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

Abstract

This information processing device comprises a processor. The processor: acquires a past document; extracts items corresponding to elements constituting the past document; generates a format corresponding to the past document, on the basis of the relative positional relationship of the plurality of extracted items; acquires input information; causes the format and the input information to be inputted to a document generation model; and acquires a new document outputted from the document generation model.

Description

Information Processing Apparatus, Learning Apparatus, Information Processing Method, Learning Method, Information Processing Program, and Learning Program

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

[0002] Techniques for providing a format to assist in generating a new document are known. For example, Japanese Patent Application Laid-Open No. 2016-151827 (Patent Document 1) describes a technique of analyzing a sentence, obtaining a term expression from the sentence based on the analysis result, classifying the term expression based on the usage form of the term expression, and presenting the result by converting it into a format of a predetermined form.

[0003] In the above technique, since one of a predetermined format is provided, there is a problem that it is difficult to provide an appropriate format. Therefore, there has been room for improvement in assisting the generation of a new document in the conventional technique.

[0004] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide an information processing apparatus, a learning apparatus, an information processing method, a learning method, an information processing program, and a learning program that can more assist in generating a new document than when providing a predetermined format.

[0005] To achieve the above object, the information processing apparatus of the present disclosure includes a processor. The processor acquires a past document, extracts items corresponding to elements constituting the past document, generates a format corresponding to the past document based on the relative positional relationship of the plurality of extracted items, acquires input information, inputs the format and the input information to a document generation model, and acquires a new document output from the document generation model.

[0006] To achieve the above object, the learning apparatus of the present disclosure includes a processor. The processor acquires a plurality of types of formats corresponding to past documents based on the relative positional relationship of a plurality of items corresponding to elements constituting the past document, and generates a document generation model for each type of format using learning data with the format.

[0007] Furthermore, in order to achieve the above objectives, the information processing method disclosed herein acquires past documents, extracts items corresponding to the elements constituting the past documents, generates a format corresponding to the past documents based on the relative positional relationships of the extracted items, acquires input information, inputs the format and input information into a document generation model, and acquires a new document output from the document generation model.

[0008] Furthermore, in order to achieve the above objectives, the learning method disclosed herein acquires multiple types of formats corresponding to past documents, which are based on the relative positional relationships of multiple items corresponding to the elements that constitute the past documents, and generates a document generation model for each type of format using formatted training data.

[0009] Furthermore, in order to achieve the above objectives, the information processing program disclosed herein is designed to cause a computer to perform the following processes: acquire past documents, extract items corresponding to the elements constituting the past documents, generate a format corresponding to the past documents based on the relative positional relationships of the extracted items, acquire input information, input the format and input information into a document generation model, and acquire a new document output from the document generation model.

[0010] Furthermore, in order to achieve the above objectives, the learning program of this disclosure acquires multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to the elements that constitute the past documents, and causes a computer to perform the process of generating a document generation model using formatted learning data for each type of format.

[0011] According to this disclosure, it is possible to better support the generation of new documents compared to providing a predetermined format.

[0012] This is a diagram illustrating the overview of the information processing device of the embodiment. This is a block diagram illustrating an example of the configuration of the information processing device of the embodiment. This is a functional block diagram showing an example of the configuration of the information processing device of the embodiment. This is a diagram illustrating a specific example of the past summary and format of the embodiment. This is a flowchart showing an example of the format generation process flow by the format generation unit of the embodiment. This is a diagram illustrating the format generation method. This is a flowchart showing an example of the information processing flow by the information processing device of the embodiment. This is a diagram illustrating an example of the display mode of the past summary, format, and new summary. This is a diagram illustrating modification 1. This is a diagram illustrating modification 2. This is a diagram illustrating the function as a learning device.

[0013] Embodiments of this disclosure will be described in detail below with reference to the drawings. These embodiments are not intended to limit the technology of this disclosure.

[0014] As shown in Figure 1, the information processing device of this embodiment generates a format 52 from a past summary 50, which is a past document acquired at least before the generation of a new summary, and generates a new summary 54 by inputting the format 52 and patient information 70 as input information into a document generation model 32. In this embodiment, the past summary 50 is an example of a past document in this disclosure, the patient information 70 is an example of input information in this embodiment, and the new summary 54 is an example of a new document in this embodiment.

[0015] Figure 2 shows a block diagram illustrating an example of the configuration of the information processing device 10 of this embodiment. As shown in Figure 2, the information processing device 10 of this embodiment comprises a control unit 20, a storage unit 22, a communication interface unit 24, an operation unit 26, and a display unit 28. The control unit 20, storage unit 22, communication interface unit 24, operation unit 26, and display unit 28 are connected to each other via a bus 29, such as a system bus or a control bus, enabling the exchange of various types of information.

[0016] The control unit 20 in this embodiment controls the overall operation of the information processing device 10. The control unit 20 is a processor and includes a CPU (Central Processing Unit) 20A. The control unit 20 is also connected to the storage unit 22, which will be described later. The control unit 20 may also include a GPU (Graphics Processing Unit).

[0017] The operation unit 26 is used for the user to input information related to anonymization. The operation unit 26 is not particularly limited and may include, for example, various switches, a touch panel, a stylus, a mouse, and a microphone for voice input. The display unit 28 displays the anonymized document 80 and various other information. The operation unit 26 and the display unit 28 may be integrated to form a touch panel display.

[0018] The communication interface unit 24 communicates various types of information with external devices of the information processing device 10 via the network 9 using wireless or wired communication.

[0019] The memory unit 22 includes a ROM (Read Only Memory) 22A, a RAM (Random Access Memory) 22B, and a storage unit 22C. The ROM 22A has various programs and the like that are pre-stored in it, which are executed by the CPU 20A. The RAM 22B temporarily stores various data. The storage unit 22C stores the information processing program 30 executed by the CPU 20A, and other various information. In addition, the storage unit 22C of this embodiment has a document generation model 32 pre-stored in it that takes past summaries 50 and patient information 70 as input and outputs a new summary 54. The storage unit 22C is a non-volatile memory unit, and examples include HDDs (Hard Disk Drives) and SSDs (Solid State Drives).

[0020] Furthermore, Figure 3 shows a functional block diagram of an example configuration of the information processing device 10 of this embodiment. As shown in Figure 3, the information processing device 10 includes an acquisition unit 40, a format generation unit 42, a new document generation unit 44, and a display control unit 46. As an example, in the information processing device 10 of this embodiment, the CPU 20A of the control unit 20 executes an information processing program 30 stored in the storage 22C, so that the CPU 20A functions as the acquisition unit 40, the format generation unit 42, the new document generation unit 44, the display control unit 46, and the display control unit 48.

[0021] The acquisition unit 40 acquires past summaries 50. The source from which the acquisition unit 40 acquires past summaries 50, and the specific acquisition method, are not limited. As an example, in this embodiment, the acquisition unit 40 receives user identification information to identify the user, and acquires past summaries 50 corresponding to the user identification information from a past summary database (not shown) in which multiple past summaries 50 are stored, thereby acquiring past summaries 50 created by the user in the past. The past summary database may be stored inside the information processing device 10, for example, in the storage 22C, or it may be stored in a device outside the information processing device 10. The acquisition unit 40 outputs the acquired past summaries 50 to the format generation unit 42. If there are multiple past summaries 50 associated with the user's user identification information, the acquisition unit 40 may present the multiple summaries 50, accept the user's selection, and output the selected past summary 50 to the format generation unit 42. Furthermore, if the user specifies at least a portion of the patient information 70, a past summary 50 associated with patient identification information for identifying the patient may be selected, and the selected past summary 50 may be output to the format generation unit 42.

[0022] Furthermore, the acquisition unit 40 acquires patient information 70. Patient information 70 is information related to the patient's medical information and includes, for example, at least one of electronic medical records, test results, and radiology reports. In this embodiment, patient information 70 is associated with patient identification information and stored as a patient database (not shown). That is, the acquisition unit 40 acquires a past summary 50 created in the past, relating to the medical information of at least one patient.

[0023] The source from which the acquisition unit 40 acquires patient information 70, and the specific acquisition method, are not limited. As an example, in this embodiment, the acquisition unit 40 receives identification information for identifying a patient and acquires patient information 70 corresponding to the identification information from the patient database (not shown) mentioned above, thereby acquiring patient information 70 concerning the target patient for which the user intends to generate a summary. The patient database may be stored inside the information processing device 10, for example, in the storage 22C, or it may be stored in a device outside the information processing device 10. The acquisition unit 40 outputs the acquired patient information 70 to the new document generation unit 44.

[0024] The format generation unit 42 extracts items corresponding to the elements that make up the past summary 50, and generates a format corresponding to the past summary 50 based on the relative positional relationships of the extracted items. In this embodiment, the format corresponding to the past summary 50 refers to the format generated based on the past summary 50. Figure 4 shows examples of the past summary 50 (50-1 to 50-3) and format 52 (52-1 to 52-3). In this embodiment, format 52 represents the relative positional relationships of the items corresponding to the elements that make up the past summary 50.

[0025] As shown in Figure 4, the format 52 of this embodiment can be broken down into a description format and a description content. The description format includes, for example, at least one of the following: the number of line breaks, word endings, specific expressions, and the selection of terms. The description content includes, for example, at least one of the following: the category of the content or the topic. Note that a single topic may include multiple sub-topics. For example, a topic such as "hospitalization progress" may include multiple sub-topics such as "date" and "implementation details." Furthermore, as shown in Figure 4, the format 52 of this embodiment reflects the content and elements to be included in the new summary 54, the description format such as bullet points and paragraph divisions, and the personal preferences and habits of the individual's writing, including word endings and nominalizations.

[0026] A specific example of the method for generating the format 52 by the format generation unit 42 of this embodiment will be described. Figure 5 shows a flowchart illustrating an example of the flow of the format 52 generation process executed by the format generation unit 42. The generation process shown in Figure 5 will be executed in the information processing (see Figure 7), which will be described in detail later.

[0027] In step S1 of Figure 5, the format generation unit 42 obtains candidate topic divisions in the past summary 50 based on at least one of the following: paragraph, line break, paragraph, and sentence (punctuation) divisions.

[0028] In step S2, the format generation unit 42 estimates the content of each topic. The method by which the format generation unit 42 estimates the content of each topic is not limited. For example, all topic names may be defined in advance, training data may be created and trained, or classification may be performed using zero-shot learning. Alternatively, topic names may be output using LLM (Large Language Models), or the training data may be clustered and topic names assigned to each cluster. As an example, the format generation unit 42 in this embodiment uses document analysis AI (Artificial Intelligence) to estimate the content of each topic in the acquisition unit 40. Specifically, the past summary 50 is input to the document analysis AI, and the content represented by each topic output from the document analysis AI (for example, hospitalization records) is acquired. For the estimation of topic content, the entire topic, for example, the entire text of the topic, may be used, or it may be limited to a part such as a noun.

[0029] In step S3, the format generation unit 42 treats consecutive topics as the same topic without separating them. Since topics with common content estimated in step S2 are grouped into a single topic, the separator candidates for topics with common content are determined to be the same separator, and the separator position is changed. For example, if in step S1 the text is separated into sentences a, b, and c, and in step S2 the content of sentence a is estimated to be topic A, the content of sentence b is estimated to be topic A, and the content of sentence c is estimated to be topic B, then sentences a and b have common topic content, but sentence C has different topic content, so the separator position of the topics is corrected to be between sentences b and c (the separator between sentences a and b is deleted).

[0030] In step S4, the format generation unit 42 divides each topic into units smaller than those in step S1. In other words, in step S4, the format generation unit 42 divides each topic in such a way that the number of characters after division is less than in step S1. For example, if the format generation unit 42 divided the topic into paragraph units in step S1, in step S4 it divides it into sentence units.

[0031] In step S5, the format generation unit 42 estimates the content of topics for each element. It estimates each topic with even more detail than in step S2. In other words, the format generation unit 42 estimates the content corresponding to major items in step S2, and estimates the content corresponding to minor items described for major items in step S4. For example, in step S2, the format generation unit 42 estimates major items such as hospitalization, discharge, and medical examinations, and in step S4, it estimates the content at the level of minor items such as dates and details of implementation described for each event.

[0032] In step S6, the format generation unit 42 treats repeated topics as belonging to the same topic series. For example, even if the topic "Date: Implemented Content" appears on three consecutive lines, it treats it as one topic rather than three separate topics. In other words, similar to step S3 above, it adjusts the position of the delimiter for topics with identical content.

[0033] In step S7, the format generation unit 42 determines the format of the text. The format of the text may be determined for the entire past summary 50 or for each topic. Specifically, the format generation unit 42 determines the format of each section, including nominalization, word endings, term bias, and bullet points. More specifically, the format generation unit 42 determines nominalization through part-of-speech analysis. It also estimates the matching rate of predetermined expressions, such as "desu" and "masu." It also determines the frequency of terms. For example, if word A and word B included in the past summary 50 are synonyms, the one with the higher frequency is selected.

[0034] In this embodiment, steps S1 to S3 are performed to process the first topic, and steps S4 to S6 are performed to process a second topic that is more detailed than the first topic, similar to steps S1 to S3. However, the embodiment is not limited to this. For example, between steps S6 and S7, steps S8 to S10 may be performed to add more detailed content to a third topic that is more detailed than the second topic, similar to steps S1 to S3. In other words, the same processing as steps S1 to S3 may be repeated multiple times (N≧2).

[0035] Furthermore, the format generation unit 42 replaces proper nouns and the like with symbols. The symbols to be replaced preferably include strings, for example, special tokens. Figure 6 shows an example of format 52 in which the proper nouns in the past summary 50 have been replaced with special tokens. In format 52 shown in Figure 6, the "date," "examination name," and "treatment" have been replaced with special tokens.

[0036] Through the above processing, the format generation unit 42 generates a format 52 corresponding to the past summary 50, which represents the relative positional relationship of multiple items (topics). The format generation unit 42 outputs the generated format 52 to the new document generation unit 44.

[0037] The new document generation unit 44 inputs the format 52 and patient information 70 into the document generation model 32 and acquires the new summary 54 output from the document generation model 32. The new summary 54 is a summary related to the medical information of a specific patient. The new document generation unit 44 outputs the acquired new summary 54 to the display control unit 46.

[0038] The display control unit 48 controls the display unit 28 to display the new summary 54.

[0039] Next, the operation of the information processing device 10 of this embodiment will be described with reference to the drawings. Figure 7 shows a flowchart illustrating an example of the flow of information processing performed in the information processing device 10 of this embodiment. In this embodiment, as an example, the information processing device 10 executes the information processing shown as an example in Figure 7 by having the CPU 20A of the control unit 20 execute the information processing program 30 stored in the storage 22C based on user generation instructions etc. made by the operation unit 26.

[0040] In step S100 of Figure 7, the acquisition unit 40 acquires the past summary 50. As described above, the acquisition unit 40 of this embodiment acquires the past summary 50 associated with the user identification information from the past summary database (not shown). If there are multiple past summaries 50 associated with the user identification information, one of them is acquired. If the past summaries 50 are divided into multiple types, such as for discharge, and the types are stored in the past summary database in a way that allows for type identification, the system may acquire the past summary 50 of the type specified by the user.

[0041] In the next step S102, the format generation unit 42 generates a format 52 from the past summary 50. As described above, the format generation unit 42 of this embodiment performs a generation process (see Figure 5) and generates a format 52 corresponding to the past summary 50, which represents the relative positional relationship of items (topics) corresponding to the elements that make up the past summary 50.

[0042] In the next step S104, the acquisition unit 40 acquires patient information 70. As described above, the acquisition unit 40 of the present embodiment acquires the patient information 70 corresponding to the patient identification information representing the patient specified by the user from a patient database (not shown).

[0043] In the next step S106, the new document generation unit 44 generates a new summary 54. As described above, the new document generation unit 44 of the present embodiment inputs the format 52 generated in step S102 and the patient information 70 acquired in step S104 into the document generation model 32, and acquires the new summary 54 output from the document generation model 32, thereby generating the new summary 54.

[0044] In the next step S108, the display control unit 46 causes the display unit 28 to display the new summary 54 generated in step S106. In the present embodiment, the past summary 50 and the format 52 are also displayed on the display unit 28. FIG. 8 shows an example of the display modes of the past summary 50, the format 52, and the new summary 54. As shown in FIG. 8, in the present embodiment, the format 52 and the new summary 54 are displayed so as to be comparable.

[0045] Also, as shown in FIG. 8, the display control unit 46 of the present embodiment causes the display unit 28 to display a button 60 for instructing correction of the format 52 and a button 62 for instructing change. The user can give an instruction to correct the format 52 by pressing the button 60. The method of correcting the format 52 by the user is not limited. For example, each element may be corrected for each topic. For example, in the example shown in FIG. 8, the description format of "date" from "MM (month) / DD (day)" to "YY (year) / MM (month) / DD (day)" is shown as an example of correction. Also, for example, it may be possible to correct by deleting any one of a plurality of topics. Also, for example, it may be possible to add new elements or items. Also, the user may desire another format 52 rather than correcting the format 52. In that case, an instruction to change the format 52 can be given by pressing the button 62.

[0046] As described above, since the format 52 can be modified, in the next step S110, the format generation unit 42 determines whether it has received an instruction to modify the format 52. If it has received an instruction to modify the format 52, the determination in step S110 is an affirmative determination, and the process proceeds to step S112.

[0047] In step S112, after the format generation unit 42 acquires the modification content, it returns to step S106 and repeats the processes of steps S106 to S110.

[0048] On the other hand, if it has not received a modification instruction, the determination in step S110 is a negative determination, and the process proceeds to step S114.

[0049] In step S114, the acquisition unit 40 determines whether it has received a change to the format 52. As described above, when the button 62 is pressed, the acquisition unit 40 receives a change to the format 52. If it has received a change to the format 52, the determination in step S114 is an affirmative determination, and the process returns to step S100 and repeats the processes of steps S100 to S114. In step S100 when repeated, another past summary 50 that has not yet been acquired is acquired. If there is no available past summary 50, information indicating that may be displayed. Also, when repeated, the process of step S104 may be omitted, and the patient information 70 of step S104 acquired by the previous process may be used.

[0050] On the other hand, if it has not received a change to the format 52, the determination in step S114 is a negative determination, and the information processing shown in FIG. 7 ends.

[0051] Note that the configurations and operations of the information processing apparatus 10 and the like described in each of the above embodiments are examples, and it is needless to say that they can be changed according to the situation without departing from the gist of the present invention. For example, it may be in the form of the following modification examples, and it is also needless to say that each of the above embodiments and the following modification examples may be appropriately combined.

[0052] (Modification 1) The display control unit 46 may display multiple past summaries 50 on the display unit 28, and the user may select from the displayed past summaries 50 to be used for generating the format 52. Figure 9 shows a case where past summary 50-7 is selected by the user from among the four past summaries 50 (50-4 to 50-7) displayed by the display control unit 46. When displaying multiple past summaries 50 in this way, the display control unit 46 may display them in an order according to predetermined conditions, such as prioritizing the most recent one.

[0053] The format generation unit 42 generates a format 52 corresponding to the specified past summary 50-7.

[0054] In the example shown in Figure 9, past summaries 50-4 to 50-7 are all of the same type and relate to the "same disease". In this embodiment, the types of past summaries 50 can be displayed in a pull-down menu, from which the user can select the type corresponding to the new summary 54 to be created.

[0055] (Modification 2) The format generation unit 42 may store multiple generated formats 52. For example, the format 52 may be stored in association with the corresponding past summary 50. By storing the format 52 in this way, it can be used from the next time onward, eliminating the need to generate the format 52 multiple times for the same past summary 50, thereby reducing the load associated with generation.

[0056] Furthermore, the frequency with which multiple formats 52 have been used to generate new summaries 54, i.e., the usage frequency of format 52, may be stored, and the display control unit 46 may display information according to that frequency. In this case, the system may target formats 52 used to generate new summaries 54 generated by a specific user, or it may target all formats 52. For example, in the example shown in Figure 10, the frequency with which each of formats 52-4 to 52-7 has been used to generate new summaries 54 (in "%" in Figure 10) is displayed. By displaying the frequency with which formats 52 have been used to generate new summaries 54 in this way, it is possible to know which formats 52 have been used in the past. In particular, if the system targets formats 52 used in a certain medical department, it can make it easier for new doctors, such as doctors newly assigned to that department, to select a format 52.

[0057] Furthermore, when storing the generated format 52, in order to prevent the number of stored formats 52 from becoming enormous, formats 52 stored in the database may be deleted according to their usage frequency. For example, formats 52 may be deleted from the database if their usage frequency within a predetermined period is below a threshold. Alternatively, for example, for each type of new summary 54, only a predetermined number of formats 52 may be stored in the database, in order of their usage frequency.

[0058] Furthermore, if a predetermined number or more of formats 52 and their usage frequencies are stored in the database, the user's selection of past summaries 50 may be omitted. In other words, the information processing steps S100, S102, and S114 shown in Figure 7 may be omitted, and the display unit 28 may be configured to display the format 52, usage frequency, and new summary 54, as shown in Figure 10. This simplifies the processing procedure.

[0059] (Modification 3) In the above embodiment, a configuration was described in which the format generation unit 42 generates one format 52 from one past summary 50. However, a configuration in which one format 52 is generated from multiple past summaries 50 is also possible.

[0060] The acquisition unit 40 may, for example, acquire multiple past summaries 50 to which the same user identification information is associated. However, the acquisition condition for acquiring multiple past summaries 50 is not limited to cases where the same user identification information is associated. In other words, the multiple past summaries 50 to be acquired may include past summaries 50 to which different user identification information is associated. The acquisition unit 40 may, for example, acquire multiple past summaries 50 to which the same medical department or other group to which the user belongs, or it may acquire multiple past summaries to which some of the patient information, such as diseases, associated with the past summaries 50 is the same. Furthermore, the format generation unit 42 may, for example, generate a format from each of the past summaries 50 and generate a format 52 by extracting the common part of the multiple generated formats. In this case, for example, only the part that completely matches among the multiple formats generated from each past summary 50 may be used as format 52. Alternatively, for example, a part with a certain degree of agreement may be used as format 52. Specifically, the threshold may be set to 7 items, and the portion where 7 or more items match out of a total of 10 formats may be designated as format 52.

[0061] (Modification 4) The information processing device 10 may function as a learning device for learning the document generation model 32. In this case, the information processing program 30 may include a learning program for causing the document generation device 10 to function as a learning device, or it may have a separate learning program in addition to the information processing program 30. For example, the new document generation unit 44 may function as a learning unit for learning the document generation model 32. In this case, the new document generation unit 44 acquires multiple types of formats 52 corresponding to the past summaries 50, and learns the document generation model 32 by having the document generation model 32 learn using learning data with formats 52 for each type of format 52.

[0062] For example, as shown in Figure 11, the new document generation unit 44 generates a format 52 corresponding to each past summary 50 from a set of pre-prepared past summaries 50, which are the learning data, using a formatting model for generating a format, and uses this as formatted learning data. The past summaries 50 used as learning data may be those prepared for learning, or they may include the past summaries 50 used to generate the new summary 54.

[0063] The new document generation unit 44 clusters the format 52 of the training data with format 52 and classifies it into multiple types. In the example shown in Figure 11, a configuration is shown in which the data is clustered into three types: format A, format B, and format C. The new document generation unit 44 then generates a document generation model 32 for format A by training with training data for format A, generates a document generation model 32 for format B by training with training data for format B, and generates a document generation model 32 for format C by training with training data for format C.

[0064] In this way, by training the document generation model 32 according to the type of format 52, the accuracy of the document generation model 32 can be improved.

[0065] As described above, the information processing device 10 of the above embodiment comprises an acquisition unit 40, a format generation unit 42, and a new document generation unit 44. The acquisition unit 40 acquires past summaries 50, which are past documents. The acquisition unit 40 also acquires patient information 70, which is input information. The format generation unit 42 extracts items corresponding to the elements that make up the past summary 50 and generates a format 52 corresponding to the past summary 50 based on the relative positional relationship of the extracted items. For the new summary 54, the format 52 and patient information 70 are input to the document generation model 32, and the new summary 54 output from the document generation model 32 is acquired.

[0066] As described above, in the information processing device 10 of each embodiment, the string to be anonymized is detected from the first masked document 60 after the specific information string has been masked. Therefore, even if the specific information string contains a string similar to the string to be anonymized, that string will not be detected as the string to be anonymized.

[0067] Therefore, the information processing device 10 of the above embodiment can better support the generation of new documents compared to the case where a predetermined format is provided. Furthermore, since the format generation unit 42 in the information processing device 10 of the above embodiment generates the format 52 based on the relative positional relationship of items corresponding to the elements constituting the past summary 50, the format 52 can be maintained regardless of the type or size of the medium on which the new summary 54 is output.

[0068] In the above embodiments and modifications, a configuration was described in which the past document is a past summary 50 and the input information is patient information 70. However, the types of past summaries 50 and patient information 70 are not limited to these.

[0069] Furthermore, in this embodiment, each process is executed on any computer. Alternatively, any computer may execute these processes using a processor as hardware, a program as software, or a combination thereof. In that case, the processor is configured to work in cooperation with the program to execute the various processes in this embodiment, and can function as a unit or means in this embodiment. The execution order of the processes by the processor is not limited to the order described and may be changed as appropriate. Any computer may be a general-purpose computer, a computer designed for a specific purpose, a workstation, or any other system capable of executing each process.

[0070] A processor may consist of one or more hardware components, and the type of hardware is not limited. For example, a processor may consist of programmable logic devices such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), FPGA (Field Programmable Gate Array), dedicated circuits for performing specific processing such as an ASIC (Application Specific Integrated Circuit), a GPU (Graphic Processing Unit), or an NPU (Neural Processing Unit). Furthermore, the type of hardware may be a combination of different types of hardware. When multiple hardware components are configured to execute one or more processes of a processor, these components may reside in physically separate devices or in the same device. Furthermore, in any embodiment, the order of the processes performed by the processor is not limited to the order described above and may be changed as appropriate. The hardware components are composed of electrical circuits (circuits) and the like, which are combinations of circuit elements such as semiconductor elements.

[0071] Furthermore, the program may be firmware or software such as microcode. Alternatively, the program may be, for example, a set of program modules, each function of which may be implemented by a processor configured to perform its respective function. The program may be program code or multiple code segments stored on one or more non-temporary computer-readable media (e.g., storage media or other storage). The program may be divided and stored on multiple non-temporary computer-readable media located on physically separate devices. Program code or code segments may represent any combination of procedures, functions, subprograms, routines, subroutines, modules, software packages, classes, or instructions, data structures, or program statements. Program code or code segments may be connected to other code segments or hardware circuits by sending and receiving information, data, arguments, parameters, or memory contents.

[0072] Furthermore, although the above embodiment describes an embodiment in which the information processing program 30 is pre-stored (installed) in the storage 22C of the memory unit 22, the invention is not limited to this. The information processing program 30 may be provided in the form of a recording medium such as a CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), or USB (Universal Serial Bus) memory. Alternatively, the information processing program 30 may be provided in the form of a download from an external device via a network.

[0073] Furthermore, the technology disclosed herein extends to all program products. A program product includes all forms of products for providing programs. For example, a program product includes programs provided via a network such as the Internet, and non-temporary computer-readable recording media such as CD-ROMs, DVDs, and USB memory sticks on which programs are stored. The present invention is also applicable to programs and program products.

[0074] The following additional information is disclosed regarding the above-described embodiments.

[0075] (Note 1) An information processing device comprising a processor, the processor acquires past documents, extracts items corresponding to the elements constituting the past documents, generates a format corresponding to the past documents based on the relative positional relationship of the extracted items, acquires input information, inputs the format and the input information into a document generation model, and acquires a new document output from the document generation model.

[0076] (Note 2) The information processing apparatus described in Note 1, wherein the processor generates the format based on the form of expression in the sentences contained in the past document.

[0077] (Note 3) The information processing apparatus according to Note 1 or Note 2, wherein the processor generates the format by replacing the items using at least one of a specific part of speech and a specific term in the past document with symbols.

[0078] (Note 4) The information processing apparatus according to any one of Notes 1 to 3, wherein the processor generates a plurality of formats corresponding to at least one of the creator of the past document and the type of the past document.

[0079] (Note 5) The information processing apparatus according to any one of Notes 1 to 4, wherein the processor displays the format and accepts modifications to the displayed format.

[0080] (Note 6) The information processing device described in Note 5, wherein the processor displays the new document, and if a modification is received after the display of the new document, it inputs the input information and the modified format into the document generation model, and retrieves the new document output from the document generation model again.

[0081] (Note 7) The information processing apparatus according to any one of Notes 1 to 6, wherein the processor displays the format and the new document in a comparable manner.

[0082] (Note 8) The information processing device according to any one of Notes 1 to 7, wherein the processor displays multiple past documents and generates a format corresponding to a specified past document from among the multiple past documents.

[0083] (Note 9) The processor is the information processing device described in Note 8, which displays the plurality of past documents in an order according to predetermined conditions.

[0084] (Note 10) The information processing device according to any one of Notes 1 to 7, wherein the processor acquires a plurality of past documents, displays a past document from among the acquired plurality of past documents that meets a predetermined condition, and generates a format corresponding to a specified past document from among the past documents that meet the predetermined condition.

[0085] (Note 11) The information processing apparatus according to any one of Notes 1 to 7, wherein the processor generates a plurality of formats, displays the plurality of formats according to the frequency with which they were used to generate the new document, and inputs the format selected from the plurality of formats and the input information into the document generation model.

[0086] (Note 12) The information processing device described in any one of Notes 1 to 11, wherein the past document is a summary relating to the medical information of at least one patient, the input information is the medical information of a specific patient, and the new document is a summary relating to the medical information of the specific patient.

[0087] (Note 13) A learning device comprising a processor, wherein the processor acquires multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past documents, and generates a document generation model for each type of format using the formatted learning data.

[0088] (Note 14) An information processing method which involves obtaining past documents, extracting items corresponding to the elements constituting the past documents, generating a format corresponding to the past documents based on the relative positional relationship of the extracted items, obtaining input information, inputting the format and the input information into a document generation model, and obtaining a new document output from the document generation model.

[0089] (Note 15) A learning method that acquires multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past document, and generates a document generation model for each type of format using training data with the format.

[0090] (Note 16) An information processing program that causes a computer to perform the following processes: acquire past documents, extract items corresponding to the elements constituting the past documents, generate a format corresponding to the past documents based on the relative positional relationship of the extracted items, acquire input information, input the format and the input information into a document generation model, and acquire a new document output from the document generation model.

[0091] (Note 17) A learning program that causes a computer to perform the process of obtaining multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past document, and generating a document generation model for each type of format using the formatted learning data.

[0092] (Note 18) A computer program product that includes an information processing program that causes a processor to perform the following processes: acquire past documents, extract items corresponding to the elements constituting the past documents, generate a format corresponding to the past documents based on the relative positional relationship of the extracted items, acquire input information, input the format and the input information into a document generation model, and acquire a new document output from the document generation model.

[0093] (Note 19) A computer program product that includes a learning program for a processor to perform the process of obtaining multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past documents, and generating a document generation model for each type of format using the formatted learning data.

[0094] (Note 20) A computer-readable storage medium storing an information processing program that causes the processor to perform the following processes: acquire past documents, extract items corresponding to the elements constituting the past documents, generate a format corresponding to the past documents based on the relative positional relationship of the extracted items, acquire input information, input the format and the input information into a document generation model, and acquire a new document output from the document generation model.

[0095] (Note 21) A computer-readable storage medium in which a learning program is stored, which causes the processor to acquire multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past documents, and to execute a process for generating a document generation model using the formatted learning data for each type of format.

[0096] The disclosure of Japanese Patent Application No. 2024-171854 is incorporated herein by reference in its entirety.

[0097] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

Claims

1. An information processing device comprising a processor, wherein the processor acquires past documents, extracts items corresponding to the elements constituting the past documents, generates a format corresponding to the past documents based on the relative positional relationship of the extracted items, acquires input information, inputs the format and the input information into a document generation model, and acquires a new document output from the document generation model.

2. The information processing apparatus according to claim 1, wherein the processor generates the format based on the format of expression in the sentences contained in the past document.

3. The information processing apparatus according to claim 1, wherein the processor generates the format by replacing the items using at least one of a specific part of speech and a specific term in the past document with symbols.

4. The information processing apparatus according to claim 1, wherein the processor generates a plurality of formats corresponding to at least one of the creator of the past document and the type of the past document.

5. The information processing apparatus according to claim 1, wherein the processor displays the format and accepts modifications to the displayed format.

6. The information processing apparatus according to claim 5, wherein the processor displays the new document, and if a modification is received after the display of the new document, it inputs the input information and the modified format into the document generation model, and retrieves the new document output from the document generation model again.

7. The information processing apparatus according to claim 1, wherein the processor displays the format and the new document in a comparative manner.

8. The information processing apparatus according to claim 1, wherein the processor displays a plurality of past documents and generates a format corresponding to a specified past document from among the plurality of past documents.

9. The information processing apparatus according to claim 8, wherein the processor displays the plurality of past documents in an order according to predetermined conditions.

10. The information processing apparatus according to claim 1, wherein the processor acquires a plurality of past documents, displays a past document from among the acquired plurality of past documents that meets predetermined conditions, and generates a format corresponding to a specified past document from among the past documents that meet the predetermined conditions.

11. The information processing apparatus according to claim 1, wherein the processor generates a plurality of formats, displays the plurality of formats according to the frequency with which they were used to generate the new document, and inputs the format selected from the plurality of formats and the input information into the document generation model.

12. The information processing device according to claim 1, wherein the past document is a summary relating to the medical information of at least one patient, the input information is the medical information of a specific patient, and the new document is a summary relating to the medical information of the specific patient.

13. A learning device comprising a processor, wherein the processor acquires multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past documents, and generates a document generation model for each type of format using the formatted training data.

14. An information processing method that obtains past documents, extracts items corresponding to the elements constituting the past documents, generates a format corresponding to the past documents based on the relative positional relationships of the extracted items, obtains input information, inputs the format and the input information into a document generation model, and obtains a new document output from the document generation model.

15. A learning method that acquires multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past document, and generates a document generation model for each type of format using training data with the format.

16. An information processing program that causes a computer to perform the following processes: acquire past documents, extract items corresponding to the elements constituting the past documents, generate a format corresponding to the past documents based on the relative positional relationships of the extracted items, acquire input information, input the format and the input information into a document generation model, and acquire a new document output from the document generation model.

17. A learning program that causes a computer to perform the process of obtaining multiple types of formats corresponding to past documents based on the relative positional relationships of multiple items corresponding to elements constituting the past document, and generating a document generation model for each type of format using the formatted training data.

Images