Ai inquiry device, ai inquiry method, and program

Abstract

To provide a AI inquiry device, a AI inquiry method, and a program capable of reducing a burden on a patient during AI inquiry.SOLUTION: The AI inquiry device includes an image inputting part, a disease candidate estimating part, and a disease estimating part. The image input unit receives an input of a face image of a patient. The disease candidate estimation unit extracts symptoms of the patient from the face image of the patient to estimate disease candidates of the patient based on the symptoms of the patient by using the image recognizing AI model. The disease estimation unit uses the interactive AI model to output a question to the subject, receive an input of a response of the subject to the question, and estimate a disease of the subject from candidates for the disease of the subject based on the response of the subject.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to an AI medical interview device, an AI medical interview method, and a program. [Background technology]

[0002] An AI medical interview system has been proposed in which AI (Artificial Intelligence) uses text chat to output questions to a patient to inquire about the patient's symptoms, and conducts an AI medical interview with the patient based on the patient's answers to the questions. The AI ​​medical interview system is disclosed in Patent Document 1, for example. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-027364 Summary of the Invention [Problem to be solved by the invention]

[0004] In the technology disclosed in Patent Document 1, when an AI medical interview is conducted using text chat, the AI ​​picks up too many possible illnesses of the patient, so the number of questions to the patient increases before the patient's illness is estimated. Therefore, the technology disclosed in Patent Document 1 has the problem that the AI ​​medical interview places a heavy burden on the patient.

[0005] By taking such issues into consideration, the present disclosure can provide an AI medical interview device, an AI medical interview method, and a program that can reduce the burden on patients during AI medical interviews. [Means for solving the problem]

[0006] The AI ​​interview device of the present disclosure is an image input unit that accepts input of a patient's face image; a disease candidate estimation unit that extracts symptoms of the patient from a facial image of the patient using an image recognition AI model and estimates disease candidates of the patient based on the symptoms of the patient; The system includes a disease estimation unit that uses an interactive AI model to output questions to the patient, accepts input of the patient's answers to the questions, and estimates the patient's disease from candidate diseases based on the patient's answers.

[0007] The AI ​​interview method disclosed herein includes: The computer Accepts input of a patient's facial image, extracting symptoms of the patient from a facial image of the patient using an image recognition AI model, and predicting a possible disease of the patient based on the symptoms of the patient; Using an interactive AI model, questions are output to the patient, the patient's answers to the questions are input, and the patient's disease is estimated from candidate diseases based on the patient's answers.

[0008] The program of the present disclosure is Accepts input of a patient's facial image, extracting symptoms of the patient from a facial image of the patient using an image recognition AI model, and predicting a possible disease of the patient based on the symptoms of the patient; Using an interactive AI model, a computer is caused to execute a process of outputting questions to the patient, accepting input of the patient's answers to the questions, and inferring the patient's disease from candidate diseases based on the patient's answers. [Effects of the Invention]

[0009] The present disclosure provides an AI medical interview device, an AI medical interview method, and a program that can reduce the burden on patients during AI medical interviews. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 10 is a diagram illustrating an example of a problem of the AI ​​medical interview system R1 according to the comparative example of the present disclosure. [Figure 2] 1 is a block diagram showing an example of the configuration of an AI interview device 10 according to the present disclosure. FIG. [Figure 3] FIG. 1 is a block diagram showing an example of the configuration of an AI inquiry system 2 according to the present disclosure. [Figure 4] 1 is a flowchart showing an example of the operation of the AI ​​inquiry system 2 according to the present disclosure. [Figure 5] FIG. 2 is a schematic diagram showing an example of the operation of the AI ​​inquiry system 2 according to the present disclosure. [Figure 6] FIG. 1 is a block diagram showing an example of the configuration of a computer 1000 according to the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In each drawing, the same or corresponding elements are designated by the same reference numerals, and for clarity of explanation, duplicate explanations will be omitted as necessary.

[0012] Furthermore, "disease" refers to a state in which an abnormality occurs somewhere in the body, impairing health. Examples of diseases are "pharyngoconjunctival fever (swimming pool fever)," "influenza," and "cold." In other words, disease is "illness" or "sickness." Furthermore, "symptoms" refer to abnormal physical or mental conditions that appear due to a disease. For example, symptoms of "pharyngoconjunctival fever" include "fever," "shivering," "bloodshot eyes," and "runny nose."

[0013] (Specific explanation of the problem of the present disclosure) First, in order to specifically explain the problem of the present disclosure, the configuration of an AI medical interview system R1 according to a comparative example of the present disclosure will be described with reference to FIG.

[0014] Figure 1 is a diagram showing an example of the configuration of an AI medical interview system R1 according to a comparative example of the present disclosure. As shown in Figure 1, in the AI ​​medical interview system R1, the AI ​​uses text chat to output questions to the patient inquiring about the patient's symptoms and accepts input of answers to the questions from the patient. The AI ​​conducts an AI medical interview of the patient based on the patient's answers to the questions.

[0015] Specifically, in AI medical interview system R1, the AI ​​outputs question A1 to the patient. In question A1, symptoms can be selected using checkboxes under "What symptoms are bothering you?" The patient checks the symptom "I have a fever." Since the patient is bothered by the symptom "I have a fever," the AI ​​considers that the patient's illness may be "a cold."

[0016] Next, the AI ​​outputs question A2 to the patient. In question A2, symptoms can be selected using checkboxes under "Do you have the following symptoms?" The patient checks the symptoms of "sore throat" and "runny nose." Because the patient also has the symptoms of "sore throat" and "runny nose," the AI ​​considers that the patient's illness may be "influenza."

[0017] Next, the AI ​​outputs question A3 to the patient. In question A3, symptoms can be selected using checkboxes under "Are there any other symptoms that concern you?" The patient checks the symptom "lower back pain." Because the patient also has the symptom "lower back pain," the AI ​​further considers the possibility that the patient's illness is one that is treated by the "orthopedic department." The AI ​​also considers the possibility that the patient's illness may also be "a visceral disorder."

[0018] Next, the AI ​​outputs question A4 to the patient. In question A4, symptoms can be selected using checkboxes under "Are there any other symptoms that concern you?" The patient checks the symptom "loss of appetite." Since the patient also has the symptom "loss of appetite," the AI ​​considers that the patient's illness may be "gastroenteritis."

[0019] As mentioned above, when the AI ​​medical interview system R1 uses text chat to conduct an AI interview, it picks up too many possible illnesses, and the number of questions asked to the patient increases before it can diagnose the patient's illness. As a result, the AI ​​medical interview system R1 had the problem of placing a heavy burden on the patient during the AI ​​interview.

[0020] Therefore, in the following embodiments, an AI interview device 10 (first embodiment) and an AI interview system 2 (second embodiment) that can reduce the burden on patients during AI interviews will be disclosed.

[0021] (First embodiment) First, the configuration of the AI ​​interview device 10 according to the first embodiment will be described with reference to FIG.

[0022] 2 is a block diagram showing an example of the configuration of an AI interview device 10 according to the present disclosure. As shown in FIG. 2, the AI ​​interview device 10 includes an image input unit 11, a candidate disease estimation unit 12, and a disease estimation unit 13.

[0023] The image input unit 11 accepts input of a patient's facial image. The disease candidate estimation unit 12 uses an image recognition AI model to extract the patient's symptoms from the patient's facial image and estimates the patient's disease candidate based on the patient's symptoms. The disease estimation unit 13 uses an interactive AI model to output questions to the patient and accepts input of the patient's answers to the questions. The disease estimation unit 13 estimates the patient's disease from the patient's disease candidate based on the patient's answers.

[0024] As described above, during an AI interview, the AI ​​interview device 10 of the present disclosure narrows down candidate diseases of the patient using an image recognition AI model before inferring the patient's disease from a dialogue (e.g., text chat) with the patient using an interactive AI model. By doing so, the AI ​​interview device 10 can reduce the number of questions asked to the patient during a dialogue with the patient using the interactive AI model. Therefore, the AI ​​interview device 10 can reduce the burden on the patient during an AI interview.

[0025] (Second embodiment) Next, the configuration of an AI medical inquiry system 2 according to the second embodiment will be described with reference to Fig. 3. The AI ​​medical inquiry system 2 is a specific embodiment of the AI ​​medical inquiry device 10 according to the first embodiment.

[0026] 3 is a block diagram showing an example of the configuration of an AI medical interview system 2 according to the present disclosure. As shown in FIG. 3, the AI ​​medical interview system 2 includes an AI medical interview device 20 and a patient terminal 30.

[0027] The AI ​​interview device 20 is a server such as a cloud server or an on-premise server. The patient terminal 30 is a device used by the patient, such as a PC (Personal Computer), smartphone, or tablet. The patient terminal 30 is installed, for example, in a hospital waiting room or the patient's home, or carried by the patient himself / herself. The patient terminal 30 connects to a network and communicates with the AI ​​interview device 20. The communication method can be wireless or wired. The network can include, for example, a LAN (Local Area Network), a WAN (Wide Area Network), the Internet, etc.

[0028] Specifically, the AI ​​medical interview device 20 includes an image input unit 21, a disease candidate estimation unit 22, a disease estimation unit 23, a guidance unit 24, an image recognition AI model storage unit 25, and an interactive AI model storage unit 26.

[0029] The image input unit 21 receives an input of a facial image of a patient from the patient terminal 30. A facial image is an image including a face. The image is a still image. Note that the image may be a plurality of images in time series (i.e., a video).

[0030] The disease candidate estimation unit 22 extracts the patient's symptoms from the patient's facial image using the image recognition AI model stored in the image recognition AI model storage unit 25. The extracted patient's symptoms are features for estimating disease. The extracted patient's symptoms include at least one of symptoms related to the patient's facial expression and symptoms related to the condition of the organs that make up the patient's face. Symptoms related to the patient's facial expression include, for example, "having a fever," "shivering," and "feeling fatigued." Symptoms related to the condition of the organs that make up the patient's face include, for example, "bloodshot eyes," "runny nose," "swollen lips," and "skin rash."

[0031] The image input unit 21 may receive not only a facial image of the patient but also a body image including a body part of the patient other than the face (for example, an arm, an abdomen, a leg, etc.) from the patient terminal 30. The disease candidate estimation unit 22 may extract the patient's symptoms from the body image of the patient using an image recognition AI model.

[0032] Then, the disease candidate estimation unit 22 estimates a disease candidate of the patient based on the extracted symptoms of the patient using an image recognition AI model. The image recognition AI model is trained using a dataset of data (i.e., training data) in which facial images of patients with a predetermined symptom are labeled with the disease of the patient in which the symptom appears. Note that the image recognition AI model may also be trained using a dataset of data in which body images of patients with a predetermined symptom are labeled with the disease of the patient in which the symptom appears.

[0033] The disease estimation unit 23 outputs a question to the patient terminal 30 using the interactive AI model stored in the interactive AI model storage unit 26. The question is output in text format. The question is related to the patient's candidate disease. Specifically, the question asks the patient whether any symptoms that appear in the candidate disease estimated by the candidate disease estimation unit 22 and that differ from the symptoms extracted from the patient's facial image are among the symptoms the patient is suffering from. The question may be in the form of, for example, a descriptive format in which the patient answers freely in writing, a multiple-choice format in which the patient selects one or more from multiple options, a checkbox format in which the patient selects one or more from multiple options, or a pull-down format in which the patient selects from a drop-down menu. The question may also be output as audio.

[0034] The disease estimation unit 23 then receives input of answers to the questions from the patient terminal 30. Answers are received as text input. Alternatively, answers may be received as voice input. The disease estimation unit 23 uses an interactive AI model to estimate the patient's disease from candidate diseases based on the patient's answers.

[0035] The guidance unit 24 outputs guidance based on the estimated disease of the patient to the patient terminal 30. The guidance is guidance that indicates what measures the patient should take in the future for the estimated disease of the patient.

[0036] The image recognition AI model storage unit 25 stores an image recognition AI model. The image recognition AI model is a model that recognizes objects and features from an image. The image recognition AI model is a deep learning model. The image recognition AI model is, for example, a CNN (Convolutional Neural Network). When dealing with time-series images, the image recognition AI model may be a combination of a CNN and an RNN (Recurrent Neural Network).

[0037] The conversational AI model storage unit 26 stores a conversational AI model. The conversational AI model is a model that can hold natural conversations like a human. The conversational AI model can accept text or voice input and provide an appropriate response according to the input. The conversational AI model is a model that combines technologies such as NLP (Natural Language Processing), machine learning, and deep learning.

[0038] Specifically, the patient terminal 30 includes an imaging unit 31, an input unit 32, and an output unit 33. The photographing unit 31 is, for example, a camera. The photographing unit 31 photographs an image of the patient's face. The photographing unit 31 transmits the photographed image of the patient's face to the AI ​​interview device 20. The input unit 32 is, for example, a touch panel. The input unit 32 accepts information input from the patient. The input unit 32 transmits the information accepted as input from the patient to the AI ​​interview device 20. The output unit 33 is, for example, a display or a speaker. The output unit 33 receives information from the AI ​​interview device 20. The output unit 33 outputs the received information to the patient.

[0039] Next, the operation of the AI ​​inquiry system 2 according to the second embodiment will be described with reference to FIGS.

[0040] Fig. 4 is a flowchart showing an example of the operation of the AI ​​medical inquiry system 2 according to the present disclosure. Also, Fig. 5 is a schematic diagram showing an example of the operation of the AI ​​medical inquiry system 2 according to the present disclosure.

[0041] First, the patient takes a picture of his or her face using the photographing unit 31 of the patient terminal 30. The patient terminal 30 then transmits the patient's face image to the AI ​​medical interview device 20.

[0042] As shown in FIGS. 4 and 5, in step S101, the image input unit 21 of the AI ​​medical interview device 20 receives a facial image of the patient from the patient terminal 30.

[0043] Next, in step S102, the disease candidate prediction unit 22 extracts the patient's symptoms from the patient's face image using the image recognition AI model stored in the image recognition AI model storage unit 25. For example, the extracted patient's symptoms are "fever," "bloodshot eyes," "runny nose," and "shivering."

[0044] Next, in step S103, the disease candidate estimation unit 22 estimates a disease candidate of the patient from the symptoms of the patient using an image recognition AI model. For example, the disease candidate estimated from the symptoms of the patient is "pharyngoconjunctival fever (swimming pool fever)."

[0045] After the process of step S103, the image input unit 21 and the disease candidate estimation unit 22 may repeat a series of processes (the processes of steps S101 to S103) for estimating a patient's disease candidate from the patient's symptoms using an image recognition AI model. In this case, after the process of step S103, in the process of step S101, the image input unit 21 may receive from the patient terminal 30 a face image of the patient captured from a different angle than the previous face image of the patient. This allows the AI ​​interview system 2 to improve the accuracy of estimating a patient's disease candidate. After the process of step S103, in the process of step S101, the image input unit 21 may receive from the patient terminal 30 not only a face image of the patient, but also a body image including a body part of the patient other than the face (e.g., an arm, abdomen, or leg), etc. Next, in step S102, the disease candidate estimation unit 22 may extract the patient's symptoms from the body image of the patient using an image recognition AI model. This allows the AI ​​interview system 2 to improve the accuracy of estimating a patient's disease candidate.

[0046] Next, in step S104, the disease estimation unit 23 uses the interactive AI model stored in the interactive AI model storage unit 26 to send a question related to the patient's candidate disease to the patient terminal 30.

[0047] After step S104, the patient terminal 30 outputs the question received from the AI ​​interview device 20 to the output unit 33 of the patient terminal 30. The patient then inputs an answer to the question into the input unit 32 of the patient terminal 30. For example, in the question related to the patient's possible illness that is output, under "Do you have the following symptoms?", the symptom "itchy eyes" is selectable using a check box. The patient checks the symptom "itchy eyes." The input unit 32 of the patient terminal 30 then accepts input of an answer to the question and transmits the answer to the question to the AI ​​interview device 20.

[0048] Next, in step S105, the disease inferring unit 23 receives the answer to the question from the patient terminal 30. Next, in step S106, the disease estimation unit 23 uses the interactive AI model to estimate the patient's disease from the candidate diseases based on the answers to the questions. For example, the symptom of "itchy eyes" included in the patient's answer among the candidate diseases is determined to be a symptom that appears in the disease "pharyngoconjunctival fever." Therefore, it is estimated that the patient's disease is highly likely to be "pharyngoconjunctival fever."

[0049] The disease estimation unit 23 may repeat a series of processes (the processes of steps S104 to S106) for estimating the patient's disease from candidate diseases of the patient using an interactive AI model.

[0050] In step S107, the guidance unit 24 outputs guidance based on the estimated disease of the patient to the patient terminal 30. For example, the disease estimation unit 23 outputs guidance such as "There is a possibility of pharyngoconjunctival fever. As there is a risk of infection, we will guide you to a separate room" to the patient terminal 30.

[0051] As described above, during an AI interview, the AI ​​interview system 2 of the present disclosure narrows down candidate diseases of a patient using an image recognition AI model before inferring the patient's disease from a dialogue (e.g., text chat) with the patient using an interactive AI model. By doing so, the AI ​​interview system 2 can reduce the number of questions asked to the patient during a dialogue with the patient using the interactive AI model. Therefore, the AI ​​interview system 2 can reduce the burden on the patient during the AI ​​interview.

[0052] Furthermore, the AI ​​medical interview system 2 uses the interactive AI model to output questions to the patient that relate to the patient's disease candidates narrowed down using the image recognition AI model. By doing so, the AI ​​medical interview system 2 can further reduce the number of questions asked to the patient when interacting with the patient using the interactive AI model. As a result, the AI ​​medical interview system 2 can further reduce the burden on the patient during the AI ​​medical interview.

[0053] (A variant of AI medical interview system 2) Next, a modified example of the AI ​​medical inquiry system 2 according to the second embodiment will be described. The AI ​​interview device 20 of the AI ​​interview system 2 according to the modified example may include a patient information storage unit (not shown) and an authentication unit (not shown). The patient information storage unit stores information linking each patient's facial image with a patient ID (Identification) corresponding to the patient's facial image. The authentication unit compares the patient's facial image stored in the patient information storage unit with the patient's facial image input by the image input unit 21, and if the comparison is successful, acquires the patient ID corresponding to the patient's facial image input. Thereafter, for example, the disease estimation unit 23 associates the patient ID with the estimated patient's symptoms and stores them.

[0054] Furthermore, the system configuration of the AI ​​interview system 2 is not limited to the example of the system configuration shown in Fig. 3. For example, in the AI ​​interview system 2, the AI ​​interview device 20 may include some or all of the components of the patient terminal 30. Furthermore, the patient terminal 30 may include some or all of the components of the AI ​​interview device 20. Furthermore, at least one of the AI ​​interview device 20 and the patient terminal 30 may be realized by multiple devices.

[0055] <Hardware configuration> FIG. 6 is a block diagram showing an example of the configuration of a computer 1000 according to this embodiment. The AI ​​interview device 10 and each device (AI interview device 20 and patient terminal 30) of the AI ​​interview system 2 described above are realized by the computer 1000. As shown in FIG. 6, the computer 1000 includes a processor 1001, a memory 1002, and a network interface 1003. The network interface 1003 may be used to communicate with a network node. The network interface 1003 may include, for example, a network interface card (NIC) compliant with the IEEE 802.3 series. IEEE stands for Institute of Electrical and Electronics Engineers.

[0056] The processor 1001 reads one or more programs containing instructions for causing a computer to execute the algorithm (AI questioning method) described with reference to the drawings from the memory 1002, and executes the read programs. The processor 1001 may be, for example, a microprocessor, an MPU, or a CPU. The processor 1001 may include multiple processors.

[0057] The memory 1002 is configured by a combination of volatile memory and non-volatile memory. The memory 1002 stores programs. The memory 1002 may include storage located away from the processor 1001. In this case, the processor 1001 may access the memory 1002 via an I / O (Input / Output) interface (not shown).

[0058] In the above examples, the program includes instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more functions described in the embodiments. The program may be stored on a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disc (DVD), Blu-ray® disc or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may also be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals.

[0059] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate.

[0060] Each drawing is merely an example for describing one or more embodiments. Each drawing may relate not only to one possible embodiment, but also to one or more other embodiments. As will be understood by those skilled in the art, various features or steps described with reference to any one drawing can be combined with features or steps shown in one or more other drawings to create, for example, an embodiment not explicitly shown or described. Not all features or steps shown in any one drawing are necessary to describe an exemplary embodiment, and some features or steps may be omitted. The order of steps described in any drawing may be changed as appropriate.

[0061] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes. (Appendix 1) an image input unit that accepts input of a patient's face image; a disease candidate estimation unit that extracts symptoms of the patient from a facial image of the patient using an image recognition AI model and estimates disease candidates of the patient based on the symptoms of the patient; a disease estimation unit that uses an interactive AI model to output questions to the patient, receives input of the patient's answers to the questions, and estimates the patient's disease from candidate diseases based on the patient's answers. AI interview device. (Appendix 2) The question is, and questions relating to the patient's suspected disease. The AI ​​interview device described in Appendix 1. (Appendix 3) The question is, and a question to ask the patient whether any of the symptoms the patient is suffering from are symptoms that appear in the disease candidates and are different from the symptoms extracted from the face image of the patient. The AI ​​interview device described in Appendix 2. (Appendix 4) The patient's symptoms include: The symptom includes at least one of a symptom related to the facial expression of the patient and a symptom related to the condition of the organs that constitute the face of the patient. The AI ​​interview device described in Appendix 1. (Appendix 5) a guidance unit that outputs guidance corresponding to the patient's disease to the patient; The AI ​​interview device described in Appendix 1. (Appendix 6) The image recognition AI model is It is trained by a dataset of facial images of patients with a given symptom and labeled data for the disease in which that symptom appears. The AI ​​interview device described in Appendix 1. (Appendix 7) The computer Accepts input of a patient's facial image, extracting symptoms of the patient from a facial image of the patient using an image recognition AI model, and predicting a possible disease of the patient based on the symptoms of the patient; Using an interactive AI model, a question is output to the patient, an answer to the question is received from the patient, and a disease of the patient is estimated from candidate diseases of the patient based on the answer of the patient. AI interview method. (Appendix 8) The question is, and questions relating to the patient's suspected disease. The AI ​​interview method described in Appendix 7. (Appendix 9) Accepts input of a patient's facial image, extracting symptoms of the patient from a facial image of the patient using an image recognition AI model, and predicting a possible disease of the patient based on the symptoms of the patient; Using an interactive AI model, a computer is caused to execute a process of outputting questions to the patient, receiving input of the patient's answers to the questions, and estimating the patient's disease from candidate diseases of the patient based on the patient's answers. program. (Appendix 10) The question is, and questions relating to the patient's suspected disease. 10. The program described in Appendix 9.

[0062] Some or all of the elements (e.g., configurations and functions) described in Supplementary Note 3 to Supplementary Note 6 that are dependent on Supplementary Note 1 {e.g., device} may also be dependent on Supplementary Note 7 {e.g., method} and Supplementary Note 9 {e.g., program} in the same dependency relationship as Supplementary Note 3 to Supplementary Note 6. Some or all of the elements described in any Supplementary Note may be applied to various hardware, software, recording means for recording software, systems, and methods. [Explanation of symbols]

[0063] 2. AI medical interview system 10 AI interview device 11 Image input unit 12. Disease candidate prediction unit 13 Disease Prediction Department 20 AI interview device 21 Image input unit 22 Disease candidate prediction unit 23 Disease Prediction Department 24 Information Department 25 Image recognition AI model memory section 26 Interactive AI model memory section 30 Patient terminals 31 Photography Department 32 Input section 33 Output section 1000 computers 1001 processor 1002 memory 1003 Network Interface

Claims

1. an image input unit that accepts input of a patient's face image; a disease candidate estimation unit that extracts symptoms of the patient from a facial image of the patient using an image recognition AI model and estimates disease candidates of the patient based on the symptoms of the patient; a disease estimation unit that uses an interactive AI model to output questions to the patient, receives input of the patient's answers to the questions, and estimates the patient's disease from candidate diseases based on the patient's answers. AI interview device.

2. The question is, and questions relating to the patient's suspected disease. The AI ​​inquiry device according to claim 1.

3. The question is, and a question to ask the patient whether any of the symptoms the patient is suffering from are symptoms that appear in the disease candidates and are different from the symptoms extracted from the face image of the patient. The AI ​​inquiry device according to claim 2.

4. The patient's symptoms include: The symptom includes at least one of a symptom related to the facial expression of the patient and a symptom related to the condition of the organs that constitute the face of the patient. The AI ​​inquiry device according to claim 1.

5. a guidance unit that outputs guidance corresponding to the patient's disease to the patient; The AI ​​inquiry device according to claim 1.

6. The image recognition AI model is It is trained by a dataset of facial images of patients with a given symptom and labeled data for the disease in which that symptom appears. The AI ​​inquiry device according to claim 1.

7. The computer Accepts input of a patient's facial image, Using an image recognition AI model, extracting symptoms of the patient from a facial image of the patient, and estimating a possible disease of the patient based on the symptoms of the patient; Using an interactive AI model, a question is output to the patient, an answer to the question is received from the patient, and a disease of the patient is estimated from candidate diseases of the patient based on the answer of the patient. AI interview method.

8. The question is, and questions relating to the patient's suspected disease. The AI ​​interview method according to claim 7.

9. Accepts input of a patient's facial image, Using an image recognition AI model, extracting symptoms of the patient from a facial image of the patient, and estimating a possible disease of the patient based on the symptoms of the patient; Using an interactive AI model, a computer is caused to execute a process of outputting questions to the patient, receiving input of the patient's answers to the questions, and estimating the patient's disease from candidate diseases based on the patient's answers. program.

10. The question is, and questions relating to the patient's suspected disease. The program according to claim 9.

Images