Document generation device, document generation method, and document generation program

Abstract

Provided is a document generation device including a processor. The processor acquires first input information, generates a new document on the basis of past documents generated in the past and an additional document corresponding to the first input information, and, when generating the new document, generates the additional document using a document generation model.

Description

Document Generation Device, Document Generation Method, and Document Generation Program

[0001] The present disclosure relates to a document generation device, a document generation method, and a document generation program.

[0002] Techniques for assisting a user in creating a document are known (see, for example, Japanese Patent Application Laid-Open No. 2011-123623 (Patent Document 1)).

[0003] By the way, a document generation model that generates a document based on input information is known. When attempting to generate a new document using the document generation model, the processing time may become long, and shortening the processing time is desired.

[0004] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide a document generation device, a document generation method, and a document generation program that can shorten the time required to generate a new document.

[0005] To achieve the above object, the document generation device of the present disclosure includes a processor. The processor acquires first input information, and generates a new document based on a past document generated in the past and an additional document corresponding to the first input information. When generating a new document, the processor generates the additional document using a document generation model.

[0006] Also, to achieve the above object, the document generation method of the present disclosure acquires first input information, and generates a new document based on a past document generated in the past and an additional document corresponding to the first input information. When generating a new document, the processor generates the additional document using a document generation model.

[0007] Also, to achieve the above object, the document generation program of the present disclosure is for causing a computer to execute a process of acquiring first input information, generating a new document based on a past document generated in the past and an additional document corresponding to the first input information, and generating the additional document using a document generation model when generating a new document.

[0008] According to the present disclosure, the time required to generate a new document can be shortened.

[0009] This is a diagram illustrating the overview of new document generation by the document generation device of the first embodiment. This is a block diagram showing an example of the configuration of the document generation device of the embodiment. This is a functional block diagram showing an example of the configuration of the document generation device of the embodiment. This is a flowchart showing an example of the document generation process flow by the document generation device of the first embodiment. This is a flowchart showing another example of the document generation process flow by the document generation device of the first embodiment. This is a diagram illustrating the overview of new document generation by the document generation device of the second embodiment. This is a flowchart showing an example of the document generation process flow by the document generation device of the second embodiment. This is a diagram illustrating the overview of new document generation by the document generation device of modification 1. This is a flowchart showing an example of the document generation process flow by the document generation device of modification 2.

[0010] The 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. As a specific example, the case in which a healthcare professional is the user and generates a summary will be described below.

[0011] [First Embodiment] Figure 1 shows a diagram illustrating the overview of new document generation by the document generation device of this embodiment. As shown in Figure 1, the document generation device of this embodiment is a device that generates a new document 62 from past documents 52 that were generated at least in the past before the generation of the new document, past input information 50 that served as the information source in the generation of past documents 52, and new input information 60 which is the information source for generating an additional document 54 to be added to past documents 52. In the example shown in Figure 1, the document generation device 10 inputs the past input information 50, past documents 52, and new input information 60 into the document generation model 32, and obtains the new document 62 output from the document generation model 32. The new document 62 output from the document generation model 32 is a document to which an additional document 54 has been added to past documents 52. The past input information 50 and new input information 60, which are the information sources for generating documents, are, for example, medical data and include not only text data but also image data, etc.

[0012] Figure 2 shows a block diagram illustrating an example of the configuration of the document generation device 10 of this embodiment. As shown in Figure 2, the document generation 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 control bus, enabling the exchange of various types of information.

[0013] The control unit 20 in this embodiment controls the overall operation of the document generation 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).

[0014] The operation unit 26 is used for the user to input information related to the generation of a new document. 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 information. The operation unit 26 and the display unit 28 may be integrated to form a touch panel display.

[0015] The communication interface unit 24 communicates various types of information with external devices of the document generation device 10 via a network using wireless or wired communication.

[0016] 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 document generation 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 used for generating new documents pre-stored in it. The storage unit 22C also stores generated documents and information (input information) used for generating documents. The storage unit 22C is a non-volatile memory unit, and examples include an HDD (Hard Disk Drive) and an SSD (Solid State Drive).

[0017] Furthermore, Figure 3 shows a functional block diagram of an example of the configuration of the document generation device 10 of this embodiment. As shown in Figure 3, the document generation device 10 comprises an acquisition unit 40, a generation unit 42, and a display control unit 44. As an example, in the document generation device 10 of this embodiment, the CPU 20A of the control unit 20 executes a document generation program 30 stored in the storage 22C, so that the CPU 20A functions as the acquisition unit 40, the generation unit 42, and the display control unit 44.

[0018] The acquisition unit 40 acquires new input information 60. As described above, the new input information 60 is a source of information for generating an additional document 54 to be added to the past document 52. If the new input information 60 is stored in the document database 33, the acquisition unit 40 acquires the new input information 60 from the document database 33. In this case, the acquisition unit 40 refers to the time information associated with the past document 52 and acquires the new input information 60 from the document database 33 that is associated with time information from this time onward. The time information associated with the document or input information refers to information about at least one of the month, day, and date and time when the document or input information was generated. On the other hand, if the user directly inputs the new input information 60 to the document generation device 10 when generating a new document 62, the acquisition unit 40 acquires the input new input information 60. For example, the acquisition unit 40 may acquire the new input information 60 by distinguishing and recording input information from after the past document 52 was written as new input information 60. The new input information 60 in this embodiment is an example of the first input information of this disclosure.

[0019] Furthermore, the acquisition unit 40 of this embodiment also acquires past input information 50 and past documents 52. As an example, in this embodiment, past input information 50 and past documents 52 are stored in the document database 33 in an associated state. Therefore, the acquisition unit 40 acquires past input information 50 and past documents 52 from the document database 33. The past input information 50 of this embodiment is an example of the second input information of this disclosure. In this embodiment, it is assumed that past input information 50 and past documents 52 are stored in the document database 33 in an associated state, but this is not limited to this. The acquisition unit 40 may acquire as past input information 50 input information to which time information prior to the time information associated with past documents 52 is associated.

[0020] The acquisition unit 40 outputs the acquired past input information 50, past documents 52, and new input information 60 to the generation unit 42.

[0021] The generation unit 42 uses a document generation model 32 to generate a new document 62 based on past documents 52 and additional documents 54 corresponding to new input information 60. For example, the document generation model 32 can be a Large Language Model (LLM), a type of artificial intelligence (AI). The document generation model 32 may be owned by an external device separate from the document generation device 10. The generation unit 42 may output input information such as new input information 60 to the external device possessing the document generation model 32 via a network or the like, and receive output information such as additional documents 54 from the external device possessing the document generation model 32.

[0022] As described above, the generation unit 42 inputs the new input information 60, the past input information 50, and the past document 52 into the document generation model 32. Specifically, the generation unit 42 inputs the "input document + generation instruction token ([ / INST], etc.)" followed by the token of the past document 52, which is a previously generated document, into the document generation model 32. Note that by fixing the number of characters in the past document 52, the processing time can be reduced. The document generation model 32 generates an additional document 54 using at least the new input information 60 as an information source, and outputs a new document 62 that combines the past document 52 and the generated additional document 54.

[0023] The generation unit 42 acquires the new document 62 output from the document generation model 32 and outputs it to the display control unit 44.

[0024] The display control unit 44 controls the display of the new document 62 on the display unit 28.

[0025] Next, the operation of the document generation device 10 of this embodiment will be described with reference to the drawings. Figure 4A shows a flowchart illustrating an example of the flow of the document generation process performed in the document generation device 10 of this embodiment. In this embodiment, as an example, the document generation device 10 executes the document generation process shown as an example in Figure 4A by having the CPU 20A of the control unit 20 execute the document generation program 30 stored in the storage 22C based on user instructions made by the operation unit 26.

[0026] In step S100 of Figure 4A, the acquisition unit 40 acquires the new input information 60. As described above, in this embodiment, the acquisition unit 40 acquires the new input information 60 entered by the user when the user enters the new input information 60. Also, if the new input information 60 is included in the document database 33 of the storage 22C, the acquisition unit 40 acquires the new input information 60 from the storage 22C according to the user's instructions. For example, the acquisition unit 40 acquires the new input information 60 for medical information such as electronic medical records that is associated with the patient identification information entered by the user.

[0027] In this embodiment, the acquisition unit 40 acquires the new input information 60 by acquiring it from the input information based on the patient identification information entered by the user. However, the method for acquiring the new input information 60 is not limited to this method. For example, the new input information 60 may be acquired based on new period information corresponding to the new document 62. Here, new period information is information indicating the period of input information to be included in the new document, and is defined, for example, by a start date and an end date. As described above, the new document 62 is a document that combines the past document 52 and the additional document 54. Therefore, the period indicated by the new period information includes the date and time corresponding to the past input information 50 and the date and time corresponding to the new input information 60. The new period information may be entered directly by the user, or a corresponding period, such as the hospitalization period, may be defined for each document. In this case, the acquisition unit 40 may acquire the new input information 60 from the input information corresponding to the period indicated by the period information.

[0028] Alternatively, the acquisition unit 40 may acquire the new input information 60 based on the period corresponding to the new input information 60, in other words, the additional period information corresponding to the additional document 64. In this case, the acquisition unit 40 acquires the additional period information corresponding to the new input information 60. The acquisition unit 40 may also acquire input information as the new input information 60, where the date and time indicated by the associated time information is included in the period indicated by the additional period information.

[0029] In the next step S102, the acquisition unit 40 acquires past input information 50 and past documents 52. As described above, the acquisition unit 40 acquires past input information 50 and past documents 52 from the storage 22C in accordance with the user's instructions. For example, the acquisition unit 40 acquires past documents 52 associated with patient identification information entered by the user from the storage 22C, and also acquires past input information 50 associated with past documents 52.

[0030] In this embodiment, the acquisition unit 40 acquires past input information 50 and past documents 52 based on patient identification information entered by the user. However, the method for acquiring past documents 52 and past input information 50 is not limited to this method. For example, the acquisition unit 40 may acquire new period information corresponding to a new document 62, and among a plurality of documents, acquire as past documents 52 documents whose date and time indicated by the time information associated with the document falls within the period indicated by the new period information. If there are multiple past documents 52, the acquisition unit 40 may present the multiple past documents 52 to the user, accept a selection, and acquire the past document 52 selected by the user, as well as the past input information 50 corresponding to the past document 52.

[0031] In this embodiment, the order in which the acquisition unit 40 acquires new input information 60 and then acquires past input information 50 and past documents 52 is such that new input information 60 is acquired first, followed by past documents 52 and past input information 50. However, this is not limited to this order. That is, the execution order of steps S100 and S102 in the document generation process is not limited to the order shown in Figure 4. For example, as shown in Figure 4B, the process of step S102 may be executed first, followed by the process of step S100. That is, the acquisition unit 40 may acquire past documents 52 and past input information 50 before acquiring new input information 60. In this case, for example, the acquisition unit 40 may first acquire new period information corresponding to the new document 62 as the process of step S102, and then acquire as past documents 52 documents the date and time indicated by the time information associated with the document that falls within the period indicated by the new period information. The acquisition unit 40 may also acquire as past input information 50 input information the date and time indicated by the time information associated with the input information that falls within the period indicated by the new period information. Next, as part of the process in step S100, the acquisition unit 40 may acquire new input information 60 by excluding past input information 50 from all input information.

[0032] In the next step S204, the derivation unit 46 generates a new document 62 using the document generation model 32. As described above, the derivation unit 46 inputs the past input information 50, the past document 52, and the new input information 60 into the document generation model 32 and obtains a new document 62 which is a combination of the past document 52 and the additional document 54.

[0033] In the next step, S106, the display control unit 44 determines whether or not to display the new document 62. In this embodiment, the user can choose whether to display the new document 62 immediately after generation, or to generate it first and display the generated new document 62 later. If it is selected to display the new document 62 later, the determination in step S106 is negative, and the process proceeds to step S110. On the other hand, if it is selected to display the new document 62 immediately, the determination in step S106 is negative, and the process proceeds to step S108.

[0034] In step S108, the display control unit 44 causes the new document 62 to be displayed on the display unit 28.

[0035] In the next step S110, the display control unit 44 stores the new document 62 in the storage 22C. When the processing in step S110 is completed, the document generation process shown in Figure 4A is completed.

[0036] Thus, the document generation device 10 of this embodiment can generate a new document 62 that includes the past document 52 by newly generating an additional document 54 to be added to the past document 52.

[0037] Furthermore, because the additional document 54 is combined with the past document 52, the new document 62 may become long. In this case, the generation unit 42 of the document generation device 10 may further perform processing to adjust the length of the new document 62. Whether or not to adjust the length of the new document 62 may be done depending on the type of new document 62. For example, a threshold for the number of characters may be associated and recorded for each type of new document 62, and the generation unit 42 may read the threshold according to the type of new document 62 to be generated, and adjust the length of the new document 62 so that the length of the new document 62 is shortened if the number of characters in the new document 62 is greater than or equal to the read threshold. Alternatively, the generation unit 42 may adjust the length of the new document 62 using a rule-based method such as deleting duplicate entries, or by using a summary document generation AI that shortens the length of the document.

[0038] [Second Embodiment] Figure 5 shows a diagram illustrating the overview of new document generation by the document generation device of this embodiment. As shown in Figure 5, the document generation device of this embodiment is a device that generates a new document 62 from past documents 52 and new input information 60. In the example shown in Figure 5, the document generation device 10 inputs the new input information 60 into the document generation model 32 and obtains the additional document 54 output from the document generation model 32. The document generation device 10 generates a new document 62 by combining the additional document 54 generated by the document generation model 32 with the past documents 52.

[0039] The configuration of the document generation device 10 in this embodiment is the same as that of the document generation device 10 in the first embodiment (see FIGS. 2 and 3). However, since some of the functions of the acquisition unit 40 and the generation unit 42 are different, the acquisition unit 40 and the generation unit 42 in this embodiment will be described.

[0040] The acquisition unit 40 acquires the past document 52 and the new input information 60. In this embodiment, different from the first embodiment, since the past input information 50 is not used for generating the new document 62, the acquisition unit 40 does not acquire the past input information 50. The acquisition unit 40 outputs the acquired past document 52 and new input information 60 to the generation unit 42.

[0041] The generation unit 42 in this embodiment also uses the document generation model 32 to generate a new document 62 based on the past document 52 and the additional document 54 corresponding to the new input information 60. Specifically, as described above, the generation unit 42 inputs the new input information 60 into the document generation model 32. The document generation model 32 generates an additional document 54 using the new input information 60 as an information source. The generation unit 42 generates a new document 62 by combining the past document 52 and the additional document 54 so as to add the additional document 54 after the past document 52. The generation unit 42 outputs the newly generated new document 62 to the display control unit 44.

[0042] Also, the document generation process executed in the document generation device 10 in this embodiment is different in that, instead of the step S102 of the document generation process (see FIGS. 4A and 4B) executed in the document generation device 10 in the first embodiment, the process of step S103 shown in FIG. 6 is performed.

[0043] In this embodiment, as described above, the acquisition unit 40 does not acquire the past input information 50. Therefore, as shown in FIG. 6, the acquisition unit 40 acquires only the past document 52 in step S103.

[0044] In this way, also in this embodiment, the document generation device 10 can generate a new document 62 including the past document 52 by newly generating an additional document 54 to be added to the past document 52.

[0045] Note that the configurations, operations, etc. of the document generation device 10 and the like described in each of the above embodiments are merely examples, and it goes without saying 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 also goes without saying that each of the above embodiments and the following modification examples may be appropriately combined.

[0046] (Modification Example 1) FIG. 7 shows a diagram for explaining an overview of the generation of a new document by the document generation device of the present embodiment. As shown in FIG. 7, the document generation device 10 of this modification example inputs the past document 52 and the new input information 60 to the document generation model 32, and acquires the new document 62 output from the document generation model 32. The new document 62 output from the document generation model 32 is a document in which an additional document 54 is added to the past document 52.

[0047] In the first embodiment, the past input information 50 was also input to the document generation model 32, but in this modification example, the past input information 50 is not input to the document generation model 32.

[0048] In the document generation device 10 of this modification example, the generation unit 42 gives a prompt "Please generate a sentence following the past document based on the new input information" to the document generation model 32, whereby the document generation model 32 generates a new document 62 in which the past document 52 and the additional document 54 are combined.

[0049] (Modification Example 2) In each of the above embodiments and modification examples, the mode of generating the new document 62 at the timing instructed by the user has been described, but the timing of generating the new document 62 is not limited to this mode. As a modification example of the timing of generating the new document 62, the timing based on the operation rate of the document generation device 10 will be described here.

[0050] In this modified version, the document generation device 10 generates a new document 62 based on previously accumulated new input information 60 when the operating rate of the document generation device 10 is relatively low, such as on holidays or at night. In this case, the new input information 60 is sequentially accumulated in the document database 33 of the storage 22C. Furthermore, the type of new document 62 to be generated is pre-configured for the document generation device 10. For example, it is configured to generate a hospitalization summary as a new document 62 at predetermined intervals during a patient's hospitalization.

[0051] Figure 8 shows a flowchart illustrating an example of the document generation process performed in the document generation device 10 of this embodiment. The document generation process shown in Figure 8 is executed, for example, when the power to the document generation device 10 is turned on. If the user issues an instruction to generate a new document 62, the document generation process described in each of the above embodiments (see Figures 4A and 6) is executed separately.

[0052] In step S150 of Figure 8, the generation unit 42 determines whether the operating rate of its own device (document generation device 10) is below a threshold. The operating rate of the document generation device 10 includes, for example, at least one of the CPU 20A or the memory usage rate of the storage unit 22. If the operating rate is not below the threshold, in other words, if the operating rate exceeds the threshold, the determination in step S150 becomes a negative determination, and the process proceeds to step S156. On the other hand, if the operating rate is below the threshold, the determination in step S150 becomes an positive determination, and the process proceeds to step S152.

[0053] In step S152, the generation unit 42 selects a new document 62 to generate based on priority. If there are multiple new input information 60 to generate, the generation unit 42 in this modified example generates the new input information 60 in an order based on priority.

[0054] Prioritization can be based on, for example, the elapsed time since the generation of a past document 52, or in other words, the elapsed time since the last generation of a new document 62. In this case, for example, the longer the elapsed time, the higher the priority may be. Another example is prioritization based on the probability of receiving a generation instruction from a user. In this case, the longer the elapsed time since the last generation instruction was received, the higher the probability of receiving a generation instruction. Therefore, the longer the elapsed time since the last generation instruction was received, the higher the priority may be. Yet another example is prioritization based on the amount of new input information 60 accumulated in the document database 33. In this case, the larger the amount of new input information 60 accumulated, the longer the time required to generate a new document 62, so the larger the accumulated amount, the higher the priority may be. Note that the accumulated amount and elapsed time may be set with different thresholds for each type of new document 62 to be generated, for example, for each type corresponding to at least one of the following: user, medical department, disease, and patient's treatment stage, and the priority may be set according to the threshold for each type of new document 62.

[0055] Furthermore, regarding multiple types of priorities, the priority order may be predetermined, or it may be possible for the user to configure which type takes precedence.

[0056] Furthermore, rules for the above-mentioned priority may be defined in advance, and the generation unit 42 may select the new documents 62 to be generated based on these rules. Note that rules may be generated for each type of new document 62. For example, a planned creation date may be associated with each type of new document 62, and the new documents 62 may be selected in order of proximity to the corresponding planned creation date. Specifically, in the case of a discharge summary document, the planned discharge date may be associated as the planned creation date, and new documents 62 with an approaching planned discharge date may be selected preferentially. Alternatively, for example, the generation unit 42 may use AI to derive priority and select the new documents 62 to be generated. Such AI could include, for example, an AI that takes the number of days hospitalized, document volume, and disease name as features of the new input information 60, and outputs the probability that the user will instruct the generation of a new document 62.

[0057] In the next step S154, the generation unit 42 generates a new document 62. Specifically, it executes the processes of steps S100 to S110 of the document generation process described above. In this modified example, since the user is often not present at the processing site, steps S106 and S108 may be omitted.

[0058] In the next step, S156, the generation unit 42 determines whether or not to terminate the document generation process. In this modified example, the document generation process shown in Figure 8 is terminated when a predetermined termination condition is met, such as when the power to the document generation device 10 is turned off. Therefore, the determination in step S156 is negative until the termination condition is met, and the process returns to step S150, repeating the processes from S150 to S154. On the other hand, if the termination condition is met, the determination in step S156 becomes positive, and the document generation process shown in Figure 8 is terminated.

[0059] Thus, in this modified example, by generating a new document 62 when the operating rate of the document generation device 10 is relatively low, the processing time required when generating a new document 62 in response to a user's instruction can be shortened.

[0060] (Modification 3) In each of the above embodiments and modifications, the generation unit 42 uses the past documents 52 stored in the document generation model 32 as they are, but it is also possible to generate a new document 62 using a processed past document 52. For example, if the addition of additional documents 54 is repeated, the overall quality of the document may deteriorate. For this reason, a new document 62 may be generated using a modified past document 52. For example, a past document 52 modified based on user instructions may be used. Also, if the length of the past document 52 becomes long, the generation unit 42 may use a past document 52 that has been adjusted to shorten its length, similar to when the length of the new document 62 becomes long.

[0061] (Modification 4) In the above, the portion of the new document 62 corresponding to the past document 52 utilizes the past document 52 stored in the document database 33. However, there are cases where it is preferable to generate the entire new document 62 without using the past document 52. Therefore, in this modification, the generation unit 42 of the document generation device 10 determines whether or not to generate the entire text of the new document 62 based on past input information 50 and new input information 60, instead of generating the new document 62 based on the past document 52 and the additional document 54.

[0062] The generation unit 42 determines whether or not to generate the full text of the new document 62, for example, by using a predetermined rule or a machine learning model that outputs a recommendation level for generating the full text of the new document 62. An example of a predetermined rule is that the full text of the new document 62 is generated each time the new document 62 is generated based on the past document 52 and the additional document 54 a predetermined number of times. An example of a machine learning model is a machine learning model that has been trained so that the greater the difference between the new document 62 generated as the full text and the new document 62 generated based on the past document 52 and the additional document 54, the higher the recommendation level for generating the full text.

[0063] Furthermore, for example, the system may decide whether or not to generate the full text of the new document 62 in response to user instructions. For example, the system may decide in response to user instructions based on the time required for creation. In this case, the display control unit 44 displays the estimated time required to generate the new document 62 based on the past document 52 and the additional document 54, and the estimated time required to generate the full text of the new document 62 based on the past input information 50 and the new input information 60. The generation unit 42 then decides whether or not to generate the full text of the new document 62 based on the user instructions received in response to this display. Alternatively, for example, the system may decide in response to user instructions based on the number of times the new document 62 has been generated based on the past document 52 and the additional document 54. In this case, the display control unit 44 displays the number of times the new document 62 has been generated based on the past document 52 and the additional document 54.

[0064] Furthermore, when the user reviews the newly generated document 62, they may be given the option to indicate whether or not to regenerate the entire document 62. For example, an instruction button for whether or not to regenerate the entire document may be displayed, and the status of the button's operation may be recorded. Also, if the user makes modifications to the new document 62, or if they spend a long time viewing the displayed new document 62, the system may decide not to regenerate the entire document in order to maintain the current new document 62.

[0065] Furthermore, it is preferable to ensure that users can distinguish between a new document 62 generated based on past document 52 and additional document 54, and a new document 62 that has been completely regenerated. For example, a history of which method was used for generation may be saved so that users can check (view) it at any time.

[0066] According to the document generation device 10 of this modified example, the quality of the new document 62 can be set in order to generate the entire new document 62.

[0067] (Modification 5) In the case where past input information 50 is input to the document generation model 32, if there are multiple past input information 50, the generation unit 42 may input past input information 50 that has a similarity to the past document 52 of a threshold or higher into the document generation model 32. Past input information 50 with a similarity below the threshold is presumed not to have been used in the generation of the past document 52, and by not inputting it into the document generation model 32, the processing time in the document generation model 32 can be shortened.

[0068] (Modification 6) The generation unit 42 may also be configured such that, when the past document 52 contains multiple items, the additional documents 54 generated for each item are combined with the past document 52 for each item to generate a new document 62. For example, if the past document 52 contains multiple topics, an additional document 54 may be added for each topic to generate a new document 62. Alternatively, an additional document 54 may be added according to the sentences, paragraphs, and divisions of the past document 52 to generate a new document 62. In this way, when an additional document 54 is added for each item, it is preferable to add an additional document 54 corresponding to each item. Documents may also be rearranged in chronological order. Note that the categorization of the past document 52 may be determined using AI.

[0069] (Variation 7) An example of a case in which the newly entered information 60 is information that is in the middle of the past entered information 50 in chronological order will be described. As a specific example, the case in which a past document 52 containing information for 6 days has been generated by past entered information 50 for 9 / 1, 9 / 2, 9 / 3, 9 / 4, 9 / 6, and 9 / 7, and new entered information 60 for 9 / 5 is added will be described.

[0070] For example, an additional document 54 generated from newly entered information 60 for September 5 to September 7 may be added to a past document 52 containing information for September 1 to September 4. Alternatively, for example, a new document 62 containing information up to September 5 may be generated using newly entered information 60 for September 5 in a past document 52 containing information for September 1 to September 4. Then, this new document 62 may be made into a new past document 52, and a new document 62 containing information up to September 7, i.e., 7 days' worth of information, is generated using newly entered information 60 for September 5, using newly entered information 60 for September 6 and September 7. Alternatively, for example, an additional document 54 generated from newly entered information 60 for September 5 may be added after a past document 52 containing information for 6 days. Alternatively, for example, an additional document 54 generated from newly entered information 60 for September 5 to September 7 may be added to a past document 52 containing information for September 1 to September 4.

[0071] (Modification 8) In some cases, multiple pieces of time information may be associated with a single piece of information. For example, a combination of a radiographic image and a radiographic interpretation report based on that image may be included as new input information 60.

[0072] In such cases, each piece of information may be treated as separate pieces of information, or as a single piece of information. As a specific example, let's consider information that includes a radiographic image taken on September 1st and an interpretation report that was read on September 3rd. For example, if the date of acquisition is important, the radiographic image and interpretation report may be added for September 1st, and a note indicating that the interpretation date was September 3rd may be added. Alternatively, for example, if the chronological order is important, the radiographic image may be added for September 1st, and the interpretation report may be added for September 3rd. Alternatively, for example, if the date of interpretation is important, the radiographic image and interpretation report may be added for September 3rd, and a note indicating that the acquisition date was September 1st may be added.

[0073] As described above, the document generation device 10 of each of the above embodiments and each modified example comprises an acquisition unit 40 and a generation unit 42. The acquisition unit 40 acquires new input information 60. The display control unit 44 generates a new document 62 based on past documents 52 that have been generated in the past and additional documents 54 corresponding to the new input information 60, and when generating a new document 62, it generates additional documents 54 using the document generation model 32.

[0074] Thus, with the document generation device 10, it is sufficient to generate the additional document 54, and it is not necessary to generate the entire new document 62. Therefore, the document generation device 10 can shorten the time required to generate the new document 62.

[0075] In the above embodiments and variations, the case where a medical professional is the user and the new input information 60 is a summary has been described, but the user and the type of new document 62 generated are not limited to this embodiment. Also, the past document 52 does not have to be a document created by the document generation model 32, but may be a document created by the user.

[0076] 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.

[0077] 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 (Graphics 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.

[0078] 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.

[0079] Furthermore, although the above embodiment describes a configuration in which the document generation program 30 is pre-stored (installed) in the storage 22C of the memory unit 22, the invention is not limited to this configuration. The document generation 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 document generation program 30 may be provided in the form of a download from an external device via a network.

[0080] 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.

[0081] The following additional information is disclosed regarding the above-described embodiments.

[0082] (Note 1) A document generation device comprising a processor, wherein the processor acquires first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and generates the additional documents using a document generation model when generating the new document.

[0083] (Note 2) The document generation apparatus according to Note 1, wherein the processor causes the document generation model to input the first input information, the second input information used to generate the past document, and the past document, and obtains the new document formed by combining the past document and the additional document from the document generation model.

[0084] (Note 3) The document generation apparatus according to Note 1 or Note 2, wherein the processor causes the first input information to be input to the document generation model, the additional document is obtained from the document generation model, and the past document and the additional document are combined to generate the new document.

[0085] (Note 4) The document generation device according to any one of Notes 1 to 3, wherein the processor generates the new document when it receives a generation instruction from the user, and when the operating rate of the device due to other processing is below a threshold.

[0086] (Note 5) The document generation device according to Note 4, wherein the processor is scheduled to generate a plurality of new documents, and when the operating rate of the device is below a threshold, the device generates the new documents from among the plurality of new documents to be generated in an order according to a priority determined based on at least one of the elapsed time since the generation of the past document and the probability of receiving the generation instruction.

[0087] (Note 6) The document generation device according to Note 4, wherein the processor acquires a plurality of the first input information, sequentially stores it in the storage unit, and is scheduled to generate a plurality of new documents, and when the operating rate of the device is below a threshold, the device generates the new documents from the plurality of new documents to be generated in an order according to the priority determined based on the amount of the first input information stored.

[0088] (Note 7) The document generation device according to any one of Notes 1 to 7, wherein the processor makes modifications to the past document based on at least one of the following: modifications based on user instructions and modifications to shorten the length of the past document, and generates the new document using the modified past document.

[0089] (Note 8) The document generation device according to any one of Notes 1 to 7, wherein the processor, when generating the new document, determines whether or not to generate the full text of the new document based on the first input information and the second input information used to generate the past document, instead of generating the new document based on the past document and the additional document.

[0090] (Note 9) The document generation device described in Note 8, wherein the processor uses predetermined rules or a machine learning model that outputs a recommendation level for generating the full text of the new document to determine whether or not to generate the full text of the new document.

[0091] (Note 10) The document generation device according to Note 8, wherein the processor displays the estimated time required to generate the new document based on the past document and the additional document, and the estimated time required to generate the full text of the new document based on the first input information and the second input information used to generate the past document, and determines whether or not to generate the full text of the new document based on the instructions of the user received.

[0092] (Note 11) The document generation device according to any one of Notes 1 to 10, wherein when the past document includes multiple items, the additional document generated for each item generates a new document which is combined with the past document for each item.

[0093] (Note 12) The document generation device according to Note 2, wherein, when there are multiple second input information items, the processor causes the second input information items whose similarity to the past document is equal to or greater than a threshold to be input to the document generation model.

[0094] (Note 13) The document generation device described in any one of Notes 1 to 12, wherein the processor adjusts the length of the newly generated document.

[0095] (Note 14) A document generation method that obtains first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and generates the additional documents using a document generation model when generating the new document.

[0096] (Note 15) A document generation program that obtains first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and causes a computer to execute a process to generate the additional documents using a document generation model when generating the new document.

[0097] (Note 16) A computer program product that includes a learning program that causes a processor to acquire first input information, generate a new document based on past documents generated in the past and additional documents corresponding to the first input information, and, when generating the new document, to execute a process of generating the additional documents using a document generation model.

[0098] (Note 17) A computer-readable storage medium in which a document generation program is stored for a processor that acquires first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and generates the additional documents using a document generation model when generating the new document.

[0099] The disclosure of Japanese Patent Application No. 2024-171857 is incorporated herein by reference in its entirety.

[0100] 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. A document generation device comprising a processor, wherein the processor acquires first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and generates the additional documents using a document generation model when generating the new document.

2. The document generation apparatus according to claim 1, wherein the processor inputs the first input information, the second input information used to generate the past document, and the past document into the document generation model, and obtains the new document formed by combining the past document and the additional document from the document generation model.

3. The document generation apparatus according to claim 1, wherein the processor causes the first input information to be input to the document generation model, obtains the additional document from the document generation model, and generates the new document by combining the past document and the additional document.

4. The document generation apparatus according to claim 1, wherein the processor generates a new document when it receives a generation instruction from a user, and when the operating rate of the device due to other processing is below a threshold, in at least one of the above cases.

5. The document generation apparatus according to claim 4, wherein the processor is scheduled to generate a plurality of new documents, and when the operating rate of the device is below a threshold, it generates the new documents from among the plurality of new documents scheduled to be generated in an order according to a priority determined based on at least one of the elapsed time since the generation of the previous document and the probability of receiving the generation instruction.

6. The document generation apparatus according to claim 4, wherein the processor acquires a plurality of the first input information, sequentially stores it in the storage unit, and is scheduled to generate a plurality of new documents, and when the operating rate of the apparatus is below a threshold, it generates the new documents from the plurality of new documents to be generated in an order according to the priority determined based on the amount of the first input information stored.

7. The document generation apparatus according to claim 1, wherein the processor makes modifications to the past document based on at least one of modifications based on user instructions and modifications to shorten the length of the past document, and generates the new document using the modified past document.

8. The document generation apparatus according to claim 1, wherein the processor, when generating the new document, determines whether or not to generate the entire text of the new document based on the first input information and the second input information used to generate the past document, instead of generating the new document based on the past document and the additional document.

9. The document generation apparatus according to claim 8, wherein the processor determines whether or not to generate the full text of the new document using a predetermined rule or a machine learning model that outputs a recommendation level for generating the full text of the new document.

10. The document generation device according to claim 8, wherein the processor displays an estimated time required to generate the new document based on the past document and the additional document, and an estimated time required to generate the full text of the new document based on the first input information and the second input information used to generate the past document, and determines whether or not to generate the full text of the new document based on the instructions received from the user.

11. The document generation apparatus according to claim 1, wherein, if the past document includes multiple items, the additional document generated for each item generates a new document which is combined with the past document for each item.

12. The document generation apparatus according to claim 2, wherein, when there are multiple second input pieces of information, the processor causes the second input piece having a similarity to the past document of a threshold or higher to be input to the document generation model.

13. The document generation apparatus according to claim 1, wherein the processor adjusts the length of the newly generated document.

14. A document generation method that obtains first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and generates the additional documents using a document generation model when generating the new document.

15. A document generation program that obtains first input information, generates a new document based on past documents generated in the past and additional documents corresponding to the first input information, and, when generating the new document, causes a computer to execute a process to generate the additional documents using a document generation model.

Images