Information processing device, information processing method, and program

Abstract

By utilizing generative models, we can perform patient-related estimations with high accuracy. [Solution] The information processing device comprises: acquisition means for acquiring medical information of one or more subjects; structuring means for generating structured medical information by structuring at least a portion of the medical information; first generation means for generating input information including the structured medical information and queries; and second generation means for generating answers to the queries concerning the subjects by referring to the output from a generation model into which the input information has been input.

Description

Technical Field

[0001] This disclosure relates to an information processing apparatus, an information processing method, and a program.

Background Art

[0002] Techniques for making inferences related to patients using machine learning techniques are known. For example, Non-Patent Document 1 discloses a technique for matching clinical trials and patients using a large language model (LLM).

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Generally, as the outpatient period or inpatient period of a patient becomes longer, the medical information related to the patient also increases. In the technique described in Non-Patent Document 1, there is a problem that the estimation accuracy tends to decrease because the medical information of the patient is directly input into the large language model.

[0005] This disclosure has been made in view of the above problems, and an exemplary object thereof is to provide a technique capable of making inferences related to patients with high accuracy while using a generative model.

Means for Solving the Problems

[0006] An information processing device relating to an illustrative aspect of this disclosure includes: acquisition means for acquiring medical information of one or more subjects; structuring means for generating structured medical information by structuring at least a portion of the medical information; first generation means for generating input information including the structured medical information and queries; and second generation means for generating answers to the queries concerning the subjects by referring to the output from a generation model into which the input information has been input.

[0007] An example of an information processing method relating to this disclosure includes one or more processors acquiring medical information of one or more subjects, generating structured medical information by structuring at least a portion of the medical information, generating input information including the structured medical information and queries, and generating answers to the queries concerning the subjects by referring to the output from a generative model into which the input information has been input.

[0008] An exemplary aspect of the present disclosure is a program that causes a computer to function as an information processing device, wherein the computer functions as an acquisition means for acquiring medical information of one or more subjects; a structuring means for generating structured medical information by structuring at least a portion of the medical information; a first generation means for generating input information including the structured medical information and queries; and a second generation means for generating answers to the queries concerning the subjects by referring to the output from a generation model into which the input information has been input. [Effects of the Invention]

[0009] One illustrative aspect of this disclosure demonstrates the effect of being able to perform patient-related estimations with high accuracy while utilizing generative models. [Brief explanation of the drawing]

[0010] [Figure 1] This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 2] This is a flowchart showing the flow of the information processing method related to this disclosure. [Figure 3] This is a block diagram showing the configuration of the information processing system related to this disclosure. [Figure 4] This is a diagram illustrating the information referenced by the information processing system related to this disclosure. [Figure 5] A flowchart illustrating an example of the processing flow in the information processing system related to this disclosure. [Figure 6] This diagram illustrates an example of data processing in the information processing system related to this disclosure. [Figure 7] This diagram illustrates an example of data processing in the information processing system related to this disclosure. [Figure 8] This diagram illustrates an example of data processing in the information processing system related to this disclosure. [Figure 9] This diagram illustrates an example of data processing in the information processing system related to this disclosure. [Figure 10] This diagram illustrates an example of data processing in the information processing system related to this disclosure. [Figure 11] This figure shows an example of how the information processing system related to this disclosure displays the information. [Figure 12] A flowchart illustrating an example of the processing flow in the information processing system related to this disclosure. [Figure 13] This is a block diagram showing the configuration of the information processing system related to this disclosure. [Figure 14] This is a block diagram showing the configuration of a computer that functions as an information processing device related to this disclosure. [Modes for carrying out the invention]

[0011] Hereinafter, embodiments of the present invention will be exemplified. However, the present invention is not limited to the following exemplary embodiments, and various modifications are possible within the scope shown in the claims. For example, embodiments obtained by appropriately combining the technologies (part or all of the objects or methods) employed in the following exemplary embodiments may also be included in the scope of the present invention. In addition, embodiments obtained by appropriately omitting a part of the technologies employed in the following exemplary embodiments may also be included in the scope of the present invention. Also, the effects mentioned in the following exemplary embodiments are merely examples of the effects expected in those exemplary embodiments and do not define the extension of the present invention. That is, embodiments that do not exhibit the effects mentioned in the following exemplary embodiments may also be included in the scope of the present invention.

[0012] 〔First 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 later. Note that the scope of application of each technology employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology employed in this exemplary embodiment can also be employed in other exemplary embodiments included in this disclosure as long as there are no particular technical obstacles. Also, each technology shown in the drawings referred to for explaining this exemplary embodiment can also be employed in other exemplary embodiments included in this disclosure as long as there are no particular technical obstacles.

[0013] (Configuration of Information Processing Apparatus 1) The configuration of the information processing apparatus 1 according to this exemplary embodiment will be described with reference to FIG. 1. FIG. 1 is a block diagram showing the configuration of the information processing apparatus 1. As shown in FIG. 1, the information processing apparatus 1 includes an acquisition unit 11, a structuring unit 12, a first generation unit 13, and a second generation unit 14.

[0014] (Acquisition Unit 11) The acquisition unit 11 acquires the medical information of one or more target persons. Here, the medical information may include various types of information extracted from the medical records (including electronic medical records) of the target person (target patient). As an example, the medical information may include information such as the initial consultation interview, progress record, radiology report, and nursing record of the target person. However, these examples do not limit this exemplary embodiment.

[0015] (Structuring unit 12) The structuring unit 12 generates structured medical information by structuring at least a part of the medical information acquired by the acquisition unit 11. The specific structuring process by the structuring unit 12 does not limit this exemplary embodiment, but as an example, the following processes may be performed. · Example 1: Extract a plurality of entities from at least a part of the medical information, and generate one or more triplets including the plurality of entities and the relationships between the plurality of entities as the structured medical information. · Example 2: Extract a plurality of entities from at least a part of the medical information, and generate a graph (graph structure) including the plurality of entities as nodes as the structured medical information. · Example 3: Extract a plurality of entities from at least a part of the medical information, and generate a table (tabular form) including the plurality of entities as data items as the structured medical information.

[0016] (First generation unit 13) The first generation unit 13 generates input information including the structured medical information and a query. Here, the input information is the information input to the generation model described later. The input information may also be expressed as a prompt or the like. Also, the structured medical information is generated by the structuring unit 12 as described above. On the other hand, the query may be predetermined or may be based on the information acquired by the acquisition unit 11.

[0017] For example, the acquisition unit 11 may acquire standard information relating to the clinical trial, and the first generation unit 13 may generate the input information including the standard information as a query. Alternatively, the acquisition unit 11 may acquire standard information relating to the clinical trial, the structuring unit 12 may generate structured standard information by structuring at least a part of the standard information, and the first generation unit 13 may generate the input information including the structured standard information as a query.

[0018] Furthermore, the first generation unit 13 may be configured to include text extracted from the medical record (electronic medical record) or the medical information in the input information as is (i.e., without structuring it by the structuring unit 12). For example, the first generation unit 13 may include a sentence extracted from the medical information, such as "Because of the finding of xxx, examination yyy will be performed at the next consultation," in the input information as is without structuring it.

[0019] (Second generation unit 14) The second generation unit 14 generates a response to the query regarding the subject by referring to the output from the generation model that has received the input information. Here, the generation model may be, for example, a machine learning-prepared language model such as a Large Language Model (LLM), a generation model that utilizes a graph database, or any other model. Here, GraphRAG (Graph Retrieval Augmented Generation) can be given as an example of a generation model that utilizes a graph database, but this is not limited to this exemplary embodiment. Furthermore, the generation model may be, for example, a configuration provided by the information processing device 1, or a configuration provided by another device that is communicatively connected to the information processing device 1.

[0020] Furthermore, the second generation unit 14 may either use the output from the generation model that receives the input information as the answer, or it may be configured to generate the answer by processing the output from the generation model that receives the input information.

[0021] The second generation unit 14 may, as an example, refer to the output from the generation model into which the input information has been input, and generate information as the response regarding the extent to which the subject is suitable for the clinical trial.

[0022] (Effects of Information Processing Device 1) As described above, in the information processing device 1, • Obtain medical information from one or more subjects, • Structured medical information is generated by structuring at least a portion of the aforementioned medical information. • Generate input information including the structured medical information and queries, • Refer to the output from the generative model that has received the input information and generate a response to the query regarding the subject. This configuration is adopted. In this way, the information processing device 1 generates structured medical information from medical information and generates a response by referring to the output from a generative model into which the input information including the structured medical information has been input, so that estimations related to the subject (patient) can be made with high accuracy.

[0023] (Information processing method S1 flow) Next, the flow of the information processing method S1 according to this exemplary embodiment will be explained with reference to Figure 2. Figure 2 is a flowchart showing the flow of the information processing method S1. As shown in Figure 2, the information processing method S1 includes a step (processing) S11 for acquiring medical information, a step (processing) S12 for generating structured medical information from the medical information, a step (processing) S13 for generating input information including structured medical information and queries, and a step (processing) S14 for generating answers to queries.

[0024] (Step S11) In step S11, the acquisition unit 11 acquires medical information for one or more subjects. A more detailed explanation of the acquisition unit 11 has been given above, so it will be omitted here.

[0025] (Step S12) In step S12, the structuring unit 12 generates structured medical information by structuring at least a portion of the medical information acquired by the acquisition unit 11 in step S11. A more detailed explanation of the structuring unit 12 has been given above, so it will be omitted here.

[0026] (Step S13) In step S13, the first generation unit 13 generates input information including the structured medical information and the query. A more detailed explanation of the first generation unit 13 has been given above, so it will be omitted here.

[0027] (Step S14) In step S14, the second generation unit 14 generates a response to the query regarding the subject by referring to the output from the generation model into which the input information was received. A more detailed explanation of the second generation unit 14 has been given above, so it will be omitted here.

[0028] (Effects of information processing method S1) As described above, in the information processing method S1, • Obtain medical information from one or more subjects, • Structured medical information is generated by structuring at least a portion of the aforementioned medical information. • Generate input information including the structured medical information and queries, • Refer to the output from the generative model that has received the input information and generate a response to the query regarding the subject. This configuration is employed. According to the above configuration, the same effect as that of the information processing device 1 is achieved.

[0029] [Second 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 function as those described in the above-described exemplary embodiment are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise.

[0030] (Configuration of Information Processing System 100A) The configuration of the information processing system 100A according to this exemplary embodiment will be described with reference to Figure 3. Figure 3 is a block diagram showing the configuration of the information processing system 100A. As shown in Figure 3, the information processing system 100A comprises an information processing device 1A and a medical record management device 50 and a server device 60 connected to the information processing device 1A via a network N. Here, the specific configuration of the network N is not limited to this exemplary embodiment, but as an example, a wireless LAN (Local Area Network), a wired LAN, a WAN (Wide Area Network), a public telephone network, a mobile data communication network, or a combination of these networks can be used. Note that the configuration shown in Figure 3 is merely an example. For example, the information processing device 1A may have a configuration in which some or all of the configurations of the medical record management device 50 and the server device 60 are provided. More specifically, the information processing device 1A may have a configuration in which at least one of the electronic medical record and the generation model GM described later is stored in the storage unit 20A described later.

[0031] (Medical record management device 50) The medical record management device 50 manages the electronic medical records of multiple subjects (patients, clinical trial candidates). Each subject's electronic medical record contains their medical information. Each patient's electronic medical record or medical information is acquired and referenced by the information processing device 1A.

[0032] (Server device 60) As shown in Figure 3, the server device 60 comprises a control unit 61, a storage unit 62, and a communication unit 63. The communication unit 63 communicates with devices outside the server device 60. For example, the communication unit 63 communicates with the information processing device 1A of the information processing system 100A. The communication unit 63 transmits data supplied from the control unit 61 to the information processing device 1A and supplies data received from the information processing device 1A to the control unit 61. The data received by the communication unit 63 from the information processing device 1A may include input information (prompts) generated by the information processing device 1A. In addition, the data provided by the communication unit 53 to the information processing device 1A may include output information generated by the generation model GM, described later, based on the input information.

[0033] The memory unit 62 stores a generative model GM. For example, the memory unit 62 stores several parameters that define the generative model GM. These parameters are, for example, parameters that have been pre-learned by machine learning (parameters that have undergone update processing by machine learning), but this does not limit the present exemplary embodiment. A large-scale language model that has been trained by machine learning can be used as the generative model GM. Alternatively, a generative model that utilizes a graph database may be used as the generative model GM, or other models may be used. Here, an example of a generative model that utilizes a graph database is GraphRAG (Graph Retrieval Augmented Generation), but this does not limit the present exemplary embodiment.

[0034] The control unit 61 acquires information generated by the generation model GM by using the generation model GM. For example, the control unit 61 acquires output information (generation result) generated by the generation model GM based on input information (prompt) received from the information processing device 1A, which includes structured medical information described later. The control unit 61 also provides this output information to the information processing device 1A via the communication unit 63.

[0035] In this exemplary embodiment, the server device 60 is shown as a separate device from the information processing device 1A, but this does not limit this exemplary embodiment. The control unit 61 of the server device 60, or the function of the generation model execution unit in the control unit 61, may be provided by the control unit of the information processing device 1A. Similarly, the generation model GM stored in the storage unit 62 of the server device 60 may be stored in the storage unit of the information processing device 1A, and the information processing device 1A itself may be able to execute the generation model GM.

[0036] (Configuration of Information Processing Device 1A) Next, the configuration of the information processing device 1A according to this exemplary embodiment will be described with reference to Figure 3. As shown in Figure 3, the information processing device 1A includes a control unit 10, a storage unit 20, a communication unit 30, and an input / output unit 40.

[0037] (Communications Section 30) The communication unit 30 communicates with devices outside the information processing device 1A. For example, the communication unit 30 communicates with the medical record management device 50 and the server device 60. The communication unit 30 transmits data supplied from the control unit 10 to the server device 60 and supplies data received from the medical record management device 50 and the server device 60 to the control unit 10. The data that the communication unit 30 receives from the medical record management device 50 may include the electronic medical records or medical information of multiple subjects (patients, clinical trial candidates). The data that the communication unit 30 transmits to the server device 60 may include input information (prompts) generated by the first generation unit 13, which will be described later. The data that the communication unit 30 receives from the server device 60 may include the generation results (output information) based on the input information by the generation model GM.

[0038] (Input / output section 40) The input / output unit 40 is configured to include at least one of the following input / output devices: a keyboard, mouse, display, printer, touch panel, etc. Alternatively, the input / output unit 40 may be configured to have input / output devices such as a keyboard, mouse, display, printer, touch panel, etc. connected to it. In this configuration, the input / output unit 40 receives various types of information from the connected input device to the information processing device 1A. The input / output unit 40 also outputs various types of information to the connected output device under the control of the control unit 10. An interface such as USB (Universal Serial Bus) can be used as an example of the input / output unit 40.

[0039] (Storage unit 20) The storage unit 20 stores various data referenced by the control unit 10, as well as various data generated by the control unit 10. For example, the storage unit 20 stores: • Medical information MI for each participant (patient, clinical trial candidate) Structured Data SD • Standard Information CI • Input information IN • Output information OUT • Structured Model SM The following are stored. Here, the medical information MI may include various types of information extracted from each subject's electronic medical record. For example, the medical information MI may include information such as the subject's initial consultation interview, progress records, radiology reports, and nursing records. However, these examples do not limit this exemplary embodiment. Note that a group of medical information including the medical information MI of multiple subjects may also be referred to as the target data TD.

[0040] Structured data (SD) is data generated from each subject's medical information (MI) by the structuring unit 12, which will be described later. Specific examples of structured data (SD) will be described later. Note that a data set containing structured data (SD) for multiple subjects is sometimes referred to as structured data group (SDG).

[0041] Criteria information (CI) is, for example, criteria information related to a clinical trial. Criteria information (CI) may include, for example, at least one of the eligibility criteria (criteria that a clinical trial participant must meet) and exclusion criteria (criteria that a clinical trial participant must not meet). Figure 4 shows an example of criteria information (CI).

[0042] In the example shown in Figure 4, the criterion information CI for clinical trial 1) is set as exclusion criterion 1) "history of hypertension". The criterion information CI for clinical trial 2) is set as eligibility criterion 2) "female aged 20 years or older" and exclusion criterion 2) "having multiple cancers". The criterion information CI for clinical trial 3) is set as eligibility criterion 3) "PS (Performance status) [ECOG (Eastern cooperation oncology group)] is 0 or 1". The criterion information CI for clinical trial 4) is set as exclusion criterion 4) "patients currently participating in another clinical trial or those who have not completed another trial for more than 3 months". However, these examples are not limiting to the exemplary embodiments described herein.

[0043] The input information IN is information generated by the first generation unit 13, which will be described later, and is used as input to the generation model GM described above. Specific examples of input information IN will be described later.

[0044] The output information OUT is information generated by the second generation unit 14, described later, by referring to the output from the generation model GM, which received the input information IN. Specific examples of the output information OUT will be described later.

[0045] The structured model SM is a model referenced by the first generation unit 13 and is used to generate the structured data SD described above. There may be multiple structured models SM. Specific examples of structured models SM will be described later.

[0046] (Control Unit 10) As shown in Figure 3, the control unit 10 includes an acquisition unit 11, a structuring unit 12, a first generation unit 13, a second generation unit 14, and a learning unit 15.

[0047] (Acquisition part 11) The acquisition unit 11 acquires medical information MI for one or more subjects. Here, the medical information may include various types of information extracted from each subject's electronic medical record managed by the medical record management device 50. Specific examples of medical information MI have been described above, so a redundant explanation will be omitted.

[0048] (structuring part 12) The structuring unit 12 generates structured medical information by structuring at least a portion of the medical information MI acquired by the acquisition unit 11. Here, the structured medical information is an example of the structured data SD described above. Similar to the exemplary embodiment 1, the structuring unit 12 can be configured to generate structured medical information in at least one of the data formats of triplets, graphs, and tables.

[0049] Furthermore, the structuring unit 12 may be configured to select one or more structured models from a plurality of structured models SM that have been trained (machine learning) using different training data, by referring to the criteria information for the clinical trial, and to generate the structured medical information using the selected one or more structured models SM. More specific examples of processing by the structuring unit 12 will be described later with different reference drawings.

[0050] (First generation unit 13) The first generation unit 13 generates input information IN, which includes the structured medical information and a query. Here, the input information IN is the information input to the generation model GM as described above. The query may be predetermined or it may be based on information acquired by the acquisition unit 11.

[0051] For example, the acquisition unit 11 may acquire standard information CI related to the clinical trial, and the first generation unit 13 may generate the input information IN including the standard information CI as the query. Alternatively, the acquisition unit 11 may acquire standard information CI related to the clinical trial, the structuring unit 12 may generate structured standard information by structuring at least a part of the standard information CI, and the first generation unit 13 may generate the input information including the structured standard information as the query. Here, the structured standard information can be configured to generate information in at least one of the data formats of triplets, graphs, and tables, similar to the structured clinical information described above.

[0052] Furthermore, the first generation unit 13 may be configured to include text extracted from the electronic medical record or the medical information MI in the input information IN as is (i.e., without structuring it by the structuring unit 12). For example, the first generation unit 13 may include a sentence extracted from the medical information, such as "Because of the finding of xxx, examination yyy will be performed at the next consultation," in the input information as is without structuring it.

[0053] (Second generation unit 14) The second generation unit 14 generates an answer (output information OUT) to the query regarding the subject by referring to the output from the generation model GM that has received the input information IN. The second generation unit 14 may use the output from the generation model GM that has received the input information IN as the answer (output information OUT) as is, or it may be configured to generate the answer (output information OUT) by processing the output from the generation model GM that has received the input information IN. In addition, as an example, the second generation unit 14 may refer to the output from the generation model GM that has received the input information IN and generate information regarding the degree to which the subject is suitable for the clinical trial as the answer (output information OUT).

[0054] (Learning Section 15) The learning unit 15 trains structured models SM. For example, the learning unit 15 trains each of multiple structured models SM using training data for each medical department or category (machine learning). A specific example of the learning process by the learning unit 15 will be described later with a different diagram.

[0055] (Example 1 of the processing flow by information processing system 100A) Next, with reference to Figure 5, we will explain Example 1 of the processing flow by the information processing system 100A. Figure 5 is a flowchart showing Example 1 of the processing flow by the information processing system 100A (information processing method S1A).

[0056] (Steps S111, S112) In step S111, the acquisition unit 11 acquires the electronic medical records of one or more subjects from the medical record management device 50. Then, in step S112, the acquisition unit 111 extracts the medical information MI of the subject from the electronic medical records. Steps S111 and S112 may correspond to step S11 described in exemplary embodiment 1.

[0057] (Step S12) Next, in step S12, the structuring unit 12 generates structured medical information (structured data SD) from medical information MI using the structured model SM.

[0058] (Steps S131, S132) Next, in step S131, the acquisition unit 11 acquires the clinical trial reference information CI as a query. Then, in step S132, the first generation unit 13 generates input information IN, which includes the structured medical information (structured data SD) generated in step S12 and the query. Steps S131 and S132 may correspond to step S13 described in exemplary embodiment 1.

[0059] (Steps S141, S142) Next, in step S141, the second generation unit 14 inputs the input information IN generated in step S132 to the generation model GM via the communication unit 30. Then, in step S142, the second generation unit 14 obtains the output (generation result) from the generation model GM to which the input information IN was input, via the communication unit 30. Then, the second generation unit 14 generates output information OUT by referring to the output (generation result). Here, the output information OUT includes the answer to the query obtained in step S131. As an example, the output information OUT includes information on how well the subject is suited to the clinical trial.

[0060] As described above, in the information processing device 1A, • Obtain the medical information MI of one or more subjects (steps S111, S112), - Structured medical information (structured data SD) is generated by structuring at least a portion of the aforementioned medical information MI (step S12), - Generate input information IN including the structured medical information and queries (steps S131, S132), - Referencing the output from the generation model GM that received the input information IN, the system generates the answer to the query regarding the subject (output information OUT) (steps S141, S142). This configuration is adopted. In this way, the information processing device 1A generates structured medical information from medical information MI, and generates a response by referring to the output from the generative model GM, which receives input information IN including the structured medical information. Therefore, it is possible to make estimations related to the subject (patient, clinical trial candidate) with high accuracy. In addition, because it uses input information IN including the structured medical information, it is possible to reduce the processing cost of the generative model GM.

[0061] Conventionally, a configuration in which patient medical information is directly input into a large-scale language model has been known. However, in such a configuration, the amount of information input into the large-scale language model may be too large, making it impossible to process all the input information, which may result in a decrease in accuracy. On the other hand, with the information processing device 1A configured as described above, structured medical information is generated from medical information MI, and the response is generated by referring to the output from the generative model GM into which the input information IN, including the structured medical information, is input. As a result, processing is suitably performed in the generative model GM, and as described above, estimations related to the subject (patient, clinical trial candidate) can be made with high accuracy.

[0062] (Data processing example 1) Next, with reference to Figure 6, a more specific example of data processing by the information processing device 1A, Example 1, will be explained. In the example shown in Figure 6, first, the acquisition unit 11 acquires the medical information MI, [Medical history] Hypertension, appendicitis (surgery at age 57) This is obtained (corresponding to steps S111 and S112 described above).

[0063] Then, in the structuring unit 12, multiple entities (hypertension, appendicitis, 57 years old, surgery) are extracted from the medical information MI, and multiple triplets are created that include these multiple entities and the relationships between them. (Hypertension, category, past history) (Appendicitis, category, past history) (Surgery performed at age 57) (Surgery, site, appendicitis) This is generated as the structured medical information SD (corresponding to step S12 described above).

[0064] Here, the triplet generated by the structuring unit 12 is structured as follows, as in the example above: (Entity1, Relation, Entity2), with the elements being a first entity (Entity1), a second entity (Entity2), and a relationship (Relation) between the first entity and the second entity. Here, the relationship (Relation) may be an oriented concept or an unoriented concept.

[0065] The structuring unit 12 may perform the above structuring process using the structured model SM described above, or it may perform the above structuring process using the generative model GM.

[0066] Next, the first generation unit 13 acquires "Exclusion Criterion 1: History of hypertension" as the criterion information CI (corresponding to step S131 described above), and generates input information IN which includes the structured medical information SD and a query corresponding to the criterion information CI (corresponding to step S132 described above).

[0067] More specifically, the first generation unit 13 generates input information IN as a query corresponding to the reference information CI, including "Does the following patient have a history of hypertension?" as shown in Figure 6. Here, the query may also include an instruction to include the rationale in the answer (as shown in Figure 6, "Please also state the rationale"). In addition, the input information IN generated by the first generation unit 13 includes, as shown in Figure 6, the structured medical information SD generated by the structuring unit 12 as patient information: (Hypertension, category, past history) (Appendicitis, category, past history) (Surgery performed at age 57) (Surgery, site, appendicitis) This is included. Furthermore, the first generation unit 13 may also perform the process of including the text extracted from the electronic medical record or the medical information MI as is (i.e., without structuring it by the structuring unit 12), as described above.

[0068] Next, the second generation unit 14 inputs the above input information IN as a prompt to the generation model GM (corresponding to step S141 described above). Then, the second generation unit 14 acquires the output (generation result, estimation result) of the generation model GM. Then, it generates output information OUT by referring to the acquired generation result (corresponding to step S142 described above). In the example shown in Figure 6, the second generation unit 14, as a result of generation by the generation model GM, Result: History of hypertension Reason: Because of a history of hypertension. The system obtains the data and uses the generated result as output information (OUT). In this example, the query includes instructions to include justification in the response, and the response (output information (OUT) generated by the second generation unit 14 includes information to support the user's decision-making regarding the extent to which the subject is suitable for the clinical trial, as well as the justification for that information.

[0069] As in this example, by generating triplets containing entities extracted from medical information MI as structured medical information SD, it is possible to improve the estimation accuracy of the generative model GM and reduce the processing cost of the generative model GM. Furthermore, as in this example, by instructing the generative model GM to include the basis for the estimation in the generated result (estimation result), it is possible to support the decision-making of users (physicians and healthcare professionals) through the output information OUT.

[0070] (Data processing example 2) Next, with reference to Figure 7, a more specific example of data processing by the information processing device 1A, Example 2, will be described. The example shown in Figure 7 differs from Data Processing Example 1 shown in Figure 6 in the following respects, but is otherwise the same as Data Processing Example 1.

[0071] In other words, in the data processing example 2 shown in Figure 7, when the first generation unit 13 generates the input information IN in step S132, The structuring unit 12 extracts only the triplets related to the reference information CI from the multiple triplets it generates. • Include the extracted triplets as patient information in the input information (IN). This process is performed. More specifically, as shown in Figure 7, the first generation unit 13 refers to the reference information CI and the multiple triplets generated by the structuring unit 12. (Hypertension, category, past history) (Appendicitis, category, past history) (Surgery performed at age 57) (Surgery, site, appendicitis) From this, the triplet (hypertension, category, medical history) related to the reference information CI is extracted, and input information IN containing the extracted triplet as patient information is generated. In this way, by extracting structured medical information related to the reference information CI (in other words, the query) from multiple structured medical information generated by the structuring unit 12, and generating input information IN containing the extracted structured medical information as patient information, it is possible to improve the estimation accuracy by the generative model GM and reduce the computational cost by the generative model GM.

[0072] (Data processing example 3) Next, with reference to Figure 8, a more specific example of data processing by the information processing device 1A, Example 3, will be described. The example shown in Figure 8 differs from Data Processing Example 1 shown in Figure 6 in the following respects, but is otherwise the same as Data Processing Example 1.

[0073] In other words, in the data processing example 3 shown in Figure 8, in step S12, the structuring unit 12 extracts multiple entities from at least a portion of the medical information MI and generates a table (tabular format) containing these multiple entities as data items as structured medical information SD.

[0074] Then, in the data processing example 3, the first generation unit 13 generates input information IN, which includes the table-formatted structured medical information SD, as shown in Figure 8, and inputs it into the generation model GM (corresponding to steps S131 and S141). More specifically, the first generation unit 13 generates patient information, {Diagnosis: Hypertension, Fact: Present, Past medical history: Yes} {Diagnosis: Appendicitis, Fact: Yes, Treatment: Surgery, Age: 57, Past medical history: Yes} The input information IN, which includes the specified value, is generated and input into the generation model GM.

[0075] As in this example, even when structured clinical information in a table format is generated as structured data (SD), it is possible to improve the estimation accuracy using the generative model (GM) and reduce the computational cost using the generative model (GM).

[0076] (Data processing example 4) Next, with reference to Figure 9, a more specific example of data processing by the information processing device 1A, Example 4, will be described. The example shown in Figure 9 differs from Data Processing Example 1 shown in Figure 6 in the following respects, but is otherwise the same as Data Processing Example 1.

[0077] In other words, in the data processing example 4 shown in Figure 9, in step S12, the structuring unit 12 extracts multiple entities from at least a portion of the medical information MI and generates a graph (graph structure) containing these multiple entities as nodes as structured medical information SD.

[0078] In the data processing example 4, the first generation unit 13 generates input information IN, which includes the structured medical information SD in the graph format described above, as shown in Figure 9, and inputs it into the generation model GM (corresponding to steps S131 and S141). The graph generated by the structuring unit 12 may be a directed graph in which each link has an orientation, an undirected graph in which each link does not have an orientation, or a combination of the two. When the structuring unit 12 generates a directed graph, the orientation of each link is determined according to the relationships between entities (between nodes) extracted from the medical information MI.

[0079] As in this example, even when structured clinical information in graph format is generated as structured data (SD), it is possible to improve the estimation accuracy using the generative model (GM) and reduce the computational cost using the generative model (GM).

[0080] (Data processing example 5) Next, with reference to Figure 10, a more specific example of data processing by the information processing device 1A, Example 5, will be explained. In the example shown in Figure 10, the acquisition unit 11 acquires the initial consultation interview, progress records, image interpretation reports, and nursing records as medical information MI from the electronic medical record managed by the medical record management device 50. The structuring unit 12 then applies a structured model SM individually to each of the initial consultation interview, progress records, image interpretation reports, and nursing records to generate structured data SD.

[0081] As an example, the structuring unit 12 may be configured to select a structured model SM to apply to the initial consultation interview included in the medical information MI from among multiple models by referring to the medical information MI. Here, the structured model SM to apply to the initial consultation interview is, as an example, a model that has been machine-trained using training data that includes sets of multiple initial consultation interviews and the correct labels (correct labels related to structuring) attached to each initial consultation interview.

[0082] Similarly, the structuring unit 12 may be configured to select a structured model SM to apply to the progress records included in the medical information MI from among multiple models by referring to the medical information MI. Here, the structured model SM applied to the progress records is, as an example, a machine learning model that uses training data including sets of multiple progress records and the correct labels (correct labels related to structuring) attached to each progress record. The same applies to image interpretation reports and nursing records.

[0083] Furthermore, the training of the structured model SM described above can be performed in advance by the learning unit 15. In other words, the learning unit 15 may be configured to perform machine learning on each of the multiple structured models SM using training data for each category (training data that is different from each other). Here, examples of such categories include the initial consultation interview, progress records, radiology reports, and nursing records mentioned above.

[0084] Furthermore, the learning unit 15 may be configured to perform machine learning on each of the multiple structured models SM using training data specific to each medical department. In this configuration, for example, separate training data (different training data from each other) can be prepared for each medical department, such as cardiology, gastroenterology, and respiratory medicine, and the structured model SM for each medical department can be trained using this training data.

[0085] Thus, the structuring unit 12 can be configured to select one or more structured models from a plurality of structured models SM that have been machine-trained using different training data, by referring to the criteria information CI related to the clinical trial, and to generate the structured clinical information using the selected one or more structured models SM.

[0086] In this example, the input information IN generated by the first generation unit 13 includes structured medical information SD (in table format, for example) generated from the initial consultation interview, progress records, image interpretation reports, and nursing records, respectively. Here, each table may be in an independent tabular format or an integrated tabular format. It is possible that one table and another tables may contain conflicting information. To prepare for such cases, the first generation unit 13 may assign a priority to each table (the priority of each table may be written in the input information IN), and the generation model GM may refer to each table according to that priority. Alternatively, the input information IN may include instruction information indicating that the generation model GM will also check for consistency, such as, "If one table and another tables contain conflicting information, please determine which information is appropriate and then generate the answer."

[0087] Furthermore, in this example, as shown in Figure 10, the second generation unit 14 uses the generation model GM to generate output information OUT1, which includes estimated results (judgment results) for each of the multiple criteria included in the criterion information CI for a given subject. In addition, in this example, the second generation unit 14 refers to the output information OUT1 to generate the final answer (output information) OUT2, which indicates whether or not the subject is suitable for the clinical trial.

[0088] More specifically, as shown in Figure 10, the second generation unit 14 is, • Refer to the judgment results by the generative model GM for each criterion (eligibility criteria 1-5, exclusion criteria 1-3) included in the output information OUT1. • Determine whether or not there are any non-conforming standards. • If there are no disqualifying criteria, the system generates output information OUT2 indicating that the subject is suitable for the clinical trial. This configuration is also acceptable. According to the processing in this example, it is possible to further improve the estimation accuracy of the generative model GM and further reduce the computational cost of the generative model GM.

[0089] (Example display) Figure 11 shows an example of how the output information OUT generated by the second generation unit 14 is displayed via the display provided by the input / output unit 40. As shown in Figure 11, the second generation unit 14 may also display the output information OUT via the input / output unit 40, which includes information on which of the multiple clinical trial candidates is suitable for the clinical trial. The output information OUT in this example is also an example of information that supports the decision-making of the user (physician or healthcare professional).

[0090] Furthermore, the second generation unit 14 may be configured to include supplementary information in the output information OUT, in addition to the generation results generated by the generation model GM. For example, regarding subjects who have been determined to be suitable for the clinical trial in the generation results generated by the generation model GM, information regarding the subject's intention to participate in the clinical trial (whether they wish to participate or not) may be obtained from the battery medical record or other database and included in the output information OUT. However, the supplementary information is not limited to this example.

[0091] (Example 2 of the processing flow by information processing system 100A) Next, with reference to Figure 12, we will explain Example 2 of the processing flow by the information processing system 100A. Figure 12 is a flowchart showing Example 2 of the processing flow by the information processing system 100A (information processing method S1B).

[0092] As shown in Figure 12, the process in this example differs from Example 1 of the process flow shown in Figure 5 in the following respects, and is otherwise the same as Example 1 of the process flow. Specifically, the process in this example includes steps S133 and S134 instead of step S132 shown in Figure 5.

[0093] (Step S133) In step S133, structured criteria information is generated from the criteria information CI obtained in step S131. This structured criteria information may be generated by the structuring unit 12, or by the generation model GM. Furthermore, the structured criteria information can be in the form of at least one of the data formats of a triplet, graph, or table, similar to the structured clinical information SD described above. More specifically, the following processing may be performed. Example 1: Multiple entities are extracted from at least a portion of the standard information CI, and one or more triplets, including the multiple entities and the relationships between them, are generated as the structured standard information. Example 2: Multiple entities are extracted from at least a portion of the aforementioned standard information CI, and a graph (graph structure) containing the multiple entities as nodes is generated as the structured standard information. Example 3: Extract multiple entities from at least a portion of the standard information CI, and generate a table (tabular format) containing the multiple entities as data items as the structured standard information.

[0094] (Step S134) Next, in step S134, the first generation unit 13 generates input information IN which includes the structured medical information (structured data SD) generated in step S12 and the structured criteria information generated in step S133 as a query.

[0095] Thus, in the process described in this example, structured reference information is generated from reference information CI, and input information IN that includes the structured reference information as a query is generated, thereby improving the estimation accuracy of the generative model GM and reducing the computational cost of the generative model GM.

[0096] [Third 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 function as those described in the above-described exemplary embodiments are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical hindrance occurs.

[0097] (Configuration of Information Processing System 100B) The configuration of the information processing system 100B according to this exemplary embodiment will be described with reference to Figure 13. Figure 13 is a block diagram showing the configuration of the information processing system 100B. As shown in Figure 13, the information processing system 100B comprises an information processing device 1A and a medical record management device 50, a server device 60, and a clinical trial management device 70 connected to the information processing device 1A via a network N. The information processing device 1A, the medical record management device 50, and the server device 60 are the same as in exemplary embodiment 2 and have already been described, so redundant explanations will be omitted.

[0098] (Clinical trial management device 70) The clinical trial management device 70 manages the conduct of the clinical trial. For example, the clinical trial management device 70: • Data on the type of clinical trial • Data on the drugs used in each clinical trial. • Data on candidates (clinical trial candidates) for each clinical trial • Pre-trial data regarding the subjects (trial participants) of each clinical trial. • Data on participants in each clinical trial during the trial. • Post-trial data regarding the participants in each clinical trial. It acquires and manages the following. Furthermore, the clinical trial management device 70 may, for example, be configured to generate output data (clinical trial reports, etc.) by referring to the aforementioned data. Also, the information processing system 100B may be equipped with multiple information processing devices 1A. For example, such a system configuration can be adopted when centrally managing information processing devices 1A installed in each hospital. In such a configuration, the clinical trial management device 70, • Receives output information from each hospital's information processing device 1A. • Aggregate output information from each hospital and generate a list of candidates. The configuration may also include processing such as the above.

[0099] In this exemplary embodiment, the second generation unit 14 outputs the generated output information OUT to the clinical trial management device 70. Here, it is preferable that the second generation unit 14 includes information (such as a patient ID) that identifies a clinical trial candidate suitable for the target clinical trial in the output information OUT.

[0100] The clinical trial management device 70 then refers to the output information OUT supplied from the information processing device 1A and performs processing related to the clinical trial. For example, the clinical trial management device 70 refers to the ID of the clinical trial candidate included in the output information OUT and obtains data related to the clinical trial candidate from the medical record management device 50.

[0101] According to the information processing system 100B of this exemplary embodiment, • Obtain medical information from one or more subjects, • Structured medical information is generated by structuring at least a portion of the aforementioned medical information. • Generate input information including the structured medical information and queries, ·By referring to the output from the generative model into which the input information has been received, the system generates an answer to the query regarding the subject (information on whether the subject is suitable for the clinical trial), • The output information OUT, including the response, is supplied to the clinical trial management device 70. This configuration is adopted. Therefore, it is possible to suitably conduct the clinical trial on subjects who are suitable for the trial.

[0102] [Examples of implementation using software] Some or all of the functions of the information processing devices 1, 1A, and 1B (hereinafter also referred to as "the above devices") may be implemented by hardware such as integrated circuits (IC chips) or by software.

[0103] In the latter case, each of the above devices is implemented, for example, by a computer that executes instructions for a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as Computer C) is shown in Figure 14. Figure 14 is a block diagram showing the hardware configuration of Computer C, which functions as each of the above devices.

[0104] Computer C comprises at least one processor C1 and at least one memory C2. Memory C2 stores a program P that causes computer C to operate as each of the above-mentioned devices. In computer C, processor C1 reads program P from memory C2 and executes it, thereby realizing each of the above-mentioned devices.

[0105] For processor C1, for example, a CPU (Central Processing Unit), GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Point Number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination thereof can be used. For memory C2, for example, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof can be used.

[0106] Computer C may also be equipped with RAM (Random Access Memory) for loading program P at runtime and for temporarily storing various data. Furthermore, computer C may be equipped with communication interfaces for sending and receiving data with other devices. Additionally, computer C may be equipped with input / output interfaces for connecting input / output devices such as keyboards, mice, displays, and printers.

[0107] Furthermore, program P can be recorded on a non-temporary, tangible recording medium M that is readable by computer C. Such a recording medium M could be, for example, tape, disk, card, semiconductor memory, or programmable logic circuitry. Computer C can acquire program P via such a recording medium M. Program P can also be transmitted via a transmission medium. Such a transmission medium could be, for example, a communication network or broadcast waves. Computer C can also acquire program P via such a transmission medium.

[0108] Furthermore, each of the above functions of each of the above devices may be implemented by a single processor in a single computer, by multiple processors in a single computer working together, or by multiple processors in each of multiple computers working together. In addition, the programs for implementing each of the above functions in each of the above devices may be stored in a single memory in a single computer, distributed and stored in multiple memories in a single computer, or distributed and stored in multiple memories in each of multiple computers.

[0109] [Additional Note A] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0110] (Note A1) A means for acquiring medical information of one or more subjects, A structuring means for generating structured medical information by structuring at least a portion of the aforementioned medical information, A first generation means for generating input information including the structured medical information and queries, A second generation means generates a response to the query concerning the subject by referring to the output from the generation model into which the input information has been input. An information processing device equipped with the following features.

[0111] (Appendix A2) The aforementioned acquisition means further acquires standard information related to clinical trials, The first generation means generates the input information including the reference information as the query, The second generation means generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing device described in Appendix A1.

[0112] (Note A3) The aforementioned acquisition means further acquires standard information related to clinical trials, The structuring means further generates structured reference information by structuring at least a portion of the reference information, The first generation means generates the input information including the structured criteria information as the query, The second generation means generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing device described in Appendix A1.

[0113] (Note A4) The aforementioned query includes instructions to include justification in the response. The response generated by the second generation means includes, as information to support the user's decision-making, information on the extent to which the subject is suitable for the clinical trial and the basis for such information. The information processing device described in Appendix A2 or A3.

[0114] (Note A5) The structuring means extracts a plurality of entities from at least a portion of the medical information, One or more triplets including the plurality of entities and the relationships between the plurality of entities, Graph containing the aforementioned multiple entities as nodes This is generated as the aforementioned structured medical information. An information processing device as described in any one of the appendices A2 to A4.

[0115] (Note A6) The structuring means is, Multiple entities are extracted from at least a portion of the aforementioned medical information, A table containing the aforementioned multiple entities as data items is generated as the structured medical information. An information processing device as described in any one of the appendices A2 to A4.

[0116] (Note A7) The structuring means is, From multiple structured models that have been machine-trained using different training data, one or more structured models are selected by referring to the criteria information for the clinical trial. The structured medical information is generated using the selected one or more structured models. An information processing device as described in any one of the appendices A2 to A6.

[0117] (Note A8) The system further includes a learning method for machine learning each of the aforementioned multiple structured models using training data for each medical department or category. The information processing device described in Appendix A7.

[0118] [Additional Note B] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0119] (Note B1) At least one processor performs an acquisition process to obtain medical information for one or more subjects, The at least one processor performs a structuring process that generates structured medical information by structuring at least a portion of the medical information, The at least one processor performs a first generation process that generates input information including the structured medical information and queries, The at least one processor performs a second generation process which generates a response to the query regarding the subject by referring to the output from the generation model into which the input information has been input. An information processing method that includes this.

[0120] (Note B2) In the acquisition process described above, the at least one processor further acquires reference information relating to the clinical trial, The first generation process generates the input information including the reference information as the query, The second generation process generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing method described in Appendix B1.

[0121] (Note B3) In the acquisition process described above, the at least one processor further acquires reference information relating to the clinical trial, In the structuring process, the at least one processor further generates structured reference information by structuring at least a portion of the reference information. The first generation process generates the input information including the structured criteria information as the query, The second generation process generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing method described in Appendix B1.

[0122] (Note B4) The aforementioned query includes instructions to include justification in the response. The response generated by the second generation process includes, as information to support the user's decision-making, information on the extent to which the subject is suitable for the clinical trial and the basis for that information. The information processing method described in Appendix B2 or B3.

[0123] (Note B5) In the structuring process described above, the at least one processor extracts a plurality of entities from at least a portion of the medical information, One or more triplets including the plurality of entities and the relationships between the plurality of entities, Graph containing the aforementioned multiple entities as nodes This is generated as the aforementioned structured medical information. The information processing method described in any one of the appendices B2 to B4.

[0124] (Note B6) In the structuring process, the at least one processor, Multiple entities are extracted from at least a portion of the aforementioned medical information, A table containing the aforementioned multiple entities as data items is generated as the structured medical information. The information processing method described in any one of the appendices B2 to B4.

[0125] (Note B7) In the structuring process, the at least one processor, From multiple structured models that have been machine-trained using different training data, one or more structured models are selected by referring to the criteria information for the clinical trial. The structured medical information is generated using the selected one or more structured models. The information processing method described in any one of the appendices B2 through B6.

[0126] (Note B8) The at least one processor further includes a learning process that causes each of the plurality of structured models to perform machine learning using training data for each medical department or category. The information processing method described in Appendix B7.

[0127] [Additional Note C] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0128] (Note C1) A program that makes a computer function as an information processing device. The aforementioned computer, A means for acquiring medical information of one or more subjects, A structuring means for generating structured medical information by structuring at least a portion of the aforementioned medical information, A first generation means for generating input information including the structured medical information and queries, A second generation means generates a response to the query concerning the subject by referring to the output from the generation model into which the input information has been input. An information processing program that functions as such.

[0129] (Note C2) The aforementioned acquisition means further acquires standard information related to clinical trials, The first generation means generates the input information including the reference information as the query, The second generation means generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing program described in Appendix C1.

[0130] (Note C3) The aforementioned acquisition means further acquires standard information related to clinical trials, The structuring means further generates structured reference information by structuring at least a portion of the reference information, The first generation means generates the input information including the structured criteria information as the query, The second generation means generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing program described in Appendix C1.

[0131] (Note C4) The aforementioned query includes instructions to include justification in the response. The response generated by the second generation means includes, as information to support the user's decision-making, information on the extent to which the subject is suitable for the clinical trial and the basis for such information. The information processing program described in Appendix C2 or C3.

[0132] (Note C5) The structuring means extracts a plurality of entities from at least a portion of the medical information, One or more triplets including the plurality of entities and the relationships between the plurality of entities, Graph containing the aforementioned multiple entities as nodes This is generated as the aforementioned structured medical information. An information processing program described in any one of the appendices C2 to C4.

[0133] (Appendix C6) The structuring means is, Multiple entities are extracted from at least a portion of the aforementioned medical information, A table containing the aforementioned multiple entities as data items is generated as the structured medical information. An information processing program described in any one of the appendices C2 to C4.

[0134] (Note C7) The structuring means is, From multiple structured models that have been machine-trained using different training data, one or more structured models are selected by referring to the criteria information for the clinical trial. The structured medical information is generated using the selected one or more structured models. An information processing program described in any one of the appendices C2 through C6.

[0135] (Note C8) The aforementioned computer, Each of the aforementioned structured models is further made to function as a learning tool for machine learning using training data for each medical department or category. The information processing program described in Appendix C7.

[0136] [Additional Note D] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0137] (Note D1) It comprises at least one processor, and the at least one processor is A process for obtaining medical information of one or more subjects, A structuring process that generates structured medical information by structuring at least a portion of the aforementioned medical information, A first generation process that generates input information including the structured medical information and queries, A second generation process generates a response to the query regarding the subject by referring to the output from the generation model into which the aforementioned input information has been input. An information processing device that performs the following actions.

[0138] The information processing device may also include memory. Furthermore, the memory may store a program that causes at least one processor to execute each of the aforementioned processes.

[0139] (Note D2) In the acquisition process described above, the at least one processor further acquires reference information relating to the clinical trial, The first generation process generates the input information including the reference information as the query, The second generation process generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing device described in Appendix D1.

[0140] (Note D3) In the acquisition process described above, the at least one processor further acquires reference information relating to the clinical trial, In the structuring process, the at least one processor further generates structured reference information by structuring at least a portion of the reference information. The first generation process generates the input information including the structured criteria information as the query, The second generation process generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing device described in Appendix D1.

[0141] (Note D4) The aforementioned query includes instructions to include justification in the response. The response generated by the second generation process includes, as information to support the user's decision-making, information on the extent to which the subject is suitable for the clinical trial and the basis for that information. The information processing device described in Appendix D2 or D3.

[0142] (Note D5) In the structuring process described above, the at least one processor extracts a plurality of entities from at least a portion of the medical information, One or more triplets including the plurality of entities and the relationships between the plurality of entities, Graph containing the aforementioned multiple entities as nodes This is generated as the aforementioned structured medical information. An information processing device as described in any one of the appendices D2 to D4.

[0143] (Note D6) In the structuring process, the at least one processor, Multiple entities are extracted from at least a portion of the aforementioned medical information, A table containing the aforementioned multiple entities as data items is generated as the structured medical information. An information processing device as described in any one of the appendices D2 to D4.

[0144] (Note D7) In the structuring process, the at least one processor, From multiple structured models that have been machine-trained using different training data, one or more structured models are selected by referring to the criteria information for the clinical trial. The structured medical information is generated using the selected one or more structured models. An information processing device as described in any one of the appendices D2 to D6.

[0145] (Note D8) The aforementioned at least one processor, Further training is performed to enable machine learning in each of the aforementioned structured models using training data for each medical department or category. The information processing device described in Appendix D7.

[0146] [Additional Note E] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0147] (Note E1) A program that makes a computer function as an information processing device. To the aforementioned computer, A process for obtaining medical information of one or more subjects, A structuring process that generates structured medical information by structuring at least a portion of the aforementioned medical information, A first generation process that generates input information including the structured medical information and queries, A second generation process generates a response to the query regarding the subject by referring to the output from the generation model into which the aforementioned input information has been input. A non-temporary recording medium that stores an information processing program that executes that program. [Explanation of Symbols]

[0148] 1, 1A, 1B Information Processing Device 100A, 100B Information Processing System 11 Acquisition unit (acquisition means) 12 Structuring part (structuring means) 13. First generation unit (first generation means) 14. Second generation unit (second generation means) 15. Learning Section (Learning Methods)

Claims

1. A means for acquiring medical information of one or more subjects, A structuring means for generating structured medical information by structuring at least a portion of the aforementioned medical information, A first generation means for generating input information including the structured medical information and queries, A second generation means generates a response to the query concerning the subject by referring to the output from the generation model into which the input information has been input. An information processing device equipped with the following features.

2. The aforementioned acquisition means further acquires standard information related to clinical trials, The first generation means generates the input information including the reference information as the query, The second generation means generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing apparatus according to claim 1.

3. The aforementioned acquisition means further acquires standard information related to clinical trials, The structuring means further generates structured reference information by structuring at least a portion of the reference information, The first generation means generates the input information including the structured criteria information as the query, The second generation means generates, as the response, information regarding the extent to which the subject is suitable for the clinical trial. The information processing apparatus according to claim 1.

4. The aforementioned query includes instructions to include justification in the response. The response generated by the second generation means includes, as information to support the user's decision-making, information on the extent to which the subject is suitable for the clinical trial and the basis for that information. The information processing apparatus according to claim 2 or 3.

5. The structuring means extracts a plurality of entities from at least a portion of the medical information, One or more triplets including the plurality of entities and the relationships between the plurality of entities, Graph containing the aforementioned multiple entities as nodes This is generated as the aforementioned structured medical information. The information processing apparatus according to claim 2 or 3.

6. The structuring means is, Multiple entities are extracted from at least a portion of the aforementioned medical information, A table containing the aforementioned multiple entities as data items is generated as the structured medical information. The information processing apparatus according to claim 2 or 3.

7. The structuring means is, From multiple structured models that have been machine-trained using different training data, one or more structured models are selected by referring to the criteria information for the clinical trial. The structured medical information is generated using the selected one or more structured models. The information processing apparatus according to claim 2 or 3.

8. The system further includes a learning method for machine learning each of the aforementioned multiple structured models using training data for each medical department or category. The information processing apparatus according to claim 7.

9. One or more processors, Obtaining medical information from one or more subjects, The method involves generating structured medical information by structuring at least a portion of the aforementioned medical information, The process involves generating input information including the aforementioned structured medical information and queries, The process involves referencing the output from the generative model into which the aforementioned input information is received to generate a response to the query regarding the subject. An information processing method that includes this.

10. A program that makes a computer function as an information processing device. The aforementioned computer, A means for acquiring medical information of one or more subjects, A structuring means for generating structured medical information by structuring at least a portion of the aforementioned medical information, A first generation means for generating input information including the structured medical information and queries, A second generation means generates a response to the query concerning the subject by referring to the output from the generation model into which the input information has been input. A program that makes it function as such.

Images