Information processing device, information processing system, information processing method, and program

Abstract

This information processing device comprises: an extraction unit for extracting a plurality of unique expressions from a medical record; an identification unit for referencing the extracted plurality of unique expressions and identifying the degree of causal relationship among the plurality of unique expressions; and a generation means for referencing the degree of causal relationship to generate a summary of the medical record.

Description

Information processing device, information processing system, information processing method, and program

[0001] The present disclosure relates to an information processing device, an information processing system, an information processing method, and a program.

[0002] Techniques relating to document summarization are known. For example, Patent Document 1 describes a summary creation program that creates a summary of the examination history and medical record of a target patient.

[0003] The summary creation program described in Patent Document 1 performs a time-series correlation analysis between test items in the test history and words contained in sentences in the medical record as items, calculates a first score value indicating the degree of correlation over time, and creates a summary based on test results corresponding to characteristic items whose first score value satisfies a predetermined criterion and sentences in the medical record that contain words corresponding to the characteristic items.

[0004] Japanese Patent Application Publication No. 2020-77290

[0005] For example, medical records that record medical procedures such as examinations, diagnoses, and treatments performed by medical professionals include the rationale for performing the medical procedures. In other words, medical records record causal relationships related to medical procedures, and it is preferable that such causal relationships be taken into consideration when preparing summaries.

[0006] However, the summary created by the summary creation program described in Patent Document 1 is based only on the chronological correlation between items, which means there is a problem that it may not be possible to generate an appropriate summary of medical records.

[0007] The present disclosure has been made in view of the above problems, and an exemplary purpose thereof is to provide a technique for generating an appropriate summary of medical records.

[0008] An information processing device according to an exemplary aspect of the present disclosure includes an extraction means for extracting a plurality of named entities from a medical record, an identification means for identifying the degree of causal relationship between the plurality of named entities by referring to the extracted plurality of named entities, and a generation means for generating a summary of the medical record by referring to the degree of causal relationship.

[0009] An information processing device according to an exemplary aspect of the present disclosure includes an acquisition means for acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and a learning unit for using the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0010] An information processing system according to an exemplary aspect of the present disclosure includes an extraction means for extracting a plurality of named entities from medical records, an identification means for identifying the degree of causal relationship between the extracted plurality of named entities by referring to the plurality of extracted named entities, and a generation means for generating a summary of the medical records by referring to the degree of causal relationship.

[0011] An information processing system according to an exemplary aspect of the present disclosure includes an acquisition means for acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and a learning unit for using the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0012] An information processing method according to an exemplary aspect of the present disclosure is an information processing method executed by a computer, and includes extracting a plurality of named entities from medical records, referring to the extracted plurality of named entities to determine the degree of causal relationship between the plurality of named entities, and generating a summary of the medical records by referring to the degree of causal relationship.

[0013] An information processing method according to an exemplary aspect of the present disclosure is an information processing method executed by a computer, and includes obtaining training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and using the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0014] A program according to an exemplary aspect of the present disclosure is a program that causes a computer to function as an information processing device, and causes the computer to perform the following operations: extracting multiple named entities from medical records; referring to the extracted multiple named entities to determine the degree of causal relationship between the multiple named entities; and generating a summary of the medical records by referring to the degree of causal relationship.

[0015] A program according to an exemplary aspect of the present disclosure is a program that causes a computer to function as an information processing device, and the program causes the computer to acquire training data that includes labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and to use the training data to train a model that identifies the degree of causal relationships between multiple named entities by referring to the multiple named entities.

[0016] According to an exemplary aspect of the present disclosure, an exemplary effect can be achieved in that a technique for generating an appropriate summary of medical records can be provided.

[0017] FIG. 1 is a block diagram showing a configuration of an information processing device according to the present disclosure. FIG. 2 is a block diagram showing a configuration of an information processing system according to the present disclosure. FIG. 3 is a flow diagram showing a flow of an information processing method according to the present disclosure. FIG. 4 is a block diagram showing a configuration of an information processing device according to the present disclosure. FIG. 5 is a block diagram showing a configuration of an information processing system according to the present disclosure. FIG. 6 is a flow diagram showing a flow of an information processing method according to the present disclosure. FIG. 7 is a block diagram showing a configuration of an information processing device according to the present disclosure. FIG. 8 is a diagram showing an example of a medical record and summary according to the present disclosure. FIG. 9 is a diagram showing an example of a causal relationship sequence according to the present disclosure. FIG. 10 is a block diagram showing a configuration of an information processing device according to the present disclosure. FIG. 11 is a block diagram showing a configuration of a computer that functions as an information processing device and an information processing system according to the present disclosure.

[0018] The following are examples of embodiments of the present invention. However, the present invention is not limited to the exemplary embodiments shown below, and various modifications are possible within the scope of the claims. For example, embodiments obtained by appropriately combining the technical means employed in the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, embodiments obtained by appropriately omitting some of the technical means employed in the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, the effects mentioned in the exemplary embodiments shown below are examples of effects expected in the exemplary embodiments, and do not define the scope of the present invention. In other words, embodiments that do not exhibit the effects mentioned in the exemplary embodiments shown below may also be included in the scope of the present invention.

[0019] [First Exemplary Embodiment] A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is a basic form for each of the exemplary embodiments described below. Note that the scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical obstacles arise. Furthermore, each technical means shown in the drawings referenced to explain this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical obstacles arise.

[0020] (Configuration of information processing device 1) The configuration of the information processing device 1 will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the configuration of the information processing device 1. As shown in Fig. 1, the information processing device 1 includes an extraction unit 11, an identification unit 12, and a generation unit 13. In this exemplary embodiment, the extraction unit 11, the identification unit 12, and the generation unit 13 respectively realize an extraction means, an identification means, and a generation means.

[0021] The extraction unit 11 extracts a plurality of named entities from the medical records and supplies the extracted plurality of named entities to the identification unit 12.

[0022] Examples of medical records in this disclosure include a patient's symptoms, a history of tests, diagnoses, and treatments performed on the patient by a medical professional. Other examples of medical records in this disclosure include inspections and diagnoses performed on a building by a building inspector (e.g., checking for cracks in the exterior walls), and a history of building repairs by an architect.

[0023] Although the named entities in the present disclosure are not particularly limited, examples include named entities related to medical treatment. In this case, specific named entities include symptoms, examinations, diagnoses, and treatments.

[0024] The identifying unit 12 identifies the degree of causal relationship between the extracted named entities by referring to the extracted named entities, and supplies the identified degree of causal relationship to the generating unit 13.

[0025] The generating unit 13 generates a summary of the medical record by referring to the identified degree of causal relationship.

[0026] (Effects of information processing device 1) As described above, the information processing device 1 employs a configuration including an extraction unit 11 that extracts multiple named entities from medical records, an identification unit 12 that identifies the degree of causal relationship between the multiple extracted named entities by referring to the multiple extracted named entities, and a generation unit 13 that generates a summary of the medical records by referring to the identified degree of causal relationship.

[0027] Therefore, according to the information processing device 1, a summary is generated by referring to the degree of causal relationship between multiple named entities extracted from medical records, thereby achieving the effect of generating an appropriate summary of medical records.

[0028] (Configuration of Information Processing System 2) The configuration of the information processing system 2 will be described with reference to FIG. 2. FIG. 2 is a block diagram showing the configuration of the information processing system 2. As shown in FIG. 2, the information processing system 2 includes an extraction unit 11, an identification unit 12, and a generation unit 13. In this exemplary embodiment, the extraction unit 11, the identification unit 12, and the generation unit 13 respectively realize an extraction means, an identification means, and a generation means. Also, as shown in FIG. 2, the extraction unit 11, the identification unit 12, and the generation unit 13 are connected to each other via a network N so as to be able to communicate with each other.

[0029] The specific configuration of the network N does not limit this exemplary embodiment, but as an example, a wireless LAN (Local Area Network), a wired LAN, a WAN (Wide Area Network), a public line network, a mobile data communication network, or a combination of these networks can be used.

[0030] The extraction unit 11 extracts a plurality of named entities from the medical records and supplies the extracted plurality of named entities to the identification unit 12.

[0031] The identifying unit 12 identifies the degree of causal relationship between the extracted named entities by referring to the extracted named entities, and supplies the identified degree of causal relationship to the generating unit 13.

[0032] The generating unit 13 generates a summary of the medical record by referring to the identified degree of causal relationship.

[0033] (Effects of Information Processing System 2) As described above, the information processing system 2 employs a configuration including an extraction unit 11 that extracts multiple named entities from medical records, an identification unit 12 that identifies the degree of causal relationship between the multiple extracted named entities by referring to the multiple extracted named entities, and a generation unit 13 that generates a summary of the medical records by referring to the identified degree of causal relationship. Therefore, the information processing system 2 can achieve the same effects as the information processing device 1 described above.

[0034] (Flow of Information Processing Method S1) The flow of information processing method S1 will be described with reference to Fig. 3. Fig. 3 is a flow diagram showing the flow of information processing method S1. As shown in Fig. 3, information processing method S1 includes extraction processing S11, identification processing S12, and generation processing S13.

[0035] (Extraction Process S11) In extraction process S11, the extraction unit 11 extracts a plurality of named entities from the medical records. The extraction unit 11 supplies the extracted plurality of named entities to the identification unit 12.

[0036] In the determination process S12, the determination unit 12 determines the degree of causal relationship between the extracted named entities by referring to the extracted named entities. The determination unit 12 supplies the determined degree of causal relationship to the generation unit 13.

[0037] (Generation Process S13) In the generation process S13, the generation unit 13 generates a summary of the medical record by referring to the identified degree of causal relationship.

[0038] (Effects of Information Processing Method S1) As described above, the information processing method S1 employs a configuration including an extraction process S11 in which the extraction unit 11 extracts multiple named entities from the medical records, an identification process S12 in which the identification unit 12 refers to the extracted multiple named entities and identifies the degree of causal relationship between the multiple named entities, and a generation process S13 in which the generation unit 13 generates a summary of the medical records by referring to the identified degree of causal relationship. Therefore, the information processing method S1 can achieve the same effects as the information processing device 1 described above.

[0039] Second Exemplary Embodiment A second exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0040] (Configuration of information processing device 3) The configuration of the information processing device 3 will be described with reference to Fig. 4. Fig. 4 is a block diagram showing the configuration of the information processing device 3. As shown in Fig. 4, the information processing device 3 includes an acquisition unit 31 and a learning unit 32. The acquisition unit 31 and the learning unit 32 are components that respectively realize an acquisition means and a learning means in this exemplary embodiment.

[0041] The acquisition unit 31 acquires learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records. Examples of medical records are as described above. The acquisition unit 31 supplies the acquired learning data to the learning unit 32.

[0042] The learning unit 32 uses the learning data acquired by the acquisition unit 31 to train a model that refers to a plurality of named entities and identifies the degree of causal relationship between the plurality of named entities.

[0043] (Effects of information processing device 3) As described above, the information processing device 3 employs a configuration including an acquisition unit 31 that acquires learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and a learning unit 32 that uses the learning data acquired by the acquisition unit 31 to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0044] Therefore, the information processing device 3 can learn a model that identifies the degree of causal relationship between multiple named entities, which is referenced when generating a summary. Therefore, the information processing device 3 can produce an effect of being able to generate an appropriate summary of a medical record by referring to the degree of causal relationship between multiple named entities extracted from the medical record.

[0045] (Configuration of Information Processing System 4) The configuration of the information processing system 4 will be described with reference to FIG. 5. FIG. 5 is a block diagram showing the configuration of the information processing system 4. As shown in FIG. 5, the information processing system 4 includes an acquisition unit 31 and a learning unit 32. In this exemplary embodiment, the acquisition unit 31 and the learning unit 32 respectively realize an acquisition means and a learning means. Also, as shown in FIG. 5, the acquisition unit 31 and the learning unit 32 are connected to each other so as to be able to communicate with each other via a network N. The network N is as described above.

[0046] The acquiring unit 31 acquires learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records. The acquiring unit 31 supplies the acquired learning data to the learning unit 32.

[0047] The learning unit 32 uses the learning data acquired by the acquisition unit 31 to train a model that refers to a plurality of named entities and identifies the degree of causal relationship between the plurality of named entities.

[0048] (Effects of Information Processing System 4) As described above, the information processing system 4 employs a configuration including an acquisition unit 31 that acquires training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and a learning unit 32 that trains a model that refers to a plurality of named entities and identifies the degree of causal relationships between the plurality of named entities, using the training data acquired by the acquisition unit 31. Therefore, the information processing system 4 can achieve the same effects as the information processing device 3 described above.

[0049] (Flow of Information Processing Method S3) The flow of information processing method S3 will be described with reference to Fig. 6. Fig. 6 is a flow diagram showing the flow of information processing method S3. As shown in Fig. 6, information processing method S3 includes an acquisition process S31 and a learning process S32.

[0050] In the acquisition process S31, the acquisition unit 31 acquires learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records. The acquisition unit 31 supplies the acquired learning data to the learning unit 32.

[0051] (Learning process S32) In the learning process S32, the learning unit 32 uses the learning data acquired by the acquisition unit 31 to learn a model that refers to a plurality of named entities and identifies the degree of causal relationship between the plurality of named entities.

[0052] (Effects of Information Processing Method S3) As described above, the information processing method S3 employs a configuration including an acquisition process S31 in which the acquisition unit 31 acquires training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records, and a learning process S32 in which the learning unit 32 trains a model that identifies the degree of causal relationships between a plurality of named entities by referring to the plurality of named entities, using the training data acquired by the acquisition unit 31. Therefore, the information processing method S3 can achieve the same effects as the information processing device 3 described above.

[0053] [Third Exemplary Embodiment] A third exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0054] (Information Processing Device 100A) The information processing device 100A will be described with reference to Fig. 7. Fig. 7 is a block diagram showing the configuration of the information processing device 100A.

[0055] The information processing device 100A is a device that generates a summary ABS of a medical record MR. An example of the medical record MR is as described above. As shown in FIG. 7 , the information processing device 100A also functions as an information processing system 1A. As shown in FIG. 7 , the information processing device 100A includes a control unit 10A, a storage unit 20A, a communication unit 50, and an input / output unit 60.

[0056] Data is stored in the memory unit 20A. Examples of data stored in the memory unit 20A include medical records MR, elements EM, causality degrees DCR, causality sequences CRS, summaries ABS, and trained models LM. Examples of medical records MR are as described above. Furthermore, the trained models LM may be configured such that parameters defining the trained models LM are stored in the memory unit 20A.

[0057] 7, the element EM includes at least one of a named entity PR and a word or phrase WD, and one or more words or phrases WD are associated with the named entity PR. Examples of the named entity PR are as described above. The word or phrase WD is a word or phrase described in the item indicated by the named entity PR. As an example, the named entity "symptoms" is associated with the word "suspected lung cancer" and the word "tumor disappearance." As another example, the named entity "examination" is associated with the word "chest X-ray." As shown in FIG. 7, the causal relationship string CRS is associated with a summary ABS. The degree of causal relationship DCR, the causal relationship string CRS, and the learned model LM will be described later.

[0058] Examples of the storage unit 20A include, but are not limited to, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), or a combination thereof.

[0059] The communication unit 50 is an interface that transmits and receives data via a network. For example, the communication unit 50 transmits data supplied from the control unit 10A to other devices, and supplies data received from other devices to the control unit 10A. An example of data transmitted by the communication unit 50 is an abstract ABS. An example of data received by the communication unit 50 is a medical record MR.

[0060] Examples of the communication unit 50 include, but are not limited to, communication chips for various communication standards such as Ethernet (registered trademark), Wi-Fi (registered trademark), and wireless communication standards for mobile data communication networks, and USB-compliant connectors.

[0061] The input / output unit 60 is an interface that receives and outputs data. When the input / output unit 60 receives data, it supplies the received data to the control unit 10A. The input / output unit 60 also outputs data supplied from the control unit 10A.

[0062] Examples of the input / output unit 60 include, but are not limited to, a keyboard, a mouse, a touchpad, a microphone, and a liquid crystal display.

[0063] (Control Unit 10A) The control unit 10A controls each component included in the information processing device 100A.

[0064] 7, the control unit 10A includes an extraction unit 11, an identification unit 12, and a generation unit 13. In this exemplary embodiment, the extraction unit 11, the identification unit 12, and the generation unit 13 respectively realize an extraction means, an identification means, and a generation means.

[0065] The extraction unit 11 extracts a plurality of named entities PR from the medical record MR stored in the storage unit 20A. The extraction unit 11 stores the extracted plurality of named entities PR in the storage unit 20A. The extraction unit 11 also extracts words and phrases WD included in the medical record MR and associated with the named entities PR. The extraction unit 11 stores the extracted words and phrases WD in the storage unit 20A in association with the named entities PR. An example of the processing executed by the extraction unit 11 will be described later.

[0066] The identification unit 12 identifies the degree of causal relationship DCR between the multiple named entities PR by referring to the multiple named entities PR extracted by the extraction unit 11. For example, when the extraction unit 11 extracts the named entities PR1, PR2, and PR3, the identification unit 12 identifies the degree of causal relationship DCR1 between the named entities PR1 and PR2, the degree of causal relationship DCR2 between the named entities PR2 and PR3, and the degree of causal relationship DCR3 between the named entities PR1 and PR3. As an example, the stronger the causal relationship, the larger the value of the degree of causal relationship DCR.

[0067] As an example, the identification unit 12 identifies the degree of causality DCR between multiple named entities PR using a trained model LM that is generated by machine learning with reference to training data and that is stored in the storage unit 20A. The identification unit 12 stores the identified degree of causality DCR in the storage unit 20A.

[0068] Furthermore, the identification unit 12 refers to the degree of causality DCR to identify one or more causal relationship sequences CRS, each of which is composed of a plurality of elements EM that have a causal relationship with one another. The identification unit 12 stores the identified one or more causal relationship sequences CRS in the storage unit 20A. An example of the processing executed by the identification unit 12 will be described later.

[0069] The generation unit 13 generates a summary ABS of the medical record MR by referring to the degree of causality DCR. As an example, the generation unit 13 generates a summary ABS of the medical record MR for each of one or more causal relationship columns CRS stored in the storage unit 20A. For example, the generation unit 13 generates a summary ABS that supports human decision-making. The generation unit 13 associates the generated summary ABS with the referenced causal relationship column CRS and stores it in the storage unit 20A. An example of the processing performed by the generation unit 13 will be described later.

[0070] (Example of Processing Executed by Extraction Unit 11) An example of processing performed by the extraction unit 11 to extract a plurality of named entities PR from a medical record MR will be described with reference to Fig. 8. Fig. 8 is a diagram showing an example of a medical record MR and a summary ABS of a certain patient.

[0071] The extraction unit 11 may extract a plurality of named entities PR from the medical records MR using a trained language processing model. The trained language model may be stored in the storage unit 20A or in a device capable of communicating with the information processing device 100A.

[0072] The trained language model is a model that has been machine-trained to take text written in a natural language such as Japanese or English as input and output multiple named entities contained in the text. An example of a trained language model is BERT (Bidirectional Encoder Representations from Transformers), but is not limited to this.

[0073] The extraction unit 11 inputs the medical record MR into a trained language model, thereby extracting a plurality of named entities PR included in the medical record MR. For example, when the extraction unit 11 inputs the medical record MR shown in the upper left of Fig. 8 into the trained language model, named entities PR1 to PR5 are output from the trained language model, as shown in the lower part of Fig. 8. Furthermore, the trained language model may output words WD1 to WD5 associated with the named entities PR1 to PR5, respectively.

[0074] For example, in Figure 8, since the named entities PR1 and PR5 are both "symptoms," the extraction unit 11 associates the named entity "symptoms" with the phrases WD1 "suspected lung cancer" and WD5 "tumor disappearance," which are associated with the named entities PR1 and PR5, respectively, and stores them in the memory unit 20A.

[0075] With this configuration, the extraction unit 11 can suitably extract a plurality of named entities PR included in the medical record MR.

[0076] (Example 1 of Processing Executed by Identifying Unit 12) An example of processing in which the identifying unit 12 identifies one or more causal relationship sequences CRS will be described with reference to Fig. 9. Fig. 9 is a diagram illustrating an example of a causal relationship sequence CRS.

[0077] For example, if the degree of causality DCR between the named entities PR1 and PR2 is higher than a predetermined value, the identifying unit 12 determines that an element EM1 including the named entity PR1 and an element EM2 including the named entity PR2 have a causal relationship. The identifying unit 12 identifies a plurality of elements EM having a causal relationship as a causal relationship sequence.

[0078] For example, as shown in Fig. 9, it is assumed that the medical record MR includes elements 1 to 7. In this case, the identification unit 12 identifies the degree of causal relationship DCR between each of elements 1 to 7. Next, the identification unit 12 connects elements having a causal relationship in chronological order.

[0079] For example, as shown in Example 1 at the top of Figure 9, if elements 1, 3, 5, and 6 have a causal relationship with each other, the identification unit 12 identifies a causal relationship sequence CRS that connects elements 1, 3, 5, and 6 in chronological order.

[0080] Furthermore, as shown in Example 1 at the top of Figure 9, when elements 2, 4, and 7 have a causal relationship with each other, the identification unit 12 identifies a causal relationship sequence CRS that connects elements 2, 4, and 7 in chronological order.

[0081] As another example, consider a case where element 1 has a causal relationship with each of elements 2 through 7, but elements 2 and 3 do not have a causal relationship, and element 4 does not have a causal relationship with either element 2 or element 3.

[0082] In this case, the identification unit 12 first connects elements 1 and 2, and elements 1 and 3, which have a causal relationship with each other, in chronological order, because element 1 has a causal relationship with element 2 and element 3, but does not have a causal relationship with element 2 and element 3. That is, the identification unit 12 generates a causal relationship sequence CRS that branches from element 1 to element 2 and element 3, as shown in example 2 in the center of Figure 9.

[0083] Next, since element 4 has no causal relationship with either element 2 or element 3, the specifying unit 12 does not include element 4 in any of the causal relationship sequences CRS.

[0084] Next, the identification unit 12 identifies a causal relationship sequence CRS that chronologically connects elements 2, 5, and 6, which have a causal relationship with each other. Similarly, as shown in the center of Figure 9, the identification unit 12 identifies a causal relationship sequence CRS that chronologically connects elements 3 and 7, which have a causal relationship with each other.

[0085] As yet another example, consider a case where element 1 has a causal relationship with elements 3, 4, 5, and 7, element 2 has a causal relationship with elements 6 and 7, elements 1 and 2 do not have a causal relationship, and element 4 does not have a causal relationship with either element 2 or element 3.

[0086] In this case, the identification unit 12 first connects element 1 and element 3, which have a causal relationship with each other, in chronological order, as shown in Example 3 at the bottom of Fig. 9. Next, since element 4 does not have a causal relationship with element 3, the identification unit 12 does not include element 4 in the causal relationship sequence CRS.

[0087] Next, the identification unit 12 chronologically connects elements 3 and 5, and elements 5 and 7, which have a causal relationship with each other. Similarly, the identification unit 12 chronologically connects elements 2 and 6, and elements 6 and 7, which have a causal relationship with each other.

[0088] As described above, the generating unit 13 generates a summary ABS of the medical record MR for each of one or more causal relationship sequences CRS.

[0089] For example, in the causal relationship sequence CRS shown in Example 1 of Figure 9, the generation unit 13 generates a summary ABS corresponding to the causal relationship sequence CRS consisting of elements 1, 3, 5, and 6, and a summary ABS corresponding to the causal relationship sequence CRS consisting of elements 2, 4, and 7.

[0090] Furthermore, for the causal relationship sequence CRS shown in Example 2 of Figure 9, the generation unit 13 generates a summary ABS corresponding to the causal relationship sequence CRS consisting of elements 1, 2, 5, and 6, and a summary ABS corresponding to the causal relationship sequence CRS consisting of elements 1, 3, and 7.

[0091] Furthermore, for the causal relationship sequence CRS shown in Example 3 of Figure 9, the generation unit 13 generates a summary ABS corresponding to the causal relationship sequence CRS consisting of elements 1, 3, 5, and 7, and a summary ABS corresponding to the causal relationship sequence CRS consisting of elements 2, 6, and 7.

[0092] In this way, the identification unit 12 refers to the degree of causality DCR to identify one or more causal relationship sequences CRS composed of multiple elements EM that have a causal relationship with each other, and the generation unit 13 generates a summary ABS of the medical record MR for each of the one or more causal relationship sequences CRS. With this configuration, the information processing device 100A can generate a summary ABS including elements EM that have a causal relationship, and further can generate a summary ABS that takes into account the causal relationships of each of the multiple elements EM included in the medical record MR.

[0093] (Second Example of Processing Executed by Identification Unit 12) Another example of processing performed by the identification unit 12 to identify the degree of causality DCR will be described with reference again to FIG.

[0094] The identifying unit 12 may identify the degree of causality DCR between a plurality of named entities PR by inferring the cause from the result.

[0095] As an example, the identification unit 12 identifies the degree of causal relationship between multiple named entities PR recorded in the medical record MR by performing a process of estimating a named entity PR that is located earlier than a named entity PR that is located later in time or in the sequence on the medical record MR, from among the multiple named entities PR recorded in the medical record MR.

[0096] Here, a named entity PR that is later in time or in the sequence on the medical record MR is denoted as named entity PR1, and a named entity PR that is earlier in time or in the sequence on the medical record MR than the named entity PR1 is denoted as named entity PR0. In this case, the named entity PR1 can correspond to a result caused by the named entity PR0. Conversely, the named entity PR0 can correspond to the cause that induced the result of the named entity PR1.

[0097] Therefore, the above processing by the identification unit 12 can be expressed as processing of "identifying the degree of causal relationship between multiple named entities by inferring the cause from the result."

[0098] For example, assume that named entities PR1 to PR7 are extracted by the extraction unit 11, and the elements containing named entities PR1 to PR7 are set as elements 1 to 7, respectively, as in example 2 shown in the center of Fig. 9. Note that elements 6 and 7 are simultaneous in terms of time or in terms of the sequence on the medical record MR.

[0099] In this case, the identification unit 12 identifies the degree of causality DCR for each of elements 1 to 5, which are located earlier than element 7, from element 7, which is located later in time or in the sequence on the medical record MR. Here, the degree of causality DCR between element 7 and each of elements 2, 4, and 5 is low. Therefore, the identification unit 12 does not identify the degree of causality DCR between element 2, element 4, and element 5 and each of elements 1 and 3.

[0100] Next, the identification unit 12 identifies the degree of causality DCR for each of elements 1 to 5, which are located earlier than element 6, from element 6, which is located later in time or in the sequence on the medical record MR. Here, the degree of causality DCR between element 6 and each of elements 3 and 4 is low. Therefore, the identification unit 12 does not identify the degree of causality DCR between element 3 and element 4 and each of elements 1 and 2.

[0101] Here, element 4 has no causal relationship with either element 6 or element 7, which are located later in time or in the sequence on the medical record MR, and therefore the identifying unit 12 does not need to identify the degree DCR of the causal relationship between element 4 and other elements.

[0102] On the other hand, suppose that the identification unit 12 identifies the degree of causal relationship DCR for each of element 1, which is located earlier in time or in the sequence on the medical record MR, and elements 2 to 7, which are located later than element 1.

[0103] In this case, since element 1 has a causal relationship with each of elements 2 to 7, the identification unit 12 next assumes that it identifies the degree of causal relationship DCR for each of elements 4 to 7 from each of elements 2 and 3, which are located next to element 1 in terms of time or in the sequence on the medical record MR.

[0104] Thus, even though element 4 has no causal relationship with elements 6 and 7, which are the last elements in the sequence in time or on the medical record MR, the identification unit 12 needs to identify the degree DCR of the causal relationship between element 4 and other elements.

[0105] Therefore, in a configuration in which the identification unit 12 identifies the degree of causality DCR between multiple named entities PR by inferring the cause from the result, there is no need to identify unnecessary degrees of causality DCR, thereby reducing the processing load.

[0106] (Example 3 of Processing Executed by Identification Unit 12) Yet another example of processing in which the identification unit 12 identifies the degree of causality DCR will be described.

[0107] As described above, the identification unit 12 may identify the degree of causal relationship DCR between multiple named entities PR using a trained model LM that is stored in the memory unit 20A and is trained by machine learning using training data.

[0108] As an example of the training data in this configuration, the training data may include training data including labels regarding causal relationships assigned by a human to at least one of the medical records MR and the summary ABS of the medical records MR. For example, a medical professional assigns labels regarding causal relationships to at least one of the medical records MR and the summary ABS of the medical records MR. Furthermore, the labels regarding causal relationships may be labels indicating the presence or absence of a causal relationship (binary labels) or may be labels indicating the degree of causal relationship in three or more levels (multiple-valued labels of three or more).

[0109] With this configuration, the identification unit 12 identifies the degree of causal relationship DCR between multiple named entities PR using a trained model LM that has been trained using training data to which labels have been assigned by a person with specialized knowledge for highly specialized medical records MR and summaries ABS. Therefore, the identification unit 12 can suitably identify the degree of causal relationship DCR between multiple named entities PR.

[0110] As another example of the training data in this configuration, the training data may include training data including labels relating to causal relationships mechanically assigned to a summary ABS created by a human from the medical record MR. For example, a label relating to causal relationships is assigned to a summary ABS created by a human from the medical record MR using a rule-based algorithm that outputs a named entity PR2 that has a causal relationship with a named entity PR1.

[0111] With this configuration, training data using human-created summary ABSs is mechanically generated, and the identification unit 12 can use a trained model LM that has been trained using a large amount of training data generated from appropriate summary ABSs. Therefore, the identification unit 12 can suitably identify the degree of causality DCR between multiple named entities PR.

[0112] In this way, the identification unit 12 identifies the degree of causality DCR between a plurality of named entities PR by using the trained model LM based on machine learning that references the training data. Therefore, the identification unit 12 can suitably identify the degree of causality DCR between a plurality of named entities PR.

[0113] (Example of Processing Executed by Generator 13) An example of processing performed by the generator 13 to generate a summary ABS of a medical record MR by referring to the degree of causality DCR will be described with reference to Figs. 8 and 9 again.

[0114] As an example, the generation unit 13 generates a summary ABS of the medical record MR using a large language model (LLM).

[0115] As an example of this configuration, the generation unit 13 first extracts a plurality of sentences including named entities with a high degree of causality DCR identified by the identification unit 12 (e.g., named entities whose value of the degree of causality DCR is equal to or greater than a predetermined value). Next, the generation unit 13 sets the extracted plurality of sentences as an input prompt for the large-scale language model. For example, the generation unit 13 may itemize the extracted plurality of sentences and use the itemized list as an input prompt instructing the user to summarize the sentences. The generation unit 13 may also set the number of characters after the summary as the prompt.

[0116] The generation unit 13 generates a summary ABS of the medical record MR by inputting the input prompt into the large-scale language model and obtaining a summary from the large-scale language model.

[0117] As another example, the generation unit 13 generates a summary ABS of the medical record MR by referring to the causal relationship sequence CRS generated based on the degree of causal relationship DCR. For example, as shown in Example 1 in the upper part of Figure 9, the generation unit 13 refers to the causal relationship sequence CRS and acquires element 1, element 3, element 5, and element 6.

[0118] As an example, the generation unit 13 generates a summary corresponding to the causal relationship sequence CRS using a summary template stored in the storage unit 20A (not shown in Fig. 7). The summary template is a template for generating a sentence in which the words and phrases WD included in the elements EM are appropriately used. That is, the generation unit 13 references the summary template and generates a summary ABS that uses the words and phrases WD included in elements 1, 3, 5, and 6.

[0119] As yet another example, the generation unit 13 generates a summary corresponding to the causal relationship sequence CRS by referring to a reference document stored in the storage unit 20A (not shown in FIG. 7 ). An example of the reference document is a summary ABS created by a person. In this case, the generation unit 13 extracts sentences from the reference document that include the words WD contained in element 1, element 3, element 5, and element 6. Then, the generation unit 13 uses the sentences extracted from the reference document to generate a summary ABS that uses the words WD contained in element 1, element 3, element 5, and element 6.

[0120] With this configuration, the generating unit 13 can generate a summary ABS of the medical record MR, for example, as shown in the upper right of FIG.

[0121] (Effects of information processing device 100A) As described above, in the information processing device 100A, multiple named entities PR are extracted from the medical record MR, and by referring to the extracted multiple named entities PR, the degree of causal relationship DCR between the multiple named entities PR is identified, and by referring to the degree of causal relationship DCR, a summary ABS of the medical record MR is generated.

[0122] Therefore, the information processing device 100A generates a summary ABS based on the causal relationships of multiple named entities PR included in the medical record MR, and therefore can generate an appropriate summary ABS of the medical record MR.

[0123] [Fourth Exemplary Embodiment] A fourth exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be denoted by the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technical means employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technical means employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technical means shown in each drawing referenced to describe this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0124] (Information Processing Device 300A) The information processing device 300A will be described with reference to Fig. 10. Fig. 10 is a block diagram showing the configuration of the information processing device 300A.

[0125] The information processing device 300A is a device that uses training data TD to train a trained model LM that identifies the degree of causal relationship between multiple named entities PR. The named entities PR, the trained model LM, and the training data TD are as described above. As shown in FIG. 10 , the information processing device 300A also includes a control unit 30A, a memory unit 40A, a communication unit 50, and an input / output unit 60. The communication unit 50 and the input / output unit 60 are as described above.

[0126] 10 , the information processing device 100A functions as at least a part of an information processing system 3A. More specifically, the information processing system 3A is configured to include an information processing device 300A and an information processing device 100A. As an example, in the information processing system 3A, the information processing device 300A uses a trained model LM to identify the degree of causality DCR between multiple named entities PR output from the information processing device 100A.

[0127] Data is stored in the storage unit 40A. Examples of the data stored in the storage unit 40A include training data TD1, training data TD2, and a trained model LM. Examples of the storage unit 40A include, but are not limited to, a flash memory, an HDD, an SSD, or a combination thereof.

[0128] 10, the training data TD1 includes medical records MR containing named entities PR and causal relationship labels CRL, which will be described later. The training data TD2 includes summaries ABS containing named entities PR and causal relationship labels CRL.

[0129] (Control Unit 30A) The control unit 30A controls each component included in the information processing device 300A.

[0130] 10, the control unit 30A includes an acquisition unit 31, a learning unit 32, an extraction unit 11, and a label assignment unit 33. In this exemplary embodiment, the acquisition unit 31 and the learning unit 32 respectively realize an acquisition means and a learning means.

[0131] The acquisition unit 31 acquires learning data TD including causal relationship labels CRL relating to causal relationships assigned to at least one of the medical records MR and the summary ABS of the medical records MR. The acquisition unit 31 stores at least one of the medical records MR and the summary ABS of the medical records MR, and the acquired causal relationship labels CRL in the storage unit 40A.

[0132] The learning unit 32 uses the training data TD to train a learned model LM that refers to a plurality of named entities PR and identifies the degree of causality DCR between the plurality of named entities PR.

[0133] As an example, the training data TD includes training data including causal relationship labels CRL assigned by a human to at least one of the medical records MR and the summaries ABS of the medical records MR.

[0134] With this configuration, the learning unit 32 can train the learned model LM using learning data that has been labeled by a person with specialized knowledge for highly specialized medical records MR and summary ABS.

[0135] As another example, the training data TD includes training data including causal relationship labels CRL mechanically assigned to summaries ABS created by a person from medical records MR.

[0136] With this configuration, training data is mechanically generated using human-created summary ABSs, so that the training unit 32 can train the trained model LM using a large amount of training data generated from appropriate summary ABSs.

[0137] The extraction unit 11 extracts named entities PR from the medical records MR. Examples of the processing executed by the extraction unit 11 are as described above. The extraction unit 11 stores the extracted named entities PR in the storage unit 40A.

[0138] The label assignment unit 33 assigns a causal relationship label CRL. As one example, the label assignment unit 33 refers to operation information acquired via the input / output unit 60 and assigns the causal relationship label CRL to at least one of the medical record MR and the summary ABS of the medical record MR. As another example, the label assignment unit 33 mechanically assigns the causal relationship label CRL to the summary ABS created manually from the medical record MR. An example of the process in which the label assignment unit 33 mechanically assigns the causal relationship label CRL is as described above.

[0139] (Effects of the information processing device 300A) As described above, the information processing device 300A uses training data TD including a causal relationship label CRL assigned to at least one of the medical record MR and the summary ABS of the medical record MR to train a learned model LM that refers to multiple named entities PR and identifies the degree of causal relationship DCR between the multiple named entities PR.

[0140] Therefore, the information processing device 300A can preferably train the trained model LM that identifies the degree of causality DCR between multiple named entities PR. Furthermore, because the information processing device 300A can preferably train the trained model LM, the information processing device 100A can generate a summary ABS of the medical record MR that references the degree of causality DCR.

[0141] [Example of implementation by software] Some or all of the functions of the information processing devices 1, 3, 100A, 300A (hereinafter also referred to as "the above-mentioned devices") and the information processing systems 1A, 2, 3A, 4 (hereinafter also referred to as "the above-mentioned systems") may be implemented by hardware such as an integrated circuit (IC chip), or by software.

[0142] In the latter case, the above-mentioned devices and systems are realized, for example, by a computer that executes instructions from a program, which is software that realizes the functions. An example of such a computer (hereinafter referred to as computer C) is shown in Figure 11. Figure 11 is a block diagram showing the hardware configuration of computer C that functions as the above-mentioned devices and systems.

[0143] The computer C includes at least one processor C1 and at least one memory C2. The memory C2 stores a program P for causing the computer C to function as each of the above-mentioned devices and systems. In the computer C, the processor C1 reads and executes the program P from the memory C2, thereby realizing the functions of each of the above-mentioned devices and systems.

[0144] The processor C1 may be, for example, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a micro processing unit (MPU), a floating point number processing unit (FPU), a physics processing unit (PPU), a tensor processing unit (TPU), a quantum processor, a microcontroller, or a combination thereof. The memory C2 may be, for example, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), or a combination thereof.

[0145] The computer C may further include a RAM (Random Access Memory) for expanding the program P during execution and for temporarily storing various data. The computer C may also include a communication interface for transmitting and receiving data to and from other devices. The computer C may also include an input / output interface for connecting input / output devices such as a keyboard, a mouse, a display, and a printer.

[0146] The program P can also be recorded on a non-transitory, tangible recording medium M that can be read by the computer C. Such a recording medium M can be, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit. The computer C can acquire the program P via such a recording medium M. The program P can also be transmitted via a transmission medium. Such a transmission medium can be, for example, a communications network or broadcast waves. The computer C can also acquire the program P via such a transmission medium.

[0147] [Appendix A] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0148] (Appendix A1) An information processing device comprising: an extraction means for extracting a plurality of named entities from a medical record; an identification means for identifying the degree of causal relationship between the plurality of named entities by referring to the extracted plurality of named entities; and a generation means for generating a summary of the medical record by referring to the degree of causal relationship.

[0149] (Appendix A2) The information processing device described in Appendix A1, wherein the identification means identifies one or more causal relationship sequences consisting of multiple elements that have a causal relationship with each other by referring to the degree of causal relationship, and the generation means generates a summary of the medical record for each of the one or more causal relationship sequences.

[0150] (Supplementary Note A3) The information processing device according to Supplementary Note A1 or A2, wherein the specifying means specifies a degree of causal relationship between the plurality of named entities by inferring a cause from a result.

[0151] (Supplementary Note A4) The information processing device according to any one of Supplementary Notes A1 to A3, wherein the identifying means identifies a degree of causal relationship between the plurality of named entities using a model trained by machine learning with reference to training data.

[0152] (Appendix A5) The information processing device according to any one of Appendices A1 to A4, wherein the learning data includes learning data including labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

[0153] (Appendix A6) The information processing device according to any one of Appendices A1 to A5, wherein the training data includes training data including labels relating to causal relationships mechanically assigned to summaries created by a person from medical records, and the generating means generates the summaries to support the person in making decisions.

[0154] (Appendix A7) An information processing device comprising: an acquisition means for acquiring learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit for using the learning data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0155] (Supplementary Note A8) The information processing device according to Supplementary Note A7, wherein the learning data includes learning data including labels relating to causal relationships assigned by a person to at least one of medical records and summaries of the medical records.

[0156] (Supplementary Note A9) The information processing device according to Supplementary Note A7 or A8, wherein the training data includes training data including labels relating to causal relationships mechanically assigned to summaries manually created from medical records.

[0157] (Appendix A10) An information processing system comprising: an extraction means for extracting a plurality of named entities from medical records; an identification means for identifying the degree of causal relationship between the plurality of named entities by referring to the extracted plurality of named entities; and a generation means for generating a summary of the medical records by referring to the degree of causal relationship.

[0158] (Appendix A11) An information processing system comprising: an acquisition means for acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit for using the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0159] [Appendix B] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0160] (Appendix B1) An information processing method executed by a computer, comprising: extracting a plurality of named entities from medical records; referring to the extracted plurality of named entities to identify the degree of causal relationship between the plurality of named entities; and generating a summary of the medical records by referring to the degree of causal relationship.

[0161] (Appendix B2) An information processing method as described in Appendix B1, wherein, in the identifying step, the computer identifies one or more causal relationship sequences consisting of multiple elements that have a causal relationship with each other by referring to the degree of causality; and, in the generating step, the computer generates a summary of the medical record for each of the one or more causal relationship sequences.

[0162] (Supplementary Note B3) The information processing method according to Supplementary Note B1 or B2, wherein in the identifying, the computer identifies a degree of a causal relationship between the plurality of named entities by inferring a cause from a result.

[0163] (Appendix B4) The information processing method according to any one of Appendices B1 to B3, wherein in the identifying, the computer identifies the degree of causal relationship between the plurality of named entities using a model trained by machine learning with reference to training data.

[0164] (Appendix B5) The information processing method according to any one of Appendices B1 to B4, wherein the training data includes training data including labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

[0165] (Appendix B6) The information processing method of any one of Appendices B1 to B5, wherein the training data includes training data including labels relating to causal relationships that have been mechanically assigned to summaries created by a person from medical records, and in the generating, the computer generates the summaries to support the person in making decisions.

[0166] (Appendix B7) An information processing method including: the computer acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and using the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0167] (Supplementary Note B8) The information processing method according to Supplementary Note B7, wherein the training data includes training data including labels relating to causal relationships assigned by a person to at least one of medical records and summaries of the medical records.

[0168] (Supplementary Note B9) The information processing method according to Supplementary Note B7 or B8, wherein the training data includes training data including labels relating to causal relationships mechanically assigned to summaries created by a person from medical records.

[0169] [Appendix C] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0170] (Appendix C1) A program that causes a computer to function as an information processing device, the program causing the computer to perform the following: extracting multiple named entities from medical records; referring to the extracted multiple named entities to identify the degree of causal relationship between the multiple named entities; and generating a summary of the medical records by referring to the degree of causal relationship.

[0171] (Appendix C2) The program described in Appendix C1, wherein the program causes the computer to: in the identifying step, identify one or more causal relationship sequences consisting of multiple elements that have a causal relationship with each other by referring to the degree of causality; and in the generating step, generate a summary of the medical record for each of the one or more causal relationship sequences.

[0172] (Appendix C3) The program according to appendix C1 or C2, wherein the program causes the computer to, in the identifying, identify a degree of causal relationship between the plurality of named entities by inferring a cause from a result.

[0173] (Appendix C4) The program according to any one of Appendices C1 to C3, wherein the program causes the computer to, in the identifying step, identify the degree of causal relationship between the plurality of named entities using a trained model through machine learning that references training data.

[0174] (Supplementary Note C5) The program according to any one of Supplementary Notes C1 to C4, wherein the training data includes training data including labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

[0175] (Appendix C6) The program of any one of Appendices C1 to C5, wherein the training data includes training data containing labels relating to causal relationships that have been mechanically assigned to summaries created by a person from medical records, and the program causes the computer to generate the summaries that, in generating them, assist the person in making decisions.

[0176] (Appendix C7) The program causes the computer to execute the following steps: acquire training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and use the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0177] (Supplementary Note C8) The program according to Supplementary Note C7, wherein the training data includes training data including labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

[0178] (Appendix C9) The program according to appendix C7 or C8, wherein the training data includes training data including labels relating to causal relationships mechanically assigned to summaries created by a person from medical records.

[0179] [Appendix D] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0180] (Appendix D1) An information processing device comprising at least one processor that executes: an extraction process that extracts a plurality of named entities from medical records; an identification process that refers to the extracted plurality of named entities to identify the degree of causal relationship between the plurality of named entities; and a generation process that generates a summary of the medical records by referring to the degree of causal relationship.

[0181] (Appendix D2) The information processing device described in Appendix D1, wherein in the identification process, the at least one processor identifies one or more causal relationship sequences consisting of multiple elements that have a causal relationship with each other by referring to the degree of causality, and in the generation process, the at least one processor generates a summary of the medical record for each of the one or more causal relationship sequences.

[0182] (Supplementary Note D3) The information processing device according to Supplementary Note D1 or D2, wherein in the identification process, the at least one processor identifies a degree of causal relationship between the plurality of named entities by inferring a cause from a result.

[0183] (Appendix D4) The information processing device according to any one of Appendices D1 to D3, wherein in the identification process, the at least one processor identifies the degree of causal relationship between the plurality of named entities using a model trained by machine learning with reference to training data.

[0184] (Appendix D5) The information processing device according to any one of appendices D1 to D4, wherein the learning data includes learning data including labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

[0185] (Appendix D6) The information processing device of any one of Appendices D1 to D5, wherein the training data includes training data including labels relating to causal relationships mechanically assigned to summaries created by a person from medical records, and in the generation process, the at least one processor generates the summaries to support the person's decision-making.

[0186] (Appendix D7) An information processing device, wherein the at least one processor further executes: an acquisition process for acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit for using the training data to train a model that identifies the degree of causal relationships between multiple named entities by referring to the multiple named entities.

[0187] (Supplementary Note D8) The information processing device according to Supplementary Note D7, wherein the learning data includes learning data including labels relating to causal relationships assigned by a person to at least one of medical records and summaries of the medical records.

[0188] (Supplementary Note D9) The information processing device according to Supplementary Note D7 or D8, wherein the training data includes training data including labels relating to causal relationships mechanically assigned to summaries created by a person from medical records.

[0189] (Appendix D10) An information processing system comprising at least one processor that executes: an extraction process that extracts a plurality of named entities from medical records; an identification process that refers to the extracted plurality of named entities to identify the degree of causal relationship between the plurality of named entities; and a generation process that generates a summary of the medical records by referring to the degree of causal relationship.

[0190] (Appendix D11) An information processing system comprising at least one processor, the at least one processor comprising: an acquisition process for acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit for using the training data to train a model that identifies the degree of causal relationship between multiple named entities by referring to the multiple named entities.

[0191] [Appendix E] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0192] (Appendix E1) A non-transitory recording medium having recorded thereon a program that causes a computer to function as an information processing device, the program causing the computer to execute the following: an extraction process that extracts multiple named entities from medical records; an identification process that refers to the extracted multiple named entities and identifies the degree of causal relationship between the multiple named entities; and a generation process that generates a summary of the medical records by referring to the degree of causal relationship.

[0193] (Appendix E1) A non-transitory recording medium having recorded thereon a program that causes a computer to function as an information processing device, the program causing the computer to execute: an acquisition process that acquires training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit that uses the training data to train a model that refers to a plurality of named entities and identifies the degree of causal relationships between the plurality of named entities.

[0194] 1, 3, 100A, 300A Information processing device 1A, 2, 3A, 4 Information processing system 11 Extraction unit 12 Identification unit 13 Generation unit 31 Acquisition unit 32 Learning unit ABS Summary CRL Causal relationship label CRS Causal relationship sequence DCR Degree of causal relationship EM Element LM Trained model MR Medical record PR Named entity TD Training data WD Word

Claims

1. An information processing device comprising: an extraction means for extracting a plurality of named entities from medical records; an identification means for identifying the degree of causal relationship between the plurality of named entities by referring to the extracted plurality of named entities; and a generation means for generating a summary of the medical records by referring to the degree of causal relationship.

2. The information processing device according to claim 1, wherein the identification means identifies one or more causal relationship strings consisting of multiple elements having a causal relationship with each other by referring to the degree of causal relationship, and the generation means generates a summary of the medical record for each of the one or more causal relationship strings.

3. The information processing device according to claim 1 or 2, wherein the identification means identifies the degree of causal relationship between the plurality of named entities by inferring a cause from a result.

4. An information processing device according to any one of claims 1 to 3, wherein the identification means identifies the degree of causal relationship between the plurality of named entities using a trained model through machine learning with reference to training data.

5. The information processing device according to claim 4, wherein the learning data includes learning data containing labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

6. The information processing device according to claim 4, wherein the training data includes training data containing labels relating to causal relationships mechanically assigned to summaries created by a person from medical records, and the generating means generates the summaries to assist the person in making decisions.

7. An information processing device comprising: an acquisition means for acquiring learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit for using the learning data to train a model that identifies the degree of causal relationships between multiple named entities by referring to the multiple named entities.

8. The information processing device according to claim 7, wherein the learning data includes learning data containing labels relating to causal relationships assigned by a human to at least one of medical records and summaries of the medical records.

9. The information processing device according to claim 7 or 8, wherein the training data includes training data containing labels relating to causal relationships mechanically assigned to manually created summaries of medical records.

10. An information processing system comprising: an extraction means for extracting multiple named entities from medical records; an identification means for identifying the degree of causal relationship between the multiple named entities by referring to the extracted multiple named entities; and a generation means for generating a summary of the medical records by referring to the degree of causal relationship.

11. An information processing system comprising: an acquisition means for acquiring learning data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and a learning unit for using the learning data to train a model that identifies the degree of causal relationships between multiple named entities by referencing the multiple named entities.

12. An information processing method implemented by a computer, comprising: extracting a plurality of named entities from medical records; referring to the extracted plurality of named entities to determine a degree of causal relationship between the plurality of named entities; and generating a summary of the medical records by referring to the degree of causal relationship.

13. An information processing method executed by a computer, comprising: acquiring training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and using the training data to train a model that references a plurality of named entities and identifies the degree of causal relationships between the plurality of named entities.

14. A program that causes a computer to function as an information processing device, the program causing the computer to perform the following operations: extracting a plurality of named entities from medical records; referring to the extracted plurality of named entities to determine the degree of causal relationship between the plurality of named entities; and generating a summary of the medical records by referring to the degree of causal relationship.

15. A program that causes a computer to function as an information processing device, the program causing the computer to execute the following steps: acquire training data including labels relating to causal relationships assigned to at least one of medical records and summaries of the medical records; and use the training data to train a model that refers to a plurality of named entities and identifies the degree of causal relationships between the plurality of named entities.

Images