Conversation system, conversation control device, conversation control method, and program

Abstract

The present invention provides a conversation system, conversation control device, conversation control method, and program that reduce the time and mental burden on caregivers in facilities and homes by dealing with symptoms specific to people with dementia, such as repetitive questions and desires to return home, and that enable people with dementia to converse more easily. [Solution] A conversation system comprising: an audio input / output device having a shape similar to that of a telephone handset and equipped with at least an audio input / output unit capable of inputting and outputting audio; and a conversation control device that generates response content suitable for conversation with a person with dementia based on audio representing the content of speech of a person with dementia input from the audio input / output device, and outputs audio representing the response content from the audio input / output device.

Description

Technical Field

[0001] The present invention relates to a conversation system, a conversation control device, a conversation control method, and a program.

Background Art

[0002] Conventionally, in facilities and homes, in communication with dementia patients, there has been a problem that dealing with specific symptoms such as repeated questions and the desire to return home places a great mental and time burden on caregivers.

[0003] In recent years, due to the development of AI (Artificial Intelligence) technology, various technologies that enable conversation between humans and machines have been proposed. For example, Patent Document 1 below discloses a technology that enables conversation between a human and AI via a user terminal such as a smartphone, a tablet terminal, or a personal computer.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, when using terminals such as smartphones, tablet terminals, or personal computers as in the technology described in Patent Document 1 above, many dementia patients cannot understand where the voice of the AI output from these terminals is coming from and whether it is the timing to start a conversation in the first place. Therefore, when using these terminals, caregiver assistance is required, and thus the time and mental burden on caregivers remains unchanged in dealing with the specific symptoms of dementia patients and background factor hearing.

[0006] In view of the above-mentioned problems, the object of the present invention is to reduce the time and mental burden on caregivers in facilities and homes caused by responding to symptoms specific to people with dementia, such as repetitive questions and desires to return home, and to provide a conversation system, conversation control device, conversation control method, and program that enable people with dementia to converse more easily. [Means for solving the problem]

[0007] To solve the above-mentioned problems, a conversation system according to one aspect of the present invention is a conversation system comprising: an audio input / output device having a shape similar to that of a telephone handset and equipped with at least an audio input / output unit capable of inputting and outputting audio; and a conversation control device that generates response content suitable for conversation with a person with dementia based on audio representing the content of speech of a person with dementia input from the audio input / output device, and outputs audio representing the response content from the audio input / output device.

[0008] A conversation control device according to one aspect of the present invention is a conversation control device comprising: a voice recognition unit that recognizes voice representing the content of speech of a person with dementia input from a voice input / output device having a shape similar to that of a telephone handset and equipped with at least a voice input / output unit capable of inputting and outputting voice; a response generation unit that generates response content suitable for conversation with the person with dementia based on the recognized voice; a voice synthesis unit that converts the generated response content into voice; and an output control unit that causes the converted voice to be output from the voice input / output device.

[0009] A conversation control method according to one aspect of the present invention is a conversation control method performed by a computer, which includes: a speech recognition process for recognizing speech representing the content of speech spoken by a person with dementia, input from a speech input / output device having at least a speech input / output unit having a shape similar to that of a telephone handset and capable of inputting and outputting speech; a response generation process for generating response content suitable for conversation with the person with dementia based on the recognized speech; a speech synthesis process for converting the generated response content into speech; and an output control process for outputting the converted speech from the speech input / output device.

[0010] A program according to one aspect of the present invention is a program for causing a computer to function as: a speech recognition means for recognizing speech representing the content of speech spoken by a person with dementia, input from a speech input / output device having a shape similar to that of a telephone handset and equipped with at least a speech input / output unit capable of inputting and outputting speech; a response generation means for generating response content suitable for conversation with the person with dementia based on the recognized speech; a speech synthesis means for converting the generated response content into speech; and an output control means for outputting the converted speech from the speech input / output device. [Effects of the Invention]

[0011] Firstly, the present invention can reduce the burden on caregivers in responding to repetitive questions and desires to return home from people with dementia. Secondly, it can reduce the workload on caregivers (staff and personnel) in background factor interviews for BPSD assessment work, and collect information with a certain level of accuracy. Therefore, the present invention can reduce the time and mental burden on caregivers in facilities and homes from responding to symptoms specific to people with dementia, such as repetitive questions and desires to return home, and can enable people with dementia to converse more easily. [Brief explanation of the drawing]

[0012] [Figure 1] This block diagram shows an example of the configuration of the conversation system according to this embodiment. [Figure 2] This is a block diagram showing an example of the functional configuration of the conversation control device according to this embodiment. [Figure 3] This flowchart shows an example of the pre-configuration process according to this embodiment. [Figure 4] This flowchart shows an example of the flow of conversation control processing according to this embodiment. [Figure 5] This flowchart shows an example of the report generation process flow according to this embodiment. [Modes for carrying out the invention]

[0013] Embodiments of the present invention will be described in detail below with reference to the drawings.

[0014] <1. Conversation System Configuration> The configuration of the conversation system according to this embodiment will be described with reference to Figure 1. Figure 1 is a block diagram showing an example of the configuration of the conversation system according to this embodiment.

[0015] The conversation system 1 shown in Figure 1 is a system in which AI (Artificial Intelligence) converses on behalf of a human to address various problems encountered in caregiving for people with dementia, both in facilities and at home, through communication with caregivers and family members, particularly symptoms specific to people with dementia such as repetitive questions and desires to return home. Furthermore, in facilities, conversation system 1 enables efficient and accurate interviews to understand the background factors of BPSD (Behavioral and Psychological Symptoms of Dementia). The target people with dementia are not particularly limited; they include those living in facilities, those living at home, those using facilities to receive services such as day care, day rehabilitation, and short-term stays, and those not using such facilities.

[0016] Conversation System 1 reduces the mental and time burden on caregivers by supporting communication with people with dementia, and also improves the efficiency and accuracy of assessment work. This system reduces the burden on caregivers in facilities and homes by having AI handle responses to symptoms specific to people with dementia, such as repetitive questions and desires to return home. Furthermore, by having AI efficiently and accurately conduct interviews about the background factors of BPSD (Behavioral and Psychological Symptoms of Dementia) in facilities, the burden of information gathering by staff is reduced, improving the quality of care plans. It is expected that by reducing the burden on caregivers through this system, an environment will be created where they can focus on providing more appropriate care.

[0017] As shown in FIG. 1, the conversation system 1 includes a conversation control device 10, a user terminal 20, and a voice input / output device 30. For the configuration for information transfer in the network NW, for example, LAN (Local Area Network), WAN (Wide Area Network), telephone network (mobile phone network, fixed telephone network, etc.), regional IP (Internet Protocol) network, Internet, etc. are applicable.

[0018] (1) Conversation control device 10 The conversation control device 10 is a device for controlling the conversation between a person with dementia and AI. The conversation control device 10 is constituted by, for example, one or more PCs (Personal Computers) or server devices (for example, cloud servers). The conversation control device 10 is connected to the user terminal 20 via the network NW so as to be able to transmit and receive various information.

[0019] Based on the voice indicating the speech content of the person with dementia input from the voice input / output device 30 (hereinafter also referred to as "speech voice"), the conversation control device 10 generates a response content suitable for the conversation with the person with dementia, and outputs the voice indicating the response content (hereinafter also referred to as "response voice") from the voice input / output device 30. The speech voice input to the voice input / output device 30 is input to the conversation control device 10 as its voice data. Also, the response voice generated by the conversation control device 10 is output from the voice input / output device 30 when the voice data is input to the voice input / output device 30.

[0020] (2) User terminal 20 The user terminal 20 is a terminal used by the user. The user includes, for example, dementia patients and caregivers. Caregivers include, for example, facility staff and the families of dementia patients. The user terminal 20 is a terminal such as a smartphone, a tablet terminal, or a PC. The user terminal 20 is connected to the conversation control device 10 via the network NW so as to be able to transmit and receive various information. Also, the user terminal 20 is connected to the voice input / output device 30 so as to be able to transmit and receive voice data by a wired connection or a wireless connection. The wireless connection is, for example, short-range wireless communication such as Bluetooth (registered trademark) or Wi-Fi (registered trademark).

[0021] When a dementia patient and a caregiver are in the same space and the dementia patient uses the conversation system 1 with the assistance of the caregiver, at least one user terminal 20 is sufficient. For example, when a dementia patient uses the conversation system 1 with the assistance of facility staff at a facility or when a dementia patient uses the conversation system 1 with the assistance of family at home, at least one of the terminals owned by the facility or the family or the terminal owned by the dementia patient can be used as the user terminal 20. On the other hand, when a dementia patient and a caregiver are in different spaces and the dementia patient uses the conversation system 1 with the assistance of the caregiver, a plurality of user terminals 20 may be provided. This is mainly the case when a dementia patient and a caregiver live separately in a home care environment. In this case, the caregiver uses the user terminal 20 owned by the caregiver to remotely operate the conversation system 1, thereby performing various settings so that the dementia patient can use the conversation system 1. As a result, the caregiver can support the communication between the dementia patient and the AI from a remote location. Also, the dementia patient can use the conversation system 1 alone at home using the user terminal 20 owned by the dementia patient.

[0022] (3) Voice input / output device 30 The voice input / output device 30 is a device for receiving speech input from a person with dementia and outputting AI response speech. The voice input / output device 30 is connected to the user terminal 20 by a wired or wireless connection, enabling it to send and receive voice data. The wireless connection is, for example, a short-range wireless communication such as Bluetooth® or Wi-Fi®. As shown in Figure 1, the audio input / output device 30 comprises an audio input / output unit 31, a main unit 32, and a shooting unit 33.

[0023] (3-1) Audio input / output unit 31 The audio input / output unit 31 has a shape similar to that of a telephone handset and is a device capable of inputting and outputting audio. The audio input / output unit 31 has a structure in which the transmitter (microphone) and receiver (speaker) are integrated, similar to that of a telephone handset, and is designed to be intuitively operated by people with dementia. The voice input / output unit 31 has the function of inputting the voice of a person with dementia and outputting an AI response. This allows the person with dementia to input their voice through the microphone by speaking into the microphone part of the voice input / output unit 31. In addition, the person with dementia can hear the AI ​​response voice through the speaker by placing their ear to the receiver part of the voice input / output unit 31.

[0024] (3-2) Main body 32 The main unit 32 is a housing that has the same shape as the main body of a telephone. The telephone is, for example, a rotary dial telephone. Many people with dementia are elderly and are not familiar with smartphones, tablets, PCs, etc., and sometimes they were unable to recognize the AI ​​voice output from these devices, making conversation impossible. In contrast, rotary dial telephones are familiar to the elderly generation, so many of them were able to recognize the AI ​​voice output from the handset and participate in conversations without confusion.

[0025] (3-3) Photography Department 33 The camera unit 33 is equipped with a camera for capturing the facial expressions and movements of people with dementia and is integrated into the main unit 32 of the audio input / output device 30. The camera unit 33 is connected to the conversation control device 10 via a network NW and can transmit video data. The camera unit 33 can also be used to understand the situation of people with dementia as auxiliary information when the AI ​​generates response content. As a result, the conversation system 1 can respond flexibly to conversations with people with dementia based on information including not only audio but also video.

[0026] <2. Functional Configuration of the Conversation Control Device> The configuration of the conversation system 1 according to this embodiment has been described above. Next, the functional configuration of the conversation control device 10 according to this embodiment will be described with reference to Figure 2. Figure 2 is a block diagram showing an example of the functional configuration of the conversation control device 10 according to this embodiment. As shown in Figure 1, the conversation control device 10 comprises a communication unit 110, a storage unit 120, and a control unit 130.

[0027] (1) Communications Department 110 The communication unit 110 has the function of sending and receiving various types of information. The communication unit 110 is connected to the user terminal 20 via a network NW and sends and receives various types of information.

[0028] (2) Storage section 120 The memory unit 120 has the function of storing various types of information. The memory unit 120 is composed of storage media provided as hardware by the conversation control device 10, such as an HDD (Hard Disk Drive), SSD (Solid State Drive), flash memory, EEPROM (Electrically Erasable Programmable Read Only Memory), RAM (Random Access Read / Write Memory), ROM (Read Only Memory), or any combination of these storage media. The memory unit 120 may also be capable of saving or synchronizing information to a cloud environment connected via a network NW, and by using this cloud environment, information can be shared among multiple conversation control devices 10.

[0029] (3) Control unit 130 The control unit 130 has the function of controlling the overall operation of the conversation control device 10. The control unit 130 is implemented, for example, by causing the CPU (Central Processing Unit) or GPU (Graphics Processing Unit) provided as hardware in the conversation control device 10 to execute a program. As shown in Figure 2, the control unit 130 includes a setting processing unit 131, an audio processing unit 132, an output control unit 133, a detection unit 134, a notification unit 135, a storage processing unit 136, a conversation classification unit 137, and a report generation unit 138.

[0030] (3-1) Setting processing unit 131 The configuration processing unit 131 has the function of configuring the AI. For example, a generative AI capable of generating text is used as the AI. Below, as an example, an example in which a Large Language Model (LLM) is used as the AI ​​model for the generative AI is described. Note that the AI ​​model is not limited to a large language model; any AI model capable of generating text may be used. The configuration processing unit 131 configures the AI ​​based on configuration information entered by the user. The user who performs this configuration is basically a caregiver, but a person with dementia may also configure it themselves. The configuration information includes, for example, conceptual information, event-specific response information, and individual configuration information. The configuration processing unit 131 applies the entered configuration information to the AI ​​so that the large language model can generate appropriate response content based on that information.

[0031] Conceptual information refers to information that represents the concepts of conversational skills to be set in a large-scale language model. Conceptual information also refers to information that represents the concepts of validation methods, which are communication techniques for people with dementia, and information that represents the concepts of behavioral and psychological symptoms of dementia.

[0032] The configuration processing unit 131 can apply information illustrating the concept of validation methods to the AI, thereby reflecting conversational skills for dementia in a large-scale language model. These conversational skills for dementia, for example, enable conversations that do not cause stress to people with dementia.

[0033] Among the validation methods, this invention includes conversational skills related to conversation, such as rephrasing (repeating the words of the person with dementia to show empathy) and reminiscing (encouraging them to share nostalgic memories), with the aim of not causing stress to the person with dementia during the conversation.

[0034] Furthermore, the configuration processing unit 131 can apply information representing concepts for responding to behavioral and psychological symptoms of dementia (BPSD) to the AI, thereby reflecting dementia-specific conversation skills in a large-scale language model. In addition, conversation skills that enable background factor interviews for BPSD assessment can be reflected in the large-scale language model. With conversation skills that enable background factor interviews for BPSD assessment, for example, it becomes possible to ask questions about the lifestyle and hobbies of people with dementia, which are background factors for BPSD.

[0035] Event-specific response information is set up to appropriately respond to unexpected events that may occur during conversations with people with dementia. This information is categorized based on, for example, physiological needs, psychological factors, and communication considerations, and appropriate response methods are set for each situation. For example, for physiological needs, in response to the event "I need to go to the toilet," the response is set to "Pause the conversation and encourage them to go to the toilet." For psychological factors, in response to the event "I can't follow the topic," the response is set to "Change to a simple and specific topic." For communication considerations, in response to the event "I forget what was said," the response is set to "Gently summarize and help them remember."

[0036] Individual settings information is used to customize the AI ​​for each individual with dementia. This includes, for example, setting the conversation format according to the progression of the dementia in the individual (e.g., question format or everyday conversation format), setting the response policy for specific problematic behaviors in communication with the individual with dementia (specifically, "event: repeatedly expressing a desire to go home, response: not directly responding to the desire to go home, but stimulating memory to shift the anxiety to another topic"), and setting BPSD+Q / BPSD25Q.

[0037] The BPSD+Q / BPSD25Q is a 25-item questionnaire used to assess BPSD (Behavioral and Psychological Symptoms of Dementia), and is an assessment index developed to quantitatively evaluate the behavioral and psychological symptoms of people with dementia.

[0038] (3-2) Audio processing unit 132 The voice processing unit 132 has the function of performing processing to generate the AI's response voice. As shown in Figure 2, the voice processing unit 132 comprises a voice recognition unit 1321, a response generation unit 1322, and a voice synthesis unit 1323.

[0039] (3-2-1) Speech recognition unit 1321 The speech recognition unit 1321 has the function of performing speech recognition processing. When speech is input from the speech input / output unit 31 by a person with dementia, the speech recognition unit 1321 acquires the speech data of the speech received by the communication unit 110 from the speech input / output device 30 and performs speech recognition processing on the said speech data. The speech recognition unit 1321 generates text data representing the speech content from the speech data by converting the speech representing the recognized speech content into text. The speech recognition unit 1321 outputs the generated text data to the storage processing unit 136.

[0040] (3-2-2) Response generation unit 1322 The response generation unit 1322 has the function of generating response content to the speech of a person with dementia using a large-scale language model. When text data representing the speech content recognized by the speech recognition unit 1321 is generated, the response generation unit 1322 performs a response generation process on the text data. Through the response generation process, the response generation unit 1322 generates text data representing the response content from the text data representing the speech content.

[0041] The response generation unit 1322 generates response content suitable for conversation with a person with dementia using a large-scale language model, based on pre-set conceptual information about dementia. For example, if the conceptual information is information representing the concept of validation methods, the response generation unit 1322 generates response content that enables conversation without causing stress to the person with dementia, using conversation skills tailored for people with dementia.

[0042] The response generation unit 1322 generates response content tailored to each person with dementia using a large-scale language model, based on individual setting information that is pre-configured for each person with dementia. For example, the response generation unit 1322 generates response content tailored to the progression of dementia of each person with dementia, based on individual setting information that indicates the progression of dementia, which is pre-configured for each person with dementia. Furthermore, the response generation unit 1322 generates response content corresponding to specific problem behaviors using a large-scale language model, based on individual setting information that indicates a response policy for specific problem behaviors in communication with a person with dementia. For example, if an event indicated by a response policy occurs during a conversation with a person with dementia, the response generation unit 1322 generates response content to carry out the response content defined for that event.

[0043] The response generation unit 1322 generates response content corresponding to an event using a large-scale language model based on pre-configured event-specific response information. For example, if an event specified in the event-specific response information occurs during a conversation with a dementia patient, the response generation unit 1322 generates response content to perform the response content specified for that event.

[0044] The response generation unit 1322 generates response content corresponding to the state of the person with dementia as shown in the video data, based on the video data acquired by the shooting unit 33, using a large-scale language model. For example, the response generation unit 1322 grasps the emotions of the person with dementia in real time and generates response content corresponding to the emotions of the person with dementia. The emotions of the person with dementia can be identified by, for example, recognizing the person's facial expressions and actions through image recognition processing of the video data, and then identifying the emotions associated with the recognized facial expressions and actions.

[0045] (3-2-3) Speech synthesis unit 1323 The speech synthesis unit 1323 has the function of generating audio data for the AI's response voice. The speech synthesis unit 1323 performs speech synthesis processing on the response content generated by the response generation unit 1322.

[0046] (3-3) Output control unit 133 The output control unit 133 has the function of controlling the output of various types of information. The output control unit 133 transmits the voice data generated by the voice synthesis unit 1323 to the voice input / output device 30 via the communication unit 110, and causes the voice input / output unit 31 to output the AI's response voice.

[0047] (3-4) Detection unit 134 The detection unit 134 has the function of performing various detections during conversations between people with dementia and the AI. As shown in Figure 2, the detection unit 134 includes a pause detection unit 1341, a call termination detection unit 1342, and an anomaly detection unit 1343.

[0048] (3-4-1) Pause detection unit 1341 The pause detection unit 1341 has a function to detect when it is time to pause a conversation with a person with dementia. The pause detection unit 1341 detects when it is time to pause a conversation with a person with dementia, for example, based on the content of the person's speech. As an example, when a person with dementia speaks, the pause detection unit 1341 refers to event-specific response information, and if the response to the event corresponding to the content of the person's speech indicates that the conversation should be paused, it detects that it is time to pause the conversation.

[0049] (3-4-2) End of call detection unit 1342 The end-of-conversation detection unit 1342 has a function to detect when it is time to end a conversation with a person with dementia. The end-of-conversation detection unit 1342 detects when it is time to end a conversation with a person with dementia, for example, based on the content of the person's speech. For example, when a person with dementia speaks, the end-of-conversation detection unit 1342 refers to event-specific response information, and if the response to the event corresponding to the content of the person's speech indicates that the conversation should be ended, it detects that it is time to end the conversation.

[0050] In addition, the end-of-conversation detection unit 1342 may detect that it is time to end the conversation with the person with dementia when the "end conversation" button is pressed on the web screen, or when a pre-set time has elapsed. Furthermore, the end-of-conversation detection unit 1342 may also detect that it is time to end the conversation with the person with dementia based on AI judgment. For example, the AI ​​judgment may be based on the determination that the topic has run out, or that sufficient background factors of BPSD have been heard.

[0051] (3-4-3) Anomaly detection unit 1343 The anomaly detection unit 1343 has the function of detecting anomalies in conversations with persons with dementia. The anomaly detection unit 1343 detects when an anomaly occurs in a person with dementia, for example, based on the content of the person's speech. Anomalies that occur in a person with dementia include, for example, shouting, using abusive language, or becoming confused. If an anomaly is detected, the anomaly detection unit 1343 temporarily suspends the conversation with the person with dementia.

[0052] (3-5) Notification section 135 The notification unit 135 has the function of providing various notifications. For example, if the call termination detection unit 1342 or the anomaly detection unit 1343 detects something during a conversation with a person with dementia, the notification unit 135 sends a real-time notification via chat to the user terminal 20 (caregiver's terminal) or registered contacts. The notification content includes the time of termination or anomaly occurrence and a summary of the detection content.

[0053] (3-6) Storage and processing unit 136 The storage processing unit 136 has the function of saving conversation history information that shows the conversation history with a person with dementia. The storage processing unit 136 registers text data as conversation history, obtained by removing (or masking) personal information from text data showing the content of speech uttered by the person with dementia, generated by the speech recognition unit 1321, and text data showing the content of responses from the AI, generated by the response generation unit 1322. The storage processing unit 136 removes (or masks) personal information from the text data, for example, using a large-scale language model.

[0054] (3-7) Conversation Classification Unit 137 The conversation classification unit 137 has the function of classifying the content of conversations with persons with dementia. Based on the conversation history with persons with dementia stored by the storage processing unit 136, the conversation classification unit 137 classifies the content of conversations with persons with dementia in a Q&A format into categories of background factors for behavioral and psychological symptoms of dementia (e.g., living environment and sense of security, personal needs and difficulties, daily life and social interaction, things one likes, etc.). The conversation classification unit 137 performs the classification (categorization) of conversation content, for example, using a large-scale language model.

[0055] (3-8) Report generation unit 138 The report generation unit 138 has the function of generating a report of interviews regarding the background factors of BPSD. The report generation unit 138 generates a report that can be used for dementia assessment work, for example, based on the conversation history with a person with dementia. The report displays user information, factors, BPSD+Q / BPSD25Q, etc. The report generation unit 138 may also display the conversation history in the report in QA format, for example, using the conversation history after classification by the conversation classification unit 137.

[0056] <3. Processing Flow> The functional configuration of the conversation control device 10 according to this embodiment has been described above. Next, the processing flow according to this embodiment will be described with reference to Figures 3 to 5.

[0057] (1) Flow of pre-configuration process Referring to Figure 3, the flow of the pre-configuration process according to this embodiment will be explained. Figure 3 is a flowchart showing an example of the flow of the pre-configuration process according to this embodiment. The pre-configuration process described below is performed, for example, by a caregiver, but may also be performed by a person with dementia if possible.

[0058] As shown in Figure 3, first, the user terminal 20 and the audio input / output device 30 are connected (step S101). When used in a facility or when the caregiver and the person with dementia live together, either the user terminal 20 owned by the facility or the caregiver, or the user terminal 20 owned by the person with dementia, is connected to the audio input / output device 30. When the caregiver and the person with dementia live separately, the user terminal 20 owned by the person with dementia is connected to the audio input / output device 30.

[0059] Next, an account for using conversation system 1 is issued via the web, and the user logs in to conversation system 1 (step S102).

[0060] Next, a user ID for the dementia patient who will become a user is issued and registered on the administration screen (Step S103).

[0061] Next, the symptoms of the dementia patient, who is the user, are pre-registered (step S104). For example, the symptoms of the dementia patient are registered in BPSD+Q / BPSD25Q format.

[0062] Next, pre-registration for the conversation design is performed (Step S105). In the conversation design, it is possible to set the conversation format according to the progression of the dementia of the person with dementia (for example, question format or everyday conversation format), set the response policy for specific problematic behaviors in communication with the person with dementia (for example, how to respond when the person expresses a desire to go home), and set names to address the person.

[0063] (2) Flow of conversation control processing Referring to Figure 4, the flow of the conversation control process according to this embodiment will be explained. Figure 4 is a flowchart showing an example of the flow of the conversation control process according to this embodiment.

[0064] As shown in Figure 4, the conversation is first started by pressing the start button on the user terminal 20 (step S201). Pressing the start button initiates the conversation, with the AI ​​speaking.

[0065] If the person with dementia speaks (step S202 / YES), the conversation control device 10 proceeds to step S203. On the other hand, if the person with dementia does not speak (step S202 / NO), the conversation control device 10 waits until the person with dementia speaks.

[0066] If the process proceeds to step S203, the speech recognition unit 1321 of the conversation control device 10 performs speech recognition processing on the speech data representing the speech of a person with dementia (step S203).

[0067] Next, the detection unit 134 of the conversation control device 10 performs various detections based on the speech recognition results (step S204). If the anomaly detection unit 1343 does not detect anything (step S204 / NO), the process proceeds to step S205. On the other hand, if the anomaly detection unit 1343 detects something (step S204 / YES), the process proceeds to step S209.

[0068] If the detection process results in the process proceeding to step S205 and the conversation continuing, the response generation unit 1322 of the conversation control device 10 executes a response generation process for the utterance recognized by the speech recognition unit 1321 (step S205).

[0069] Next, the speech synthesis unit 1323 of the conversation control device 10 performs speech synthesis processing on the generated response content and generates speech data of the response voice (step S206).

[0070] Next, the output control unit 133 of the conversation control device 10 transmits the generated voice data to the voice input / output device 30, causing the voice input / output unit 31 to output the AI's response voice (step S207).

[0071] If the end of the conversation with the person with dementia is detected (step S208 / YES), the process proceeds to step S209. On the other hand, if the end of the conversation with the person with dementia is not detected (step S208 / NO), the process repeats from step S202.

[0072] If the process proceeds to step S209, the notification unit 135 of the conversation control device 10 issues various notifications (step S209) and terminates the conversation. If the process proceeds from step S204 to step S209, the notification unit 135 notifies the user terminal 20 or registered contact of the content detected by the anomaly detection unit 1343. If the process proceeds from step S208 to step S209, the notification unit 135 sends a notification to the user terminal 20 or registered contacts regarding the content detected by the call termination detection unit 1342.

[0073] (3) Flow of report generation process Referring to Figure 5, the flow of the report generation process according to this embodiment will be explained. Figure 5 is a flowchart showing an example of the flow of the report generation process according to this embodiment.

[0074] As shown in Figure 5, the storage processing unit 136 stores conversation history information showing the conversation history with the person with dementia (step S301). At this time, the storage processing unit 136 uses a large-scale language model to remove (or mask) personal information from the text data.

[0075] Next, the conversation classification unit 137 classifies the content of conversations with dementia patients based on the conversation history information stored by the storage processing unit 136 (step S302). At this time, the conversation classification unit 137 uses a large-scale language model to classify the content of conversations with dementia patients into categories of background factors for behavioral and psychological symptoms of dementia in a Q&A format.

[0076] Next, the report generation unit 138 generates an assessment report on the background factors of BPSD based on the classification results from the conversation classification unit 137 (step S303).

[0077] The processing flow according to this embodiment has been described above. As described above, the conversation system 1 according to this embodiment includes an audio input / output device 30 having at least an audio input / output unit 31 that has a shape similar to that of a telephone handset and is capable of inputting and outputting audio, and a conversation control device 10 that generates response content suitable for conversation with a person with dementia based on the audio content of the person with dementia input from the audio input / output device 30, and outputs audio indicating the response content from the audio input / output device 30.

[0078] This configuration allows individuals with dementia to use a device they are more familiar with and comfortable using, making it easier for them to initiate conversations with the AI. This reduces stress for individuals with dementia when conversing with the AI, and decreases the need for assistance from caregivers. Firstly, conversation system 1 reduces the burden on caregivers when communicating with individuals with dementia regarding repetitive questions and their desire to return home. Secondly, conversation system 1 reduces the workload on caregivers (staff) when conducting background factor interviews for BPSD assessment, while also enabling the collection of information with a certain level of accuracy. Therefore, the conversation system 1 according to this embodiment reduces the time and mental burden on caregivers in facilities and homes by reducing the need to respond to symptoms specific to people with dementia, such as repetitive questions and the desire to return home, and enables people with dementia to converse more easily.

[0079] <4. Variation> The embodiments have been described above. Next, modifications of the embodiments will be described. The modifications described below may be applied to the embodiments individually or in combination. Furthermore, the modifications may be applied in place of the configuration described in the embodiments, or they may be applied in addition to the configuration described in the embodiments.

[0080] In the embodiments described above, an example was given in which the voice input / output device 30 has a shape similar to that of a rotary phone, but the invention is not limited to this example. For example, the voice input / output device 30 may be a smartphone, tablet terminal, or PC. In this case, the function of inputting and outputting voice that the voice input / output device 30 has may be realized by the function of inputting and outputting voice that the user terminal 20 has. That is, the user terminal 20 and the voice input / output device 30 may be realized by a single device.

[0081] Furthermore, although the above-described embodiment described an example in which the voice input / output device 30 comprises a voice input / output unit 31, a main unit 32, and a camera unit 33, it is not limited to such an example. For example, the voice input / output device 30 only needs to have at least a voice input / output unit 31, and does not necessarily need to have a main unit 32. In this case, the voice input / output device 30 would only have a telephone handset. Thus, even if the voice input / output device 30 does not have a telephone main unit 32, as long as it has a handset, a person with dementia can easily participate in a conversation without confusion, compared to when AI response voices are output from a smartphone, tablet terminal, or PC.

[0082] Furthermore, while the above-described embodiment explains an example in which a caregiver or a person with dementia operates the screen displayed on the user terminal 20 to start a conversation with the AI, the invention is not limited to such an example. For example, the conversation control device 10 may be configured so that a conversation with the AI ​​is started when a person with dementia picks up the handset (voice input / output unit 31) of a rotary phone (voice input / output device 30). To start a conversation with a person with dementia, the output control unit 133 may output the phone's ringtone from the voice input / output device 30. This allows the person with dementia to start a conversation with the AI ​​using a familiar operation.

[0083] Furthermore, although the above-described embodiment described an example in which the shape of the audio input / output unit 31 is that of a telephone handset, the invention is not limited to this example. The shape of the audio input / output unit 31 may be other than that of a telephone handset, for example, it may be the shape of a hearing aid, earphones, headset, etc.

[0084] Furthermore, while the embodiments described above described an example in which the AI ​​always responds alone in a conversation with a person with dementia, the invention is not limited to such an example. For example, when a third party such as a caregiver or family member joins the conversation, the conversation control device 10 may recognize the third party's voice and switch the conversation between the AI ​​and the third party according to the content of the conversation, or cooperate to advance the conversation.

[0085] The above describes some modified examples of the embodiments. Furthermore, some or all of the functions of the conversation system 1, conversation control device 10, user terminal 20, and voice input / output device 30 in the above-described embodiment may be implemented by a computer. In that case, the functions may be implemented by recording a program for implementing these functions on a computer-readable recording medium, loading the program recorded on this recording medium into the computer system, and executing it. Here, "computer system" includes hardware such as the OS and peripheral devices. Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into the computer system. Furthermore, "computer-readable recording media" may include those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs over networks such as the Internet or communication lines such as telephone lines, as well as those that hold programs for a certain period of time, such as volatile memory inside computer systems that act as servers or clients in such cases. Furthermore, the above program may be for the purpose of implementing some of the functions described above, or it may be for the purpose of implementing the above functions in combination with a program already recorded in the computer system, or it may be implemented using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0086] Although embodiments of this invention have been described in detail above with reference to the drawings, the specific configuration is not limited to those described above, and various design changes can be made without departing from the spirit of this invention. [Explanation of symbols]

[0087] 1...Conversation system, 10...Conversation control device, 20...User terminal, 30...Voice input / output device, 31...Voice input / output unit, 32...Main unit, 33...Shooting unit, 110...Communication unit, 120...Storage unit, 130...Control unit, 131...Setting processing unit, 132...Voice processing unit, 133...Output control unit, 134...Detection unit, 135...Notification unit, 136...Storage processing unit, 137...Conversation classification unit, 138...Report generation unit, 1321...Voice recognition unit, 1322...Response generation unit, 1323...Voice synthesis unit, 1341...Pause detection unit, 1342...End of call detection unit, 1343...Anomaly detection unit, NW...Network

Claims

1. An audio input / output device having at least an audio input / output section that has a shape similar to that of a telephone handset and is capable of inputting and outputting audio, A conversation control device that generates response content suitable for conversation with a person with dementia based on audio representing the speech content of a person with dementia input from the audio input / output device, and outputs audio representing the response content from the audio input / output device. A conversational system equipped with the following features.

2. The aforementioned conversation control device, A response generation unit that generates response content suitable for conversation with the person with dementia using an AI model based on pre-set conceptual information about dementia. The conversation system according to claim 1, comprising:

3. The aforementioned AI model includes a large-scale language model, The response generation unit uses the large-scale language model to generate the response content suitable for conversation with the person with dementia. The conversation system according to claim 2.

4. The aforementioned conceptual information represents the concept of the validation method, which is a communication technique for people with dementia, and reflects conversational skills that do not cause stress to people with dementia in the content of the responses. The conversation system according to claim 2.

5. The aforementioned conceptual information represents a concept for responding to the behavioral and psychological symptoms of dementia, and reflects conversational skills that do not cause stress to the person with dementia in the content of the response. The conversation system according to claim 2.

6. The response generation unit generates the response content corresponding to the person with dementia using the AI ​​model, based on the individual setting information that is set in advance for each person with dementia. The conversation system according to claim 2.

7. The aforementioned individual setting information includes information indicating the progression of dementia for each person with dementia, The response generation unit generates the response content according to the degree of dementia progression. The conversation system according to claim 6.

8. The aforementioned individual setting information includes information for registering response policies regarding specific problematic behaviors in the communication of the person with dementia, The response generation unit generates the response content corresponding to the specific problematic behavior based on the registered response policy. The conversation system according to claim 6.

9. The response generation unit generates the response content corresponding to the event based on the AI ​​model, using the event-specific response information that is set in advance for each event that is expected to occur in the dialogue with the person with dementia. The conversation system according to claim 2.

10. The aforementioned conversation control device, A pause detection unit detects the timing to pause the conversation with the person with dementia based on the content of the person's speech. The conversation system according to claim 1, further comprising:

11. The aforementioned conversation control device, A conversation termination detection unit detects the timing to end a conversation with the person with dementia based on the content of the person's speech. The conversation system according to claim 1, further comprising:

12. The aforementioned conversation control device, An abnormality detection unit detects when an abnormality occurs in the person with dementia based on the content of the person's speech. The conversation system according to claim 1, further comprising:

13. The aforementioned conversation control device, Based on the conversation history with the person with dementia, a conversation classification unit classifies the content of the conversation with the person with dementia into categories of background factors for the behavioral and psychological symptoms of dementia in a Q&A format. The conversation system according to claim 1, further comprising:

14. The aforementioned conversation control device, A report generation unit generates a report that can be used for dementia assessment work based on the conversation history with the aforementioned person with dementia. The conversation system according to claim 1, further comprising:

15. The aforementioned conversation control device, An output control unit that causes the telephone's ringtone to be output from the voice input / output device in order to initiate a conversation with the person with dementia. The conversation system according to claim 1, further comprising:

16. The voice input / output device has a main body having the same shape as the main body of the telephone, The conversation system according to claim 1, further comprising:

17. The aforementioned telephone is a black rotary phone. The conversation system according to claim 1 or claim 16.

18. The aforementioned telephone is A camera unit for recording the face of the person with dementia on video in order to initiate a conversation with the person with dementia. Furthermore, The conversation control device grasps the emotions of the person with dementia in real time based on the video data acquired by the camera unit, and generates the response content corresponding to the emotions of the person with dementia. The conversation system according to claim 1.

19. A voice recognition unit that recognizes the speech content of a person with dementia input from a voice input / output device that has a shape similar to a telephone handset and is equipped with at least a voice input / output unit capable of inputting and outputting voice, A response generation unit that generates response content suitable for conversation with the person with dementia based on the recognized voice, A speech synthesis unit that converts the generated response content into speech, An output control unit that outputs the converted audio from the audio input / output device, A conversation control device equipped with the following features.

20. A speech recognition process that recognizes speech representing the content of speech spoken by a person with dementia, which is input from a speech input / output device having at least a speech input / output section that has a shape similar to that of a telephone handset and is capable of inputting and outputting speech, A response generation process that generates response content suitable for conversation with the person with dementia based on the recognized voice, A speech synthesis process that converts the generated response content into speech, An output control process that causes the converted audio to be output from the audio input / output device, A computer-based conversation control method, including the following:

21. Computers, A voice recognition means for recognizing speech representing the content of speech spoken by a person with dementia, which is input from a voice input / output device that has a shape similar to a telephone handset and is equipped with at least a voice input / output unit capable of inputting and outputting voice, A response generation means that generates response content suitable for conversation with the person with dementia based on the recognized voice, A speech synthesis means that converts the generated response content into speech, Output control means for outputting the converted audio from the audio input / output device, A program designed to function as such.

Images