Information processing device, selection assistance method, and selection assistance program

Abstract

To improve the manner in which oral care products are presented, an information processing device (1) comprises an acquisition unit (101) that acquires explanatory text that indicates the state of the oral cavity of a subject, an inference unit (103) that inputs a prompt that has been generated on the basis of the acquired explanatory text and instructs inference of an oral care product that is suited to the state of the oral cavity of the subject from among prescribed candidates into a language model (3) to cause output of inference results, and a presentation control unit (104) that presents the oral care product indicated by the inference results.

Description

Information Processing Apparatus, Selection Support Method, and Selection Support Program

[0001] The present invention relates to an information processing apparatus and the like that facilitate the selection of oral care products.

[0002] Oral care is generally aimed at maintaining the oral hygiene status and oral functions, such as preventing tooth decay, periodontal disease, or oral frailty. Oral care products are items used for oral care. Oral care products include a wide variety of items. For example, toothbrushes, tufted brushes, electric toothbrushes, interdental brushes (wire type, rubber type), dental floss, flossers, dentifrices, liquid dentifrices, liquid mouthwashes, mouthwashes, oral pharyngeal medications, oral fresheners, etc. are examples of oral care products.

[0003] Various oral care products have different characteristics. For example, among dentifrices, some have a high tooth whitening effect on the tooth surface, while others have an effect of improving tooth sensitivity. Thus, since various oral care products have different characteristics, users need to select oral care products suitable for the state of their own oral cavity. However, because there are a wide variety of oral care products, it is not easy to select oral care products suitable for the state of one's own oral cavity.

[0004] As a technology for solving such problems, for example, there is an oral care recommendation system described in Patent Document 1 below. According to the oral care recommendation system described in Patent Document 1, a user can receive a presentation of oral care items according to their own oral condition by simply inputting predetermined input information.

[0005] Japanese Patent Application Laid-Open No. 2022-068484

[0006] There is room for improvement in the technology for suggesting oral care products that are suitable for the oral condition of a subject, including the oral care recommendation system described in Patent Document 1. For example, the oral care recommendation system described in Patent Document 1 requires input of an oral image taken of the user's mouth or information about the user's oral care as input information. Furthermore, the information about oral care cannot be just any information; it must be information specified in a pre-stored correspondence table (information showing the correspondence between input information and estimated information). Therefore, if an oral image cannot be obtained, it may not be possible to suggest oral care items. Also, if there is no input information in the correspondence table that corresponds to the user's oral condition, there is a possibility that an oral care item unsuitable for the user will be suggested.

[0007] This invention has been made in view of these problems, and its purpose is to improve the technology for presenting oral care products that are suitable for the oral condition of the target person.

[0008] To solve the above problems, an information processing device according to one aspect of the present invention includes: an acquisition unit that acquires a descriptive text describing the oral condition of a subject; an estimation unit that inputs a prompt to a language model trained in natural language, which instructs the model to estimate an oral care product suitable for the oral condition of the subject from among a predetermined number of candidates for oral care products, generated based on the descriptive text acquired by the acquisition unit, and outputs an estimation result; and a presentation control unit that presents the oral care product shown in the estimation result.

[0009] Furthermore, in order to solve the above problems, an information processing device according to another aspect of the present invention includes: an acquisition unit that acquires a descriptive text indicating the oral condition of a subject; a question generation control unit that inputs a prompt instructing a language model trained in natural language to generate a question for inputting further information to estimate oral care products suitable for the oral condition of the subject, based on the descriptive text acquired by the acquisition unit, and generates a question; and a presentation control unit that presents the question generated under the control of the question generation control unit.

[0010] Furthermore, in order to solve the above problems, an information processing device according to yet another aspect of the present invention includes an acquisition unit that acquires a descriptive text indicating the oral condition of a subject, and a diagnostic unit that determines which of the multiple categories the oral condition of a subject falls under based on the output obtained by inputting a prompt, which is generated based on the descriptive text acquired by the acquisition unit and instructs the language model trained in natural language to determine which of the multiple categories the oral condition of the subject falls under.

[0011] Furthermore, in order to solve the above problems, an information processing device according to yet another aspect of the present invention includes: an acquisition unit that acquires a descriptive text indicating the oral condition of a subject; a question generation control unit that inputs a prompt to a language model trained in natural language, which instructs the language model to generate a question for inputting further information to determine which of a predetermined plurality of types the oral condition of the subject falls under, based on the descriptive text acquired by the acquisition unit, and generates a question; and a presentation control unit that presents the question generated under the control of the question generation control unit.

[0012] Furthermore, in order to solve the above problems, an information processing device according to yet another aspect of the present invention includes: an acquisition unit that acquires feedback on oral care products recommended to a subject; an estimation unit that inputs a prompt, generated based on the feedback acquired by the acquisition unit, to a language model trained in natural language, instructing it to estimate an oral care product suitable for the subject's oral condition from among a predetermined number of candidates for oral care products, and outputs an estimation result; and a presentation control unit that presents the oral care products shown in the estimation result.

[0013] Furthermore, in order to solve the above problems, a selection support method according to one aspect of the present invention is a selection support method for oral care products executed by one or more information processing devices, comprising: an acquisition step of acquiring a descriptive text indicating the oral condition of a subject; an estimation step of inputting a prompt, generated based on the descriptive text acquired in the acquisition step, to a language model trained in natural language, instructing it to estimate an oral care product suitable for the oral condition of the subject from among a predetermined number of candidates for oral care products, and outputting an estimation result; and a presentation control step of presenting the oral care products indicated in the estimation result.

[0014] According to each of the above embodiments of the present invention, it becomes possible to improve the technology for presenting oral care products that are suitable for the oral condition of the target person.

[0015] This figure shows an example of the main components of an information processing device according to one embodiment of the present invention 1. This figure illustrates the outline of a selection support system including the information processing device shown in Figure 1. This figure shows examples of prompts used when estimating oral care products suitable for the oral condition of a subject using the judgment results of the diagnostic unit, and examples of the output of a language model for those prompts. This figure shows examples of prompts for generating questions, examples of generated questions, and examples of prompts for estimating oral care products considering the answers, and examples of the output of a language model for those prompts. This figure shows an example of oral care product presentation based on feedback. This is a flowchart of an example of processing performed by the information processing device shown in Figure 1. This is a flowchart of an example of type narrowing processing. This is a flowchart of an example of oral care product narrowing processing. This is a flowchart of an example of oral care product re-presentation processing.

[0016] [Overview of the Selection Support System] The overview of the selection support system 5 according to this embodiment will be explained based on Figure 2. Figure 2 is a diagram illustrating the overview of the selection support system 5. The selection support system 5 is a system for supporting the selection of oral care products (hereinafter referred to as OC products). As shown in the figure, the selection support system 5 includes an information processing device 1, a database 2, and a language model 3.

[0017] Information processing device 1 is a device that performs various processes to support the selection of OC (Open Campus) supplies. Information processing device 1 may be a local device used by individual users of the OC supply selection support service, or it may be a server that provides the service to multiple users. In the following, we will describe an example in which the person receiving OC supply selection support is themselves a user of the selection support system 5.

[0018] Database 2 is a database that records relevant information related to the oral condition of subjects to assist in the selection of oral hygiene products. As will be explained in detail later, the relevant information is information related to the oral condition of the subjects and does not need to be information that can be used to estimate suitable oral hygiene products for the subjects.

[0019] Language model 3 is a machine learning model that has been trained on natural language. Here, training on natural language means learning the arrangement of constituent elements (such as words) in natural language sentences and the arrangement of sentences in texts. Examples of such language models include GPT (Generative Pre-trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers). By inputting a command sentence written in natural language (called a prompt) into such language model 3, it is possible to output a natural language sentence that is the response to that command sentence.

[0020] In the selection support system 5, when assisting in the selection of oral hygiene products, it is not necessary for the subject to input predetermined information; they only need to describe their oral condition in natural language. For example, in the example in Figure 2, subject P inputs a description A1 into the information processing device 1, explaining in natural language the points of concern regarding their oral condition. The description A1 may be input into the information processing device 1 as audio data or as text data. In the former case, the input audio data is recognized by the information processing device 1 or another device and converted into text data. Furthermore, the description A1 may be input directly into the information processing device 1 or via another device. For example, subject P may input the description into a terminal device they possess. In this case, the description A1 is input into the information processing device 1 via that terminal device.

[0021] Next, the information processing device 1 acquires relevant information related to the oral condition of subject P. In the example in Figure 2, the information processing device 1 acquires relevant information A2 from the database 2, which shows the results of subject P's dental examinations and product purchase history. Such relevant information can be registered in advance for each user of the oral care product selection support service.

[0022] Next, the information processing device 1 uses the explanatory text A1 and related information A2 to generate a prompt A3 that instructs the device to estimate an oral cavity product that is suitable for the oral cavity condition of the subject P from among a predetermined number of candidates for oral cavity products.

[0023] Prompt A3, more specifically, instructs the user to select an oral contraceptive (OC) product from the product list that is recommended for a patient with the specified symptoms, based on the relevant information provided. Prompt A3 also includes explanatory text A1 to indicate the "symptoms" mentioned above, and relevant information A2 obtained from the database as the "relevant information."

[0024] Furthermore, the "Product List" above shows multiple OC products such as "Product X" and "Product Y," along with a description explaining the characteristics of each OC product. The inclusion of OC products in the "Product List" is arbitrary; for example, each OC product manufactured by a specific manufacturer may be included in the "Product List." This allows the information processing device 1 to select the OC product recommended to the user from among the OC products manufactured by that manufacturer. The description of the OC product may include, for example, the characteristics of the OC product, product category, selling points, active ingredients, uses, size, and material. The description of the OC product may also include ingredients other than the active ingredient, or it may include all ingredients, including the active ingredient and other ingredients.

[0025] It is not mandatory to include descriptions of OC products in the "Product List." For example, if a language model 3 that has been trained on various OC products and their characteristics is used, the descriptions of the OC products can be omitted. Alternatively, the descriptions of each OC product may be provided to the language model 3 as related information, and in this case as well, it is not necessary to include the descriptions of each OC product in prompt A3. Furthermore, for example, the information processing device 1 may obtain the descriptions of OC products by searching using the names of the OC products, and in this case as well, it is not necessary to include the descriptions of each OC product in prompt A3.

[0026] Furthermore, prompt A3 includes the sentence, "You are a dentist." While including such a sentence is not mandatory, it allows for the estimation of oral hygiene products recommended from a dentist's perspective. The same effect can be expected by using terms such as "dental hygienist" instead of "dentist." Such a prompt A3 can be generated, for example, by incorporating explanatory text A1 and related information A2 into a predetermined template.

[0027] Next, the information processing device 1 inputs the generated prompt A3 to the language model 3 and outputs the estimation result A4. The estimation result A4 shows product X and product Y as recommended OC products. The language model 3 may be stored in the information processing device 1 or in another device. In the latter case, the information processing device 1 sends prompt A3 to the other device that stores the language model 3 and outputs the estimation result A4, and the output estimation result A4 is retrieved from the other device.

[0028] The information processing device 1 then presents the OC products shown in the estimation result A4 to the subject P. For example, as shown in the example in Figure 2, the information processing device 1 may present the text of the estimation result A4 as recommended information A5, indicating the recommended OC products, to the subject P. The manner in which the recommended information A5 is presented is arbitrary. For example, the information processing device 1 may output the recommended information A5 as audio, display it, or print it.

[0029] As described above, in the selection support system 5, the information processing device 1 inputs a prompt A3 into a language model 3 trained on natural language, instructing it to estimate an oral care product that is suitable for the oral condition of subject P from among a predetermined number of candidates of oral care products, which is generated based on the explanatory text A1 describing the oral condition of subject P. The prompt A3 outputs an estimation result and presents the oral care product shown in the output estimation result. This makes it possible to improve the technology for presenting oral care products that are suitable for the oral condition of the subject. More specifically, it has the effect of increasing the degree of freedom of the input information necessary to present oral care products.

[0030] [Configuration of Information Processing Device 1] The configuration of the information processing device 1 will be explained based on Figure 1. Figure 1 is a block diagram showing the configuration of the information processing device 1. As shown in the figure, the information processing device 1 includes a control unit 10 that controls all parts of the information processing device 1, and a storage unit 11 that stores various data used by the information processing device 1. The information processing device 1 also includes a communication unit 12 for the information processing device 1 to communicate with other devices, an input unit 13 that receives input to the information processing device 1, and an output unit 14 for the information processing device 1 to output data. The control unit 10 includes an acquisition unit 101, a diagnosis unit 102, an estimation unit 103, a presentation control unit 104, and a question generation control unit 105. Details of the diagnosis unit 102 and the question generation control unit 105 will be described later.

[0031] The acquisition unit 101 acquires various information necessary to support the selection of oral hygiene products. For example, the acquisition unit 101 acquires a descriptive text indicating the oral condition of the subject. In addition, the acquisition unit 101 may acquire related information related to the oral condition of the subject, as well as feedback on the presented oral hygiene products. It is also possible to provide different processing blocks for each type of information to be acquired. For example, there may be a descriptive text acquisition unit for acquiring descriptive texts, a related information acquisition unit for acquiring related information, and a feedback acquisition unit for acquiring feedback.

[0032] The estimation unit 103 estimates oral care products suitable for the subject's oral condition. More specifically, the estimation unit 103 generates a prompt instructing the system to estimate an oral care product suitable for the subject's oral condition from among a predetermined number of candidates, based on the explanatory text acquired by the acquisition unit 101. The estimation unit 103 then inputs the generated prompt into a language model 3 trained in natural language to output the estimation result.

[0033] For example, the estimation unit 103 generates prompt A3 as shown in the example in Figure 2. As described above, prompt A3 can be generated by incorporating the explanatory text and related information acquired by the acquisition unit 101 into a predetermined template. Note that the generation of prompts may be performed by a processing block different from the estimation unit 103. For example, a processing block called a prompt generation unit may be provided in the control unit 10, and prompts may be generated by this prompt generation unit.

[0034] Furthermore, the estimation unit 103 may generate a prompt instructing the user to respond with a degree of suitability indicating the degree to which each candidate oral care product is suitable for the user. Such a prompt also falls under the category of "a prompt instructing the user to estimate oral care products that are suitable for the user's oral condition from among the candidate oral care products." In this case, the estimation unit 103 should estimate the oral care products that are suitable for the user's oral condition based on the outputted degree of suitability. For example, the estimation unit 103 may estimate that oral care products with a degree of suitability above a threshold are suitable for the user's oral condition. Alternatively, for example, the estimation unit 103 may estimate that a predetermined number of oral care products with high degrees of suitability are suitable for the user's oral condition.

[0035] The estimation unit 103 may estimate multiple oral care products that are suitable for the subject's oral condition. For example, the estimation unit 103 may recommend a predetermined number of oral care products for each product category. In that case, the estimation unit 103 only needs to generate prompts instructing the user to select a predetermined number of oral care products for each product category.

[0036] The presentation control unit 104 controls the presentation of various information to the user of the information processing device 1. For example, the presentation control unit 104 controls the presentation of OC products shown in the estimation results of the estimation unit 103 to the target person. The manner of presentation is arbitrary. For example, the presentation control unit 104 may present the information by having the audio output device output the information as sound. Alternatively, the presentation control unit 104 may have the information displayed on a display device, or printed on a printing device. Furthermore, the presentation control unit 104 may output the information to an external device of the information processing device 1, or to the output unit 14.

[0037] As described above, the information processing device 1 includes an acquisition unit 101 that acquires a descriptive text describing the oral condition of a subject, an estimation unit 103 that inputs a prompt to a language model 3 trained in natural language, instructing it to estimate an oral care product suitable for the subject's oral condition from among a predetermined number of candidates of oral care products generated based on the descriptive text acquired by the acquisition unit 101, and outputs an estimation result, and a presentation control unit 104 that presents the oral care product shown in the output estimation result.

[0038] According to the above configuration, by having the information processing device 1 acquire a descriptive text indicating the oral condition of the subject, oral hygiene products suitable for the subject's oral condition are presented. Therefore, the information processing device 1 allows for increased flexibility in the input information required to present oral hygiene products, thereby improving the technology for presenting oral hygiene products suitable for the subject's oral condition.

[0039] Furthermore, the display control unit 104 may not only display recommended OC products, but also provide information to support the user in easily purchasing OC products. For example, if the recommended OC products are sold online, the display control unit 104 may provide a link to the sales site. Alternatively, it may provide information on stores that sell the recommended OC products and how to access those stores. In addition, the display control unit 104 may display information that motivates the user to purchase OC products, such as coupons that can be used to purchase the recommended OC products.

[0040] Furthermore, as described above, the acquisition unit 101 may acquire a descriptive text indicating the oral condition of the subject, as well as related information concerning the oral condition of the subject. The estimation unit 103 may then input a prompt to the language model 3, instructing it to estimate an oral care product suitable for the subject's oral condition from among multiple candidates, based on the descriptive text and related information acquired by the acquisition unit 101, and output an estimation result. This makes it possible to present an appropriate oral care product while also considering related information concerning the subject's oral condition.

[0041] [Regarding related information] As mentioned above, related information is any information related to the oral condition of the subject that can be used to estimate suitable oral hygiene products for the subject. For example, related information can include medical data such as the results of the subject's dental checkup, health checkup, past dental treatment history, and interview results, as well as purchase history of oral hygiene products, data showing dietary tendencies, and physical data such as age, height, and weight. The results of the dental checkup may include, for example, the condition of the teeth, gums, dentition, occlusion, temporomandibular joint, and plaque. The results of the health checkup may also include the presence or absence of lifestyle-related diseases and values ​​related to lifestyle-related diseases (blood pressure, blood sugar levels, etc.).

[0042] Such health check-related data can be generated, for example, by processing images of health check-result forms using OCR (Optical Character Reader). Alternatively, this data can be entered in advance by the individuals concerned. Furthermore, the purchase history of health check-up supplies can be identified, for example, from ID-POS (Pint of Sales) or purchase data from online shopping sites. It can also be identified from images of purchased health check-up supplies or collected by reading barcodes or other codes attached to the supplies.

[0043] Furthermore, related information may include, for example, information related to the beauty of the area around the mouth. For instance, related information may include information obtained by analyzing the facial images of the subjects that shows the cosmetic condition of the area around the mouth and neck (e.g., wrinkles, skin texture, dullness, moisture content, sebum content, apparent age, facial impression evaluation, etc.).

[0044] Further, for example, the action history of the subject may be used as related information. The action to be targeted may be any action related to the oral condition. For example, history information indicating the implementation date and time of oral care, the duration, the care products used, etc. may be used as related information. Such history information can be obtained, for example, from a smart toothbrush (which may be an electric brush or a manual brush as long as it has a function of recording the action history), a smart mirror, a smart watch, a toothbrush stand with a built-in sensor, etc.

[0045] Also, for example, information collected using application software for oral care or healthcare (hereinafter referred to as an app) can be used as related information. For example, the history of brushing teeth, the check results of oral functions, the history of exercise, weight, blood pressure, blood glucose level, heart rate, sleep, blood oxygen concentration, and the history of diet, etc. may be collected by the app and those information may be used as related information.

[0046] Note that a part of the above related information can also be included in the description text indicating the oral condition of the subject. For example, the acquisition unit 101 may generate a description text indicating the oral condition of the subject using the above related information, or may obtain from another device a description text indicating the oral condition of the subject generated by the other device using the above related information of the subject.

[0047] [Diagnosis of Oral Condition] The diagnosis unit 102 determines which of a plurality of predetermined types the oral condition of the subject corresponds to. The types of oral conditions may be determined in advance according to the characteristics of the recommended OC products, etc. For example, the presence or absence of gingival swelling / bleeding, gingivitis, periodontitis, gingival recession, staining, tongue coating, dental plaque, tooth sensitivity, festoon, bony swelling, etc. may be defined in advance as types.

[0048] Specifically, the diagnosis unit 102 generates a prompt that instructs to determine which of a plurality of predetermined types the oral condition of the subject corresponds to based on the description text acquired by the acquisition unit 101. Further, the diagnosis unit 102 inputs the generated prompt into the language model 3 that has learned natural language. Then, the diagnosis unit 102 determines which of the plurality of types the oral condition of the subject corresponds to based on the output of the language model 3.

[0049] The determination result of the diagnosis unit 102 is used for the estimation of the OC supplies by the estimation unit 103. Hereinafter, based on FIG. 3, the determination by the diagnosis unit 102 and the estimation of the OC supplies using the determination result will be described. FIG. 3 is a diagram showing an example of a prompt used when estimating OC supplies suitable for the oral condition of the subject using the determination result of the diagnosis unit 102 and an example of the output of the language model 3 for those prompts.

[0050] The prompt B1 shown in FIG. 3 is an example of a prompt generated by the diagnosis unit 102. The prompt B1 is configured to instruct to determine which of a plurality of predetermined types the oral condition of the subject corresponds to based on the description text acquired by the acquisition unit 101. The prompt B1 includes the description text (specifically, the description text A1 shown in FIG. 2) acquired by the acquisition unit 101 as indicating the "symptoms" of the subject, and a plurality of predetermined types are shown as a "type list". Each type is shown in the "type list". Further, the description text of each type may be shown in the "type list".

[0051] It is not mandatory to include descriptions of the types in the "Type List." For example, descriptions can be omitted for types whose characteristics the language model 3 has already learned. Alternatively, the descriptions of each type may be provided to the language model 3 as related information, in which case the descriptions of each type can also be omitted from prompt B1. In addition, for example, the information processing device 1 may obtain the descriptions of the types by searching using the type names, in which case the descriptions of each type can also be omitted from prompt B1. Furthermore, the aforementioned related information may be included in prompt B1. In this case, prompt B1 should be set to instruct the device to make a determination by referring to the related information.

[0052] Furthermore, the diagnostic unit 102 may generate prompts instructing the user to respond with a degree of fit indicating the extent to which the subject's oral condition conforms to each type. Such prompts also fall under the category of "prompts instructing the user to determine which of a predetermined number of types the subject's oral condition falls into." In this case, the diagnostic unit 102 can determine which type the subject's oral condition corresponds to based on the outputted degree of fit. For example, the diagnostic unit 102 may determine that the subject's oral condition corresponds to any type whose degree of fit is above a threshold. Alternatively, the diagnostic unit 102 may determine that the subject's oral condition corresponds to any predetermined number of types with the highest degree of fit.

[0053] Prompt B1 can be generated by incorporating the explanatory text acquired by the acquisition unit 101 into a predetermined template, similar to the prompt for estimating an oral cavity product suitable for the subject's oral condition (for example, prompt A3 in Figure 2).

[0054] In the example in Figure 3, output data B2 is generated by inputting prompt B1 to the language model 3. Output data B2 is text data indicating that the subject's oral condition is highly likely to fall under type 1 of the types shown in the type list of prompt B1, and also likely to fall under type 2. The diagnostic unit 102 may use output data B2 as the judgment result as is, or it may determine the type to which the subject's oral condition falls based on output data B2. For example, in the example in Figure 3, the diagnostic unit 102 may determine that both type 1 and type 2 are the types to which the subject's oral condition falls, or it may determine that either type 1 or type 2 is the type to which the subject's oral condition falls.

[0055] Next, the estimation unit 103 generates a prompt instructing the system to estimate from among several candidates an oral hygiene product that matches the type indicated in the diagnosis result of the diagnostic unit 102. Prompt B3 shown in Figure 3 is an example of a prompt generated by the estimation unit 103. Prompt B3 shows the diagnosis result of the diagnostic unit 102, gingivitis (type 1) and tooth discoloration (type 2), as the "oral condition," and also shows a "product list" of candidate oral hygiene products to recommend to the subject. Such a prompt B3 can also be generated by incorporating the diagnosis result of the diagnostic unit 102 into a predetermined template.

[0056] As mentioned above, the diagnosis by the diagnostic unit 102 is performed using the explanatory text acquired by the acquisition unit 101. Therefore, prompt B3, which instructs the system to estimate suitable OC products for the subject using the judgment result of the diagnostic unit 102, can be said to be a prompt generated based on the explanatory text acquired by the acquisition unit 101.

[0057] The estimation unit 103 then inputs the generated prompt B3 to the language model 3. In the example in Figure 3, the language model 3 outputs estimation result B4 as a response to prompt B3. Estimation result B4 recommends product X, which has been shown to improve gingivitis, and product Y, which has a high effect in removing stains from the tooth surface. These oral care products are then presented to the subject by the presentation control unit 104.

[0058] Generally, for each type of oral condition (for example, swollen or bleeding gums, discoloration, bad breath, etc.), there are oral hygiene products suitable for that type. Therefore, as described above, the information processing device 1 is equipped with a diagnostic unit 102 that determines which of the predetermined types of oral conditions the subject's oral condition falls under based on the output obtained by inputting a prompt to the language model 3, which is generated based on the explanatory text acquired by the acquisition unit 101 and instructs the language model 3 to determine which of the predetermined types the subject's oral condition falls under.

[0059] When the diagnostic unit 102 makes the above determination, the estimation unit 103 inputs a prompt to the language model 3 instructing it to estimate an OC product that matches the type indicated by the diagnostic unit 102's determination result from among multiple candidates, and outputs the estimation result. This increases the likelihood that an appropriate OC product will be presented.

[0060] Furthermore, the presentation control unit 104 may present the judgment result of the diagnostic unit 102 to the subject. For example, the presentation control unit 104 may present the judgment result of the diagnostic unit 102 to the subject along with the oral hygiene products estimated by the estimation unit 103. This makes it possible to make the subject aware of the condition of their oral cavity and to increase the reliability of the recommended oral hygiene products.

[0061] As mentioned above, there is technical significance in providing the diagnostic unit 102, but it is not essential. If the diagnostic unit 102 is omitted, the estimation unit 103 generates a prompt that instructs the system to estimate an oral hygiene product suitable for the subject's oral condition from among a predetermined number of candidates for oral hygiene products, using the explanatory text acquired by the acquisition unit 101, and inputs the generated prompt into the language model 3 to output the estimation result.

[0062] [Question Generation] Since there are no particular restrictions on the information that the subject inputs into the information processing device 1, the subject can freely input a descriptive statement about the condition of their oral cavity. However, if the degree of freedom of input information is increased in this way, the information processing device 1 may not be able to sufficiently narrow down the recommended oral hygiene products or the type to which the subject's oral cavity condition belongs, or it may not be able to estimate appropriate oral hygiene products or types, due to the input descriptive statement being insufficient. In addition, the subject may not like the oral hygiene products recommended by the information processing device 1.

[0063] To address these issues, the information processing device 1 is equipped with a question generation control unit 105. The question generation control unit 105 generates questions to prompt the input of further information regarding the subject's oral condition. The question generation control unit 105 will be described below with reference to Figure 4. Figure 4 shows an example of a prompt for generating questions and an example of a generated question. Figure 4 also shows an example of a prompt for estimating oral hygiene products considering the answer, and an example of the output of a language model in response to that prompt.

[0064] The prompt C1 shown in Figure 4 is an example of a prompt generated by the question generation control unit 105. Prompt C1 instructs the system to generate a question to prompt the input of further information in order to estimate an oral cavity product suitable for the subject's oral condition, based on the explanatory text acquired by the acquisition unit 101. Prompt C1 includes the explanatory text acquired by the acquisition unit 101 (specifically, explanatory text A1 shown in Figure 2) as an indication of the subject's "symptoms," and also shows candidate oral cavity products as a "product list." The aforementioned related information may also be included in prompt C1. In this case, prompt C1 should be instructed to generate a question by referring to the related information. Prompt C1 can be generated by incorporating the explanatory text acquired by the acquisition unit 101 into a predetermined template, similar to the prompt for estimating an oral cavity product suitable for the subject's oral condition (for example, prompt A3 in Figure 2).

[0065] Furthermore, the question generation control unit 105 may generate different prompts depending on the circumstances when generating questions. For example, if the question generation control unit 105 is unable to narrow down the number of recommended oral care products to a predetermined number, it may generate a prompt instructing the system to generate questions that include each of the oral care products that are subject to narrowing down to a predetermined number. Also, for example, if the recommended oral care products do not suit the subject's needs, the question generation control unit 105 may generate a prompt instructing the system to generate questions that include the recommended oral care products and elicit information necessary to identify oral care products other than those recommended that are suitable for the subject's oral condition.

[0066] In the example shown in Figure 4, the question generation control unit 105 inputs prompt C1 to the language model 3, which generates question C2. Question C2 is text data asking whether the subject uses interdental brushes or mouthwash. The presentation control unit 104 presents this question C2 to subject P, who responds that they use interdental brushes but not mouthwash (response C3 in the figure).

[0067] The acquisition unit 101 acquires the subject P's response C3. This allows the estimation unit 103 to re-estimate OC products suitable for the subject based on the information shown in response C3. Specifically, the estimation unit 103 generates a prompt instructing the system to estimate OC products suitable for the subject's oral condition, taking response C3 into consideration.

[0068] The prompt C4 shown in Figure 4 is an example of a prompt that instructs the system to estimate an oral cavity (OC) product suitable for the subject's oral condition, taking into account the answer to the question. Prompt C4 instructs the system to estimate an OC product suitable for the subject's oral condition, based on the explanatory text acquired by the acquisition unit 101 and the answer to the question generated by the question generation control unit 105. Specifically, prompt C4 includes an explanatory text acquired by the acquisition unit 101 (specifically, explanatory text A1 shown in Figure 2) that indicates the subject's "symptoms," and a list of candidate OC products presented as a "product list." It also includes a question text C2 generated by the question generation control unit 105 as the "question," and an answer C3 acquired by the acquisition unit 101 as the "answer."

[0069] The aforementioned related information may also be included in prompt C4. In this case, prompt C4 should instruct the system to estimate oral care products by referring to the related information. Prompt C4 can be generated by incorporating the explanatory text acquired by the acquisition unit 101, the question C2 generated by the question generation control unit 105, and the answer C3 to the question C2 into a predetermined template, similar to the prompt for estimating oral care products suitable for the subject's oral condition (for example, prompt A3 in Figure 2).

[0070] The estimation unit 103 then inputs prompt C4 to the language model 3. In the example in Figure 4, the language model 3 outputs estimation result C5 as a response to prompt C4. Estimation result C5 recommends mouthwash (product X), which is an oral hygiene product that the subject does not currently use. This oral hygiene product is then presented to the subject by the presentation control unit 104.

[0071] As described above, the information processing device 1 includes a question generation control unit 105 that inputs a prompt to the language model 3 instructing it to generate questions for inputting further information to estimate OC products suitable for the oral condition of the subject, based on the explanatory text acquired by the acquisition unit 101, and the presentation control unit 104 presents the questions generated under the control of the question generation control unit 105.

[0072] According to the above configuration, questions are generated and presented to prompt the input of further information in order to estimate the oral cavity product suitable for the subject's oral condition. This makes it possible to input further information in order to estimate the oral cavity product suitable for the subject's oral condition. Therefore, according to the above configuration, even if the previously obtained explanatory text is not sufficient to estimate the oral cavity product suitable for the subject's oral condition, it becomes possible to estimate the appropriate oral cavity product.

[0073] Furthermore, the question generation control unit 105 can also generate questions to prompt the input of further information to determine which type the subject's oral condition falls into. For example, the question generation control unit 105 can generate a prompt that instructs the system to generate questions to prompt the input of further information to determine which type the subject's oral condition falls into by incorporating the explanatory text and related information acquired by the acquisition unit 101 into the "symptoms" and "related information" sections of a template like the one shown below. The question generation control unit 105 can then generate questions by inputting the generated prompt into the language model 3. "You are a dentist. The following symptoms and related information have been obtained to help you diagnose which of the following types the patient's oral condition falls into. Consider what you should confirm for diagnosis and create questions to elicit that information. Symptoms: Related information: List of types: ・Type 1: Gingivitis. A condition in which the gums are inflamed. Typical symptoms of gingivitis include redness, swelling, and bleeding of the gums. ・Type 2: Tooth discoloration. A condition in which pigments such as tannins, nicotine, tar, etc. adhere to the surface of the teeth. The surface of the teeth becomes yellowish or dull. ・..." In this way, the question generation control unit 105 may input a prompt to the language model 3 instructing it to generate questions that require input of further information to determine which of the multiple types the subject's oral condition falls into, based on the explanatory text acquired by the acquisition unit 101. The presentation control unit 104 may then present the questions generated under the control of the question generation control unit 105.

[0074] According to the above configuration, questions are generated and presented to prompt the user to input further information in order to determine which category the subject's oral condition falls into. This makes it possible to input further information in order to determine which category the subject's oral condition falls into. Therefore, according to the above configuration, even if the previously obtained descriptive text is insufficient to determine the category of the subject's oral condition, it becomes possible to appropriately determine the category of the subject's oral condition.

[0075] After a question is presented, the acquisition unit 101 acquires the subject's response to the presented question. Then, the diagnosis unit 102 re-determines the type to which the subject's oral condition corresponds, based on the new information regarding the subject's oral condition shown in the acquired response. Specifically, the diagnosis unit 102 generates a prompt instructing the language model 3 to determine the type to which the subject's oral condition corresponds, taking the acquired response into consideration, and inputs the generated prompt to the language model 3 to output the determination result.

[0076] For example, the diagnostic unit 102 can generate a prompt that instructs the system to determine the type to which the subject's oral condition corresponds, taking into account the acquired answers. This prompt can be generated by incorporating the explanatory text and related information acquired by the acquisition unit 101 into the "symptoms" and "related information" sections of a template like the one below, and by incorporating the questions and answers generated by the question generation control unit 105 into the "questions" and "answers" sections. "You are a dentist. Please answer which type to which the patient's oral condition corresponds. The following symptoms and related information have been obtained as judgment material, and the following answers have been obtained for the following questions. Symptoms: Related information: Type list: Question: Answer:" The diagnostic unit 102 then inputs the generated prompt into the language model 3 and determines which type the subject's oral condition corresponds to based on the output from the language model 3. Subsequently, the estimation unit 103 estimates from multiple candidates an oral care product that matches the type shown in the new judgment result of the diagnostic unit 102.

[0077] [Feedback Reception] The acquisition unit 101 may acquire feedback on the oral care products presented by the presentation control unit 104. In this case, the estimation unit 103 may input a new prompt to the language model 3, which is generated based on the feedback acquired by the acquisition unit 101, instructing it to estimate an oral care product suitable for the subject's oral condition from among multiple candidates, and output a new estimation result. The presentation control unit 104 may then present the oral care product shown in the new estimation result. This makes it possible to present new oral care products that take into account the feedback on previously presented oral care products. Feedback can be freely entered in natural language.

[0078] An example of suggesting oral hygiene products based on feedback will be explained using Figure 5. Figure 5 is a diagram illustrating an example of suggesting oral hygiene products based on feedback. In the example in Figure 5, subject P, who was suggested mouthwash as a suitable oral hygiene product, gives feedback D1 requesting that a toothbrush or toothpaste be recommended instead of mouthwash.

[0079] This feedback D1 is acquired by the acquisition unit 101. The estimation unit 103 then generates a new prompt that instructs the estimation unit 103 to estimate an oral care product suitable for the subject's oral condition from among several candidates, based on the feedback D1 acquired by the acquisition unit 101. Prompt D2 shown in Figure 5 is an example of the new prompt.

[0080] Prompt D2 instructs the system to estimate an oral care product suitable for the subject's oral condition from among several candidates, based on the content of Feedback D1. Prompt D2 includes the explanatory text and Feedback D1 acquired by the acquisition unit 101 as "symptoms" and "feedback," respectively, and also shows the previously recommended oral care product as "recommended product." Furthermore, Prompt D2 shows a list of candidate oral care products to recommend to the subject as "product list." Such a Prompt D2 can also be generated by incorporating the explanatory text and feedback acquired by the acquisition unit 101 and the previously recommended oral care product from the presentation control unit 104 into a predetermined template.

[0081] The estimation unit 103 then inputs the generated prompt D2 to the language model 3. In the example in Figure 5, the language model 3 outputs the estimation result D3 as a response to prompt D2. The estimation result D3 recommends product Y, which is toothpaste. This consumer care product is then presented to the target person by the presentation control unit 104.

[0082] [Variations of Input Information and Related Information] As described above, if a descriptive text describing the oral condition of the subject is entered, the information processing device 1 can be made to recommend oral hygiene products. However, instead of a descriptive text, an image showing the oral condition of the subject may be entered.

[0083] In this case, the control unit 10 of the information processing device 1 can be equipped with an image analysis unit that generates a descriptive text about the subject's oral condition from an image showing the subject's oral condition. In this case, the estimation unit 103 inputs a prompt to the language model 3 instructing it to estimate an oral care product suitable for the subject's oral condition from among multiple candidates, based on the descriptive text generated by the image analysis unit, and outputs the estimation result. With this configuration, it becomes possible to suggest an appropriate oral care product simply by inputting an image.

[0084] The method for generating a descriptive text from an image is arbitrary. For example, the image analysis unit may generate the descriptive text using a generative model created by machine learning to generate a descriptive text for an input image. Alternatively, for example, the image analysis unit may classify the oral cavity condition using a classification model that has been machine-learned to classify the oral cavity condition into multiple types, and generate the descriptive text using the results of that classification.

[0085] Furthermore, for example, the image analysis unit may generate explanatory text using a model such as a VLM (Vision and Language Model) that can accept both text and images as input. In this case, the subject would only need to input an image of their oral cavity or face along with a description of the condition of their oral cavity.

[0086] Images showing the condition of the subject's oral cavity only need to show the condition of the subject's oral cavity. For example, images of the subject's oral cavity or face will reflect the condition of the subject's oral cavity, so the image analysis unit can generate a descriptive text of the oral cavity from such images. Images of the oral cavity may be taken by the subject using their own terminal device, or by a dentist or other professional. Images taken with the mouth open using a mouth opener may also be used, as may images taken with the degree of plaque accumulation emphasized using a plaque disclosing agent. Furthermore, the input images may be single or multiple. For example, images taken with different mouth opening conditions may be input.

[0087] Furthermore, since the condition of the subject's oral cavity is reflected in images of oral hygiene products used by the subject, and images of oral hygiene products after use, the image analysis unit can generate a descriptive text about the oral cavity condition from such images. For example, it is possible to estimate what the subject is concerned about from the oral hygiene products they are using (for example, if the oral hygiene products they are using have a preventive effect against gingivitis, it can be estimated that the subject is concerned about gingivitis). Also, in the case of brushes such as toothbrushes, interdental brushes, and tuft brushes, the condition of the oral cavity is reflected in the degree of bristle spread, wear, and disorder. The images to be analyzed may also be X-ray images, CT (Computed Tomography) images, 3D scan images, etc., taken at a dental clinic.

[0088] The acquisition unit 101 may also acquire images showing the oral condition of the subject as related information. In this case, if a VLM or similar model that can accept both text and images is used as the language model 3, the acquisition unit 101 can generate a prompt including the acquired image and the input explanatory text and input it to the language model 3. If a language model 3 that cannot accept images is used, the image analysis unit can generate an explanatory text for the image and include the generated explanatory text in the prompt as related information. Alternatively, the process of generating an explanatory text from an image may be performed by another device different from the information processing device 1. In that case, the acquisition unit 101 can acquire the explanatory text generated by the other device from images showing the oral condition of the subject as the explanatory text showing the oral condition of the subject.

[0089] [Processing Flow: Overall] The processing flow executed by the information processing device 1 will be explained based on Figure 6. Figure 6 is a flowchart showing an example of the processing executed by the information processing device 1. Note that the processing in Figure 6 includes each step of the selection support method according to this embodiment.

[0090] In step S1 (acquisition step), the acquisition unit 101 acquires a descriptive text describing the oral condition of the subject. As mentioned above, the descriptive text can be freely entered by the subject in natural language. In addition, in step S1, the acquisition unit 101 may also acquire the subject's identification information (for example, the user ID of the selection support system 5).

[0091] In S2, the acquisition unit 101 acquires relevant information related to the oral condition of the subject. For example, the acquisition unit 101 may use the identification information described above to acquire relevant information about the subject from the database 2 shown in Figure 2.

[0092] In S3, the diagnostic unit 102 uses the explanatory text obtained in S1 and the related information obtained in S2 to generate a prompt instructing the system to determine which of several predetermined categories the subject's oral condition falls into, by referring to the related information.

[0093] In S4, the diagnostic unit 102 inputs the prompt generated in S3 to the language model 3 and determines which type the subject's oral condition falls into based on the output obtained. If, in S4, there are multiple types to which the subject's oral condition may fall (for example, if there are multiple types whose degree of fit with the subject's oral condition is above a threshold), the diagnostic unit 102 proceeds to S5 with those multiple types as the determination result.

[0094] In S5, the diagnostic unit 102 determines whether or not to narrow down the categories. If the result in S5 is YES, the process proceeds to S6; if the result in S5 is NO, the process proceeds to S7. The conditions for narrowing down the categories can be predetermined. For example, if there is an upper limit on the applicable categories, the diagnostic unit 102 may determine whether to narrow down the categories if the results of S4 include categories that exceed that upper limit.

[0095] In S6, the type filtering process is performed, and then the process returns to S5. Details of S6 will be described later based on Figure 7.

[0096] In S7, the estimation unit 103 generates a prompt instructing the system to estimate an oral cavity product that is suitable for the subject's oral condition from among a predetermined number of candidates for oral cavity products. More specifically, the estimation unit 103 generates a prompt instructing the system to estimate an oral cavity product that is suitable for the type determined in S4 (or the type after narrowing down if narrowing down has been performed) from among the number of candidates. As mentioned above, the explanatory text obtained in S1 is used to determine the type, so in S7, the estimation unit 103 generates the above prompt based on the explanatory text obtained in S1.

[0097] In S8 (estimation step), the estimation unit 103 inputs the prompt generated in S7 to the language model 3 and outputs an estimated result of oral hygiene products that are suitable for the subject's oral condition (more precisely, the type of oral condition determined by the diagnostic unit 102).

[0098] In S9, the estimation unit 103 determines whether or not to narrow down the recommended OC products. If the result in S9 is YES, the process proceeds to S10; if the result in S9 is NO, the process proceeds to S11. The conditions for narrowing down the products can be predetermined. For example, if there is an upper limit on the number of recommended OC products, the estimation unit 103 may determine whether to narrow down the products if the estimation result in S8 includes OC products exceeding that upper limit.

[0099] In S10, the OC supplies are narrowed down, and then the process returns to S9. Details of S10 will be described later based on Figure 8.

[0100] In S11 (presentation control step), the presentation control unit 104 presents the target person with the OC products (or the narrowed-down OC products if narrowing down has been performed) indicated in the estimation results output in S8.

[0101] In S12, the acquisition unit 101 determines whether or not feedback is present. If the result in S12 is YES, the process proceeds to S13; if the result in S12 is NO, the illustrated process ends. In S13, the OC supplies re-presentation process is performed, and then the process returns to S12. Details of S13 will be described later based on Figure 9.

[0102] As described above, the method for supporting the selection of oral care products according to this embodiment includes an acquisition step of acquiring a descriptive text indicating the oral condition of a subject; an estimation step of inputting a prompt, generated based on the descriptive text acquired in the acquisition step, into a language model 3 trained on natural language, instructing it to estimate an oral care product suitable for the subject's oral condition from among a predetermined number of candidates, and outputting an estimation result; and a presentation control step of presenting the oral care products indicated in the outputted estimation result. Thus, it becomes possible to improve the technology for presenting oral care products suitable for the oral condition of a subject.

[0103] [Processing Flow: Category Refinement Process] The flow of the category refinement process performed in S6 of Figure 6 will be explained based on Figure 7. Figure 7 is a flowchart showing an example of the category refinement process.

[0104] In S61, the question generation control unit 105 generates a prompt instructing the system to generate questions that prompt the input of further information to determine which of several categories the subject's oral condition falls under, based on the explanatory text acquired in S1 of Figure 6. For example, the question generation control unit 105 may generate a prompt instructing the system to generate questions that include the explanatory text acquired in S1 of Figure 6, the related information acquired in S2 of the same figure, and each category determined in S4 of the same figure, in order to elicit the information necessary to determine which of those categories the subject's oral condition falls under.

[0105] In S62, the question generation control unit 105 inputs the prompt generated in S61 into the language model 3 to generate a question. Next, in S63, the presentation control unit 104 presents the question generated in S62 to the subject. Then, in S64, the acquisition unit 101 acquires the subject's answer to the question presented in S63.

[0106] In S65, the diagnostic unit 102 generates a new prompt instructing it to determine which of several types the subject's oral condition falls under. This prompt includes the answer obtained in S64. For example, the diagnostic unit 102 may use the explanatory text obtained in S1 of Figure 6, the types determined in S4 of the same figure, the question presented in S63, and the answer obtained in S64 to generate a new prompt instructing it to determine, based on the question and answer, which of the types the subject's oral condition falls under the symptoms shown in the explanatory text.

[0107] In S66, the diagnostic unit 102 inputs the prompt generated in S65 to the language model 3 to output a judgment result and determines the type to which the subject's oral condition corresponds. This completes the process shown in Figure 7. Once the process in Figure 7 is complete, the process in S5 of Figure 6 is performed.

[0108] In the example shown in Figure 7, a question is generated, and after obtaining the answer, a new prompt is generated. However, the question generation control unit 105 may repeatedly generate a question and the acquisition unit 101 may repeatedly obtain the answer. The diagnostic unit 102 may then use the multiple answers obtained through these repetitions to generate a new prompt. This is also the case in the example shown in Figure 8, which will be described later.

[0109] [Processing Flow: OC Supplies Filtering Process] The flow of the OC supplies filtering process performed in S10 of Figure 6 will be explained based on Figure 8. Figure 8 is a flowchart showing an example of the OC supplies filtering process.

[0110] In S101, the question generation control unit 105 generates a prompt instructing the system to generate questions to elicit further information for estimating which oral care product is suitable for the subject's oral condition, based on the explanatory text acquired in S1 of Figure 6. For example, the question generation control unit 105 may generate a prompt instructing the system to generate questions that include the explanatory text acquired in S1 of Figure 6, the related information acquired in S2 of the same figure, and each oral care product shown in the estimation result of S8 of the same figure, and to elicit the information necessary to estimate which of these oral care products is suitable for the subject's oral condition. For example, the question generation control unit 105 may generate a prompt like prompt C1 in Figure 4.

[0111] In S102, the question generation control unit 105 inputs the prompt generated in S101 into the language model 3 to generate a question. Next, in S103, the presentation control unit 104 presents the question generated in S102 to the subject. Then, in S104, the acquisition unit 101 acquires the subject's answer to the question presented in S103.

[0112] In S105, the estimation unit 103 generates a new prompt instructing the system to estimate oral care products suitable for the subject's oral condition. This prompt includes the answer obtained in S104. For example, the estimation unit 103 may use the explanatory text obtained in S1 of Figure 6, the oral care products shown in the estimation results of S8 in the same figure, the question presented in S103, and the answer obtained in S104 to generate a new prompt like prompt C4 in Figure 4.

[0113] In S106, the estimation unit 103 inputs the prompt generated in S105 to the language model 3 and outputs the estimation result. This completes the process shown in Figure 8. Once the process in Figure 8 is complete, the process in S9 of Figure 6 is performed.

[0114] [Processing Flow: OC Supplies Re-presentation Process] The flow of the OC supplies re-presentation process performed in S13 of Figure 6 will be explained based on Figure 9. Figure 9 is a flowchart showing an example of the OC supplies re-presentation process.

[0115] In S131, the acquisition unit 101 acquires feedback from the subject. As mentioned above, the feedback can be freely entered in natural language.

[0116] In S132, the estimation unit 103 generates a new prompt that instructs the system to estimate an oral care product suitable for the subject's oral condition from among several candidates, based on the feedback obtained in S131. For example, the estimation unit 103 may generate a new prompt like prompt D2 in Figure 5 using the explanatory text obtained in S1 in Figure 6, the oral care product presented in S11 in Figure 6, and the feedback obtained in S131.

[0117] In S133, the estimation unit 103 inputs the new prompt generated in S132 to the language model 3 and outputs a new estimation result for oral hygiene products that are suitable for the subject's oral condition.

[0118] In S134, the presentation control unit 104 presents the OC product shown in the new estimation result output in S133 to the subject. This completes the process shown in Figure 9. Once the process in Figure 9 is complete, the process shown in S12 of Figure 6 is performed.

[0119] Furthermore, if the feedback received in S131 is insufficient or unclear, the new OC (Open Community) products presented in S134 may not be in line with the subject's wishes. For this reason, after the feedback is received in S131, the question generation control unit 105 may have the language model 3 generate questions asking the subject about matters that should be confirmed regarding the feedback. Next, the presentation control unit 104 presents the generated questions to the subject and has the acquisition unit 101 acquire the answers to those questions. Then, the estimation unit 103 estimates new OC products based on those answers. This increases the likelihood of recommending OC products that are in line with the subject's wishes.

[0120] [Modification] In the above embodiment, one language model 3 is used for multiple purposes (e.g., estimation of oral hygiene products, diagnosis of oral condition, generation of questions, etc.), but different language models may be used for each purpose. In that case, a general-purpose language model may be fine-tuned for each purpose. This makes it possible to improve the accuracy of the language model's output.

[0121] Furthermore, although the above embodiment describes an example in which the language model 3 generates questions for the subject, the questions for the subject may be prepared in advance or generated using a predetermined template. For example, if the judgment result of the diagnostic unit 102 indicates multiple types, the presentation control unit 104 may present a question asking which of those multiple types the subject is concerned about. Such a question can be generated by combining a standard phrase such as "Please tell us which of the following types you are concerned about" with the multiple types indicated by the judgment result of the diagnostic unit 102. Based on the subject's answer to such a question, the diagnostic unit 102 can narrow down the types to which the subject's oral condition corresponds. Note that if the answer to the above question indicates that the subject is concerned about all types, the diagnostic unit 102 does not need to narrow down the types. In that case, the estimation unit 103 only needs to estimate one or more OC products that are effective for those multiple types.

[0122] Furthermore, the presentation control unit 104 may pre-determine what questions to present in what situations. This makes it possible to elicit necessary information from the subject according to the situation, determine an appropriate type, and accurately decide which OC products to recommend. For example, questions corresponding to the diagnostic results of the diagnostic unit 102 may be pre-determined, and additional questions corresponding to the answers to those questions may also be pre-determined.

[0123] For example, if the diagnostic result from the diagnostic unit 102 falls under the category of "swollen gums," it may be predetermined to ask the question, "Do you experience bleeding after brushing?" If the answer to this question indicates bleeding, it may be predetermined to ask the question, "When did the gum swelling start?" This allows for more detailed information about the gum swelling to be elicited from the subject. The information obtained can then be used to accurately determine which oral hygiene products should be recommended (for example, by including it as reference information in a prompt to estimate which oral hygiene products should be recommended).

[0124] Furthermore, information indicating the subject's current and past oral care status is useful for estimating appropriate oral care products for the subject. For this reason, the presentation control unit 104 may present questions to elicit information indicating the subject's oral care status. For example, the presentation control unit 104 may present questions corresponding to the diagnostic results of the diagnostic unit 102, which are intended to elicit information indicating the subject's oral care status.

[0125] For example, if the diagnostic result from the diagnostic unit 102 falls under the category of "swollen gums," it may be predetermined to ask the question, "Are you already using oral care products for periodontal disease prevention or performing interdental care?" If the answer to this question is positive, it may be predetermined to ask further questions about the oral care situation (for example, how often oral care products are used, the names of the products used, etc.). This allows information about the subject's oral care situation to be elicited. The information obtained can then be used to accurately determine which oral care products should be recommended (for example, by including it as reference information in a prompt to estimate which oral care products should be recommended).

[0126] Furthermore, if the above-mentioned questions are presented, it may be found from the answers to those questions that the subject is already using the oral contraceptive (OC) product estimated by the estimation unit 103. In that case, the presentation control unit 104 may provide advice regarding the use of the OC product. For example, suppose the diagnosis result from the diagnostic unit 102 falls under the category of "swollen gums," and the subject is using an OC product that is effective in suppressing the swelling of the gums, but the frequency of use is lower than the ideal frequency. In this case, the presentation control unit 104 may provide advice recommending that the frequency of use of the OC product be increased to the ideal frequency.

[0127] Furthermore, it is possible that the subject is already using oral care products corresponding to the type indicated by the diagnostic results of the diagnostic unit 102, and that their usage is appropriate. In this case, it is also possible that the subject's oral condition is poor due to factors other than oral care. For this reason, if the subject is using oral care products appropriately according to the type indicated by the diagnostic results of the diagnostic unit 102, the presentation control unit 104 may present questions asking about factors other than oral care.

[0128] For example, in the above-mentioned case, the presentation control unit 104 may present the subject with a standard question such as, "Do you suffer from a systemic disease that exacerbates periodontal disease, such as diabetes?" If the answer to this question is positive, the presentation control unit 104 may then present information indicating that diabetes and periodontal disease exacerbate each other, and encourage the subject to consult a doctor or dentist appropriately.

[0129] Here, the input information to the information processing device 1 may be unclear. Examples of unclear input information include, for example, input information that is inconsistent with each other, input information that contradicts each other, or input information that is unclear. For example, the description entered by the subject may include a statement that they are concerned about swollen gums, while the dental examination record of the subject obtained as related information may state that the condition of their gums is normal.

[0130] Therefore, the information processing device 1 may perform a process to determine whether the input information is clear, and if it is determined that the input information is unclear, it may perform a process to present questions asking about the unclear points and obtain answers to those questions. This makes it possible to improve the diagnostic accuracy by reflecting the obtained answers in the diagnosis by the diagnostic unit 102, and to improve the estimation accuracy of appropriate OC supplies by reflecting the obtained answers in the estimation by the estimation unit 103.

[0131] The process of determining whether the input information is clear can be performed, for example, using the language model 3. Questions regarding unclear points may be generated using a predetermined template, or they may be generated by the language model 3. The questions can then be presented by the presentation control unit 104.

[0132] For example, the information processing device 1 may include input information (e.g., explanatory text and related information acquired by the acquisition unit 101), determine whether the input information contains inconsistent, contradictory, or unclear items, and generate a prompt instructing the device to output a question to confirm those items if they are present. By inputting this prompt to the language model 3, it is possible to output a question to confirm unclear points in the input information. For example, if the explanatory text entered by the subject contains a description that they are concerned about swollen gums, but the dental check-up record of the subject acquired as related information states that the condition of their gums is normal, it is possible to generate a question such as, "The dental check-up indicated that the condition of your gums is normal, but did the swelling of your gums start to bother you after the dental check-up?" It is also possible to present the generated question to the subject, obtain their answer, and reflect it in the diagnosis by the diagnosis unit 102 or the estimation by the estimation unit 103. For example, by generating a prompt that includes the above question and its answer, and instructing the device to perform a diagnosis or estimation by referring to the question and answer, the above question and answer can be reflected in the diagnosis or estimation. Furthermore, among the matters that are determined to be contradictory, those for which the subject responded that the content was valid (or for which no response indicating that the content was invalid was obtained) may be used as input information, while those for which the subject responded that the content was invalid (or for which no response indicating that the content was valid was obtained) may not be used as input information.

[0133] Furthermore, the input information may be divided into multiple parts, and each divided piece of information may be used by the diagnostic unit 102 to perform a diagnosis. If the diagnostic results do not match, questions may be generated and presented to confirm the diagnostic result. The criteria for dividing the input information can be determined as appropriate. For example, in the above embodiment, the input information is divided into explanatory text and related information. Alternatively, the input information may be divided based on chronological order. In this case, for example, if the input information includes the results of a past dental examination and a recently taken oral image, a diagnosis using the dental examination results and a diagnosis using the oral image will be performed separately. If there is a contradiction between the chronological information, the newer information may be used in chronological order, and the older information may not be used.

[0134] Regarding the "questions to confirm the diagnosis result" mentioned above, for example, suppose the description entered by the subject yields a diagnosis of gum swelling, but the oral image obtained as related information yields a diagnosis of no gum swelling. In this case, the presentation control unit 104 may present the question, "Do you feel pain in your gums when brushing?" This question may be prepared in advance as a standard question to confirm the presence or absence of gum swelling, or it may be generated by the language model 3. The diagnosis unit 102 should then determine the diagnosis result as gum swelling if the subject's answer to the above question is positive, and as gum swelling not present if the answer is negative.

[0135] Similarly, the input information may be divided into multiple parts, and each part of the divided information may be used by the estimation unit 103 to perform estimations. If the estimation results do not match, the subject may be presented with questions to elicit information necessary to narrow down the recommended oral care products. For example, if the description entered by the subject suggests an oral care product effective against staining, but the results of a dental check-up obtained as related information suggest an oral care product effective against gingivitis, the presentation control unit 104 may present a question asking whether the subject prioritizes addressing staining or preventing gingivitis.

[0136] Furthermore, the entity executing each process described in the above-described embodiment is arbitrary and is not limited to the examples given above. In other words, the same functions as the information processing device 1 can be realized by multiple information processing devices that can communicate with each other. For example, each of the processes shown in Figures 6 to 9 may be divided and executed by multiple information processing devices.

[0137] [Reference Example 1] An information processing device according to a reference example of the present invention comprises: an acquisition unit 101 that acquires a descriptive text describing the oral condition of a subject; a question generation control unit 105 that inputs a prompt to a language model 3 trained in natural language to generate a question for inputting further information to estimate oral care products suitable for the oral condition of the subject, based on the descriptive text acquired by the acquisition unit 101, and generates a question; and a presentation control unit 104 that presents the question generated under the control of the question generation control unit 105.

[0138] According to the above configuration, even if the previously obtained explanatory text is insufficient to estimate an appropriate oral hygiene product for the subject's oral condition, it becomes possible to estimate an appropriate oral hygiene product. Therefore, by using the above information processing device, it becomes possible to easily select an oral hygiene product that is suitable for the subject's oral condition. Thus, the information processing device in this reference example makes it possible to improve the technology for presenting oral hygiene products that are suitable for the subject's oral condition.

[0139] Furthermore, the information processing device in this reference example only needs to include at least the acquisition unit 101, the presentation control unit 104, and the question generation control unit 105, which are components of the information processing device 1 shown in Figure 1, and either or both of the diagnosis unit 102 and the estimation unit 103 can be omitted.

[0140] [Reference Example 2] An information processing device according to a reference example of the present invention comprises an acquisition unit 101 that acquires a descriptive text describing the oral condition of a subject, and a diagnosis unit 102 that determines which of the multiple categories the oral condition of a subject falls under based on the output obtained by inputting a prompt, which is generated based on the descriptive text acquired by the acquisition unit 101 and instructs the language model 3 that has been trained in natural language, to determine which of the multiple categories the oral condition of a subject falls under.

[0141] According to the above configuration, the subject's oral condition is determined by having the information processing device acquire a descriptive text indicating the subject's oral condition. Therefore, by using the above information processing device, it becomes possible to easily determine the subject's oral condition. Thus, the information processing device in this reference example makes it possible to improve the technology for presenting oral care products that are suitable for the subject's oral condition.

[0142] Furthermore, the information processing device in this reference example only needs to include at least the acquisition unit 101 and the diagnosis unit 102 among the components of the information processing device 1 shown in Figure 1, and at least one of the estimation unit 103, presentation control unit 104, and question generation control unit 105 can be omitted.

[0143] [Reference Example 3] An information processing device according to one reference example of the present invention comprises: an acquisition unit 101 that acquires a descriptive text describing the oral condition of a subject; a question generation control unit 105 that inputs a prompt to a language model 3 trained in natural language to generate a question for inputting further information to determine which of a predetermined plurality of types the oral condition of the subject falls under, based on the descriptive text acquired by the acquisition unit 101, and generates a question; and a presentation control unit 104 that presents the question generated under the control of the question generation control unit 105.

[0144] According to the above configuration, even if the previously obtained explanatory text is insufficient to determine the type of oral condition of the subject, it becomes possible to appropriately determine the type of oral condition of the subject. Therefore, by using the above information processing device, it becomes possible to easily select oral care products that are suitable for the subject's oral condition. Thus, the information processing device in this reference example makes it possible to improve the technology for presenting oral care products that are suitable for the subject's oral condition.

[0145] Furthermore, the information processing device in this reference example only needs to include at least the acquisition unit 101, the presentation control unit 104, and the question generation control unit 105, which are components of the information processing device 1 shown in Figure 1, and either or both of the diagnosis unit 102 and the estimation unit 103 can be omitted.

[0146] [Reference Example 4] An information processing device according to one reference example of the present invention comprises: an acquisition unit 101 that acquires feedback on OC products recommended to a subject; an estimation unit 103 that inputs a prompt, generated based on the feedback acquired by the acquisition unit 101, to a language model 3 trained in natural language, instructing it to estimate an OC product suitable for the subject's oral condition from among a predetermined number of candidates for OC products, and outputs an estimation result; and a presentation control unit 104 that presents the OC products shown in the estimation result.

[0147] According to the above configuration, it is possible to present new oral hygiene products based on feedback received regarding previously presented oral hygiene products. Therefore, by using the above information processing device, it becomes possible to easily select oral hygiene products that are suitable for the oral condition of the subject. Thus, the information processing device in this reference example makes it possible to improve the technology for presenting oral hygiene products that are suitable for the oral condition of the subject.

[0148] Furthermore, the information processing device in this reference example only needs to include at least the acquisition unit 101, the estimation unit 103, and the presentation control unit 104, which are components of the information processing device 1 shown in Figure 1, and either or both of the diagnostic unit 102 and the question generation control unit 105 can be omitted.

[0149] [Example of implementation by software] The functions of the information processing device 1 are programs that cause the computer to function as the information processing device 1, and these can be implemented by programs (selection support programs) that cause the computer to function as each control block (especially each part included in the control unit 10) of the information processing device 1.

[0150] In this case, the information processing device 1 includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the above program. By executing the above program using this control device and storage device, each of the functions described in the above embodiment is realized.

[0151] The above program may be recorded on one or more computer-readable recording media, rather than temporarily. This recording media may or may not be provided by the information processing device 1. In the latter case, the program may be supplied to the information processing device 1 via any wired or wireless transmission medium.

[0152] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0153] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0154] 1. Information Processing Device 101. Acquisition Unit 102. Diagnosis Unit 103. Estimation Unit 104. Presentation Control Unit 105. Question Generation Control Unit 3. Language Model

Claims

1. An information processing device comprising: an acquisition unit that acquires a descriptive text describing the oral condition of a subject; an estimation unit that inputs a prompt to a language model trained in natural language, which instructs the model to estimate an oral care product suitable for the subject's oral condition from among a predetermined number of candidates for oral care products, based on the descriptive text acquired by the acquisition unit, and outputs an estimation result; and a presentation control unit that presents the oral care product shown in the estimation result.

2. The information processing apparatus according to claim 1, comprising: a question generation control unit that inputs a prompt to the language model instructing it to generate a question for inputting further information to estimate oral care products suitable for the oral condition of the subject, based on the explanatory text acquired by the acquisition unit, and the presentation control unit presents the question generated under the control of the question generation control unit.

3. The information processing apparatus according to claim 1 or 2, comprising a diagnostic unit that determines which of the multiple categories the subject's oral condition falls under based on the output obtained by inputting a prompt to the language model that instructs it to determine which of the multiple categories the subject's oral condition falls under, generated based on the explanatory text acquired by the acquisition unit, and the estimation unit inputs a prompt to the language model that instructs it to estimate an oral care product that fits the category shown in the determination result of the diagnostic unit from among the multiple candidates, and outputs an estimation result.

4. The information processing apparatus according to claim 3, comprising: a question generation control unit that inputs a prompt to the language model instructing it to generate a question for inputting further information to determine which of the plurality of types the oral condition of the subject falls under, based on the explanatory text acquired by the acquisition unit, and the presentation control unit presents the question generated under the control of the question generation control unit.

5. The information processing apparatus according to claim 1 or 2, wherein the acquisition unit acquires feedback on the oral care products presented by the presentation control unit; the estimation unit inputs a new prompt to the language model, which is generated based on the feedback acquired by the acquisition unit and instructs the language model to estimate from the plurality of candidates an oral care product suitable for the oral condition of the subject, and outputs a new estimation result; and the presentation control unit presents the oral care products shown in the new estimation result.

6. The information processing apparatus according to claim 1 or 2, comprising an image analysis unit that generates a descriptive text of the oral condition of the subject from an image showing the oral condition of the subject, and the estimation unit inputs a prompt to the language model that instructs it to estimate an oral care product suitable for the oral condition of the subject from among a plurality of candidates, based on the descriptive text generated by the image analysis unit, and outputs an estimation result.

7. The information processing apparatus according to claim 1 or 2, wherein the acquisition unit acquires the explanatory text and related information relating to the oral condition of the subject, and the estimation unit inputs a prompt to the language model instructing it to estimate from the plurality of candidates an oral care product suitable for the oral condition of the subject, which is generated based on the explanatory text and related information acquired by the acquisition unit, and outputs an estimation result.

8. An information processing device comprising: an acquisition unit that acquires a descriptive text describing the oral condition of a subject; a question generation control unit that inputs a prompt to a language model trained in natural language to generate a question for inputting further information to estimate oral care products suitable for the oral condition of the subject, based on the descriptive text acquired by the acquisition unit, and generates a question; and a presentation control unit that presents the question generated under the control of the question generation control unit.

9. An information processing device comprising: an acquisition unit that acquires a descriptive text describing the oral condition of a subject; and a diagnostic unit that determines which of the multiple categories the oral condition of a subject falls under based on the output obtained by inputting a prompt, generated based on the descriptive text acquired by the acquisition unit and instructing the language model trained in natural language, to determine which of the multiple categories the oral condition of the subject falls under.

10. An information processing device comprising: an acquisition unit that acquires a descriptive text describing the oral condition of a subject; a question generation control unit that inputs a prompt to a language model trained in natural language, instructing it to generate a question for inputting further information to determine which of a predetermined number of categories the oral condition of the subject falls under, based on the descriptive text acquired by the acquisition unit, and generates a question; and a presentation control unit that presents the question generated under the control of the question generation control unit.

11. An information processing device comprising: an acquisition unit for acquiring feedback on oral care products recommended to a subject; an estimation unit for inputting a prompt, generated based on the feedback acquired by the acquisition unit, into a language model trained in natural language, instructing the model to estimate an oral care product suitable for the subject's oral condition from among a predetermined number of candidates, and outputting an estimation result; and a presentation control unit for presenting the oral care products shown in the estimation result.

12. A method for supporting the selection of oral care products, which is performed by one or more information processing devices, comprising: an acquisition step of acquiring a descriptive text indicating the oral condition of a subject; an estimation step of inputting a prompt, generated based on the descriptive text acquired in the acquisition step, to a language model trained in natural language, instructing it to estimate an oral care product suitable for the oral condition of the subject from among a predetermined number of candidates for oral care products, and outputting an estimation result; and a presentation control step of presenting the oral care products indicated in the estimation result.

13. A selection support program for oral care products, which causes a computer to function as an information processing device according to claim 1, wherein the computer functions as the acquisition unit, the estimation unit, and the presentation control unit.

Images