with a receipt ID (1011) 2222. Then, from the receipts of the receipt IDs 2001 and 2222, the dispensing quantity relating to the Iueo tablet is totaled. Since the administration of Aiweo tablets is 5 and 10 in April 2006 and 10 in May, the total amount is 25, and the condition “≧ 20” of the receipt extraction rule No. 11 is met.

  Next, the inspection necessity determination means 105 calculates the expected number of cases and the expected acceptance ratio from the retrieved number of receipts and the rule index (1504). For example, in the case of rule No (801) No. 1, if there are 50 receipts that match the rule, the acceptance rate is 0.5, so the expected number of matches is 25, and the average acceptance score is 300, so 25 x 300 is 7500 points. . Step 1503 and step 1504 are executed for all rules (1505). When the execution of all the rules is completed, the expected number of cases and the acceptance ratio are displayed on the output unit (1506). Here, the screen of FIG. 12 is displayed on the output unit. In the screen of Fig. 12, the number of receipts that match each receipt extraction rule, the expected number of hits 1206, and the expected acceptance score 1207 are displayed. Choose. Then, for example, a rule relating to the keyword “Aiweo” is selected in the selection field 1211 and the check button 1212 is pressed (1507). When the inspection button 1212 is pressed, the inspection necessity determination means 105 displays an inspection screen as shown in FIG. 13 on the output unit 103, and displays one of the receipts that conforms to the selected receipt extraction rule on the screen ( 1508). The displayed receipt indicates a dispensing receipt that complies with the April and May 2006 rules of the insured ID 111111. In addition, the portion related to the Iueo lock that was the rule condition is highlighted (1303, 1304) to alert the inspector. The time-series receipt display field 1305 indicates that there is a receipt of the same person for one month ago and that of the hospital B. Select this to check whether a reexamination request is made with reference to the medical receipt that issued the prescription. When requesting a reexamination, the inspector inputs a request text in the request text input field 1309 and presses a reexamination request button 1310. For example, enter “Is the Aueo Tablet not limited to 20 days”? When the reexamination request button 1310 is pressed (1509), the inspection necessity determination means 105 creates reexamination request data from the request text (1510). Here, a form to be sent to the examination / payment institution is created, and a new record is added to the re-examination record data management means 123 to record the request wording. As a sequence, a request text storage request is sent from the receipt check support terminal to the database (1805), stored in the database, and a request text storage notification is returned (1806). On the other hand, when the inspector determines that it is not an error especially by looking at the medical receipt or the like, the inspector presses the payment button 1311. If the payment button is pressed (1509), the process proceeds to the next process without doing anything. The inspection necessity determination means 105 determines the end in the next process (1511), and if there is an uninspected receipt in the selected rule, the process returns to step 1506 to inspect the next receipt. As a result of the end determination, when all the receipts are checked for the selected receipt extraction rule (1511), the process returns to step 1506 to display the screen of FIG. 12 and check the execution of the check for another receipt extraction rule. Let the staff input. In addition, when the process is completed, for example, when all the rules have been checked, the process is completed.

  As described above, the receipt inspection support system according to the present invention uses the rule creation means to create the receipt extraction rule using the result data of the reexamination, so there is an error related to an important inspection point from the past results. There is an effect that a receptor can be extracted. In particular, because the claim word search means determines the keyword to be searched from the claim word that expresses the content of the incorrect part of the receipt and inquires about the correct answer, the automatically created rule creates a meaningful rule for the inspector There is an effect that can. Further, as described above, since the receipt extraction rule is automatically created based on the billing word input in the daily inspection work, it is not necessary to perform maintenance on the rule. By regularly updating the rules, such as once a month or once a year, it is possible to respond to changes in the examination trend. In addition, when a new medicine comes out, it is possible to deal with a new medicine just by creating a keyword from the code and name of the medicine and registering it in the receipt code keyword management means.

  Furthermore, the keyword management means 122 manages keywords related to the period, and further manages a method (604, 605) for converting it to a retroactive period, so that the retroactive period of the receipt search is determined from the invoiced wording corresponding to the general inspection. There is an effect that can be done.

  Further, the receipt inspection support system of the present invention manages keywords separately in the receipt code keyword management means 121 and the wording keyword management means 122, and further creates rules from the combination patterns managed by the combination pattern management means 124. The keywords appearing on the receipt are separated from the keywords appearing only in the wording, and the search condition can be created from the keywords appearing on the receipt. In addition, by managing the combination pattern and clearly indicating the type of error indicated by the combination pattern, the inspector can know what type of error may be in the receipt that conforms to the rule. effective.

  Furthermore, in the receipt inspection support system of the present invention, the rule index calculation means 109 calculates an index such as an acceptance ratio and an average acceptance score when a reexamination request is made for a receipt extraction rule, and an inspection necessity determination means 105 Using these indicators, the number of cases and predicted acceptance points corresponding to the receipt inspection rule are calculated and displayed. As a result, there is an effect that the inspector can select and inspect a receipt having an effect of suppressing expenditure for errors and a high efficiency.

  Also, in the receipt inspection support system of the present invention, the inspection necessity determination means 105 displays a plurality of human receipts that conform to the receipt extraction rule as shown in FIG. Receipt inspection can be performed easily. Furthermore, since the keyword of the receipt extraction rule is highlighted and displayed (1303, 1304), the inspector can immediately find a keyword to be noticed, and the inspection can be performed efficiently. Furthermore, in the receipt time series display column, the same person's receipts are displayed for each medical institution in chronological order, and the reception to be displayed can be changed by this selection. The receipt can be easily compared and checked. Furthermore, by highlighting 1308 a receipt that conforms to the receipt extraction rule in the receipt time series display field, it is possible to easily find and inspect a receipt to be inspected from among a plurality of receipts.

  Next, an example will be described in which the rule creation means 106 uses past receipt data to improve the receipt extraction rule. FIG. 19 is a diagram illustrating an example of a configuration of a receipt inspection support system in the case where improvement is performed on a receipt extraction rule using past receipt data. In contrast to the rule creation means 106 of FIG. 1, in FIG. 19, after the receipt extraction rule generation means, the receipt of the re-examination results that is the basis of the rule created by the receipt extraction rule generation means is obtained to determine whether the rule is appropriate. Rule generation source receipt confirmation means 1901 to be determined, paid receipt search means 1902 for searching for past paid receipts using the generated receipt extraction rule, a receipt obtained and searched by the paid receipt search means 1901, and rule generation Additional rule analysis means 1903 for analyzing the difference between the receipts acquired by the original receipt acquisition means 1902 and additional rules created by the additional rule analysis means 1903 are added to the rules created by the receipt extraction rule generation means 110 to manage the receipt extraction rules. The receipt extraction rule correction means 1904 stored in the means 126 is added. With this configuration, it is possible to modify the rule using the receipt data for the receipt extraction rule created from the claim wording.

  Next, an example of the flow of processing when making improvements to a receipt extraction rule using past receipt data will be described using a flowchart and a sequence diagram. FIG. 20 is a diagram showing an example of the flow of processing when making improvements to the receipt extraction rule using past receipt data. FIG. 21 is a diagram showing an example of a sequence between the receipt inspection support system and the database at this time.

  The flowchart of FIG. 20 performs the same processing from step 1401 to step 1406 as the flowchart of FIG. Re-examination record data, receipt code keyword, and wording keyword are obtained (1401), the receipt code keyword and wording keyword are searched from the re-examination history data request wording (1402), and the combination pattern is obtained (1403). Search for the occurrence of the receipt code keyword and the wording keyword and the matching of the combination pattern (1404), and for the keyword combination that matches the combination pattern, the index is calculated from the acceptance score as a result of the re-examination performance data (1405). ), A receipt extraction rule is created from a combination of keywords matching the combination pattern (1406). The sequence up to here is also performed from 1701 to 1706 in the same manner as in FIG. At this point, it is assumed that the receipt extraction rule as shown in FIG.

  Subsequently, in this embodiment, the rule is corrected by using a past receipt for the created receipt extraction rule. First, the rule generation source receipt confirmation means 1901 refers to the reexamination record data used for creating the receipt extraction rule for one of the created receipt extraction rules, and obtains a receipt corresponding to the receipt ID (2001). . The case where the rule No. 801 in FIG. 8 is No. 2 will be described as an example. The rule generator receipt confirmation means 1901 is the reexamination result as shown in FIG. 4 managed by the reexamination result data management means 123. From No. 401, data with No (401) No. 21 is acquired as reexamination record data based on the second rule. Since the 21st receipt ID (402) of the reexamination record is “0104”, the receipt data with the receipt ID (911) “0104” is acquired from the receipt data of FIG. 9 managed by the receipt data management means 125. Since the receipt data includes a basic information table 901, a disease name table 902, and a summary column table 903, data is acquired from each. Also, since a prescription is issued in the summary column 932 of the receipt with the receipt ID “0104”, a dispensing receipt relating to this is also acquired. In the dispensing receipt shown in FIG. 10, the medical institution 1014 is “B hospital”, the insured ID 1015 is “98776543”, and the medical treatment date is “2006/03” and the reception ID (1011, 1021) is “2224”. To get. In the sequence diagram, the database returns the corresponding receipt data (2102) in response to the receipt data request (2101) that is the basis of the rule from the receipt inspection support terminal.

  Next, the rule generation source receipt confirmation unit 1901 confirms whether or not a search can be performed using the receipt extraction rule created from the acquired receipt and the reexamination record (2002). Here, because the rule with No (801) No. 2 is the keyword “Tatsutetsu”, it is possible to search for a receipt with a Receipt Receipt ID (1011) of “2224”, and issued a prescription from there. Since the receipt with the medical receipt ID (911) “0104” can be acquired, it is determined that the rule is satisfied. At this time, if the rule that was created from the claim text cannot find the receipt that was the basis for the reexamination request, the rule generator receipt confirmation means 1901 indicates that the rule made from the claim text was an inappropriate rule. Determine, delete the rule (2003), and return to the processing of the next rule. In this example, since the receipt that is the basis for the reexamination request can be extracted from the rule, the process proceeds to the next process.

  Next, the paid receipt search means 1902 searches and acquires past receipt data by the receipt extraction rule (2004). Here, the paid-receipt search means 1902 uses the rule No. (801) of No. 2 to search the medical receipt of FIG. 9 and the dispensing receipt of FIG. 10 from the receipt data management means 125. In this case, the dispensing information table includes “tactetsu”, and both the dispensing receipt and the medical receipt have no difference in the number of claims 1016 and 1017, and the 2223 of the dispensing receipt and the 0103 of the medical receipt are acquired. To do. Since the claim point 915 is a claim from a medical institution and the decision point 916 is the amount paid by the insurer, a receipt without this difference is considered to be a receipt with no error. In the sequence diagram, the receipt inspection support terminal requests a receipt corresponding to the receipt extraction rule (2103), and the database returns the receipt data corresponding to the receipt extraction rule (2104).

  Next, the additional rule analysis means 1903 learns the difference between the receipt used to create the receipt extraction rule and the retrieved receipt with no error (2005). Here, there is an error in the combination of the medical receipt ID 0104 and the dispensing receipt ID 2224, and the combination of the medical receipt ID 0103 and the dispensing receipt ID 2223 has no error. Looking at the difference between these receipts, the dispensing receipt is basically the same. Name 922 has diabetes only. From this, it can be seen that an erroneous receptor does not have high blood pressure in the name of the wound.

Next, the receipt extraction rule correction means 1904 adds the learned result to the receipt extraction rule (2006). Here, the result that there is no hypertension in the wound name is added to the rule. An example of a rule corrected by the receipt extraction rule correction means 1904 is shown in FIG. In the rule added by No. (801) No. 3 in FIG. 22, a rule for searching for a receipt that does not have the keyword “high blood pressure” is added. The rule No. 3 indicates that “2” is written in the And connection (803), and the rule “No. hypertension” is indicated by No. 2 and “No.
Further, the receipt extraction rule correction means 1904 searches the receipt data from the receipt data management means using the learned rule, and corrects the acceptance ratio based on the number of results (2007). The acceptance ratio so far has been the ratio of the number of acceptances to the number of reexamination requests, but here it is changed to the ratio of the number of acceptances to the rules that can be searched by the receipt extraction rule. Thereby, the efficiency of inspection when this rule is applied to a new receipt can be obtained. In the sequence diagram, the receipt inspection support terminal searches for the receipt data with the corrected receipt extraction rule (2105), and returns the searched receipt data (2106). Then, the receipt extraction rule correction means 1904 stores the corrected receipt extraction rule in the receipt extraction rule management means 126 (2008). In the sequence diagram, the receipt extraction rule storage request (2107) is stored in the database receipt extraction rule management means 126, and the completion is notified (2108). The processing from step 2001 to step 2009 is repeated in the same manner until all the receipt extraction rules are completed (2009), and the processing is terminated when the confirmation / correction of all the receipt extraction rules is completed. In this embodiment, a receipt extraction rule is created by the above procedure.

  Using the receipt extraction rule created in the above procedure, the inspection necessity determination unit 105 extracts a receipt that has a high possibility of an error from a new receipt. The processing procedure of the inspection necessity determination unit 105 is performed in the flowchart of FIG. 15 as in the above-described embodiment. It is possible to improve the inspection efficiency by learning from the past receipts a rule that cannot be understood only from the billing wording and correcting the receipt extraction rule. In the above example, if a receipt having the same contents as the receipt IDs (1001) 2223 and 2224 in FIG. 10 is input to the inspection necessity determination means 105 as a new receipt, only the rule No. 801 before correction is the second rule. In this case, both erroneous receipt IDs 2223 and 2224 are searched, and after looking at both, it is necessary to find that there is an error. After the correction, No (801) in Fig. 22 is a rule that combines No. 2 and No. 3, so only Receipt ID 2224 is subject to inspection, reducing the number of receipts that should be seen by the inspector, and there is an error. The efficiency of extracting the receptor can be increased.

  In addition, in the above-described example, the additional rule analysis unit 1903 has been described as an example in which the difference is the correction content from the difference between the two receipts, but when analyzing from the actual receipt, There will be many errors-free receipts. Therefore, a machine learning technique such as analysis of a decision tree or correlation rule is applied here, and an item that best separates a receipt having an error from a receipt having no error is learned as the content of the additional rule. In the case of learning, learning is performed such that the acceptance ratio is higher and the acceptance score is higher than an index such as the acceptance ratio and the acceptance score of the original receipt extraction rule. As explained in the conventional method, if the analysis of decision trees and correlation rules is simply applied to the receipt data, it is not possible to determine the period to go back to deal with the general inspection. In addition, it is necessary to perform learning after making all the keywords appearing on the reception a learning item and collecting all the parameters. For this reason, prior processing is difficult, and a rule that is not related to an actual inspection item may appear in the output result. On the other hand, when creating only from the billing wording, the billing wording is the content for pointing out / inquiring about the content of the error on the receipt, so a rule indicating the point of inspection can be created. However, depending on the content of the claim wording, there may be a case where a rule for specifying an error receipt such as a name of a sickness cannot be made. In the present invention, the claim word search means 107 extracts a keyword from the request word of the reexamination request, the receipt extraction rule generation means 110 creates a receipt extraction rule from the keyword and the combination pattern, and payment is made using this receipt extraction rule. The receipt search means 1902 obtains a receipt that matches the rule created from the claim wording, and learns the erroneous receipt that is the basis of the rule obtained by the rule generation source receipt confirmation means 1901 and the additional rule analysis means 1903, The learned contents are reflected in the receipt extraction rule by the receipt extraction rule correction means 1904. Compared to applying machine learning as it is, this method can create a rule that accurately reflects the inspection items based on the invoiced words, and specifies the items to be learned, the retroactive period of learning, items to be aggregated, etc. There is an effect that can be done. In addition to creating a rule from only the billing wording, learning from the receipt and adding the result to the rule reflects the difference between a receipt with an error that cannot be understood only by the billing wording and a receipt without an error. There is an effect that can be done.

  Further, in the embodiments described so far, the description has been made using the example of the receipt and the reexamination record of the receipt. However, the present method can be applied to other than the reexamination of the receipt. In the case where there is business data, query text related to the data, and information on the judgment result at the query query destination based on the text, the data keyword appearing in the data and query text By managing the keywords that appear only in the wording of the query and the combination pattern of the data keyword and the wording keyword, rules can be created in the same way. For example, a rule can be created in the same way using slip data between two departments and information on a query related to a slip.

  In the embodiment described so far, the receipt data (FIG. 9, FIG. 10) managed by the receipt data management means 124, the wound name / sickness 222, 922, summary fields 232, 932, new receipt (FIG. 2, FIG. 3) The drug names 324 and 1024 have been described by taking the case where the names are written in text as examples, but the names may be expressed in codes. Even in such a case, the receipt inspection support system of the present invention manages the relationship between the code ID 501 written in the receipt and the keyword 502 written in the claim text in the receipt code keyword management means 121 as shown in FIG. Therefore, when the inspection necessity determination unit 105 searches for a receipt using the receipt extraction rule, the receipt can be searched by converting the keyword 807 into the code ID 501 and searching.

  In addition, in the case of drugs, the drug name may be the same but the quantity and dosage form may be different. In such a case, the receipt code keyword table of FIG. 5 managed by the receipt code keyword management means 121 may be divided into a code correspondence table 2601 and a keyword table 2602 as shown in FIG. The keyword ID field 2603 associates the keyword 502 with the code 501. For example, “Iueo Tablets” have two types of medicines with different amounts, and “Kakikukeko” may have different shapes from tablets, capsules, and injections. Even in that case, there may be cases where the request text is written only with keywords without distinguishing such differences. In this way, by managing the correspondence between keywords and codes, it is possible to deal with receipt code keywords having the same keyword.

  Furthermore, in the receipt inspection support system of the present invention, the receipt code keyword management means 121 designates the type of receipt to be searched according to the type of code as in the type of receipt 504 in FIG. Thereby, in the case of searching for a medicine code in the receipt extraction rule as shown in FIG. 8, both the medical receipt and the dispensing receipt can be searched. Further, if the keyword included in the rule includes a drug and the dispensing receipt conforms to the rule, the inspection necessity determination means 105 determines the medical department concerned from the medical treatment date 212 of the dispensing receipt, the medical institution 213, and the insured person ID 214. By searching for the receipt, it is possible to cope with the dispensing check. In addition, the paid receipt search means 1902 obtains the related medical receipt from the medical treatment date 1012 of the dispensing receipt, the medical institution 1014, and the insured person ID 1015 when the receipt retrieved by the receipt extraction rule is a dispensing receipt. Since the analysis is performed by the additional rule analysis means, it is possible to perform learning using both items of the dispensing receipt and the prescription issuance receipt corresponding to the dispensing check.

In the embodiments described so far, the inspection necessity determination means 105 compares and displays the indicators according to the rules as shown in FIG. 12, and selects the rule from the expected number of hits and the expected number of acceptable points to determine the receipt to be inspected. Although the method has been described, a receipt to be checked by another method may be selected. For example, the rules may be arranged in descending order according to the expected acceptable score so that the inspection can be performed in descending order of the score. As another method, when multiple rules are applicable to a single receipt, the multiple expected acceptance scores are summed, and the list of receipts is displayed in descending order of the expected acceptance scores. You may make it do. By these methods, it is possible to inspect from the one having a higher acceptable score, and the efficiency of the inspection can be improved.
Next, another method for displaying the result extracted by the receipt extraction rule will be described. FIG. 23 is a diagram illustrating an example of a screen of the priority inspection receipt extraction result displayed by the inspection necessity determination unit 105. When a new receipt for one month is input from the input unit 101, the inspection necessity determination unit 105 extracts a receipt to be inspected using the receipt extraction rule acquired from the receipt extraction rule management unit 126. A screen like FIG. 23 is displayed. In the screen of FIG. 23, an extraction result graph 2301 and an inspection effect prediction graph 2302 are displayed.

  The extraction result graph 2301 shows the acceptance ratio of the receipt extraction rule on the vertical axis, the average acceptance score of the receipt extraction rule on the horizontal axis, and the circles (2314 to 2318) on the graph show the receipts extracted from one receipt extraction rule. Show. The size of the circle (2314 to 2318) on the graph is selected from the rule hit count 2311 and the total allowable score 2312. Here, since the number 2311 of rule applicable cases is selected, the number of receipts corresponding to the receipt inspection rule among the receipts in the inspection target month is shown. The inspector can select the receipt of the extraction result of the receipt extraction rule that is present in the upper right on the graph having a high acceptance ratio and high average acceptance score and has a large number of cases and a large circle as a receipt with high inspection efficiency. In addition, if you select 2312 for the total number of acceptable points, the average number of acceptable points (the number of acceptable points per acceptable receipt) and the value of the total number of acceptable points calculated from the product of the number of acceptable receipts and the percentage of acceptance are displayed as the size of the circle To do. The inspector can select the receipt of the extraction result of the receipt extraction rule that is present in the upper right on the graph with a high acceptance ratio and high average acceptance score and has a large acceptance score and a large circle as a receipt with high inspection efficiency. Further, the acceptance ratio display 2313 displays circles 2314 to 2318 on the graph as a pie chart, and displays the acceptance ratio. By making a pie chart, the inspector can select a circle with a large area as an acceptance ratio as a receptor for inspecting, and select as a receptor with high inspection efficiency.

  Further, the inspection effect prediction graph 2302 is a graph showing the prediction results of the effect of the receipt extraction and the inspection with respect to the number of receipts 2321 and the score 2322 of the inspection target month. The total number 2326 of the graph shows the total number of receipts for May 2006 and the total number of claims. The rule hit 2325 indicates the total of the number of cases corresponding to the rule and the number of claims for the claim corresponding to the rule as a result of using the receipt extraction rule acquired from the receipt extraction rule management unit 126. Further, the acceptance expectation 2324 indicates the product of the number of applicable cases for each receipt extraction rule and the acceptance ratio, and the product of the average number of acceptance points (the number of acceptance points per acceptance receipt), the number of receipts corresponding to the rule and the acceptance ratio. Further, the selection 2323 indicates the range of the result (2318) of the selected receipt extraction rule in the acceptance expectation 2324. As a result, it is possible to see the extraction result of the priority inspection receipt based on the receipt extraction rule, and the influence on the entire receipt of the month to be inspected when the result is inspected. The inspector looks at the extraction result graph 2301 and the inspection effect prediction graph 2321, selects one of the circles 2318 of the receipt extraction result, presses the inspection button 2303, and moves to the reception inspection screen shown in FIG. Inspect.

  As described above, the receipt inspection support system of the present invention displays the priority inspection receipt extraction result using the index such as the acceptance ratio and the average acceptance score of the receipt extraction rule created from the reexamination results. Since the acceptance ratio, the average number of acceptance points, and the inspection result are visually displayed on the reception subject to inspection, the inspector can select a reception with high inspection efficiency and perform inspection.

  In the embodiment described so far, FIG. 13 shows a method for inputting a billing word in the billing word input field 1309. However, a template may be used for inputting the billing wording to simplify the input. FIG. 24 is a diagram showing an example of a screen when a template is used as another example of the billing message input field 1309 of FIG. For example, when an attempt is made to input in the billing text input field of FIG. 13, the screen of FIG. 24 is displayed to assist the input of the billing text. The template input screen includes a template selection screen 2401 and a billing text input field 2402, and the template selected in the template selection field 2401 is added to the billing text input field 2402 with an add button 2405 and edited. Here, three types of templates are illustrated, but one of them is selected (2403), an add button 2405 is pressed, and displayed in the billing text input field 2402. In the billing text input field 2404, the blank field 2404 of the template is edited, and a keyword such as a drug name is input here to create a billing text. Then, when the enter button 2406 is pressed, the screen returns to the screen in FIG. 13 and the created billing text is set in the billing text input field 1309. The cancel button 2407 returns to the screen in FIG. 13 without inputting anything. The template is managed by the combination pattern management unit 124 together with the combination pattern of the billing wording. This can be realized by adding a template column to the combination pattern table shown in FIG. 7 and preparing a template for each pattern. Create the wording of the template in advance, and create the wording keyword (Figure 6) and combination pattern (Figure 7) from the words used in the template, so that the relationship between the billing wording used, the wording keyword, and the combination pattern You can strengthen it. As a result, there is an effect that it is easy to create a receipt extraction rule from the reexamination record while suppressing variation in the billing wording. Further, when moving from the screen of FIG. 13 to the template input screen of FIG. 24, the information of the template of FIG. 7 is acquired from the pattern ID (802 of FIG. 8) of the receipt extraction rule from which the receipt to be inspected is extracted. Displays the template of the combination pattern. By doing in this way, the operation | work which an inspector selects a template can be abbreviate | omitted, and preparation of a claim wording can be made easy. Furthermore, when a keyword such as a drug name is selected on the screen in FIG. 13 and then the screen is shifted to the screen in FIG. 24, the name of the drug or the like selected on the template is automatically entered as a keyword and displayed. By doing so, it is possible to facilitate the creation of the invoice for the inspector. In addition, when a template is used to create a billing wording, a column for managing the ID indicating the used template is added to the reexamination actual data in FIG. 4 managed by the reexamination actual data management means 123, and a pattern ID is added. You may make it manage.

  Next, another example in which the inspection necessity determination unit 105 displays the result of extracting the priority inspection receipt will be described. FIG. 25 is a diagram showing another example of the display result of the priority inspection receipt displayed by the inspection necessity determination means 105, and an example in which the records corresponding to the rules are displayed in descending order of the predicted acceptance score and the corresponding rule count. FIG. The inspection necessity determination unit 105 acquires the receipt extraction rule from the receipt extraction rule management unit 126, and applies it to the receipt for May 2006 input from the input unit 101, and displays the screen of FIG. In the screen of FIG. 25, the insured person name 2501, the medical institution name 2502, the number of applicable rules 2503, and the expected acceptable score 2504 are displayed in a list form. Some receipts may correspond to multiple receipt extraction rules. Therefore, the inspection necessity determination unit 105 displays the receipts in the order of the number of corresponding rules among the rules corresponding to the receipt extraction rules, or in the order of the expected acceptable score. The expected acceptance score is a value obtained by summing up the average acceptance scores when a plurality of receipts are applicable. It is possible to select whether the rule number is in descending order of the number of rules 2503 or the descending order of expected acceptance points is 2504. When the inspection order is determined and the inspection button 2506 is pressed, the screen shifts to the screen shown in FIG. 13 and the inspection is performed. In the screen of Fig. 13, when the inspector presses the reexamination request button 1310 or the payment button 1311 and decides to process the receipt, the next receipt is displayed in the order decided on the screen of Fig. 25, and the inspection is efficiently performed in the specified order. It can be performed.

  In addition, a rule to be used may be selected and updated by providing means for evaluating the receipt extraction rule. Hereinafter, an example of evaluating and updating the receipt extraction rule will be described. FIG. 27 is a diagram showing an example of an inspection result management table managed by the reexamination result data management means 123. As shown in FIG. The inspection result management table is managed by the reexamination result data management means 123 together with the reexamination request result data of FIG. Inspection month 2701 indicates the date of the receipt subject to inspection, rule No (2702) corresponds to the number of the rule used for inspection (801), and receipt ID (2703) corresponds to the receipt ID (211, 311, etc.) . The reexamination number (2704) corresponds to the reexamination request performance data No (401). The inspection result management table is created when the inspection necessity determination unit 105 performs an inspection. In the flowchart of FIG. 15, when the user selects a rule to be checked (1507), the medical date of the new receipt is set to the check month 2701, the ID of the selected rule is set to rule No 2702, and the receipt extracted by the rule is set. Record the receipt ID in 2703. Then, when the reexamination request button 1310 is pressed on the inspection screen of FIG. 13, a new record is added to the reexamination request data in step 1510 of creating the reexamination request data, and the reexamination record number of 2704 is recorded. . Receipt extraction rule evaluation is performed using this data.

  The evaluation of the receipt extraction rule is performed by the receipt extraction rule evaluation means. Although not shown, the receipt extraction rule evaluation unit is provided in the receipt extraction unit 102 in the configuration diagrams of FIGS. 1 and 19. And it shall carry out a user's instruction | indication or regularly.

  The processing flow of the receipt extraction rule evaluation unit will be described with reference to the flowchart of FIG. First, re-examination record data (Fig. 4) and inspection record management table (Fig. 27) are read (2801). Then, the number of receipts and the number of reexamination requests for each rule are tabulated from the inspection result management table (2802). For example, for Rule No (2702) 1, there are 4 receipts and 3 reexamination requests. For Rule No (2702) 3, there are 2 receipts and 0 reexamination requests. It becomes. Next, the number of approvals and the number of acceptance points for each rule are tabulated from the inspection result management table and reexamination result data (2803). For rule No (2702) 1, there are 2 acceptances and the score is 800 points. Regarding Rule No. (2702) 3, since the re-examination request has not been made, both the number of cases accepted and the number of acceptance points are 0. Next, an acceptance ratio and an average acceptance score are calculated (2804). Here, for rule No (2702) 1, the acceptance ratio is 0.5, and the average acceptance score is 200 points. Then, the rule is deleted according to a predetermined standard (2805). Here, a rule with an acceptance ratio or an average acceptance score of 0 is deleted. In addition, a reference value may be provided for the acceptance ratio and the average acceptance score, and rules that do not reach the reference value may be deleted. In this way, the receipt extraction rule evaluation means ends the receipt extraction rule evaluation process.

  As described above, the receipt extraction rule evaluation means evaluates the receipt extraction rule using the acceptance ratio and the average acceptance score, and it is not accepted even if a rule that does not request a reexamination even if inspected or a request for reexamination, Rules with poor inspection efficiency can be deleted. Therefore, when the inspection necessity determination means next extracts a receipt, there is an effect that only a receptor having a higher inspection efficiency can be extracted. By periodically evaluating and deleting unnecessary rules, it is possible to constantly evaluate the receipt extraction rules and support efficient receipt inspection.

The figure which shows the example of 1 structure of the receipt inspection assistance system of this invention. The figure which shows an example of the medical receipt data which the input part 101 receives. The figure which shows an example of the dispensing receipt data which the input part 101 receives. The figure which shows an example of the performance data of the reexamination request which the reexamination performance data management means 123 manages. The figure which shows an example of the data of the receipt code keyword which the receipt code keyword management means 121 manages. The figure which shows an example of the data which manage the keyword which appears in the claim wording which the wording keyword management means 122 manages. The figure which shows an example of the data which manages the combination of the receipt code keyword and the wording keyword which the combination pattern management means 124 manages. The figure which shows an example of the receipt extraction rule table which the receipt extraction rule management means 126 manages. The figure which shows an example of the medical receipt of the past receipt data which the receipt data management means 125 manages. The figure which shows an example of the dispensing receipt of the past receipt data which the receipt data management means 125 manages. The figure which shows an example of the screen of the receipt extraction rule list | wrist which the inspection necessity determination means 105 displays on the output part 103. FIG. The figure which shows an example of the screen of the important inspection receipt extraction result list which the inspection necessity determination means 105 displays on the output part 103. FIG. The figure which shows an example of the receipt inspection screen which the inspection necessity determination means 105 displays on the output part 103. FIG. 7 is a flowchart showing an example of a flow of receipt extraction rule creation processing performed by the rule creation means 106. The figure which shows an example of the flow of a process when extracting the receipt which should be checked mainly from the receipt which the inspection necessity determination means 105 performs, and an inspector checks the receipt. The figure which shows an example of the flow of a receipt process, and the hardware constitutions of the receipt inspection assistance system of this invention. The figure which shows the sequence between the receipt inspection assistance terminal and database in a receipt extraction rule preparation process. FIG. 16 is a diagram showing a sequence between a receipt inspection support terminal and a database in the case of FIG. The figure which shows an example of a structure of the receipt inspection assistance system in the case of adding improvement to a receipt extraction rule using the past receipt data. The figure which shows an example of the flow of a process in the case of adding improvement to a receipt extraction rule using the past receipt data. FIG. 21 is a diagram showing an example of a sequence between a receipt inspection support system and a database in the case of FIG. The figure which shows the example of the rule which the receipt extraction rule correction means 1904 corrected. The figure which shows an example of the screen of the important inspection receipt extraction result which the inspection necessity determination means 105 displays on the output part 103. FIG. . The figure which shows an example of the screen in the case of using a template as another example of the claim text input column 1309. The figure which shows another example of the display result of the important inspection receipt which the inspection necessity determination means 105 displays on the output part 103. FIG. The figure which shows an example of the receipt code keyword which attaches keyword ID to a receipt code keyword, and manages. The figure which shows an example of the inspection performance management table managed by the reexamination performance data management means 123. The flowchart which shows an example of a process in the case of deleting a receipt extraction rule by a receipt extraction rule evaluation means.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 ... Input part, 102 ... Receptacle extraction part, 103 ... Output part, 104 ... Database, 105 ... Inspection necessity judgment means, 106 ... Rule preparation means, 107 ... Claim text search means, 108 ... Combination pattern conformity judgment means, 109 ... Rule index calculation means, 110 ... Receipt extraction rule generation means, 121 ... Receive code keyword management means, 122 ... Writing keyword management means, 123 ... Reexamination result data management means, 124 ... Combination pattern management means, 125 ... Receive data management Means, 126 ... Receipt extraction rule management means, 130 ... Receipt inspection support terminal, 201, 901 ... Receipt basic information table (medicine), 202, 902 ... Wound and disease name table, 203, 903 ... Abstract column table, 211, 221, 231 , 311, 321, 911, 921, 931 ... Receipt ID, 212, 312, 912 ... Medical treatment date, 213, 913 ... Medical institution, 314, 214, 315, 914 ... Insured person ID, 215, 316, 915 ... Claim points, 222, 922 ... Injury name, 223, 923 ... Medical treatment start date, 232, 932 ... Summary column, 233, 326, 933 ... Quantity, 301, 1001 ... Receipt basic information table (dispensing), 302, 1002 ... Dispensing information table, 313, 1013 ... Pharmacy, 322, 1022 ... Prescription date, 323, 1023 ... Dispensing date, 324,1024 ... Pharmaceutical name, 325,1025 ... Unit drug fee, 916,1017 ... Decision points, 401 ... Reexamination result data number, 402 ... Receive ID, 403 ... billing wording, 404 ... result, 405 ... score, 501 ... code ID, 502 ... keyword, 503 ... code name, 504 ... receipt type, 505 ... code type, 601 ... keyword ID, 602 ... keyword, 604 ... search period , 606 ... Condition value, 701, 802 ... Pattern ID, 702 ... Pattern name, 703 ... Pattern, 704 ... Search type, 705 ... Search period, 706 ... Total column, 707 code keyword type, 801 ... Rule number, 803 ... And Connection number, 804 ... Rule type, 805 ... Receive type, 806 ... Period, 807 ... Keyword, 808 ... Collection Weighing, 809 ... Condition, 810 ... Acceptance ratio, 811 ... Average acceptance score, 1101,1201 ... Pattern display column, 1102,1202 ... Keyword display column, 1103,1203 ... Receive type display column, 1104,1204 ... Backward months display Column, 1105, 1205 ... Condition display column, 1106 ... Acceptance ratio display column, 1107 ... Average acceptance score display column, 1111 ... Receive extraction start button, 1206 ... Expected hit count display column, 1207 ... Expected acceptance score display column, 1211 ... Check receipt selection field, 1212 ... Inspection start button, 1301, 1302 ... Receipt display field, 1303, 1304 ... Keyword emphasis display, 1305 ... Time series receipt display field, 1306, 1307 ... Displaying receipt, 1308 ... According to the receipt extraction rule Display indicating equal range, 1309 ... Request text input field, 1310 ... Reexamination request button, 1311 ... Payment button, 1601 ... Hospital, 1602 ... Clinic, 1603 ... Dispensing pharmacy, 1604 ... Examination payment institution, 1605 ... Insurer , 1606 ... Receipt inspection support system, 1611 ... Receipt check support terminal, 1612 ... Database, 1613 ... Display, 1614 ... Keyboard, 1615 ... Mouse, 1621 ... Send receipt (request from medical institution), 1622 ... Send receipt (after primary review), 1623 ... Request for reexamination , 1624 ... Reexamination result notification, 1901 ... Rule generation source receipt confirmation means, 1902 ... Paid receipt search means, 1903 ... Additional rule analysis means, 1904 ... Receipt extraction rule correction means, 2301 ... Extraction result graph, 2302 ... Inspection effect Prediction graph, 2303 ... Check button, 2311 ... Rule count display, 2312 ... Total acceptance score display, 2313 ... Acceptance ratio display, 2314, 2315, 2316, 2317, 2318 ... Receptacle extraction result, 2321 ... Number of cases graph, 2322 ... Score Graph, 2401 ... Template selection field, 2402 ... Invoice text input field, 2404 ... Keyword input, 2405 ... Add button, 2406 ... Determination button, 2407 ... Cancel button, 2501 ... Insured person name 2502 ... Medical institution, 2503 ... Number of applicable rules, 2504 ... Expected acceptance score, 2505 ... Check button, 2601 ... Code correspondence table, 2602 ... Keyword table, 2603 ... Keyword ID, 2701 ... Month of inspection, 2702 ... Rule No, 2703 ... Receipt ID, 2704 ... No re-examination.

Claims (16)

A receipt is extracted via a computer using a rule creation unit that creates a receipt extraction rule having information on the characteristics of the receipt, and an inspection necessity determination unit that extracts a receipt corresponding to the receipt extraction rule from a plurality of receipts. A method of extracting a receptor,
The rule creation unit that creates the receipt extraction rule includes reexamination record data including a claim text that is a claim when an insurer makes a request for reexamination of a receipt to an examination and payment institution, and is written on the receipt A receipt code keyword that manages a keyword comprising at least a part of at least one of the name of a wound or disease name, a medicine name, or a medical practice name, and a keyword that appears in the claim statement and is not included in the receipt code keyword And a wording keyword including a keyword indicating a unit of the period, a search period that is a period that goes back when the receipt is extracted, and a receipt extraction rule for the combination of the receipt code keyword, the wording keyword, and the search period Using a database having a combination pattern defining a method, The numerical controls how to convert the unit of the period a month for the word,
First, the receipt code keyword and the wording keyword are searched from the database using the database,
Second, confirm the match between the combination of the receipt code keyword and the word keyword retrieved from the claim word and the combination pattern;
Third, create a receipt extraction rule that searches for the receipt having the receipt code keyword defined by the combination pattern, going back to the period including the numerical value in which the wording keyword is converted into a month,
The receipt extraction method, wherein the inspection necessity determination unit extracts a corresponding receipt from a receipt stored in a database in accordance with a receipt extraction rule created by the rule creation unit . The receipt extraction method according to claim 1,
The rule creation unit , when the combination pattern matched in the second step includes the word keyword indicating the numerical value and the unit of the period and the receipt code keyword,
The third step totals the value of the quantity of medicines or medical treatments indicated by the receipt code keyword that exists during a period of searching retroactively for the receipt, and the aggregated value represents the numerical value of the wording keyword. A receipt extraction method characterized in that a search condition of a receipt extraction rule is to exceed a value obtained by converting a unit of a period into days. The method of extracting a receptor according to claim 1 or 2,
The reexamination result data stored in the database includes information indicating either “acceptance” in which the request was approved as a result of the reexamination or “as per the original referee” in which the request was not approved, and the result was “acceptance”. Has a reduced acceptance score,
The rule creation unit
When confirming the match between the combination of the receipt code keyword and the wording keyword and the combination pattern in the second step, the number of acceptances and the number of acceptance points are totaled for each combination pattern and the receipt code keyword,
When creating a receipt extraction rule in the third step, an acceptance ratio indicating an acceptance ratio with respect to the number of reexamination results by the combination pattern and the receipt code keyword and an average acceptance score per acceptance are obtained. Correlate with extraction rules,
When the inspection necessity determination unit extracts a receipt from a plurality of receipts stored in the database using the receipt extraction rule, the acceptance ratio is set to the number of cases searched by the receipt extraction rule from the plurality of receipts. A receipt extraction method, wherein a value obtained by multiplying the multiplied value and the number of receipts searched by the receipt extraction rule by the acceptance ratio and the average acceptance score is calculated. The method of extracting a receptor according to claim 3,
The database has receipt data related to the reexamination record data and receipt data that has not been requested for reexamination in the past.
The rule creation unit
Fourth, from the receipt data related to the reexamination record data, obtain the receipt data related to the reexamination record data that is the basis of the receipt extraction rule and accepted as a result of the reexamination,
Fifth, from the receipt that has not been requested for reexamination in the past, obtain the receipt data using the receipt extraction rule,
Sixth, from the receipt data that is the basis of the receipt extraction rule and the receipt data that is not requested for reexamination obtained using the receipt extraction rule, the difference between the items described is determined as the acceptance ratio or the average acceptance. Learning so that the score is higher than the receipt extraction rule,
Seventh, a receipt extraction method, wherein correction is performed by adding an item obtained as a result of the learning to the receipt extraction rule. In a receipt inspection support system that extracts a receipt corresponding to the reception extraction rule from a plurality of receptions using a reception extraction rule having information on the characteristics of the reception to be inspected, and supports the reception inspection.
The receipt inspection support system has a receipt inspection support terminal and a database,
The database is written on the receipt, and a reexamination result data management unit for managing the reexamination results including a claim text that is a claim when the insurer makes a request for reexamination of the receipt to the examination and payment institution. And a receipt code keyword management unit that manages a keyword comprising at least a part of at least one of a name of a disease or a disease, a drug name, or a medical practice name, and a keyword that appears in the claim text and is not included in the receipt code keyword A wording keyword management unit for managing wording keywords, a combination pattern management unit for managing a combination pattern that defines a method for creating a receipt extraction rule for extracting a receipt having a combination of the receipt code keyword and the wording keyword, and receipt data management unit that accumulates the receipt data, the Has a receipt extraction rule management unit for managing the concept extraction rule,
The receipt inspection support terminal includes an input unit that receives a receipt to be inspected, a receipt extraction unit that prioritizes inspection with respect to the reception subject to inspection, and an output unit that outputs the prioritized result. And
The healthcare cost extraction unit, and the receipt extraction rule management unit check necessity judgment unit to prioritize inspection of the inspection target receipts based on receipt extracted with receipt extraction rules governing the receipt extraction It consists of a rule creation section that creates rules,
The rule creation unit, and the claim language search unit keyword and the wording keyword management unit for the re-examination performance data management unit said the claim wording receipt code keyword management unit to manage to manage is to search by keyword to management, the A combination pattern matching determination unit that determines a match between a receipt code keyword, an appearance pattern of the word keyword, and a combination pattern managed by the combination pattern management unit; and the receipt code keyword and the word keyword defined by the combination pattern; A receipt extraction rule generation unit that creates a receipt extraction rule according to a receipt extraction rule creation method of extracting a receipt having
The wording keyword management unit manages a method for converting the unit of the period to the month for the wording keyword when the wording keyword is a keyword indicating a numerical value and a unit of the period,
The receipt extraction rule generating unit , when the combination pattern includes a keyword indicating the unit of the numerical value and the period, receives the period including the numerical value obtained by converting the unit into a month as a period for searching backwards for the receipt Receipt inspection support system characterized by creating rules. In the receipt inspection support system according to claim 5,
When the combination pattern determined by the combination pattern matching determination unit includes the word keyword indicating the numerical value and the unit of the period and the receipt code keyword,
The said receipt extraction rule production | generation part totals the value of the quantity of the medicine or medical treatment which the said receipt code keyword which exists during the period which searches retroactively for the said receipt, and the said total value is the numerical value of the said wording keyword. The receipt inspection support system characterized in that a search condition of the receipt extraction rule is that the unit of the period exceeds the value converted into days. In the receipt inspection support system according to claim 5 or 6,
The reexamination performance data management department shall indicate information indicating either “acceptance” for which the request was approved as a result of the reexamination or “as per the original referee” for which the request was not approved, and a reduction if the result was approved. Manage the accepted acceptance score,
The combination pattern conformity determination unit , when confirming a match between the combination of the receipt code keyword and the wording keyword and the combination pattern, totals the number of acceptances and the number of acceptance points for each combination pattern and the receipt code keyword,
The rule index calculation unit that calculates an acceptance ratio indicating an acceptance ratio with respect to the number of reexamination results for each combination pattern and the receipt code keyword and an average acceptance score per acceptance,
The said receipt extraction rule production | generation part combines the said acceptance extraction rule and the acceptance ratio calculated by the said rule parameter | index calculation part , and stores it in a receipt extraction rule management part in combination,
The inspection necessity determination unit , when performing the extraction of the input receipt, searches for the value obtained by multiplying the number of cases searched for by the receipt extraction rule from the input receipt by the acceptance ratio and the receipt extraction rule. A receipt inspection support system that performs prioritization of reception inspection using a value obtained by multiplying the number of receipts by the acceptance ratio and the average acceptance score. In the receipt inspection support system according to claim 7,
The receipt data management unit receives data related to the reexamination record data, and
Manage receipt data that has not been requested for reexamination in the past,
A rule generation source receipt confirmation unit that obtains from the receipt data management unit the receipt data that is related to the reexamination record data that is the basis of the receipt extraction rule and that has been accepted as a result of the reexamination;
Receipt data that has not been requested for reexamination, and a received receipt search unit that obtains receipt data corresponding to the receipt extraction rule from the receipt data management unit ;
Based on the receipt data that is the basis of the receipt extraction rule and the receipt data that is not requested for reexamination obtained by using the receipt extraction rule, the acceptance ratio or the average acceptance score indicates the difference between the items described. An additional rule analysis unit that learns to be higher than the extraction rule;
A receipt inspection support system, comprising: a receipt extraction rule correction unit that performs correction to add an item obtained as a result of the learning to the receipt extraction rule. In the receipt inspection support system according to claim 7 or claim 8,
The output unit is a value obtained by multiplying the number of cases to be inspected by the receipt extraction rule and the receipt extraction rule by an acceptance ratio, and the number of cases to be inspected by the receipt extraction rule is multiplied by an average acceptance score. A receipt inspection support system, characterized in that a value is displayed and displayed so that a receipt inspection rule can be selected. In the receipt inspection support system according to claim 7 or claim 8,
The output unit displays on the screen a person's receipt having a receipt that conforms to the receipt extraction rule;
A receipt inspection support system that highlights a name of a disease or a medicine or a name of a medical practice indicated by the receipt code keyword included in a receipt extraction rule for searching for the receipt. The receipt inspection support system according to claim 7 or 8, wherein the receipt displayed on the screen is a plurality of receipts, and the output unit includes a time-series display unit of the plurality of receipts, A receipt inspection support system, wherein a receipt selected by a selection unit displayed on a series display unit is displayed. 9. The receipt inspection support system according to claim 7 or 8, wherein the output unit includes an extraction result display unit that displays the extracted reception result as a pie chart on an axis of an acceptance ratio and an average acceptance score. The receipt inspection support system, wherein the size of the circle of the pie chart is changed and displayed based on the selected rule corresponding number or the total allowable score. The receipt inspection support system according to claim 12, wherein the output unit further includes an inspection effect prediction display unit that displays a ratio of the extracted receptors with respect to a receptor to be inspected for each number of cases and / or points. A receipt inspection support system, wherein a ratio in a circle selected from circles of the pie chart of the extraction result display unit is displayed on the inspection effect prediction display unit. 8. The receipt inspection support system according to claim 7, wherein the priority ranking performed by the inspection necessity determination unit is a priority order based on an expected acceptance score or a corresponding rule number, and the expected acceptance score. Or the receipt inspection support system which displays a receipt extraction result in order of an applicable rule. In receipt inspection support system according to claim 5, wherein the output unit, receipt inspection support system characterized in that it has a claim language input unit for inputting billing wording against receipts extraction results. 16. The receipt inspection support system according to claim 15, wherein the output unit includes a template selection unit that selects a word to be input to the billing word input unit, and the billing word input unit applies the selected template to the selected template. A receipt inspection support system that allows a claim word to be input.

Images