Dispensing document inspection device

Abstract

To provide a medicine preparation document audit device that conducts an audit as to whether there is a mismatch in the details of description for each of a plurality of types of medicine preparation related documents used when providing a prescribed drug for every patient in a pharmacy.SOLUTION: A medicine preparation document audit device comprises: a camera photographing unit 11 that photographs a plurality of types of medicine preparation related documents 2 used when providing a prescribed drug for every patient; an image data analysis unit 12 that extracts a character region for each of one or more pieces of image data acquired by the camera photographing unit 11; a document identification unit 13 that collates character data included in the character region and a feature quantity 17 defined in advance for every type of medicine preparation related document 2 with each other, and specifies the plurality of types of medicine preparation related documents 2 included in each of the pieces of image data; a character identification unit 14 that identifies information on a 'patient's name' by using a learning model 16 for each of the plurality of types of medicine preparation related documents 2; and a character collation unit 15 that performs the collation about the information on a 'patient's name' identified by the character identification unit 14 for each of the plurality of types of medicine preparation related documents 2 to conduct an audit as to whether there is a mismatch in a result of the collation.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a technology for auditing whether there are any inconsistencies in the contents of multiple types of dispensing-related documents used when dispensing prescription drugs to each patient at a dispensing pharmacy. [Background technology]

[0002] At dispensing pharmacies, medicines are dispensed based on prescriptions issued by medical institutions for each patient. The dispensing-related documents used in this process include prescriptions, dispensing records, dispensing details, drug information sheets, medicine envelopes, medicine notebook stickers, and receipts, and it is necessary that these dispensing-related documents are provided correctly for each patient.

[0003] However, since human intervention is required when preparing the multiple dispensing-related documents required for each patient at a dispensing pharmacy, there is a possibility of human error occurring. For example, if patient A's dispensing-related documents are mixed with those of patient B, it could become a serious incident that could lead to a medical accident. Therefore, a system is needed to prevent such incidents caused by human error from occurring. Summary of the Invention [Problem to be solved by the invention]

[0004] When pharmacies prepare the multiple types of dispensing-related documents required for each patient, such as prescriptions, dispensing records, dispensing details, drug information sheets, medicine envelopes, medicine notebook stickers, and receipts, it is important to audit each dispensing-related document for inconsistencies in the contents to prevent incidents caused by human error.

[0005] The object of the present invention is to provide a dispensing document inspection device that inspects whether there are any inconsistencies in the contents of multiple types of dispensing-related documents used when providing prescription drugs to each patient at a dispensing pharmacy, particularly by verifying the information on 'patient name' and inspecting whether the verification results contain any inconsistencies. [Means for solving the problem]

[0006] In order to achieve the above object, a dispensing document inspection device according to a first aspect of the present invention comprises: A device for auditing the contents of prescription-related documents, including "prescriptions," "prescription records," "prescription details," "drug information sheets," "medication envelopes," "medication notebook stickers," and "receipts," which are documents used when providing prescription drugs to each patient at a dispensing pharmacy, a camera photographing unit that photographs the dispensing-related documents used when providing prescription drugs to each patient; an image data analysis unit that analyzes the layout of each of the one or more image data acquired by the camera photographing unit and extracts a plurality of character regions; a document identification unit that identifies, for each character area extracted by the image data analysis unit, the character data included in the character area and a feature amount that is predefined for each type of the dispensing-related document, and identifies the multiple types of the dispensing-related document included in the image data; a character identification unit that uses a learning model to identify information about 'patient name' included in the dispensing-related document for each of the multiple types of dispensing-related documents identified by the document identification unit; a character collation unit that collates the information on the 'patient name' identified by the character identification unit with each of the multiple types of dispensing-related documents and audits whether the collation results include any inconsistencies; Equipped with The learning model is For each type of dispensing-related document, one or more specific characters, including "patient" necessary to identify the information on "patient name," and a specific layout type or context for identifying the information on "patient name" from the specific characters are included, which are generated by machine learning. It is characterized by:

[0007] A dispensing document inspection device according to a second aspect of the present invention is the first aspect of the present invention, The learning model includes a model generated by machine learning for each type of dispensing-related document based on one or more specific characters, including "prescription" necessary to identify information on "prescription drugs," and a specific layout type and context for identifying information on "prescription drugs" from the specific characters; The character identification unit a means for identifying, for each of the plurality of types of dispensing-related documents identified by the document identification unit, information on 'prescription drugs' included in the dispensing-related documents using the learning model; The character matching unit The device is characterized by having a means for comparing the information on 'prescription drugs' identified by the character recognition unit with each of the multiple types of dispensing-related documents and auditing whether the comparison results include any inconsistencies.

[0008] A dispensing document inspection device according to a third aspect of the present invention is the second aspect of the present invention, The character matching unit is characterized by having a means for calculating the quantity of "medicine bags" based on the information of "prescription drugs" identified by the character recognition unit, and comparing that quantity with the quantity of "medicine bags" identified by the document recognition unit, and auditing whether the matching results include any discrepancies.

[0009] A dispensing document inspection device according to a fourth aspect of the present invention is the device according to the first aspect of the present invention, In a configuration in which a management server is arranged to collectively manage feature definition files defined according to the formats of the dispensing-related documents used for each dispensing pharmacy store, The dispensing document inspection device according to claim 1 is a feature acquisition unit that downloads the feature definition file corresponding to each store from the management server; The document identification unit is characterized in that it identifies multiple types of dispensing-related documents included in the image data using the features defined in the feature definition file acquired by the feature acquisition unit. [Effects of the Invention]

[0010] According to the present invention, by auditing whether there are any inconsistencies in the contents of multiple types of dispensing-related documents used when providing prescription drugs to each patient at a dispensing pharmacy, particularly by checking the information on 'patient name' and auditing whether the results of the check contain any inconsistencies, it becomes possible to prevent incidents caused by human error from occurring.

[0011] Furthermore, by incorporating the prescription document auditing device into information terminals such as smartphones and tablet devices, it will be possible to audit prescription-related documents in real time with simple operations. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a configuration diagram illustrating an example of a prescription document inspection device according to an embodiment of the present invention.

[0023] FIG. [Figure 2] FIG. 2 is a functional configuration diagram illustrating an example of a functional configuration of the dispensing document auditing device according to the embodiment of the present invention. [Figure 3] FIG. 10 is a flowchart illustrating an example of processing executed by the dispensing document inspection device according to the embodiment of the present invention. [Figure 4] 1 is a schematic diagram showing an example of captured images of multiple types of prescription-related documents taken by a camera in a prescription document inspection device according to an embodiment of the present invention. FIG. [Figure 5] FIG. 1 is a schematic diagram illustrating an example in which a dispensing document auditing device according to an embodiment of the present invention is mounted on a smartphone with a camera. [Figure 6] 1 is a system configuration diagram showing an example of a system configured by introducing a dispensing document inspection device according to an embodiment of the present invention into a plurality of dispensing pharmacies. DETAILED DESCRIPTION OF THE INVENTION

[0013] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. However, the present invention is not limited to the following embodiments.

[0014] Fig. 1 shows an example of an image of a dispensing document auditing device according to an embodiment of the present invention used in the operations of a dispensing pharmacy. As shown in Fig. 1, the dispensing document auditing device 1 has a camera function and also has a function of taking pictures of multiple types of dispensing-related documents 2 used when providing prescription drugs to each patient at a dispensing pharmacy with the camera, identifying the dispensing-related documents 2 based on the captured images, and auditing for inconsistencies in the contents of each type of dispensing-related document 2. Here, the dispensing-related documents 2 include "prescriptions," "dispensing records," "dispensing details," "drug information sheets," "medication envelopes," "medication notebook stickers," and "receipts."

[0015] Furthermore, when photographing multiple types of dispensing-related documents 2 with a camera, it is assumed that the multiple types of dispensing-related documents 2 will be photographed together in one shot, or that they will be photographed separately in multiple shots.

[0016] Furthermore, by incorporating the dispensing document auditing device 1 into an information terminal such as a smartphone or tablet terminal, it is possible to audit dispensing-related documents in real time with simple operations.

[0017] Next, Fig. 2 is a functional configuration diagram showing an example of the functional configuration of the dispensing document auditing device 1. As shown in Fig. 2, the dispensing document auditing device 1 includes a camera unit 11, an image data analysis unit 12, a document identification unit 13, a character identification unit 14, a character matching unit 15, a learning model 16, and a feature definition file 17.

[0018] The camera photographing unit 11 photographs multiple types of dispensing-related documents 2 used when providing prescription drugs to each patient. Here, the dispensing-related documents 2 include "prescriptions," "dispensing records," "dispensing details," "drug information sheets," "medication envelopes," "medication notebook stickers," and "receipts." When photographing multiple types of dispensing-related documents 2, there are cases where multiple types of dispensing-related documents 2 are lined up and photographed at once, and cases where they are photographed separately multiple times.

[0019] The image data analysis unit 12 analyzes the layout of each of the one or more image data acquired by the camera shooting unit 11 and extracts a plurality of character regions. In extracting the character regions, for example, an OCR (Optical Character Recognition) program can be run to read character information for each of the image data.

[0020] For each character region extracted by the image data analysis unit 12, the document identification unit 13 compares the character data included in the character region with features predefined for each type of dispensing-related document 2, thereby identifying multiple types of dispensing-related document 2 included in each piece of image data. That is, for each of the multiple character regions included in each piece of image data, the document identification unit 13 repeatedly compares the character data included in the character region with features predefined for each of the dispensing-related documents 2, such as "prescription," "dispensing record," "dispensing specification," "drug information sheet," "medication envelope," "medication notebook sticker," and "receipt," and if a matching feature is detected, the document identification unit 13 identifies the type of dispensing-related document 2 from the feature.

[0021] Here, the above features are predefined and stored in the feature definition file 17 in order to identify the type of dispensing-related document 2. For example, to identify a "prescription" document, character strings such as "prescription" and "prescription" are defined as features. Also, for example, to identify a "medicine envelope" document, character strings such as "oral medication" and "external medication" are defined as features.

[0022] Furthermore, because the formats of dispensing-related documents 2 are not standardized according to uniform standards across the country, it is common for different dispensing pharmacy stores to use different document formats. That is, there are cases where different dispensing pharmacy stores use different document formats for dispensing-related documents 2. Therefore, when defining the above features, customizing the features according to the document formats used by each dispensing pharmacy store can improve the accuracy of identifying the type of dispensing-related document 2.

[0023] Furthermore, when the dispensing document inspection device 1 is installed in multiple dispensing pharmacies, for example, a configuration may be adopted in which feature definition files 17 customized and defined for each dispensing pharmacy are centrally managed on a management server and downloaded and used by each dispensing pharmacy. In this case, although not shown in FIG. 2 , the dispensing document inspection device 1 includes a feature acquisition unit that downloads feature definition files 17 corresponding to each store from the management server, and the document identification unit 13 identifies the multiple types of dispensing-related documents included in the image data using the feature amounts defined in the feature definition file 17 acquired by the feature acquisition unit. This configuration allows each store to obtain the latest information in real time, which is expected to lead to improved business efficiency and reduced management costs.

[0024] Here, the type of dispensing-related document 2 is identified using predefined features for each type of dispensing-related document 2, but this method is not limited to this, and for example, a method using machine learning using a neural network may also be used.

[0025] The character recognition unit 14 uses the learning model 16 to recognize the information of 'patient name' included in the dispensing-related document 2 for each of the multiple types of dispensing-related document 2 identified by the document recognition unit 13.

[0026] The character matching unit 15 checks the 'patient name' information identified by the character recognition unit 14 against multiple types of dispensing-related documents 2, and audits whether the matching results include any mismatches. If the matching results include any mismatches, measures such as issuing an alert are taken.

[0027] The learning model 16 includes models generated by machine learning for each type of dispensing-related document 2 based on one or more specific characters, including "patient," necessary to identify the information on "patient name," and a specific layout type and context for identifying the information on "patient name" from the specific characters.

[0028] In the above functional configuration, we have explained that the information on 'patient name' is compared for each of multiple types of dispensing-related documents 2 and the comparison results are audited to see if there are any discrepancies. However, another functional configuration will be described below.

[0029] The learning model 16 includes models generated by machine learning for each type of dispensing-related document 2 based on one or more specific characters, including "prescription," necessary to identify information on "prescription drugs," and specific layout types and contexts for identifying information on "prescription drugs" from the specific characters.

[0030] The character recognition unit 14 has means for identifying information on 'prescription drugs' contained in the dispensing-related document 2 using the learning model 16 for each of the multiple types of dispensing-related document 2 identified by the document recognition unit 13.

[0031] The character matching unit 15 has a means for checking whether or not the matching results include any inconsistencies by matching the information on 'prescription drugs' identified by the character recognition unit 14 with each of the multiple types of dispensing-related documents 2. If the matching results include any inconsistencies, a means for issuing an alert, for example, is executed.

[0032] Furthermore, the character matching unit 15 has a means for calculating the quantity of "medicine envelopes" based on the information of the 'prescription drugs' identified by the character recognition unit 14, and for checking whether the result of the comparison includes any discrepancies by comparing the calculated quantity with the quantity of "medicine envelopes" identified by the document recognition unit 13. If the result of the comparison includes any discrepancies, a means for issuing an alert, for example, is executed.

[0033] Next, a process executed in the dispensing document inspection device 1 will be described with reference to the flowchart shown in Fig. 3. Fig. 3 is a flowchart showing an example of processing executed in the dispensing document inspection device 1.

[0034] In step S10, image data is acquired by capturing images of multiple types of dispensing-related documents 2 used when providing prescription drugs to each patient. Here, the dispensing-related documents 2 include "prescriptions," "dispensing records," "dispensing details," "drug information sheets," "medication envelopes," "medication notebook stickers," and "receipts." When capturing images of multiple types of dispensing-related documents 2, there are cases where multiple types of dispensing-related documents 2 are lined up and captured at once, and cases where images are captured in separate batches.

[0035] Figure 4 shows an example of images taken with a camera of multiple types of dispensing-related documents. The example in Figure 4 shows image data 1 to 4, which were taken four times for multiple types of dispensing-related documents 2 for each patient, with image data 1 including a "prescription," "dispensing details," "drug information sheet," and "receipt," image data 2 including a "medication notebook sticker," image data 3 including a "medication envelope," and image data 4 including a "dispensing record."

[0036] Next, in step S20, the layout of one or more pieces of image data captured in step S10 is analyzed to extract multiple character regions for each piece of image data. In the example of Figure 4, four pieces of captured image data are analyzed, and multiple character regions are extracted from image data 1. Similarly, the layouts of image data 2, image data 3, and image data 4 are analyzed, and multiple character regions are extracted.

[0037] Next, in step S30, for each character region extracted in step S20, the character data included in the character region is compared with features predefined for each type of dispensing-related document 2 to identify multiple types of dispensing-related document 2 included in each piece of image data. That is, for each of the multiple character regions included in each piece of image data, the character data included in the character region is repeatedly compared with features predefined for each of the dispensing-related documents 2, such as "prescription," "dispensing record," "dispensing specification," "drug information sheet," "medication envelope," "medication notebook sticker," and "receipt," and when a matching feature is detected, the type of dispensing-related document 2 is identified from the feature.

[0038] Here, the above features are predefined and stored in the feature definition file 17 in order to identify the type of dispensing-related document 2. For example, to identify a "prescription" document, character strings such as "prescription" and "prescription" are defined as features. Also, to identify a "medicine envelope" document, character strings such as "oral medication" and "external medication" are defined as features.

[0039] In the example of Figure 4, for image data 1, "prescription," "dispensing details," "drug information sheet," and "receipt" are identified as dispensing-related documents 2. Furthermore, for image data 2, "medication notebook sticker" is identified as dispensing-related documents 2, for image data 3, "medication envelope" is identified as dispensing-related documents 2, and for image data 4, "dispensing record" is identified as dispensing-related documents 2.

[0040] Here, the type of dispensing-related document 2 is identified using predefined features for each type of dispensing-related document 2, but this method is not limited to this, and for example, a method using machine learning using a neural network may also be used.

[0041] Next, in step S40, for each of the multiple types of prescription-related documents 2 identified in step S30, the learning model 16 is used to identify the information on 'patient name' contained in the prescription-related document 2.

[0042] Here, the learning model 16 includes, for each type of dispensing-related document 2, one or more specific characters, including "patient", which are necessary to identify the information on "patient name", and a specific layout type and context for identifying the information on "patient name" from the specific characters, generated by machine learning.

[0043] Next, in step S50, the 'patient name' information identified in step S40 is checked against multiple types of dispensing-related documents 2 to determine whether the check results include any inconsistencies. Although not shown in the flowchart, if the check results include any inconsistencies, it is also possible to configure the system so that a means such as issuing an alert is executed.

[0044] As mentioned above, by auditing whether there are any inconsistencies in the contents of the multiple types of dispensing-related documents 2 used when dispensing prescription drugs to each patient at a dispensing pharmacy, and in particular by checking the information on 'patient name' and auditing whether the results of the check contain any inconsistencies, it is possible to prevent incidents caused by human error from occurring.

[0045] Furthermore, with regard to the above steps S40 and S50, it has been explained that the information on 'patient name' is checked and the check results are audited to see if there are any discrepancies, but another process not shown in the flowchart will be explained below.

[0046] In step S40, for each of the multiple types of dispensing-related documents 2 identified in step S30, the learning model 16 is used to identify the information on 'prescription drugs' contained in the dispensing-related documents 2.

[0047] Here, the learning model 16 includes, for each type of dispensing-related document 2, one or more specific characters, including "prescription" necessary to identify information on "prescription drugs," and a specific layout type and context for identifying information on "prescription drugs" from the specific characters, generated by machine learning.

[0048] Next, in step S50, the information on the 'prescription drug' identified in step S40 is checked against multiple types of dispensing-related documents 2, and the check results are audited to determine whether any discrepancies are included. Although not shown in the flowchart, if the check results include any discrepancies, it is also possible to configure the system so that measures such as issuing an alert are executed.

[0049] Furthermore, the quantity of "medicine envelopes" is calculated based on the information of the 'prescription drugs' identified in step S40, and this quantity is compared with the quantity of "medicine envelopes" identified by the document recognition unit 14 to check whether the comparison results include any discrepancies. If the comparison results include any discrepancies, it is also possible to configure the system so that measures such as issuing an alert are executed.

[0050] By executing steps S10 to S50 described above, it is possible to audit whether there are any inconsistencies in the contents of each of the multiple types of dispensing-related documents 2 used when providing prescription drugs to each patient at a dispensing pharmacy, in particular by comparing information on 'patient name' and 'prescribed drug' and auditing whether the comparison results contain any inconsistencies, thereby making it possible to prevent incidents caused by human error from occurring.

[0051] Furthermore, the above-described steps S10 to S50 can be configured to be executed by, for example, a computer program.

[0052] Next, an embodiment in which the dispensing document inspection device 1 is mounted on a smartphone with a camera will be described with reference to FIG.

[0053] Example 1 In the embodiment shown in FIG. 5, the prescription-related documents to be provided to Patient A include the following 12 documents. Prescription 1, Prescription 2, Prescription 3 Medicine bag 1, Medicine bag 2, Medicine bag 3, Medicine bag 4 Dispensing Record 1 Dispensing Schedule 1, Dispensing Schedule 2, Dispensing Schedule 3, Dispensing Schedule 4 The dispensing document inspection device 1 mounted on a camera-equipped smartphone inspects these 12 dispensing-related documents to check for inconsistencies in the contents of each document. Here, the device verifies the patient name information and inspects whether the verification results include any inconsistencies. This embodiment will be explained for each step in the flowchart shown in Figure 3.

[0054] <Step S10> The dispensing document inspection device 1 mounted on a camera-equipped smartphone acquires image data captured by the camera of the above 12 dispensing-related documents to be provided to Patient A. It is expected that these dispensing-related documents will be photographed multiple times.

[0055] <Step S20> The dispensing document inspection device 1 analyzes the layout of each of the plurality of image data captured in step S10 and extracts a plurality of character regions.

[0056] <Step S30> For each character region extracted in step S20, the dispensing document inspection device 1 compares the character data contained in the character region with features predefined for each type of dispensing-related document to identify multiple types of dispensing-related documents contained in the image data. That is, for each of the multiple character regions contained in the image data, the device repeatedly compares the character data contained in the character region with features predefined for each type of dispensing-related document, such as "prescription," "dispensing record," "dispensing specification," "drug information sheet," "medication envelope," "medication notebook sticker," and "receipt." If a matching feature is detected, the device identifies the type of dispensing-related document from the feature. Here, the 12 dispensing-related documents mentioned above are identified.

[0057] <Step S40> The dispensing document inspection device 1 uses the learning model to identify the information of 'patient name' included in the dispensing-related document for each of the 12 dispensing-related documents identified in step S30.

[0058] <Step S50> The dispensing document inspection device 1 checks the 'patient name' information identified in step S40 against each of the 12 dispensing-related documents and inspects whether the inspection results include any inconsistencies. In other words, it inspects whether the 12 dispensing-related documents are the ones to be provided to patient A. Figure 5 shows an example of the display of the inspection results.

[0059] In the example of the audit results shown in Figure 5, the following is displayed: Prescription 3 / 3 Medicine bag 1 / 4 ·Pharmacy Record 1 / 1 Dispensing details 4 / 4 For prescriptions 1 to 3, dispensing record 1, and dispensing details 1 to 4, the 'patient name' information all matches, but for medicine bags 1 to 4, there is only one where the 'patient name' information matches, and for the remaining three, it is displayed that the 'patient name' information does not match.

[0060] From the above, it will be possible to easily check the dispensing-related documents provided to Patient A for inconsistencies and audit whether the check results contain any inconsistencies, thereby preventing incidents caused by human error.

[0061] Next, an embodiment of a system in which the dispensing document inspection device 1 is installed in a plurality of dispensing pharmacies will be described with reference to FIG.

[0062] <Example 2> Fig. 6 is a system configuration diagram showing an example of a system configured when the dispensing document auditing device 1 is installed in three dispensing pharmacies. In the dispensing document auditing system 31 shown in Fig. 6, the dispensing document auditing device 1 is installed in each store, and the feature definition file 17 is centrally managed on the management server 32. The dispensing document auditing device 1 installed in each dispensing pharmacy downloads and uses the feature definition file 17 that has been customized and defined for each store from the management server 32 via the network.

[0063] The dispensing document inspection device 1 includes a feature acquisition unit that downloads a feature definition file 17 corresponding to each store from the management server 32. The document identification unit 13 identifies multiple types of dispensing-related documents 2 included in the image data by using the features defined in the feature definition file 17 acquired by the feature acquisition unit.

[0064] As mentioned above, it is common for different dispensing pharmacies to use different document formats. That is, different dispensing pharmacies may use different formats for dispensing-related documents 2. Therefore, when defining features, customizing the features according to the document formats used by each dispensing pharmacy can improve the accuracy of identifying the type of dispensing-related document 2. Furthermore, by centrally managing the feature definition file 17 on the management server 32, each store can obtain the latest information in real time, which is expected to improve business efficiency and reduce management costs. [Explanation of symbols]

[0065] 1...Pharmacy document inspection device 2. Dispensing-related documents 11...Camera Department 12...Image data analysis unit 13...Document Identification Section 14...Character recognition section 15...Character matching section 16...Learning Model 17...Feature definition file 31...Pharmacy Document Audit System 32...Administrative server

Claims

1. A device for auditing the contents of prescription-related documents, including "prescriptions," "prescription records," "prescription details," "drug information sheets," "medication envelopes," "medication notebook stickers," and "receipts," which are documents used when providing prescription drugs to each patient at a dispensing pharmacy, a camera photographing unit that photographs the dispensing-related documents used when providing prescription drugs to each patient; an image data analysis unit that analyzes the layout of each of the one or more image data acquired by the camera photographing unit and extracts a plurality of character regions; a document identification unit that identifies, for each character area extracted by the image data analysis unit, the character data included in the character area and a feature amount that is predefined for each type of the dispensing-related document, and identifies the multiple types of the dispensing-related document included in the image data; a character identification unit that uses a learning model to identify information about 'patient name' included in the dispensing-related document for each of the multiple types of dispensing-related documents identified by the document identification unit; a character collation unit that collates the information on the 'patient name' identified by the character identification unit with each of the multiple types of dispensing-related documents and audits whether the collation results include any inconsistencies; Equipped with The learning model is For each type of dispensing-related document, one or more specific characters, including "patient" necessary to identify the information on "patient name," and a specific layout type or context for identifying the information on "patient name" from the specific characters are included, which are generated by machine learning. A prescription document auditing device characterized by the above.

2. The prescription document inspection device according to claim 1, The learning model includes a model generated by machine learning for each type of dispensing-related document based on one or more specific characters, including "prescription" necessary to identify information on "prescription drugs," and a specific layout type and context for identifying information on "prescription drugs" from the specific characters; The character identification unit a means for identifying, for each of the plurality of types of dispensing-related documents identified by the document identification unit, information on 'prescription drugs' included in the dispensing-related documents using the learning model; The character matching unit A dispensing document auditing device characterized by having a means for comparing the information on 'prescription drugs' identified by the character recognition unit with each of multiple types of dispensing-related documents and auditing whether the comparison results include any inconsistencies.

3. The prescription document inspection device according to claim 2, The character matching unit calculates the quantity of "medicine bags" based on the information of "prescription drugs" identified by the character recognition unit, and compares this quantity with the quantity of "medicine bags" identified by the document recognition unit, and has a means for auditing whether the comparison results include any discrepancies.

4. In a configuration in which a management server is arranged to collectively manage feature definition files defined according to the formats of the dispensing-related documents used for each dispensing pharmacy store, The dispensing document inspection device according to claim 1, a feature acquisition unit that downloads the feature definition file corresponding to each store from the management server; The document identification unit identifies multiple types of dispensing-related documents included in the image data using the features defined in the feature definition file acquired by the feature acquisition unit.

Images