Information processing system, information processing method, and program

Abstract

To reduce the work burden on medical workers in creating medical documents.SOLUTION: An information processing system for supporting creation of medical documents includes a data acquisition unit and a medical document generation unit. The data acquisition unit is configured to obtain conversation data and medical data. The conversation data includes the content of a conversation between a medical worker and a patient. The medical data includes medicine data relating to the patient. The medical document generation unit is configured to generate the medical document by inputting the conversation data and the medicine data into a learning model and making the learning model output the medical document.SELECTED DRAWING: Figure 1

Description

Technical Field

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

Background Art

[0002] Medical workers (for example, pharmacists in a pharmacy) have tasks of obtaining and managing patient data through conversations with patients in addition to dispensing the drugs prescribed by doctors. Conventionally, technologies for supporting the work of medical workers have been proposed. Patent Document 1 discloses a system for improving the quality of the task of medication guidance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The work of medical workers includes creating medical documents to record interactions with patients. For example, in a situation where there are other tasks that require attention, such as dispensing and medication guidance, creating medical documents is a burden for medical workers, but it is a necessary task and cannot be omitted.

[0005] An object of the present invention is to suppress the workload of medical workers in creating medical documents.

Means for Solving the Problems

[0006] [1] According to the present invention, there is provided an information processing system for assisting in the creation of medical texts, comprising a data acquisition unit and a medical text generation unit. The data acquisition unit is configured to be able to acquire conversation data and medical data. The conversation data includes the content of conversations between medical practitioners and patients, and the medical data includes pharmaceutical data related to the patients. The medical text generation unit is configured to generate the medical text by inputting the conversation data and the pharmaceutical data into a learning model and causing the learning model to output the medical text.

[0007] According to the present invention, since the above configuration is provided, it is possible to reduce the workload of medical practitioners in creating medical texts.

[0008] Hereinafter, various embodiments of the present invention will be exemplified. The embodiments shown below can be combined with each other. [2] Preferably, in the information processing system according to [1], the medical text generation unit has an output generation unit. The output generation unit generates a text generation prompt to be output to a large language model as the learning model. The text generation prompt includes the conversation data and the medical data as character data, and an instruction to generate the medical text. The medical text generation unit is configured to generate the medical text by inputting the text generation prompt into the large language model and causing the large language model to output the medical text. [3] Preferably, in the information processing system according to [1], the information processing system further includes a speech-to-text conversion unit. The speech-to-text conversion unit is configured to convert voice conversation data corresponding to the content of the conversation into character data to generate the conversation data. The data acquisition unit acquires the conversation data from the speech-to-text conversion unit. [4] Preferably, it is the information processing system described in [3], wherein the medical text generation unit has an output generation unit, the output generation unit generates an accuracy improvement prompt to be output to the large language model as the learning model, the accuracy improvement prompt has pre - processing character data and the medical data, and has an instruction to generate post - processing character data with reference to the medical data and an instruction to improve the accuracy of the generation of the medical text. The pre - processing character data is the conversation data obtained from the speech - to - text conversion unit, the post - processing character data is the conversation data output from the large language model based on the accuracy improvement prompt, and the medical text generation unit is configured to generate the post - processing character data by inputting the accuracy improvement prompt into the large language model and causing the large language model to output the post - processing character data. An information processing system is provided. [5] Preferably, it is the information processing system described in [4], wherein the output generation unit generates a text generation prompt to be output to the large language model, the text generation prompt has the conversation data of the post - processing character data and the medical data, and has an instruction to generate the medical text. The medical text generation unit is configured to generate the medical text by inputting the text generation prompt into the large language model and causing the large language model to output the medical text. An information processing system is provided. [6] Preferably, it is the information processing system described in [4] or [5], wherein the instruction to improve the accuracy of the generation of the medical text in the accuracy improvement prompt has an instruction to generate the post - processing character data by distinguishing between the content spoken by the medical staff and the content spoken by the patient. The conversation data of the post - processing character data has the content spoken by the medical staff and the content spoken by the patient distinguished from each other. An information processing system is provided. [7] Preferably, it is the information processing system described in any one of [1] to [5], wherein the conversation data has the content spoken by the medical staff and the content spoken by the patient distinguished from each other. An information processing system. [8] Preferably, there is provided an information processing system according to any one of [1] to [7], wherein the medical data further includes basic data of the patient, the basic data includes the name of the patient and the gender of the patient, and the medical text generation unit further generates the medical text based on the basic data. [9] Preferably, there is provided an information processing system according to any one of [1] to [8], wherein the medical data further includes medication history data related to the patient, and the medical text generation unit generates the medical text by inputting the conversation data, the pharmaceutical data, and the medication history data into the learning model and outputting the medical text from the learning model.

[10] Preferably, there is provided an information processing system according to any one of [1] to [9], wherein the medical text is generated with subjective information, objective information, evaluation, and plan being distinguished.

[11] From another perspective in the embodiment, there is provided an information processing method for assisting in creating a medical text, including an acquisition step and a generation step. In the acquisition step, conversation data and medical data are acquired. The conversation data includes the content of the conversation between a medical staff member and a patient, and the medical data includes pharmaceutical data related to the patient. In the generation step, the medical text is generated by inputting the conversation data and the pharmaceutical data into a learning model and outputting the medical text from the learning model.

[12] There is provided a program for causing a computer to execute the information processing method according to

[11] .

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The various characteristic matters shown in the following embodiments can be combined with each other. Also, an invention can be established independently for each characteristic matter.

[0011] 1. System Configuration of Information Processing System 100 The information processing system 100 can assist in creating medical texts that need to be created by medical workers such as pharmacists. That is, the information processing system 100 has a function of suppressing the workload of medical workers in creating medical texts by generating medical texts.

[0012] As shown in FIG. 1, the information processing system 100 includes an information processing device 1, a large language model server 2 (an example of a large language model), a speech-to-text server 3, a terminal 4, and a voice acquisition device 5. These are connected via a communication network 6 (for example, the Internet, etc.) so that information can be exchanged. Note that part or all of the communication network 6 may be a closed network separated from the Internet. Each component of the information processing system 100 such as the information processing device 1 has one or more functions (functional units). Each component may be configured as a single device as shown in FIG. 1, or may be configured by a plurality of independent devices capable of exchanging information. The same applies to each functional unit such as the control unit 12 of the information processing device 1 described later. Hereinafter, each component included in the information processing system 100 will be further described.

[0013] 1-1. Information Processing Device 1 As shown in FIG. 2, the information processing device 1 has a communication unit 10, a storage unit 11, a control unit 12, an output unit 13, and an input unit 14, and these components are electrically connected via a communication bus 15 inside the information processing device 1. The control unit 12 has a data acquisition unit 121, a medical text generation unit 122, and an encoding unit 123. Further, the medical text generation unit 122 has an output generation unit 122a and an adjustment unit 122b.

[0014] Each of the above-described components may be implemented by software or by hardware. When implemented by software, various functions can be realized by a CPU executing a computer program. The program may be stored in a non-transitory computer-readable recording medium, may be provided so as to be downloadable from an external server, or may also be realized by so-called cloud computing that reads out a program stored in an external storage unit to realize functions. When implemented by hardware, it can be realized by various circuits such as ASIC, FPGA, or DRP. In the embodiments, various information and concepts including the same are handled, and these are represented by the high and low of signal values or quantum bits as a set of binary bits composed of 0 or 1, and communication and calculation can be executed by the above-described software or hardware aspects. Note that the software may be a general-purpose OS or a dedicated OS.

[0015] The communication unit 10 can adopt wired communication means such as, for example, USB, IEEE1394, Thunderbolt (registered trademark), wired LAN network communication, etc. Note that the communication unit 10 may adopt a configuration in which it is connected to the communication network 6 via wireless communication means such as, for example, wireless LAN network communication, mobile communication such as 3G / LTE / 5G, Bluetooth (registered trademark) communication, etc. Further, the communication unit 10 may be configured to use both the above-described wired communication means and wireless communication means in combination.

[0016] The storage unit 11 stores various values such as, for example, various programs, constants, variables, and setting values of the information processing apparatus 1 executed by the control unit 12. As the storage unit 11, a storage device such as a solid state drive (SSD) or a random access memory (RAM) that stores information (arguments, arrays, etc.) temporarily necessary for program calculation can be adopted. Further, in addition to the storage unit 11, the information processing apparatus 1 may also use an external storage unit (for example, an external storage medium, cloud, etc.) in combination.

[0017] The control unit 12 is configured to execute processing and control related to the information processing of the information processing apparatus 1. The control unit 12 can be constituted by, for example, a central processing unit (CPU). In the embodiment, the control unit 12 is an example of a processor capable of executing a program related to each step of the flowchart described later. The control unit 12 realizes various functions related to the information processing apparatus 1, for example, by reading out a program stored in the storage unit 11. Further, the information processing of the software in the information processing apparatus 1 is realized, for example, by various programs stored in the storage unit 11 being processed by the control unit 12 as hardware.

[0018] The output unit 13 is, for example, the display unit of the information processing apparatus 1. The output unit 13 may be included in the housing of the information processing apparatus 1 or may be externally attached. The output unit 13 displays a screen of a graphical user interface (GUI) operable by the user. The output unit 13 may employ, for example, a display device such as a CRT display, a liquid crystal display, an organic EL display, a plasma display, an electronic paper display, or other display devices such as a lit light or a projector. Note that whether or not the information processing apparatus 1 includes the output unit 13 is optional. For example, the output of the information processing apparatus 1 may be displayed on a display unit at a location independent of the location where the information processing apparatus 1 is installed. Further, the output unit 13 may have a device that outputs in voice.

[0019] The input unit 14 is configured to receive an operation input made by the user of the information processing apparatus 1, for example. The input unit 14 may be included in the housing of the information processing apparatus 1 or may be externally attached. The input unit 14 may employ, for example, a touch panel, a switch button, a mouse, a keyboard, etc. Note that whether or not the information processing apparatus 1 includes the input unit 14 is optional. For example, the operation input to the information processing apparatus 1 may be received by the information processing apparatus 1 via an information processing terminal at a location independent of the location where the information processing apparatus 1 is installed.

[0020] 1-2. Large Language Model Server 2 and Speech-to-Text Server 3 The large language model server 2 and the speech-to-text server 3 are communicably connected to the information processing apparatus 1 via the communication network 6. The large language model server 2 has a text generation unit 21. The text generation unit 21 has, for example, a neural network trained using a large number of teacher data, and also has a function of generating an answer to the data (prompt) output from the information processing apparatus 1 and outputting it to the information processing apparatus 1. Note that the large language model server 2 is not particularly limited, but for example, it is possible to adopt ChatGPT of OpenAI or the like.

[0021] The speech-to-text server 3 has a speech recognition unit 31. The speech recognition unit 31 is configured to generate text conversation data that converts the content of the conversation into text using the voice conversation data. The speech recognition unit 31 has, for example, a neural network trained using a large number of teacher data (voice teacher data). When the speech recognition unit 31 receives the data (voice data) output from the information processing apparatus 1, it has a function of outputting it to the information processing apparatus 1 as text data. In the embodiment, the speech recognition unit 31 has been described as being composed of a trained neural network, but it is not limited to this, and for example, a configuration in which voice data is analyzed and converted into text according to a predetermined algorithm may be used. Note that the speech-to-text server 3 is not particularly limited, but for example, it is possible to adopt Whisper (registered trademark) of OpenAI or the like.

[0022] 1-3. Terminal 4 and Voice Acquisition Device 5 The terminal 4 is an information processing device (e.g., a personal computer) arranged in, for example, a pharmacy or the like, and is configured such that medical staff can input patient data. Also, the medical staff can display patient data on the display unit (not shown) of the terminal 4 so as to be able to acquire the patient data using the terminal 4. Note that the patient data exchanged in the information processing system 100 will be described in detail later. The voice acquisition device 5 has a device such as a microphone and is configured to acquire conversation data between medical staff and patients made in a pharmacy. The conversation data is voice conversation data. Note that the voice acquisition device 5 may be integrated with the terminal 4 or may be separate. The voice acquisition device 5 can employ, for example, a dedicated voice recorder or a mobile phone (e.g., a smartphone or the like) on which a voice acquisition application is installed.

[0023] In an example of the embodiment, when the voice acquisition device 5 acquires conversation data (voice data), the voice data is transmitted from the voice acquisition device 5 to the terminal 4. Then, the terminal 4 outputs the patient data and the conversation data together to the information processing device 1. As a result, the patient data related to the current prescription and the conversation data related to the current prescription are not separated and can be appropriately processed by the information processing device 1. Note that the configuration of data transmission is not limited to this, and when the voice acquisition device 5 acquires conversation data, it may be configured to output from the voice acquisition device 5 to the information processing device 1 via the communication network 6 and the information processing device 1 side summarizes the information.

[0024] 2. Functional configuration Referring to FIG. 3, the functional configuration of the information processing device 1 according to the present embodiment will be described. Information processing by software stored in the storage unit 11 is specifically realized by the control unit 12, which is an example of hardware, so that each functional unit included in the control unit 12 is executed.

[0025] The data acquisition unit 121 is configured to be able to acquire various data from the information processing device 1, the large language model server 2, the speech-to-text server 3, and the like. In the embodiment, the data acquisition unit 121 is configured to be able to acquire conversation data and medical data. Further, the data acquisition unit 121 is configured to be able to acquire text conversation data from the speech-to-text unit 31. These data will be described in detail later.

[0026] The medical text generation unit 122 is configured to generate a medical text based on conversation data (text conversation data) and pharmaceutical data. The output generation unit 122a of the medical text generation unit 122 has a function of generating various prompts (for example, a text generation prompt and an accuracy improvement prompt described later). The adjustment unit 122b of the medical text generation unit 122 has a post-processing function. That is, the adjustment unit 122b has a function of adding information to the medical text MT1 generated by the large language model server 2 (for example, adding information such as a name that can identify an individual to the medical text MT1) and adjusting the content of the medical text MT1. Whether the information processing device 1 has the adjustment unit 122b is optional.

[0027] The encoding unit 123 has a function of encoding some or all of the non-output medical data among the data related to the patient. In other words, the encoding unit 123 has a function of encoding so that the content of the non-output medical data is anonymized or abstracted. Note that the non-output medical data is data among the patient's data that is not output to the large language model server 2. The information processing device 1 outputs the encoded data so that an individual cannot be identified. That is, since the data related to the patient includes, for example, data that is personal information, in the embodiment, non-output medical data (data that is not output to the large language model server 2) is predetermined, but the information processing device 1 can encode the non-output medical data and include it in the output medical data.

[0028] 3. Regarding the data exchanged in the information processing system 100 In the information processing system 100 according to the embodiment, various data are exchanged among the information processing device 1, the large language model server 2, the speech-to-text server 3, and the terminal 4. The data will be described with reference to FIGS. 4A to 8B.

[0029] The data exchanged in the information processing system 100 includes medical data, conversation data, and medical texts. In an example of the embodiment, the types of medical data include the items shown in data D1 to data D19 shown in FIG. 4B. The numbers of data D1 to data D19 are arbitrarily assigned for the purpose of distinguishing these items. Also, not all of data D1 to data D19 are necessarily essential items.

[0030] 3-1. Medical Data In the information processing system 100 according to the embodiment, since the storage unit 11 functions as a database, the entire data exchanged in the information processing system 100 is accumulated in the storage unit 11. First, the types of medical data among this entire data will be listed.

[0031] As shown in FIG. 4B, in the embodiment, medical data includes items such as ID (data D1), name (data D2), gender (data D3), date of birth (data D4), age (data D5), age group (data D5t), insurance card information (data D6), allergy history (data D7), current medical history (data D8), past medical history (data D9), drug side effect history (data D10), drug combination history (data D11), information related to problems in taking drugs (data D12), pharmacy name (data D13), prescribing doctor's name (data D14), hospital name (data D15), prescription date (data D16), name of medicine (data D17), dosage, usage, and frequency of medicine (data D18), and drug history (data D19).

[0032] The ID (data D1) is a unique symbol assigned to each individual and is represented by, for example, numbers, alphabets, etc. The age (data D5t) may be obtained from the terminal 4 on the pharmacy side, or the information processing device 1 may have a function of calculating the age based on the age, and may obtain it by exerting the function. This function can be realized by, for example, the encoding unit 123. The insurance certificate information (data D6) includes the insurer number, symbol, and number. The information related to problems in taking medicine (data D12) includes, for example, information on circumstances such as unsuitable for an individual's mouth or body (e.g., wanting to avoid Chinese herbal medicine because it doesn't suit the taste), which is different from allergies.

[0033] Note that the medicine name (data D17) is an example of medicine data. This medicine name may be the name of the medicine itself or an abbreviated name. The name (data D2), gender (data D3), date of birth (data D4), age (data D5), and insurance certificate information (data D6) are examples of the patient's basic data. The medication history (data D19) is an example of medication history data.

[0034] <Stored data DT0 and medical data DT1> In an example of the embodiment, all of these 20 items of data are stored in the storage unit 11 of the information processing device 1 as a database. When a patient visits a pharmacy and a medical staff member receives a prescription, for example, the medical staff member inputs the prescription data into the terminal 4. As a result, as shown in FIG. 4A, the prescription data (medical data DT1) is output to the information processing device 1. For example, if the medical data DT1 is prescription data, it includes at least data such as the pharmacy name (data D13), medicine name (data D17), and dosage, usage, and amount of the medicine (data D18).

[0035] Also, for example, if the medical data DT1 is dispensing data, it includes data such as the medicine name (data D17) and dosage, usage, and amount of the medicine (data D18). Also, for example, if the medical data DT1 is pharmaceutical data, it includes data such as the name of the pharmaceutical product (data D17). Note that the types of data included in the medical data DT1 are not limited to prescription data, dispensing data, and pharmaceutical data, and various changes are possible. That is, the medical data DT1 may include all of the above-mentioned 20 items of data, or may include only a part of them, and various settings are possible. The information processing system 100 is preferably configured to be able to change the types of data included in the medical data.

[0036] When the medical data DT1 is output to the information processing device 1, the content of the stored data DT0 as a database is updated each time. Also, for example, when there is a change in the patient's status (e.g., name), the data related to the change is input to the terminal 4 and the stored data DT0 is updated. Note that the storage unit 11 also holds the data before the update (old data). In this way, the storage unit 11 comprehensively stores the data exchanged in the information processing system 100, and the data accumulates.

[0037] By the way, medical data can be distinguished into the above-mentioned non-output medical data and output medical data. Non-output data is data that does not include personal information and is not output to the large language model server 2, and output data is data that is output to the large language model server 2. These data will be described below.

[0038] 3-1-1. Non-output medical data The non-output medical data will be described with reference to FIG. 7A. In an embodiment, the non-output medical data may include, for example, name (data D2), date of birth (data D4), age (data D5), insurance card information (data D6), pharmacy name (data D13), prescribing doctor name (data D14), hospital name (data D15), and prescribing date (data D16). The data is not limited to those exemplified here. For example, information (date of birth and age) that is considered to be relatively highly anonymized compared to the name etc. may be included in the output medical data and can be appropriately changed.

[0039] 3-1-2. Output Medical Data The output medical data will be described with reference to FIG. 7A. In an embodiment, the output medical data may include, for example, ID (data D1), gender (data D3), age group (data D5t), allergy history (data D7), current medical history (data D8), past medical history (data D9), drug side effect history (data D10), drug combination history (data D11), information related to problems in taking drugs (data D12), drug name (data D17), dosage, usage, and frequency of drugs (data D18), and drug history (data D19). The data is not limited to those exemplified here, and other information may be added or reduced as necessary.

[0040] <Medical Data DT2> Here, the medical data DT2 shown in FIG. 5 and the medical data DT3 shown in FIG. 6 will be described. In FIG. 5, in order to improve the accuracy (correctness) of the text of the pre-processing character data d2 described later, the state in which the medical data DT2 is output to the large language model server 2 is shown. The medical data DT2 may include all or part of the above-described output medical data. Note that the medical data DT2 includes a drug name (data D17). This is because many drug names are long and many are names that are not commonly used, making it difficult to appropriately perform character recognition. Preferably, the medical data DT2 includes at least one of the dosage, usage, and frequency of medicine (data D18) and the medication history (data D19). In addition, the medical data DT2 may include not only the content related to the current prescription (see FIG. 7A), but also the content related to past prescriptions (see FIG. 7B). The content related to past prescriptions may include those of multiple times (for example, the previous time and the time before that, etc.).

[0041] <Medical data DT3> In FIG. 6, in order to generate the medical text MT1 described later, the medical data DT3 is shown being output to the large language model server 2 together with the processed character data d3 described later. The medical data DT3 may include all or part of the above-described output medical data. Note that the medical data DT3 includes the name of a medicine (data D17), which is an example of medicine data, in order to appropriately generate the medical text MT1 output from the large language model server 2. Preferably, the medical data DT3 includes at least one of the dosage, usage, and frequency of medicine (data D18) and the medication history (data D19). More preferably, the medical data DT3 includes at least one of the ID (data D1), gender (data D3), age (data D5t), allergy history (data D7), current medical history (data D8), past medical history (data D9), history of drug side effects (data D10), and history of drug combination (data D11).

[0042] Also, similar to the medical data DT3, the medical data DT3 may include not only the content related to the current prescription (see FIG. 7A), but also the content related to past prescriptions (see FIG. 7B). The content related to past prescriptions may include those of multiple times (for example, the previous time and the time before that, etc.). Specifically, the drug name (an example of drug data) may have new drug data and old drug data. The new drug data corresponds to the conversation in the conversation data. In other words, the new drug data is, for example, the data of the drug name described in the prescription submitted by the patient this time. The old drug data is data prior to the new drug data. In other words, the old drug data is, for example, the data of the drug described in the prescription submitted by the patient before this time. Before this time may refer to the data related to the prescription one time before the prescription in the new drug data, or the data related to the prescription two times before, or even earlier. Also, the old drug data may have a plurality of data prior to this time.

[0043] 3-2. Conversation Data The conversation data includes the content of the conversation between the medical staff and the patient. The types of conversation data include the voice conversation data d1 shown in FIGS. 4A and 5 (the voice data acquired by the voice acquisition device 5 in the embodiment) and the text conversation data. Also, the types of text conversation data include the pre-processed text data d2 shown in FIG. 5 and the post-processed text data d3 shown in FIGS. 5 and 6. Here, the pre-processed text data d2 is the data obtained by performing speech recognition on the voice conversation data by the speech recognition unit 31. The post-processed text data d3 is the text conversation data used when the medical article generation unit 122 generates a medical article. In the embodiment, the post-processed text data d3 is the text data after processing the pre-processed text data d2 by the text generation unit 21 so that the accuracy of the content is improved.

[0044] Specifically, as shown in an example in FIG. 8A, the pre-processed text data d2 may have errors in the drug name etc. (see the underlined part). However, as shown in an example in FIG. 8B, the errors in the post-processed text data d3 are corrected (see the underlined part), and the accuracy as a text is improved. In addition, in the post-processed text data d3, the content spoken by the medical staff and the content spoken by the patient are distinguished. Since such post-processed text data d3 is output to the large language model server 2, the output (medical article MT1) from the large language model server 2 is optimized and the accuracy is also improved.

[0045] 3-3. Medical text In the medical text in the embodiment, the medical text MT1 output from the large language model server 2 to the information processing device 1 and the medical text MT2 output from the information processing device 1 to the pharmacy side are distinguished. As shown in FIG. 9, the medical text MT1 has a predetermined format. An example of this predetermined format is SOAP. Here, SOAP refers to a recording method (recording format) used at the medical site where subjective information (Subject), objective information (Object), assessment (Assessment), and plan (Plan) are distinguished. That is, the medical text MT1 is generated with subjective information, objective information, assessment, and plan distinguished.

[0046] In addition to the medical text MT1, the medical text MT2 is data configured to be able to identify an individual by adding at least one of the name (data D2), gender (data D3), date of birth (data D4), age (data D5), and insurance card information (data D6). In other words, the medical text MT2 is data with basic data elements added. Note that the information processing device 1 may not generate the medical text MT2 and may output the medical text MT1 itself to the pharmacy side. In this case, for example, the pharmacy-side terminal 4 may perform a process of identifying an individual.

[0047] 4. Information processing flow of the information processing system 100 The program according to the embodiment executes each step (information processing method) described below. This information processing method is a process for assisting in the creation of medical text. With reference to FIG. 10, an example of the information processing method according to the embodiment will be described.

[0048] <Start: Step S001> Medical staff receive a prescription from a patient and start dispensing according to the prescription. When the dispensing is completed, a conversation with the patient is started to hand over the medicine. Prior to this conversation, the medical staff activates the voice acquisition device 5. Medical staff at a pharmacy need to leave records as medical documents for medical acts such as dispensing and conversation, which is time-consuming. The information processing system 100 according to the embodiment can reduce this workload by automatically generating all or part of the medical document and suppress the burden on the medical staff.

[0049] <Step S002> When the conversation with the patient is completed, the medical staff stops the voice acquisition device 5 and transmits the voice conversation data d1 from the memory of the voice acquisition device 5 to the memory of the terminal 4. As a result, the terminal 4 acquires the medical data DT1. In addition, the medical staff inputs the content of the current prescription as the medical data DT1 into the terminal 4, and the memory of the terminal 4 acquires this medical data DT1. Note that the image of the prescription may be scanned and information such as the characters of the prescription may be read from the scanned data to acquire the medical data DT1.

[0050] In the description here, the patient has already used any of the pharmacies utilized by the information processing system 100, and the basic data of the patient (such as name (data D2), etc.) is already stored in the storage unit 11 of the information processing device 1 serving as a database. Therefore, in an example of the medical data DT1 in this step S001, it is not always necessary to input all the basic data (for example, age (data D5), etc.).

[0051] <Step S003> The data acquisition unit 121 of the information processing device 1 acquires the voice conversation data d1 and the medical data DT1 transmitted from the pharmacy side and stores them in the storage unit 11.

[0052] This step S003 is an example of an acquisition step.

[0053] <Step S004> The control unit 12 of the information processing apparatus 1 outputs the voice conversation data d1 stored in the storage unit 11 to the speech-to-text server 3. The speech recognition unit 31 of the speech-to-text server 3 performs speech recognition on the voice conversation data d1 to generate pre-processed character data d2 which is character data.

[0054] <Step S005> The speech-to-text server 3 outputs the pre-processed character data d2 to the information processing apparatus 1. The data acquisition unit 121 of the information processing apparatus 1 acquires the pre-processed character data d2 and stores it in the storage unit 11.

[0055] <Step S006> The control unit 12 of the information processing apparatus 1 outputs the pre-processed character data d2 stored in the storage unit 11 and the medical data DT2 to the large language model server 2, and in so doing, generates an output prompt. Specifically, the output generation unit 122a of the medical text generation unit 122 generates an accuracy improvement prompt to be output to the large language model server 2. The accuracy improvement prompt includes the pre-processed character data d2 and the medical data DT2, and has an instruction to generate post-processed character data d3 with reference to the medical data DT2. Further, the accuracy improvement prompt has an instruction to generate the post-processed character data d3 by distinguishing between the content spoken by medical staff and the content spoken by the patient (see FIG. 8B), whereby the effect of improving the generation accuracy of the medical text MT1 can be expected.

[0056] The control unit 12 of the information processing apparatus 1 outputs this accuracy improvement prompt to the large language model server 2. Specifically, the number of characters of the accuracy improvement prompt is, for example, 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1200, 1400, 1600, 1800, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 12000, 14000, and it may be within the range between any two of the values exemplified here. Further, the accuracy improvement prompt may be composed of a plurality of divided prompts.

[0057] The text generation unit 21 of the large language model server 2 refers to the accuracy improvement prompt, generates processed text data d3 by modifying the content of the pre - processed text data d2 so that the accuracy of the pre - processed text data d2 is improved, and outputs it to the information processing apparatus 1.

[0058] <Step S007> The data acquisition unit 121 of the information processing apparatus 1 acquires the processed text data d3 as the output from the large language model for the accuracy improvement prompt, and stores it in the storage unit 11.

[0059] <Step S008> Subsequently, the control unit 12 of the information processing apparatus 1 outputs the processed text data d3 stored in the storage unit 11 and the medical data DT3 to the large language model server 2. In doing so, it generates an output prompt. Specifically, the output generation unit 122a of the medical text generation unit 122 generates a text generation prompt to be output to the large language model server 2. The text generation prompt includes the processed text data d3 and the medical data DT3, and has an instruction to generate the medical text MT1. In the embodiment, this prompt has an instruction to create the content of the processed text data d3 and the medical data DT3 in a predetermined format (SOAP in the embodiment). Also, this prompt may include supplementary data. The supplementary data can include, for example, content that defines what SOAP is, content that limits the number of characters of the generated medical text MT1, and the like. Note that the medical data DT3 in this prompt excludes information by which an individual can be identified (for example, name, etc., which is an example of basic data), but may be included as encoded data (abstracted data) in which the information is encoded (abstracted) by the encoding unit 123.

[0060] The control unit 12 of the information processing apparatus 1 outputs this text generation prompt to the large language model server 2. Specifically, the number of characters in the text generation prompt is, for example, 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1200, 1400, 1600, 1800, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 12000, 14000, and it may also be within the range between any two of the numerical values exemplified here. Further, the text generation prompt may be composed of a plurality of divided prompts.

[0061] The text generation unit 21 of the large language model server 2 refers to the text generation prompt, generates a medical text MT1, and outputs it to the information processing apparatus 1.

[0062] <Step S009> The data acquisition unit 121 of the information processing apparatus 1 acquires the medical text MT1 as the output from the large language model for the text generation prompt and stores it in the storage unit 11. Here, the medical data DT3 output to the large language model server 2 is processed so as not to include personal information. Therefore, the adjustment unit 122b of the medical text generation unit 122 generates a medical text MT2 with basic data (such as name) added so that an individual can be identified in the medical text MT1 and stores it in the storage unit 11. That is, the medical text generation unit 122 generates the medical text MT2 based on the conversation data (processed character data), the pharmaceutical data (medical data), and the basic data.

[0063] Steps S008 and S009 are an example of the generation step.

[0064] <Step S010> Medical staff acquire the medical text MT2 via the terminal 4, and the medical text MT2 is stored in the memory of the terminal 4. Note that when the information processing system 100 does not have the adjustment unit 122b, the terminal 4 acquires the medical text MT1.

[0065] <End: Step S011> Medical workers can save the trouble of creating medical texts by utilizing the medical text MT2.

[0066] 5. Variations 5-1. Variation 1 In the embodiment, a configuration for generating medical texts by utilizing servers outside the information processing apparatus 1, such as the large language model server 2 and the speech-to-text server 3, has been described, but the present invention is not limited thereto. As shown in FIGS. 11A and 11B, the information processing apparatus 1 may have the functions of the large language model server 2 and the speech-to-text server 3. That is, the information processing apparatus 1 may have a text generation unit 21 and a speech recognition unit 31.

[0067] 5-2. Variation 2 In Variation 1, the information processing system 100 has been described as a system in which the medical institution (pharmacy) side and the information processing apparatus 1 arranged at a location independent of the medical institution exchange information, but the present invention is not limited thereto. As shown in FIG. 12, the information processing apparatus 1 may be arranged in the medical institution (pharmacy), and a system in which data is completed within the medical institution (pharmacy) may be used.

[0068] 5-3. Other variations In the embodiment, the medical institution has been described as a pharmacy, but the present invention is not limited thereto. For example, it can also be applied to a place that provides medicines installed in a hospital. Further, it can also be applied to fields where it is necessary to create documents in a predetermined format. For example, it can be applied not only to places where medicines are prescribed but also to the business in a hospital for creating SOAP.

[0069] The article generation unit 21 may adopt a configuration having a dedicated neural network for generating medical articles. Dedicated means specialized in a specific field. For example, when applied to a pharmacy as in the embodiment, it is the field of medicine and healthcare. For example, the article generation unit 21 can be configured to perform operations based on a predetermined learning model. Here, a learning model is a model that trains a model using a large number of teacher data and enables prediction of future outputs. As data for input, for example, a large number of medical data (such as medicine names) can be used, and as data for output, for example, a large number of medical articles can be used. Thereby, the article generation unit 21 can generate a medical article based on a learning model that inputs medical data and outputs a medical article.

[0070] In the embodiment, it has been described as having the speech-to-text conversion unit 31 within the information processing system 100, but it is not limited thereto. That is, the speech-to-text conversion unit 31 exists outside the information processing system 100, and the information processing system 100 may be configured to adopt conversation data that has already been speech-to-text converted externally. In this case, the information processing apparatus 1 can acquire the conversation data that has been speech-to-text converted externally and include this in the medical data DT3 to generate an article generation prompt.

[0071] In the above, the embodiments have been described, but these are presented as examples and are not intended to limit the scope of the invention. The novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made. The embodiments and their modifications are included in the scope and gist of the invention and are included in the invention described in the claims and its equivalent scope.

Explanation of Reference Numerals

[0072] 100: Information processing system 1: Information processing apparatus 10: Communication unit 11: Storage unit 12: Control unit 121: Data acquisition unit 122: Medical Article Generation Unit 122a: Output Generation Unit 122b: Adjustment Unit 123: Encoding Unit 13: Output Unit 14: Input Unit 15: Communication Bus 2: Large Language Model Server 21: Article Generation Unit 3: Speech-to-Text Server 31: Speech Recognition Unit 4: Terminal 5: Voice Acquisition Device 6: Communication Network DT0: Stored Data DT1: Medical Data DT2: Medical Data DT3: Medical Data MT1: Medical Article MT2: Medical Article d1: Voice Conversation Data d2: Pre-Processing Character Data d3: Post-Processing Character Data

Claims

1. An information processing system for assisting in the creation of medical articles, comprising a data acquisition unit and a medical article generation unit, wherein the data acquisition unit is configured to be able to acquire conversation data and medical data, the conversation data includes the content of conversations between medical practitioners and patients, the medical data has pharmaceutical data related to the patient, the medical article generation unit is configured to generate the medical article by inputting the conversation data and the pharmaceutical data into a learning model and causing the learning model to output the medical article.

2. The information processing system according to claim 1, wherein the medical article generation unit has an output generation unit, the output generation unit generates an article generation prompt for output to a large language model as the learning model, the article generation prompt has the conversation data and the medical data as character data, and has an instruction to generate the medical article, the medical article generation unit is configured to generate the medical article by inputting the article generation prompt into the large language model and causing the large language model to output the medical article.

3. The information processing system according to claim 1, further comprising a speech-to-text conversion unit, the speech-to-text conversion unit is configured to convert voice conversation data corresponding to the content of the conversation into character data to generate the conversation data, the data acquisition unit acquires the conversation data from the speech-to-text conversion unit.

4. The information processing system according to claim 3, wherein the medical article generation unit has an output generation unit, the output generation unit generates an accuracy improvement prompt for output to a large language model as the learning model, the accuracy improvement prompt has pre-processing character data and the medical data, and has an instruction to generate post-processing character data with reference to the medical data and an instruction to improve the accuracy of the generation of the medical article, the pre-processing character data is the conversation data acquired from the speech-to-text conversion unit, the post-processing character data is the conversation data output from the large language model based on the accuracy improvement prompt. The medical text generation unit is configured to generate the processed character data by inputting the accuracy improvement prompt into the large language model and causing the large language model to output the processed character data, the information processing system.

5. The information processing system according to claim 4, wherein the output generation unit generates a text generation prompt to be output to the large language model, the text generation prompt includes the conversation data and the medical data of the processed character data, and has an instruction to generate the medical text, The medical text generation unit is configured to generate the medical text by inputting the text generation prompt into the large language model and causing the large language model to output the medical text, the information processing system.

6. The information processing system according to claim 4, wherein the instruction for improving the accuracy of generating the medical text in the accuracy improvement prompt has an instruction to generate the processed character data by distinguishing the content spoken by the medical staff from the content spoken by the patient, the conversation data of the processed character data has the content spoken by the medical staff and the content spoken by the patient distinguished from each other, the information processing system.

7. The information processing system according to claim 1, wherein the conversation data has the content spoken by the medical staff and the content spoken by the patient distinguished from each other, the information processing system.

8. The information processing system according to any one of claims 1 to 7, wherein the medical data further includes basic data of the patient, the basic data includes the name of the patient and the gender of the patient, The medical text generation unit further generates the medical text based on the basic data, the information processing system.

9. The information processing system according to any one of claims 1 to 7, wherein the medical data further includes medication history data related to the patient, The medical text generation unit generates the medical text by inputting the conversation data, the pharmaceutical data, and the medication history data into the learning model and causing the learning model to output the medical text, the information processing system.

10. The information processing system according to any one of claims 1 to 7, wherein the medical text is generated by distinguishing subjective information, objective information, evaluation, and plan, the information processing system.

11. An information processing method for assisting in the creation of medical articles, comprising: an acquisition step and a generation step; in the acquisition step, conversation data and medical data are acquired, the conversation data includes the content of conversations between medical practitioners and patients, the medical data has pharmaceutical data related to the patient, in the generation step, the medical article is generated by inputting the conversation data and the pharmaceutical data into a learning model and causing the learning model to output the medical article.

12. A program for causing a computer to execute the information processing method according to claim 11.

Images