System

Abstract

A system is provided.SOLUTION: A system comprising data collection means, means for cleansing the collected data and transforming it into a unified format as preprocessing means, means for learning the preprocessed data using a generated AI model and training the model, means for receiving and analyzing consultation content from a user, means for generating a response using the generated AI model based on the analyzed consultation content, and means for displaying the generated response to the user.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In conventional peer support systems, employees have a high psychological barrier to seeking advice themselves, and there are concerns about confidentiality, making it difficult to achieve the original goal of creating a healthy workplace environment for both body and mind. Furthermore, there is a lack of mechanisms for early detection of mental health issues among employees and appropriate responses. The challenge is to resolve these issues, create an environment where employees can easily seek advice, and prevent mental health issues from occurring. [Means for solving the problem]

[0005] To solve these problems, the present invention provides the following means. Specifically, the system includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a generative AI model for learning from the preprocessed data and training the model, a means for receiving and analyzing consultation content from a user, a means for generating a response using the generative AI model based on the analyzed consultation content, and a means for displaying the generated response to the user. This system allows employees to easily seek advice, catches signs of mental illness early, and receives an appropriate response, thereby improving mental care and realizing a healthy work environment.

[0006] "Data collection means" refers to the functions and devices that acquire the information to be collected and input it into the system.

[0007] "Preprocessing means" is a set of processes and functions that shape collected data into an analyzable format.

[0008] A "generative AI model" is a learning model that uses artificial intelligence technology to process natural language and generate responses.

[0009] The "means for receiving consultation contents from the user" refers to a function and device for acquiring consultation contents input by the user within the system.

[0010] The "analysis means" is a process for analyzing the received consultation content using natural language processing technology to understand its meaning and intent.

[0011] The "means for generating a response" refers to a function and device for creating an appropriate response based on the analyzed consultation content.

[0012] The "means for displaying a response" refers to the functionality and devices for presenting the generated response to the user. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0014] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0015] First, the terms used in the following description will be explained.

[0016] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0017] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0018] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0019] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0020] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0021] [First embodiment]

[0022] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0023] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0024] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0025] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0026] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0027] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0028] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0029] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0031] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0032] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0033] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0034] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[0035] The program of this system performs the following processing.

[0036] 1. Data Collection

[0037] The server acquires the data to be collected, i.e., information provided by wellness centers and past consultation cases of peer supporters nationwide. This collection is performed using API calls and database queries.

[0038] 2. Data Preprocessing

[0039] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format by segmenting and tokenizing it. This preprocessing prepares the data in a format that can be analyzed.

[0040] 3. Model Training

[0041] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0042] 4. Receiving inquiries from users

[0043] The user inputs the details of the consultation into the terminal and transmits them. The terminal transmits the input data to the server.

[0044] 5. Data Analysis

[0045] The server uses natural language processing technology to analyze the received consultation content and understand its meaning and intent. This analysis prepares input data for generating an appropriate response.

[0046] 6. Response Generation

[0047] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is generated by inputting the consultation content into the model and obtaining the generated text data.

[0048] 7. Response Display

[0049] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[0050] Specific examples

[0051] Example 1: Stress consultation

[0052] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[0053] The terminal sends this input to the server.

[0054] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[0055] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0056] The terminal displays the generated response to the user, and the user can receive specific advice.

[0057] Example 2: Relationship advice

[0058] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[0059] The terminal sends this input to the server.

[0060] The server receives the consultation content, analyzes it using natural language processing, and prepares the input data to generate an appropriate response.

[0061] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0062] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0063] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[0064] The processing flow will be explained below.

[0065] Processing Steps

[0066] Step 1:

[0067] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[0068] How it works: It uses API calls to retrieve information from the wellness center database, and also uses database queries to gather peer supporter consultation cases stored locally or in cloud storage.

[0069] Step 2:

[0070] The server pre-processes the collected data.

[0071] How it works: Text data is cleansed to remove noise and unnecessary characters, then natural language processing techniques are used to segment and tokenize it, converting it into a structured format.

[0072] Step 3:

[0073] The server formats the preprocessed data into a training dataset.

[0074] How it works: Split the dataset into training and validation sets, and format them into JSON or CSV formats. Consider the balance of the data, and perform sampling or data augmentation if there is an imbalance.

[0075] Step 4:

[0076] The server trains a generative AI model (e.g., GPT-4).

[0077] How it works: The preprocessed dataset is fed into a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model. Adaptive learning rates and other optimization techniques are applied to improve the model's accuracy.

[0078] Step 5:

[0079] The server prepares to receive the consultation content from the user.

[0080] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[0081] Step 6:

[0082] The user uses the terminal to input and transmit the consultation contents.

[0083] Operation: Enter the content of your inquiry into the text field and click the "Send" button. This will cause the device to send the content of your inquiry to the server.

[0084] Step 7:

[0085] The terminal transmits the user's input data to the server.

[0086] Operation: The text data of the consultation is sent to the server via an HTTP request.

[0087] Step 8:

[0088] The server analyzes the received consultation content.

[0089] How it works: Using natural language processing technology, the text of the consultation is semantically analyzed and important keywords and context are extracted.

[0090] Step 9:

[0091] The server generates a response using a generative AI model based on the analyzed data.

[0092] What it does: It feeds the analysis results into a generative AI model to generate an appropriate response text, which is then formatted appropriately and prepared to be sent back to the user.

[0093] Step 10:

[0094] The server sends the generated response to the terminal.

[0095] Behavior: Sends the generated response to the device as an HTTP response.

[0096] Step 11:

[0097] The terminal displays the response received from the server to the user.

[0098] Behavior: Displays the received response text on the user interface.

[0099] Specific examples

[0100] Example 1: Consultation regarding stress

[0101] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice" (step 6).

[0102] The terminal transmits the consultation content to the server (step 7).

[0103] The server receives the consultation content and analyzes it using natural language processing (step 8).

[0104] The server generates a response using a generative AI model. "To reduce stress, try daily relaxation techniques, such as meditation, deep breathing, and light exercise" (Step 9).

[0105] The server sends the generated response to the terminal (step 10).

[0106] The terminal displays the response to the user (step 11).

[0107] The above is a detailed flow of processing in the system.

[0108] Example 1

[0109] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0110] It is important to detect and appropriately address mental health issues among employees early. However, due to a lack of environments where employees can easily seek advice, problems are often discovered late. Furthermore, conventional systems require a great deal of time and effort for data collection, preprocessing, and model training, making it difficult to respond in real time. For this reason, there is a need for a system that can provide effective and prompt mental health support.

[0111] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0112] In this invention, the server includes means for acquiring information to be collected, means for cleansing the collected information and converting it into a unified format, means for learning a generative AI model using the preprocessed data and training the model, means for receiving and transmitting consultation content from a user, means for analyzing the received consultation content, means for generating a response using the generative AI model based on the analyzed consultation content, and means for displaying the generated response to the user. This makes it possible to effectively utilize the collected data and provide appropriate responses in real time in response to employee consultations.

[0113] "Information to be collected" refers to the information source that the system is set up to acquire, and in this case, this refers to data such as information provided by the health support center and past consultation cases of peer supporters.

[0114] "Cleansing" is a preprocessing process to remove unnecessary characters and noise from collected data and ensure data quality.

[0115] A "uniform format" is a standardization procedure for converting data into a format suitable for analysis and learning, and is a data format that ensures consistency.

[0116] A "generative AI model" is an artificial intelligence model that learns from collected and preprocessed data and generates appropriate responses based on the user's inquiry.

[0117] "Learning" is the process by which a generative AI model uses training data to improve the accuracy of response generation.

[0118] "Training" is the process of repeatedly learning a generative AI model using collected data.

[0119] "User consultation content" is text data that system users input and send regarding mental health issues and questions.

[0120] "Analysis" refers to the process of analyzing the meaning of the received user's consultation content using natural language processing technology and understanding the intent of the consultation content.

[0121] A "response" is a text message containing advice or information generated by the generative AI model based on the analyzed content of the user's consultation.

[0122] "Display" refers to an operation or function for visually presenting the generated response on the user's terminal.

[0123] This invention relates to an AI peer support system for preventing mental health problems among employees. This system uses a generative AI model to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by health support centers.

[0124] Program processing explanation

[0125] The main processing of this system is carried out by three elements: the server, the terminal, and the user.

[0126] Data collection

[0127] The server acquires the information to be collected. Specifically, it uses API calls to collect information provided by the health support center and past consultation cases by peer supporters. This data is stored in a database such as MongoDB or PostgreSQL.

[0128] Data Preprocessing

[0129] The server cleanses the collected data and converts it into a unified format, using Python scripts to remove unnecessary data, and natural language processing libraries such as NLTK and SpaCy for segmentation and tokenization. Finally, the data is converted into formats such as CSV and Parquet.

[0130] Model learning

[0131] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). The data is split into a training dataset and a test dataset, and training is performed using a deep learning framework such as TensorFlow or PyTorch. Once training is complete, the model is saved.

[0132] Accepting inquiries from users

[0133] The user uses the device to input and send the content of their consultation. For example, they might use a web or mobile application to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device then sends this data to the server as an HTTP request.

[0134] Analyzing received data

[0135] The server analyzes the received consultation content using natural language processing technology. It obtains text data and performs tokenization and entity recognition using NLTK and SpaCy. This allows it to understand the intent of the consultation content.

[0136] Response Generation

[0137] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. For example, a prompt such as "I've been feeling more stressed at work lately, so I'd like some advice" is input into the model, and the generated text data is retrieved. The response generated is, "To reduce stress, it's effective to try daily relaxation techniques. Examples include meditation, deep breathing, and light exercise."

[0138] Response Display

[0139] The device displays the response received from the server to the user, and displays the response message in a text area on a web page or in an application, or in the chatbot's UI, allowing the user to receive appropriate advice.

[0140] Specific examples

[0141] Example 1: Stress consultation

[0142] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[0143] The terminal sends this input to the server.

[0144] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[0145] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0146] The terminal displays the generated response to the user, and the user can receive specific advice.

[0147] Example 2: Relationship advice

[0148] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[0149] The terminal sends this input to the server.

[0150] The server receives the consultation content, analyzes it using natural language processing technology, and prepares the input data to generate an appropriate response.

[0151] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0152] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0153] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[0154] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0155] Step 1:

[0156] The server uses API calls to obtain the information to be collected. Specifically, it obtains data including past consultation cases from health support centers and peer supporters from the API endpoint. The input is the API endpoint URL and authentication information, and the output is data in JSON format. This data is stored in a database such as MongoDB or PostgreSQL.

[0157] Step 2:

[0158] The server cleanses the collected data and converts it into a unified format. Specifically, it uses Python scripts to remove unnecessary characters and noise, and performs word segmentation and tokenization using natural language processing libraries such as NLTK and SpaCy. The input is the collected JSON data, and the output is structured data in CSV or Parquet format.

[0159] Step 3:

[0160] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). Specifically, the data is divided into a training set and a test set, and the model is trained using a deep learning framework such as TensorFlow or PyTorch. The input is the preprocessed CSV or Parquet data, and the output is a trained generative AI model. This model is stored on the server.

[0161] Step 4:

[0162] The user uses a terminal to input and send the content of their consultation. Specifically, they enter something like "I've been feeling more stressed at work lately, so I'd like some advice" into a form on a web or mobile application, and then press the send button. The input is text data entered by the user, and the output is sent to the server as an HTTP request.

[0163] Step 5:

[0164] The server analyzes the consultation content received from the device. Specifically, it takes the received text data and performs tokenization and entity recognition using natural language processing libraries such as NLTK and SpaCy. The input is the received text data, and the output is the tokenized analyzed data. This analysis allows the intent and gist of the consultation content to be understood.

[0165] Step 6:

[0166] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. Specifically, the analyzed data is input into the model as a prompt sentence, and the generated text data is obtained. The input is the tokenized analyzed data and the prompt sentence, and the output is the generated response text. In this example, a response such as "To reduce stress, it is effective to try daily relaxation methods. Examples include meditation, deep breathing, and light exercise" is generated.

[0167] Step 7:

[0168] The terminal displays the response from the server to the user. Specifically, it displays the received response text in a text area on a web page or in an application, or in the chatbot's UI. The input is the response text received from the server, and the output is a text display that the user can check on the screen.

[0169] (Application example 1)

[0170] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0171] Mental health problems among employees are a serious problem that can lead to reduced work efficiency and a worsening work environment. Particularly in brick-and-mortar stores, where employees often have direct contact with customers, accumulating stress and interpersonal problems can negatively impact how they treat customers. The present invention aims to provide an environment where employees working in brick-and-mortar stores can easily receive mental care, thereby improving the work environment and the mental health of employees.

[0172] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0173] In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing consultation content from a user, a means for generating a response using the generative AI model based on the analyzed consultation content, and a means for displaying the generated response on a terminal used by an employee at a physical store. This enables employees working at physical stores to easily seek mental health consultation and receive appropriate advice.

[0174] "Data collection methods" are methods for collecting information provided by wellness centers and past consultation cases of peer supporters across the country.

[0175] The "preprocessing means" is a means for cleansing the collected data, removing unnecessary characters and noise, and converting the data into a unified format.

[0176] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to employee inquiries.

[0177] The "means for receiving the consultation content from the user" is a means for transmitting the consultation content input by the employee to the terminal to the server and receiving it.

[0178] The "means for analyzing" refers to a means for analyzing the received consultation content using natural language processing technology and understanding the meaning and intent of the content.

[0179] The "means for generating a response" is a means for generating an appropriate response using a generative AI model based on the analyzed consultation content.

[0180] The "means for displaying the generated response" is a means for displaying the response sent from the server on the terminal used by the employee.

[0181] "Brick and mortar store" refers to a physical sales or service location where employees interact directly with customers.

[0182] This invention relates to an AI peer support system to prevent employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers. This system is designed especially for use by employees working in brick-and-mortar stores.

[0183] The system is configured as follows:

[0184] Hardware and Software

[0185] The system's main hardware consists of a smartphone or tablet connected to a server. The server is responsible for data collection, preprocessing, training the generative AI model, and generating responses. The smartphone or tablet acts as an interface where users input their inquiries and view responses from the server. The software primarily uses OpenAI's API, Python scripts, and the HTTP request library.

[0186] Data collection

[0187] The server collects information provided by the wellness center and past consultation cases of peer supporters nationwide through API calls or database queries. This data is used as the basis for user consultations.

[0188] Data Preprocessing

[0189] The collected data is cleansed by the server to remove unnecessary characters and noise, and then converted into a structured format using techniques such as tokenization and word segmentation. This preprocessing prepares the data in a format that can be analyzed.

[0190] Model learning

[0191] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0192] Receiving and analyzing inquiries from users

[0193] Users input and submit their consultation details using a device such as a smartphone or tablet. The device then sends this input data to a server, which then analyzes the received consultation details using natural language processing technology.

[0194] Response generation and display

[0195] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is sent to the device and displayed to the user, allowing the user to obtain appropriate countermeasures and advice.

[0196] Specific examples

[0197] Example 1: Stress consultation

[0198] User: I'm under a lot of pressure at work, how can I reduce it?

[0199] Prompt: User wants to know: I'm under a lot of pressure at work. How can I alleviate it?\nGive appropriate advice:

[0200] Example 2: Relationship Advice

[0201] User: I'm having some disagreements with a coworker and it's been a strained relationship. Is there anything I can do to improve this?

[0202] Prompt: User's question: I'm having disagreements with a colleague and it's been a strained relationship. Is there a way to improve this?\nPlease provide appropriate advice:

[0203] In this way, the AI ​​peer supporter system of the present invention allows store employees to easily seek mental health advice and receive appropriate advice in real time, enabling them to smoothly carry out their daily work and improve the quality of customer service.

[0204] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0205] Step 1:

[0206] The server collects information provided by wellness centers and past consultation cases of peer supporters nationwide using API calls or database queries. The input is the API endpoints for wellness centers and peer support data, and the output is the collected data.

[0207] Step 2:

[0208] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format using tokenization, word segmentation, etc. The input is the collected data, and the output is the preprocessed data.

[0209] Step 3:

[0210] The server uses the preprocessed data to learn and train a generative AI model (e.g., GPT-4), which improves the model's accuracy. The input is the preprocessed data, and the output is a trained generative AI model.

[0211] Step 4:

[0212] The user inputs the content of the consultation using a device such as a smartphone or tablet and sends the input to the server. The input is the content of the user's consultation, and the output is the data sent to the server.

[0213] Step 5:

[0214] The server analyzes the received consultation content using natural language processing technology (e.g., NLP technology) to understand the meaning and intent of the content. The input is the consultation content from the user, and the output is the analyzed consultation content.

[0215] Step 6:

[0216] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. The input is the analyzed consultation content and the generative AI model, and the output is the generated response.

[0217] Step 7:

[0218] The terminal displays the response sent by the server to the user. The input is the response generated, and the output is the advice or action displayed to the user.

[0219] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0220] This invention relates to an AI peer support system for preventing mental health problems among employees. This system combines generative AI and an emotion engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[0221] The program of this system performs the following processing.

[0222] 1. Data Collection

[0223] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country using API calls and database queries.

[0224] 2. Data Preprocessing

[0225] The server cleanses the collected data, removing unnecessary characters and noise, and then converts the data into a structured format through segmentation and tokenization. This preprocessing prepares the data in a form that can be analyzed.

[0226] 3. Model Training

[0227] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0228] 4. Receiving and analyzing consultation content

[0229] The user inputs the content of their consultation into the terminal and sends it. The terminal then sends the input data to the server. The server then analyzes the received content using natural language processing technology to understand the meaning and intent of the content.

[0230] 5. Emotion analysis

[0231] The server uses an emotion engine to analyze the emotion from the user's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response based on the analysis.

[0232] 6. Response Generation

[0233] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. This response is generated by inputting the consultation content and emotional data into the model and obtaining the generated text data.

[0234] 7. Response Display

[0235] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[0236] Specific examples

[0237] Example 1: Stress consultation

[0238] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice."

[0239] The terminal sends this input to the server.

[0240] The server receives the consultation content and analyzes it using natural language processing technology.

[0241] The server uses an emotion engine to analyze the user's emotions and extracts emotion data indicating "increasing stress."

[0242] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0243] The terminal displays the generated response to the user, and the user can receive specific advice.

[0244] Example 2: Relationship advice

[0245] The user uses the terminal to input, "I'm having trouble with my relationships at work. What should I do?"

[0246] The terminal sends this input to the server.

[0247] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning and intent.

[0248] The server uses an emotion engine to analyze the user's emotions and extracts data indicating that "anxiety" is strongly felt.

[0249] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0250] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0251] In this way, the AI ​​peer supporter system of the present invention, combined with an emotion engine, provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice. This makes it possible to detect signs of mental illness early and provide appropriate countermeasures, with the aim of realizing a workplace environment that is healthy both physically and mentally.

[0252] The processing flow will be explained below.

[0253] Processing Steps

[0254] Step 1:

[0255] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[0256] How it works: The server uses API calls and database queries to gather the necessary data, which is then stored locally or in cloud storage.

[0257] Step 2:

[0258] The server pre-processes the collected data.

[0259] How it works: The server cleanses the text data, removing noise and unnecessary characters, and then segments and tokenizes the data to make it parseable.

[0260] Step 3:

[0261] The server formats the preprocessed data into a training dataset.

[0262] What it does: Splits the dataset into training and validation sets, and performs sampling and data augmentation as needed to keep the data balanced.

[0263] Step 4:

[0264] The server trains a generative AI model (e.g., GPT-4).

[0265] How it works: The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model using the preprocessed dataset, applying adaptive learning rates and other optimization techniques to improve the model's accuracy.

[0266] Step 5:

[0267] The server prepares to receive the consultation content from the user.

[0268] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[0269] Step 6:

[0270] The user uses the terminal to input and transmit the consultation contents.

[0271] Operation: The user enters the content of the consultation in the text field and clicks the "Send" button. The device sends the consultation data to the server.

[0272] Step 7:

[0273] The terminal transmits the user's input data to the server.

[0274] Operation: The text data of the consultation is sent to the server using an HTTP request.

[0275] Step 8:

[0276] The server analyzes the received consultation content.

[0277] How it works: The server uses natural language processing technology to analyze the meaning and intent of the consultation, extracting important keywords and context.

[0278] Step 9:

[0279] The server uses an emotion engine to analyze the emotion from the content of the user's consultation.

[0280] How it works: The emotion engine detects emotion types (e.g., joy, sadness, fear) and their intensity from text data. This emotion data is fed into a generative AI model.

[0281] Step 10:

[0282] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data.

[0283] How it works: The server inputs the consultation content and emotional data into the generative AI model, receives the generated text response, and adjusts the tone and content of the response based on the emotional data.

[0284] Step 11:

[0285] The server sends the generated response to the terminal.

[0286] Operation: The generated response data is sent back to the terminal via an HTTP response.

[0287] Step 12:

[0288] The terminal displays the response received from the server to the user.

[0289] Behavior: Displays the received response text on the user's screen.

[0290] Specific examples

[0291] Example 1: Stress consultation

[0292] Step 1:

[0293] The server collects information provided by the wellness center and past consultation cases of peer supporters.

[0294] Step 2:

[0295] The server cleanses the collected data and puts it into an analyzable format.

[0296] Step 3:

[0297] The server formats the preprocessed data into a training dataset and trains a generative AI model.

[0298] Step 4:

[0299] The user uses the terminal to input "I've been feeling more and more stressed at work recently, so I'd like some advice," and then sends it.

[0300] Step 5:

[0301] The terminal transmits the user's input data to the server.

[0302] Step 6:

[0303] The server receives the consultation content and analyzes it using natural language processing technology.

[0304] Step 7:

[0305] The server uses an emotion engine to extract emotion data such as "stress is increasing."

[0306] Step 8:

[0307] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0308] Step 9:

[0309] The server sends the generated response to the terminal.

[0310] Step 10:

[0311] The terminal displays the response to the user, and the user can receive specific advice.

[0312] The above is the specific processing flow of a system that combines an emotion engine.

[0313] Example 2

[0314] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0315] To prevent employees from developing mental health problems, it is necessary to provide an environment where employees can easily seek advice. However, conventional systems lack the data necessary to generate appropriate responses and perform insufficient emotion analysis, making it difficult to provide personalized responses to users. As a result, employees may not receive the support they need, which could worsen their mental health.

[0316] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing the consultation content from the user, a sentiment analysis means for analyzing the sentiment from the user's consultation content and generating sentiment data for response generation, a means for generating a response using the generative AI model based on the analyzed consultation content and sentiment data, and a means for displaying the generated response to the user. This makes it possible to provide a personalized response tailored to the user's emotional state and realize an environment where employees can easily seek consultation.

[0317] "Data collection methods" are methods for collecting necessary data from past consultation cases of wellness centers and supporters.

[0318] "Preprocessing means" refers to means for cleansing collected data, removing unnecessary data and noise, and converting data into a unified format.

[0319] A "means for training using a generative AI model" is a means for training a generative AI model using preprocessed data.

[0320] The "means for receiving and analyzing the consultation content" is a means for receiving the consultation content from the user and analyzing the content using natural language processing technology.

[0321] The "emotion analysis means" is a means for analyzing emotions from the content of the user's consultation and generating emotion data.

[0322] The "means for generating a response" refers to a means for generating an appropriate response using a generative AI model based on the analyzed consultation content and emotional data.

[0323] The "means for displaying a response" is a means for displaying the generated response to the user.

[0324] A "wellness center" is a specialized institution that provides information on maintaining employee health and mental care.

[0325] A "support person" is a person or expert who serves as a source of advice for employees.

[0326] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system combines generative AI and an emotion analysis engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by specialist institutions.

[0327] The system's program is composed of the following hardware and software: The server uses Python libraries (e.g., requests and SQLAlchemy) to collect the necessary information from the wellness center and supporter databases. Additionally, as a preprocessing method, the pandas and nltk libraries are used to cleanse the collected data and convert it into a unified format.

[0328] The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to learn and train data using a generative AI model (e.g., GPT-4). It receives the user's consultation and analyzes it using natural language processing techniques (e.g., spaCy or BERT). Furthermore, it uses a sentiment analysis engine (e.g., TextBlob or VADER) to analyze the user's sentiment and generate sentiment data for response generation.

[0329] Let's explain with a concrete example. A user uses a device to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device sends this input to the server. The server receives the consultation content and performs natural language processing using spaCy to analyze the meaning and intent of the sentence. Next, the server performs sentiment analysis using the VADER library and extracts the emotional data "stress is increasing." The server uses a GPT-4 model to generate an appropriate response based on the following prompt sentence:

[0330] "The user is asking for advice about work stress. Please provide appropriate advice."

[0331] The terminal then receives the generated response and displays it to the user.

[0332] Similarly, if a user types "I'm having trouble with relationships at work. What should I do?" into their device, the same process will occur. The server will analyze the consultation content using spaCy and perform sentiment analysis using TextBlob. If "anxiety" is strongly detected, the server will use the GPT-4 model to input the following prompt:

[0333] "The user is asking for advice about relationships at work. Please provide advice on how to improve them."

[0334] The generated response is then received by the terminal and displayed to the user.

[0335] This system provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice, making it possible to catch early signs of mental health problems and provide appropriate countermeasures.

[0336] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0337] Step 1: Data collection

[0338] The server collects information from the wellness center and supporter databases. It uses Python's requests library to send API requests and retrieves data in JSON format. It also extracts supporter past consultation cases from the database using SQL queries. The input is the wellness center API and database queries, and the output is the collected raw data.

[0339] Step 2: Data Preprocessing

[0340] The server cleanses the collected data to remove noise and unnecessary characters. Specifically, it creates a data frame using Python's pandas library and filters out unnecessary data using regular expressions. It then uses nltk to tokenize the data and convert it into a unified format (e.g., JSON). The input is the collected raw data, and the output is the preprocessed data.

[0341] Step 3: Training the generative AI model

[0342] The server trains a generative AI model (e.g., GPT-4) using the preprocessed data. Specifically, it builds a model using a deep learning framework (e.g., TensorFlow or PyTorch) and runs training using the preprocessed data. The input is the preprocessed data, and the output is a trained generative AI model.

[0343] Step 4: Receiving consultation details

[0344] The user enters the content of their consultation into an input field on the terminal. The terminal then sends the entered content to the server. Specifically, an HTTP POST request is used. The input is the user's consultation content, and the output is the data to be sent to the server.

[0345] Step 5: Analysis of consultation content

[0346] The server analyzes the received consultation content and understands the meaning and intent of the text using natural language processing technology (e.g., spaCy or BERT). The input is the received consultation content, and the output is the analyzed meaning and intent of the text.

[0347] Step 6: Sentiment Analysis

[0348] The server uses an emotion analysis engine (e.g., TextBlob or VADER) to extract emotion data from the analyzed consultation content. It measures the type of emotion (joy, sadness, anger, etc.) and its intensity. The input is the analyzed meaning and intent, and the output is emotion data.

[0349] Step 7: Response Generation

[0350] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data. Specifically, it inputs a prompt sentence (e.g., "The user is asking for advice about work stress. Please provide appropriate advice.") into the generative AI model and obtains the generated response. The input is the analyzed consultation content and emotional data, and the output is the generated response.

[0351] Step 8: Display the response

[0352] The terminal receives the response sent by the server and displays it to the user, typically using a text view or a dialog box to display the response. The input is the response generated, and the output is what is displayed to the user.

[0353] (Application example 2)

[0354] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0355] There is a need for systems that can detect mental stress and mental disorders among employees early and deal with them promptly and appropriately. However, existing employee support systems lack effective emotion analysis and personalized responses, and do not provide an environment where employees can easily seek advice. Furthermore, there are many situations where an appropriate response in real time is required, so technology to solve this issue is needed.

[0356] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a preprocessing means for cleansing collected data and converting it into a unified format; a means for learning the preprocessed data using a generative AI model and training the model; a means for receiving and analyzing consultation content from a user; a means for generating a response using the generative AI model based on the analyzed consultation content; a means for displaying the generated response to the user; a means for an employee to directly input the consultation content using a smart device; an emotion analysis means for analyzing emotions based on the input consultation content; and a means for generating the emotion analysis result as a prompt sentence for the generative AI model. This enables employees to easily receive mental support and enables early detection and appropriate treatment of mental disorders.

[0357] "Data collection means" refers to the means used to collect information such as past consultation cases of employees and information provided by the wellness center.

[0358] "Preprocessing means" refers to means for cleansing collected data and converting it into a unified format.

[0359] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to user inquiries.

[0360] A "training means" is a means for training preprocessed data using a generative AI model.

[0361] The "means for receiving consultation content" is a means for receiving consultation content from employees.

[0362] The "analysis means" is a means for analyzing the received consultation content and understanding its meaning and intent.

[0363] The "response generation means" is a means for generating a personalized response using a generative AI model based on the analyzed consultation content.

[0364] A "display means" is a means for displaying the generated response on the employee's device.

[0365] "Smart devices" are electronic devices such as smartphones and tablets that employees use to input consultation details.

[0366] The "emotion analysis means" is a means for analyzing emotions based on the content of an employee's consultation and obtaining the results.

[0367] The "prompt generation means" is a means for generating prompt sentences for the generative AI model based on the results of emotion analysis.

[0368] This invention relates to an AI peer support system for preventing employee mental health problems, and provides a consultation environment by combining generative AI models and emotion analysis technology. This system creates an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers.

[0369] This system is configured as follows:

[0370] Hardware and Software

[0371] 1. Hardware: smartphones, tablets, servers.

[0372] 2. Software: Python program, OpenAI GPT-4, TextBlob.

[0373] Explanation of main processes

[0374] 1. Data collection methods:

[0375] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[0376] 2. Pretreatment methods:

[0377] The server cleanses the collected data, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format, making it parseable.

[0378] 3. Training methods:

[0379] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4) using a deep learning framework, which performs iterative learning to improve the model's accuracy.

[0380] 4. How to receive consultation details:

[0381] The user inputs and sends the consultation details using a smart device, which then sends the input to the server.

[0382] 5. Analysis methods:

[0383] The server analyzes the received consultation content using natural language processing technology to understand the meaning and intent of the consultation content.

[0384] 6. Emotion analysis means:

[0385] The server uses an emotion engine to analyze emotions from the employee's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response.

[0386] 7. Prompt Generation Method:

[0387] Based on the results of the sentiment analysis, a prompt sentence is generated for the generative AI model. For example, the following prompt sentence is generated:

[0388] text

[0389] User's question: I've been so busy at work lately that I'm feeling stressed.

[0390] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[0391] Provide specific advice to this user.

[0392] 8. Response Generation Methods:

[0393] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. A personalized response is created by inputting a prompt sentence to the generative AI model and obtaining the generated text data.

[0394] 9. Display means:

[0395] The smart device displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate countermeasures or advice.

[0396] Specific examples

[0397] Example 1: Stress consultation

[0398] The user uses the terminal to input, "I've been so busy at work recently that I'm stressed out."

[0399] The smart device sends this input to the server.

[0400] The server receives the consultation content and analyzes it using natural language processing technology.

[0401] The server uses an emotion engine to analyze emotions and extract data indicating "increasing stress."

[0402] The server uses a generative AI model to generate a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0403] The smart device displays the generated response to the user, who can then receive specific advice.

[0404] As a result, the AI ​​peer supporter system of the present invention combines an emotion engine and a generative AI model to provide personalized responses tailored to the user's own emotional state, creating an environment where employees can feel comfortable seeking advice.

[0405] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0406] Step 1:

[0407] Data collection methods

[0408] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[0409] Input: Wellness center and peer supporter data.

[0410] Output: A dataset of collected information.

[0411] Step 2:

[0412] Pretreatment means

[0413] The server cleanses the data collected in step 1, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format.

[0414] Input: Collected information dataset.

[0415] Output: A cleansed and tokenized dataset.

[0416] Step 3:

[0417] Training methods

[0418] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4), using a deep learning framework to iteratively train the model to improve its accuracy.

[0419] Input: The preprocessed dataset.

[0420] Output: A trained generative AI model.

[0421] Step 4:

[0422] Means of receiving consultation details

[0423] The user inputs the consultation details using the smart device, and the terminal transmits this input to the server.

[0424] Input: User's consultation content.

[0425] Output: Consultation content data sent to the server.

[0426] Step 5:

[0427] Analysis means

[0428] The server analyzes the consultation content received in step 4 using natural language processing technology to understand the meaning and intent of the consultation content.

[0429] Input: Received consultation data.

[0430] Output: Parsed consultation data.

[0431] Step 6:

[0432] Emotion analysis means

[0433] The server analyzes emotions using an emotion engine based on the analyzed consultation content, and detects the type and intensity of the emotion.

[0434] Input: Parsed consultation content data.

[0435] Output: Sentiment analysis data.

[0436] Step 7:

[0437] Prompt Generation Method

[0438] The server generates prompt sentences for the generative AI model based on the emotion analysis data.

[0439] Input: Sentiment analysis data.

[0440] Output: The generated prompt statement.

[0441] Examples:

[0442] text

[0443] User's question: I've been so busy at work lately that I'm feeling stressed.

[0444] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[0445] Provide specific advice to this user.

[0446] Step 8:

[0447] Response Generation Method

[0448] The server inputs the prompt sentence into a generative AI model, which then generates an appropriate response.

[0449] Input: The generated prompt statement.

[0450] Output: The generated response text.

[0451] Step 9:

[0452] Display means

[0453] The terminal displays the response sent from the server to the user, allowing the user to receive specific advice and solutions.

[0454] Input: The generated response text.

[0455] Output: The response text displayed to the user.

[0456] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0457] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0458] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0459] [Second embodiment]

[0460] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0461] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0462] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0463] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0464] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0465] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0466] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0467] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0468] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0469] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0470] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0471] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0472] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[0473] The program of this system performs the following processing.

[0474] 1. Data Collection

[0475] The server acquires the data to be collected, i.e., information provided by wellness centers and past consultation cases of peer supporters nationwide. This collection is performed using API calls and database queries.

[0476] 2. Data Preprocessing

[0477] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format by segmenting and tokenizing it. This preprocessing prepares the data in a format that can be analyzed.

[0478] 3. Model Training

[0479] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0480] 4. Receiving inquiries from users

[0481] The user inputs the details of the consultation into the terminal and transmits them. The terminal transmits the input data to the server.

[0482] 5. Data Analysis

[0483] The server uses natural language processing technology to analyze the received consultation content and understand its meaning and intent. This analysis prepares input data for generating an appropriate response.

[0484] 6. Response Generation

[0485] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is generated by inputting the consultation content into the model and obtaining the generated text data.

[0486] 7. Response Display

[0487] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[0488] Specific examples

[0489] Example 1: Stress consultation

[0490] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[0491] The terminal sends this input to the server.

[0492] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[0493] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0494] The terminal displays the generated response to the user, and the user can receive specific advice.

[0495] Example 2: Relationship advice

[0496] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[0497] The terminal sends this input to the server.

[0498] The server receives the consultation content, analyzes it using natural language processing, and prepares the input data to generate an appropriate response.

[0499] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0500] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0501] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[0502] The processing flow will be explained below.

[0503] Processing Steps

[0504] Step 1:

[0505] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[0506] How it works: It uses API calls to retrieve information from the wellness center database, and also uses database queries to gather peer supporter consultation cases stored locally or in cloud storage.

[0507] Step 2:

[0508] The server pre-processes the collected data.

[0509] How it works: Text data is cleansed to remove noise and unnecessary characters, then natural language processing techniques are used to segment and tokenize it, converting it into a structured format.

[0510] Step 3:

[0511] The server formats the preprocessed data into a training dataset.

[0512] How it works: Split the dataset into training and validation sets, and format them into JSON or CSV formats. Consider the balance of the data, and perform sampling or data augmentation if there is an imbalance.

[0513] Step 4:

[0514] The server trains a generative AI model (e.g., GPT-4).

[0515] How it works: The preprocessed dataset is fed into a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model. Adaptive learning rates and other optimization techniques are applied to improve the model's accuracy.

[0516] Step 5:

[0517] The server prepares to receive the consultation content from the user.

[0518] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[0519] Step 6:

[0520] The user uses the terminal to input and transmit the consultation contents.

[0521] Operation: Enter the content of your inquiry into the text field and click the "Send" button. This will cause the device to send the content of your inquiry to the server.

[0522] Step 7:

[0523] The terminal transmits the user's input data to the server.

[0524] Operation: The text data of the consultation is sent to the server via an HTTP request.

[0525] Step 8:

[0526] The server analyzes the received consultation content.

[0527] How it works: Using natural language processing technology, the text of the consultation is semantically analyzed and important keywords and context are extracted.

[0528] Step 9:

[0529] The server generates a response using a generative AI model based on the analyzed data.

[0530] What it does: It feeds the analysis results into a generative AI model to generate an appropriate response text, which is then formatted appropriately and prepared to be sent back to the user.

[0531] Step 10:

[0532] The server sends the generated response to the terminal.

[0533] Behavior: Sends the generated response to the device as an HTTP response.

[0534] Step 11:

[0535] The terminal displays the response received from the server to the user.

[0536] Behavior: Displays the received response text on the user interface.

[0537] Specific examples

[0538] Example 1: Consultation regarding stress

[0539] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice" (step 6).

[0540] The terminal transmits the consultation content to the server (step 7).

[0541] The server receives the consultation content and analyzes it using natural language processing (step 8).

[0542] The server generates a response using a generative AI model. "To reduce stress, try daily relaxation techniques, such as meditation, deep breathing, and light exercise" (Step 9).

[0543] The server sends the generated response to the terminal (step 10).

[0544] The terminal displays the response to the user (step 11).

[0545] The above is a detailed flow of processing in the system.

[0546] Example 1

[0547] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0548] It is important to detect and appropriately address mental health issues among employees early. However, due to a lack of environments where employees can easily seek advice, problems are often discovered late. Furthermore, conventional systems require a great deal of time and effort for data collection, preprocessing, and model training, making it difficult to respond in real time. For this reason, there is a need for a system that can provide effective and prompt mental health support.

[0549] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0550] In this invention, the server includes means for acquiring information to be collected, means for cleansing the collected information and converting it into a unified format, means for learning a generative AI model using the preprocessed data and training the model, means for receiving and transmitting consultation content from a user, means for analyzing the received consultation content, means for generating a response using the generative AI model based on the analyzed consultation content, and means for displaying the generated response to the user. This makes it possible to effectively utilize the collected data and provide appropriate responses in real time in response to employee consultations.

[0551] "Information to be collected" refers to the information source that the system is set up to acquire, and in this case, this refers to data such as information provided by the health support center and past consultation cases of peer supporters.

[0552] "Cleansing" is a preprocessing process to remove unnecessary characters and noise from collected data and ensure data quality.

[0553] A "uniform format" is a standardization procedure for converting data into a format suitable for analysis and learning, and is a data format that ensures consistency.

[0554] A "generative AI model" is an artificial intelligence model that learns from collected and preprocessed data and generates appropriate responses based on the user's inquiry.

[0555] "Learning" is the process by which a generative AI model uses training data to improve the accuracy of response generation.

[0556] "Training" is the process of repeatedly learning a generative AI model using collected data.

[0557] "User consultation content" is text data that system users input and send regarding mental health issues and questions.

[0558] "Analysis" refers to the process of analyzing the meaning of the received user's consultation content using natural language processing technology and understanding the intent of the consultation content.

[0559] A "response" is a text message containing advice or information generated by the generative AI model based on the analyzed content of the user's consultation.

[0560] "Display" refers to an operation or function for visually presenting the generated response on the user's terminal.

[0561] This invention relates to an AI peer support system for preventing mental health problems among employees. This system uses a generative AI model to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by health support centers.

[0562] Program processing explanation

[0563] The main processing of this system is carried out by three elements: the server, the terminal, and the user.

[0564] Data collection

[0565] The server acquires the information to be collected. Specifically, it uses API calls to collect information provided by the health support center and past consultation cases by peer supporters. This data is stored in a database such as MongoDB or PostgreSQL.

[0566] Data Preprocessing

[0567] The server cleanses the collected data and converts it into a unified format, using Python scripts to remove unnecessary data, and natural language processing libraries such as NLTK and SpaCy for segmentation and tokenization. Finally, the data is converted into formats such as CSV and Parquet.

[0568] Model learning

[0569] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). The data is split into a training dataset and a test dataset, and training is performed using a deep learning framework such as TensorFlow or PyTorch. Once training is complete, the model is saved.

[0570] Accepting inquiries from users

[0571] The user uses the device to input and send the content of their consultation. For example, they might use a web or mobile application to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device then sends this data to the server as an HTTP request.

[0572] Analyzing received data

[0573] The server analyzes the received consultation content using natural language processing technology. It obtains text data and performs tokenization and entity recognition using NLTK and SpaCy. This allows it to understand the intent of the consultation content.

[0574] Response Generation

[0575] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. For example, a prompt such as "I've been feeling more stressed at work lately, so I'd like some advice" is input into the model, and the generated text data is retrieved. The response generated is, "To reduce stress, it's effective to try daily relaxation techniques. Examples include meditation, deep breathing, and light exercise."

[0576] Response Display

[0577] The device displays the response received from the server to the user, and displays the response message in a text area on a web page or in an application, or in the chatbot's UI, allowing the user to receive appropriate advice.

[0578] Specific examples

[0579] Example 1: Stress consultation

[0580] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[0581] The terminal sends this input to the server.

[0582] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[0583] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0584] The terminal displays the generated response to the user, and the user can receive specific advice.

[0585] Example 2: Relationship advice

[0586] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[0587] The terminal sends this input to the server.

[0588] The server receives the consultation content, analyzes it using natural language processing technology, and prepares the input data to generate an appropriate response.

[0589] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0590] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0591] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[0592] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0593] Step 1:

[0594] The server uses API calls to obtain the information to be collected. Specifically, it obtains data including past consultation cases from health support centers and peer supporters from the API endpoint. The input is the API endpoint URL and authentication information, and the output is data in JSON format. This data is stored in a database such as MongoDB or PostgreSQL.

[0595] Step 2:

[0596] The server cleanses the collected data and converts it into a unified format. Specifically, it uses Python scripts to remove unnecessary characters and noise, and performs word segmentation and tokenization using natural language processing libraries such as NLTK and SpaCy. The input is the collected JSON data, and the output is structured data in CSV or Parquet format.

[0597] Step 3:

[0598] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). Specifically, the data is divided into a training set and a test set, and the model is trained using a deep learning framework such as TensorFlow or PyTorch. The input is the preprocessed CSV or Parquet data, and the output is a trained generative AI model. This model is stored on the server.

[0599] Step 4:

[0600] The user uses a terminal to input and send the content of their consultation. Specifically, they enter something like "I've been feeling more stressed at work lately, so I'd like some advice" into a form on a web or mobile application, and then press the send button. The input is text data entered by the user, and the output is sent to the server as an HTTP request.

[0601] Step 5:

[0602] The server analyzes the consultation content received from the device. Specifically, it takes the received text data and performs tokenization and entity recognition using natural language processing libraries such as NLTK and SpaCy. The input is the received text data, and the output is the tokenized analyzed data. This analysis allows the intent and gist of the consultation content to be understood.

[0603] Step 6:

[0604] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. Specifically, the analyzed data is input into the model as a prompt sentence, and the generated text data is obtained. The input is the tokenized analyzed data and the prompt sentence, and the output is the generated response text. In this example, a response such as "To reduce stress, it is effective to try daily relaxation methods. Examples include meditation, deep breathing, and light exercise" is generated.

[0605] Step 7:

[0606] The terminal displays the response from the server to the user. Specifically, it displays the received response text in a text area on a web page or in an application, or in the chatbot's UI. The input is the response text received from the server, and the output is a text display that the user can check on the screen.

[0607] (Application example 1)

[0608] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0609] Mental health problems among employees are a serious problem that can lead to reduced work efficiency and a worsening work environment. Particularly in brick-and-mortar stores, where employees often have direct contact with customers, accumulating stress and interpersonal problems can negatively impact how they treat customers. The present invention aims to provide an environment where employees working in brick-and-mortar stores can easily receive mental care, thereby improving the work environment and the mental health of employees.

[0610] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0611] In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing consultation content from a user, a means for generating a response using the generative AI model based on the analyzed consultation content, and a means for displaying the generated response on a terminal used by an employee at a physical store. This enables employees working at physical stores to easily seek mental health consultation and receive appropriate advice.

[0612] "Data collection methods" are methods for collecting information provided by wellness centers and past consultation cases of peer supporters across the country.

[0613] The "preprocessing means" is a means for cleansing the collected data, removing unnecessary characters and noise, and converting the data into a unified format.

[0614] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to employee inquiries.

[0615] The "means for receiving the consultation content from the user" is a means for transmitting the consultation content input by the employee to the terminal to the server and receiving it.

[0616] The "means for analyzing" refers to a means for analyzing the received consultation content using natural language processing technology and understanding the meaning and intent of the content.

[0617] The "means for generating a response" is a means for generating an appropriate response using a generative AI model based on the analyzed consultation content.

[0618] The "means for displaying the generated response" is a means for displaying the response sent from the server on the terminal used by the employee.

[0619] "Brick and mortar store" refers to a physical sales or service location where employees interact directly with customers.

[0620] This invention relates to an AI peer support system to prevent employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers. This system is designed especially for use by employees working in brick-and-mortar stores.

[0621] The system is configured as follows:

[0622] Hardware and Software

[0623] The system's main hardware consists of a smartphone or tablet connected to a server. The server is responsible for data collection, preprocessing, training the generative AI model, and generating responses. The smartphone or tablet acts as an interface where users input their inquiries and view responses from the server. The software primarily uses OpenAI's API, Python scripts, and the HTTP request library.

[0624] Data collection

[0625] The server collects information provided by the wellness center and past consultation cases of peer supporters nationwide through API calls or database queries. This data is used as the basis for user consultations.

[0626] Data Preprocessing

[0627] The collected data is cleansed by the server to remove unnecessary characters and noise, and then converted into a structured format using techniques such as tokenization and word segmentation. This preprocessing prepares the data in a format that can be analyzed.

[0628] Model learning

[0629] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0630] Receiving and analyzing inquiries from users

[0631] Users input and submit their consultation details using a device such as a smartphone or tablet. The device then sends this input data to a server, which then analyzes the received consultation details using natural language processing technology.

[0632] Response generation and display

[0633] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is sent to the device and displayed to the user, allowing the user to obtain appropriate countermeasures and advice.

[0634] Specific examples

[0635] Example 1: Stress consultation

[0636] User: I'm under a lot of pressure at work, how can I reduce it?

[0637] Prompt: User wants to know: I'm under a lot of pressure at work. How can I alleviate it?\nGive appropriate advice:

[0638] Example 2: Relationship Advice

[0639] User: I'm having some disagreements with a coworker and it's been a strained relationship. Is there anything I can do to improve this?

[0640] Prompt: User's question: I'm having disagreements with a colleague and it's been a strained relationship. Is there a way to improve this?\nPlease provide appropriate advice:

[0641] In this way, the AI ​​peer supporter system of the present invention allows store employees to easily seek mental health advice and receive appropriate advice in real time, enabling them to smoothly carry out their daily work and improve the quality of customer service.

[0642] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0643] Step 1:

[0644] The server collects information provided by wellness centers and past consultation cases of peer supporters nationwide using API calls or database queries. The input is the API endpoints for wellness centers and peer support data, and the output is the collected data.

[0645] Step 2:

[0646] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format using tokenization, word segmentation, etc. The input is the collected data, and the output is the preprocessed data.

[0647] Step 3:

[0648] The server uses the preprocessed data to learn and train a generative AI model (e.g., GPT-4), which improves the model's accuracy. The input is the preprocessed data, and the output is a trained generative AI model.

[0649] Step 4:

[0650] The user inputs the content of the consultation using a device such as a smartphone or tablet and sends the input to the server. The input is the content of the user's consultation, and the output is the data sent to the server.

[0651] Step 5:

[0652] The server analyzes the received consultation content using natural language processing technology (e.g., NLP technology) to understand the meaning and intent of the content. The input is the consultation content from the user, and the output is the analyzed consultation content.

[0653] Step 6:

[0654] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. The input is the analyzed consultation content and the generative AI model, and the output is the generated response.

[0655] Step 7:

[0656] The terminal displays the response sent by the server to the user. The input is the response generated, and the output is the advice or action displayed to the user.

[0657] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0658] This invention relates to an AI peer support system for preventing mental health problems among employees. This system combines generative AI and an emotion engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[0659] The program of this system performs the following processing.

[0660] 1. Data Collection

[0661] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country using API calls and database queries.

[0662] 2. Data Preprocessing

[0663] The server cleanses the collected data, removing unnecessary characters and noise, and then converts the data into a structured format through segmentation and tokenization. This preprocessing prepares the data in a form that can be analyzed.

[0664] 3. Model Training

[0665] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0666] 4. Receiving and analyzing consultation content

[0667] The user inputs the content of their consultation into the terminal and sends it. The terminal then sends the input data to the server. The server then analyzes the received content using natural language processing technology to understand the meaning and intent of the content.

[0668] 5. Emotion analysis

[0669] The server uses an emotion engine to analyze the emotion from the user's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response based on the analysis.

[0670] 6. Response Generation

[0671] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. This response is generated by inputting the consultation content and emotional data into the model and obtaining the generated text data.

[0672] 7. Response Display

[0673] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[0674] Specific examples

[0675] Example 1: Stress consultation

[0676] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice."

[0677] The terminal sends this input to the server.

[0678] The server receives the consultation content and analyzes it using natural language processing technology.

[0679] The server uses an emotion engine to analyze the user's emotions and extracts emotion data indicating "increasing stress."

[0680] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0681] The terminal displays the generated response to the user, and the user can receive specific advice.

[0682] Example 2: Relationship advice

[0683] The user uses the terminal to input, "I'm having trouble with my relationships at work. What should I do?"

[0684] The terminal sends this input to the server.

[0685] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning and intent.

[0686] The server uses an emotion engine to analyze the user's emotions and extracts data indicating that "anxiety" is strongly felt.

[0687] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0688] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0689] In this way, the AI ​​peer supporter system of the present invention, combined with an emotion engine, provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice. This makes it possible to detect signs of mental illness early and provide appropriate countermeasures, with the aim of realizing a workplace environment that is healthy both physically and mentally.

[0690] The processing flow will be explained below.

[0691] Processing Steps

[0692] Step 1:

[0693] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[0694] How it works: The server uses API calls and database queries to gather the necessary data, which is then stored locally or in cloud storage.

[0695] Step 2:

[0696] The server pre-processes the collected data.

[0697] How it works: The server cleanses the text data, removing noise and unnecessary characters, and then segments and tokenizes the data to make it parseable.

[0698] Step 3:

[0699] The server formats the preprocessed data into a training dataset.

[0700] What it does: Splits the dataset into training and validation sets, and performs sampling and data augmentation as needed to keep the data balanced.

[0701] Step 4:

[0702] The server trains a generative AI model (e.g., GPT-4).

[0703] How it works: The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model using the preprocessed dataset, applying adaptive learning rates and other optimization techniques to improve the model's accuracy.

[0704] Step 5:

[0705] The server prepares to receive the consultation content from the user.

[0706] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[0707] Step 6:

[0708] The user uses the terminal to input and transmit the consultation contents.

[0709] Operation: The user enters the content of the consultation in the text field and clicks the "Send" button. The device sends the consultation data to the server.

[0710] Step 7:

[0711] The terminal transmits the user's input data to the server.

[0712] Operation: The text data of the consultation is sent to the server using an HTTP request.

[0713] Step 8:

[0714] The server analyzes the received consultation content.

[0715] How it works: The server uses natural language processing technology to analyze the meaning and intent of the consultation, extracting important keywords and context.

[0716] Step 9:

[0717] The server uses an emotion engine to analyze the emotion from the content of the user's consultation.

[0718] How it works: The emotion engine detects emotion types (e.g., joy, sadness, fear) and their intensity from text data. This emotion data is fed into a generative AI model.

[0719] Step 10:

[0720] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data.

[0721] How it works: The server inputs the consultation content and emotional data into the generative AI model, receives the generated text response, and adjusts the tone and content of the response based on the emotional data.

[0722] Step 11:

[0723] The server sends the generated response to the terminal.

[0724] Operation: The generated response data is sent back to the terminal via an HTTP response.

[0725] Step 12:

[0726] The terminal displays the response received from the server to the user.

[0727] Behavior: Displays the received response text on the user's screen.

[0728] Specific examples

[0729] Example 1: Stress consultation

[0730] Step 1:

[0731] The server collects information provided by the wellness center and past consultation cases of peer supporters.

[0732] Step 2:

[0733] The server cleanses the collected data and puts it into an analyzable format.

[0734] Step 3:

[0735] The server formats the preprocessed data into a training dataset and trains a generative AI model.

[0736] Step 4:

[0737] The user uses the terminal to input "I've been feeling more and more stressed at work recently, so I'd like some advice," and then sends it.

[0738] Step 5:

[0739] The terminal transmits the user's input data to the server.

[0740] Step 6:

[0741] The server receives the consultation content and analyzes it using natural language processing technology.

[0742] Step 7:

[0743] The server uses an emotion engine to extract emotion data such as "stress is increasing."

[0744] Step 8:

[0745] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0746] Step 9:

[0747] The server sends the generated response to the terminal.

[0748] Step 10:

[0749] The terminal displays the response to the user, and the user can receive specific advice.

[0750] The above is the specific processing flow of a system that combines an emotion engine.

[0751] Example 2

[0752] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0753] To prevent employees from developing mental health problems, it is necessary to provide an environment where employees can easily seek advice. However, conventional systems lack the data necessary to generate appropriate responses and perform insufficient emotion analysis, making it difficult to provide personalized responses to users. As a result, employees may not receive the support they need, which could worsen their mental health.

[0754] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing the consultation content from the user, a sentiment analysis means for analyzing the sentiment from the user's consultation content and generating sentiment data for response generation, a means for generating a response using the generative AI model based on the analyzed consultation content and sentiment data, and a means for displaying the generated response to the user. This makes it possible to provide a personalized response tailored to the user's emotional state and realize an environment where employees can easily seek consultation.

[0755] "Data collection methods" are methods for collecting necessary data from past consultation cases of wellness centers and supporters.

[0756] "Preprocessing means" refers to means for cleansing collected data, removing unnecessary data and noise, and converting data into a unified format.

[0757] A "means for training using a generative AI model" is a means for training a generative AI model using preprocessed data.

[0758] The "means for receiving and analyzing the consultation content" is a means for receiving the consultation content from the user and analyzing the content using natural language processing technology.

[0759] The "emotion analysis means" is a means for analyzing emotions from the content of the user's consultation and generating emotion data.

[0760] The "means for generating a response" refers to a means for generating an appropriate response using a generative AI model based on the analyzed consultation content and emotional data.

[0761] The "means for displaying a response" is a means for displaying the generated response to the user.

[0762] A "wellness center" is a specialized institution that provides information on maintaining employee health and mental care.

[0763] A "support person" is a person or expert who serves as a source of advice for employees.

[0764] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system combines generative AI and an emotion analysis engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by specialist institutions.

[0765] The system's program is composed of the following hardware and software: The server uses Python libraries (e.g., requests and SQLAlchemy) to collect the necessary information from the wellness center and supporter databases. Additionally, as a preprocessing method, the pandas and nltk libraries are used to cleanse the collected data and convert it into a unified format.

[0766] The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to learn and train data using a generative AI model (e.g., GPT-4). It receives the user's consultation and analyzes it using natural language processing techniques (e.g., spaCy or BERT). Furthermore, it uses a sentiment analysis engine (e.g., TextBlob or VADER) to analyze the user's sentiment and generate sentiment data for response generation.

[0767] Let's explain with a concrete example. A user uses a device to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device sends this input to the server. The server receives the consultation content and performs natural language processing using spaCy to analyze the meaning and intent of the sentence. Next, the server performs sentiment analysis using the VADER library and extracts the emotional data "stress is increasing." The server uses a GPT-4 model to generate an appropriate response based on the following prompt sentence:

[0768] "The user is asking for advice about work stress. Please provide appropriate advice."

[0769] The terminal then receives the generated response and displays it to the user.

[0770] Similarly, if a user types "I'm having trouble with relationships at work. What should I do?" into their device, the same process will occur. The server will analyze the consultation content using spaCy and perform sentiment analysis using TextBlob. If "anxiety" is strongly detected, the server will use the GPT-4 model to input the following prompt:

[0771] "The user is asking for advice about relationships at work. Please provide advice on how to improve them."

[0772] The generated response is then received by the terminal and displayed to the user.

[0773] This system provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice, making it possible to catch early signs of mental health problems and provide appropriate countermeasures.

[0774] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0775] Step 1: Data collection

[0776] The server collects information from the wellness center and supporter databases. It uses Python's requests library to send API requests and retrieves data in JSON format. It also extracts supporter past consultation cases from the database using SQL queries. The input is the wellness center API and database queries, and the output is the collected raw data.

[0777] Step 2: Data Preprocessing

[0778] The server cleanses the collected data to remove noise and unnecessary characters. Specifically, it creates a data frame using Python's pandas library and filters out unnecessary data using regular expressions. It then uses nltk to tokenize the data and convert it into a unified format (e.g., JSON). The input is the collected raw data, and the output is the preprocessed data.

[0779] Step 3: Training the generative AI model

[0780] The server trains a generative AI model (e.g., GPT-4) using the preprocessed data. Specifically, it builds a model using a deep learning framework (e.g., TensorFlow or PyTorch) and runs training using the preprocessed data. The input is the preprocessed data, and the output is a trained generative AI model.

[0781] Step 4: Receiving consultation details

[0782] The user enters the content of their consultation into an input field on the terminal. The terminal then sends the entered content to the server. Specifically, an HTTP POST request is used. The input is the user's consultation content, and the output is the data to be sent to the server.

[0783] Step 5: Analysis of consultation content

[0784] The server analyzes the received consultation content and understands the meaning and intent of the text using natural language processing technology (e.g., spaCy or BERT). The input is the received consultation content, and the output is the analyzed meaning and intent of the text.

[0785] Step 6: Sentiment Analysis

[0786] The server uses an emotion analysis engine (e.g., TextBlob or VADER) to extract emotion data from the analyzed consultation content. It measures the type of emotion (joy, sadness, anger, etc.) and its intensity. The input is the analyzed meaning and intent, and the output is emotion data.

[0787] Step 7: Response Generation

[0788] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data. Specifically, it inputs a prompt sentence (e.g., "The user is asking for advice about work stress. Please provide appropriate advice.") into the generative AI model and obtains the generated response. The input is the analyzed consultation content and emotional data, and the output is the generated response.

[0789] Step 8: Display the response

[0790] The terminal receives the response sent by the server and displays it to the user, typically using a text view or a dialog box to display the response. The input is the response generated, and the output is what is displayed to the user.

[0791] (Application example 2)

[0792] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0793] There is a need for systems that can detect mental stress and mental disorders among employees early and deal with them promptly and appropriately. However, existing employee support systems lack effective emotion analysis and personalized responses, and do not provide an environment where employees can easily seek advice. Furthermore, there are many situations where an appropriate response in real time is required, so technology to solve this issue is needed.

[0794] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a preprocessing means for cleansing collected data and converting it into a unified format; a means for learning the preprocessed data using a generative AI model and training the model; a means for receiving and analyzing consultation content from a user; a means for generating a response using the generative AI model based on the analyzed consultation content; a means for displaying the generated response to the user; a means for an employee to directly input the consultation content using a smart device; an emotion analysis means for analyzing emotions based on the input consultation content; and a means for generating the emotion analysis result as a prompt sentence for the generative AI model. This enables employees to easily receive mental support and enables early detection and appropriate treatment of mental disorders.

[0795] "Data collection means" refers to the means used to collect information such as past consultation cases of employees and information provided by the wellness center.

[0796] "Preprocessing means" refers to means for cleansing collected data and converting it into a unified format.

[0797] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to user inquiries.

[0798] A "training means" is a means for training preprocessed data using a generative AI model.

[0799] The "means for receiving consultation content" is a means for receiving consultation content from employees.

[0800] The "analysis means" is a means for analyzing the received consultation content and understanding its meaning and intent.

[0801] The "response generation means" is a means for generating a personalized response using a generative AI model based on the analyzed consultation content.

[0802] A "display means" is a means for displaying the generated response on the employee's device.

[0803] "Smart devices" are electronic devices such as smartphones and tablets that employees use to input consultation details.

[0804] The "emotion analysis means" is a means for analyzing emotions based on the content of an employee's consultation and obtaining the results.

[0805] The "prompt generation means" is a means for generating prompt sentences for the generative AI model based on the results of emotion analysis.

[0806] This invention relates to an AI peer support system for preventing employee mental health problems, and provides a consultation environment by combining generative AI models and emotion analysis technology. This system creates an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers.

[0807] This system is configured as follows:

[0808] Hardware and Software

[0809] 1. Hardware: smartphones, tablets, servers.

[0810] 2. Software: Python program, OpenAI GPT-4, TextBlob.

[0811] Explanation of main processes

[0812] 1. Data collection methods:

[0813] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[0814] 2. Pretreatment methods:

[0815] The server cleanses the collected data, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format, making it parseable.

[0816] 3. Training methods:

[0817] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4) using a deep learning framework, which performs iterative learning to improve the model's accuracy.

[0818] 4. How to receive consultation details:

[0819] The user inputs and sends the consultation details using a smart device, which then sends the input to the server.

[0820] 5. Analysis methods:

[0821] The server analyzes the received consultation content using natural language processing technology to understand the meaning and intent of the consultation content.

[0822] 6. Emotion analysis means:

[0823] The server uses an emotion engine to analyze emotions from the employee's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response.

[0824] 7. Prompt Generation Method:

[0825] Based on the results of the sentiment analysis, a prompt sentence is generated for the generative AI model. For example, the following prompt sentence is generated:

[0826] text

[0827] User's question: I've been so busy at work lately that I'm feeling stressed.

[0828] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[0829] Provide specific advice to this user.

[0830] 8. Response Generation Methods:

[0831] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. A personalized response is created by inputting a prompt sentence to the generative AI model and obtaining the generated text data.

[0832] 9. Display means:

[0833] The smart device displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate countermeasures or advice.

[0834] Specific examples

[0835] Example 1: Stress consultation

[0836] The user uses the terminal to input, "I've been so busy at work recently that I'm stressed out."

[0837] The smart device sends this input to the server.

[0838] The server receives the consultation content and analyzes it using natural language processing technology.

[0839] The server uses an emotion engine to analyze emotions and extract data indicating "increasing stress."

[0840] The server uses a generative AI model to generate a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0841] The smart device displays the generated response to the user, who can then receive specific advice.

[0842] As a result, the AI ​​peer supporter system of the present invention combines an emotion engine and a generative AI model to provide personalized responses tailored to the user's own emotional state, creating an environment where employees can feel comfortable seeking advice.

[0843] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0844] Step 1:

[0845] Data collection methods

[0846] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[0847] Input: Wellness center and peer supporter data.

[0848] Output: A dataset of collected information.

[0849] Step 2:

[0850] Pretreatment means

[0851] The server cleanses the data collected in step 1, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format.

[0852] Input: Collected information dataset.

[0853] Output: A cleansed and tokenized dataset.

[0854] Step 3:

[0855] Training methods

[0856] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4), using a deep learning framework to iteratively train the model to improve its accuracy.

[0857] Input: The preprocessed dataset.

[0858] Output: A trained generative AI model.

[0859] Step 4:

[0860] Means of receiving consultation details

[0861] The user inputs the consultation details using the smart device, and the terminal transmits this input to the server.

[0862] Input: User's consultation content.

[0863] Output: Consultation content data sent to the server.

[0864] Step 5:

[0865] Analysis means

[0866] The server analyzes the consultation content received in step 4 using natural language processing technology to understand the meaning and intent of the consultation content.

[0867] Input: Received consultation data.

[0868] Output: Parsed consultation data.

[0869] Step 6:

[0870] Emotion analysis means

[0871] The server analyzes emotions using an emotion engine based on the analyzed consultation content, and detects the type and intensity of the emotion.

[0872] Input: Parsed consultation content data.

[0873] Output: Sentiment analysis data.

[0874] Step 7:

[0875] Prompt Generation Method

[0876] The server generates prompt sentences for the generative AI model based on the emotion analysis data.

[0877] Input: Sentiment analysis data.

[0878] Output: The generated prompt statement.

[0879] Examples:

[0880] text

[0881] User's question: I've been so busy at work lately that I'm feeling stressed.

[0882] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[0883] Provide specific advice to this user.

[0884] Step 8:

[0885] Response Generation Method

[0886] The server inputs the prompt sentence into a generative AI model, which then generates an appropriate response.

[0887] Input: The generated prompt statement.

[0888] Output: The generated response text.

[0889] Step 9:

[0890] Display means

[0891] The terminal displays the response sent from the server to the user, allowing the user to receive specific advice and solutions.

[0892] Input: The generated response text.

[0893] Output: The response text displayed to the user.

[0894] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0895] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0896] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0897] [Third embodiment]

[0898] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0899] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0900] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0901] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0902] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0903] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0904] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0905] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0906] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0907] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0908] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0909] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0910] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[0911] The program of this system performs the following processing.

[0912] 1. Data Collection

[0913] The server acquires the data to be collected, i.e., information provided by wellness centers and past consultation cases of peer supporters nationwide. This collection is performed using API calls and database queries.

[0914] 2. Data Preprocessing

[0915] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format by segmenting and tokenizing it. This preprocessing prepares the data in a format that can be analyzed.

[0916] 3. Model Training

[0917] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[0918] 4. Receiving inquiries from users

[0919] The user inputs the details of the consultation into the terminal and transmits them. The terminal transmits the input data to the server.

[0920] 5. Data Analysis

[0921] The server uses natural language processing technology to analyze the received consultation content and understand its meaning and intent. This analysis prepares input data for generating an appropriate response.

[0922] 6. Response Generation

[0923] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is generated by inputting the consultation content into the model and obtaining the generated text data.

[0924] 7. Response Display

[0925] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[0926] Specific examples

[0927] Example 1: Stress consultation

[0928] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[0929] The terminal sends this input to the server.

[0930] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[0931] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[0932] The terminal displays the generated response to the user, and the user can receive specific advice.

[0933] Example 2: Relationship advice

[0934] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[0935] The terminal sends this input to the server.

[0936] The server receives the consultation content, analyzes it using natural language processing, and prepares the input data to generate an appropriate response.

[0937] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[0938] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[0939] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[0940] The processing flow will be explained below.

[0941] Processing Steps

[0942] Step 1:

[0943] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[0944] How it works: It uses API calls to retrieve information from the wellness center database, and also uses database queries to gather peer supporter consultation cases stored locally or in cloud storage.

[0945] Step 2:

[0946] The server pre-processes the collected data.

[0947] How it works: Text data is cleansed to remove noise and unnecessary characters, then natural language processing techniques are used to segment and tokenize it, converting it into a structured format.

[0948] Step 3:

[0949] The server formats the preprocessed data into a training dataset.

[0950] How it works: Split the dataset into training and validation sets, and format them into JSON or CSV formats. Consider the balance of the data, and perform sampling or data augmentation if there is an imbalance.

[0951] Step 4:

[0952] The server trains a generative AI model (e.g., GPT-4).

[0953] How it works: The preprocessed dataset is fed into a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model. Adaptive learning rates and other optimization techniques are applied to improve the model's accuracy.

[0954] Step 5:

[0955] The server prepares to receive the consultation content from the user.

[0956] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[0957] Step 6:

[0958] The user uses the terminal to input and transmit the consultation contents.

[0959] Operation: Enter the content of your inquiry into the text field and click the "Send" button. This will cause the device to send the content of your inquiry to the server.

[0960] Step 7:

[0961] The terminal transmits the user's input data to the server.

[0962] Operation: The text data of the consultation is sent to the server via an HTTP request.

[0963] Step 8:

[0964] The server analyzes the received consultation content.

[0965] How it works: Using natural language processing technology, the text of the consultation is semantically analyzed and important keywords and context are extracted.

[0966] Step 9:

[0967] The server generates a response using a generative AI model based on the analyzed data.

[0968] What it does: It feeds the analysis results into a generative AI model to generate an appropriate response text, which is then formatted appropriately and prepared to be sent back to the user.

[0969] Step 10:

[0970] The server sends the generated response to the terminal.

[0971] Behavior: Sends the generated response to the device as an HTTP response.

[0972] Step 11:

[0973] The terminal displays the response received from the server to the user.

[0974] Behavior: Displays the received response text on the user interface.

[0975] Specific examples

[0976] Example 1: Consultation regarding stress

[0977] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice" (step 6).

[0978] The terminal transmits the consultation content to the server (step 7).

[0979] The server receives the consultation content and analyzes it using natural language processing (step 8).

[0980] The server generates a response using a generative AI model. "To reduce stress, try daily relaxation techniques, such as meditation, deep breathing, and light exercise" (Step 9).

[0981] The server sends the generated response to the terminal (step 10).

[0982] The terminal displays the response to the user (step 11).

[0983] The above is a detailed flow of processing in the system.

[0984] Example 1

[0985] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0986] It is important to detect and appropriately address mental health issues among employees early. However, due to a lack of environments where employees can easily seek advice, problems are often discovered late. Furthermore, conventional systems require a great deal of time and effort for data collection, preprocessing, and model training, making it difficult to respond in real time. For this reason, there is a need for a system that can provide effective and prompt mental health support.

[0987] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0988] In this invention, the server includes means for acquiring information to be collected, means for cleansing the collected information and converting it into a unified format, means for learning a generative AI model using the preprocessed data and training the model, means for receiving and transmitting consultation content from a user, means for analyzing the received consultation content, means for generating a response using the generative AI model based on the analyzed consultation content, and means for displaying the generated response to the user. This makes it possible to effectively utilize the collected data and provide appropriate responses in real time in response to employee consultations.

[0989] "Information to be collected" refers to the information source that the system is set up to acquire, and in this case, this refers to data such as information provided by the health support center and past consultation cases of peer supporters.

[0990] "Cleansing" is a preprocessing process to remove unnecessary characters and noise from collected data and ensure data quality.

[0991] A "uniform format" is a standardization procedure for converting data into a format suitable for analysis and learning, and is a data format that ensures consistency.

[0992] A "generative AI model" is an artificial intelligence model that learns from collected and preprocessed data and generates appropriate responses based on the user's inquiry.

[0993] "Learning" is the process by which a generative AI model uses training data to improve the accuracy of response generation.

[0994] "Training" is the process of repeatedly learning a generative AI model using collected data.

[0995] "User consultation content" is text data that system users input and send regarding mental health issues and questions.

[0996] "Analysis" refers to the process of analyzing the meaning of the received user's consultation content using natural language processing technology and understanding the intent of the consultation content.

[0997] A "response" is a text message containing advice or information generated by the generative AI model based on the analyzed content of the user's consultation.

[0998] "Display" refers to an operation or function for visually presenting the generated response on the user's terminal.

[0999] This invention relates to an AI peer support system for preventing mental health problems among employees. This system uses a generative AI model to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by health support centers.

[1000] Program processing explanation

[1001] The main processing of this system is carried out by three elements: the server, the terminal, and the user.

[1002] Data collection

[1003] The server acquires the information to be collected. Specifically, it uses API calls to collect information provided by the health support center and past consultation cases by peer supporters. This data is stored in a database such as MongoDB or PostgreSQL.

[1004] Data Preprocessing

[1005] The server cleanses the collected data and converts it into a unified format, using Python scripts to remove unnecessary data, and natural language processing libraries such as NLTK and SpaCy for segmentation and tokenization. Finally, the data is converted into formats such as CSV and Parquet.

[1006] Model learning

[1007] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). The data is split into a training dataset and a test dataset, and training is performed using a deep learning framework such as TensorFlow or PyTorch. Once training is complete, the model is saved.

[1008] Accepting inquiries from users

[1009] The user uses the device to input and send the content of their consultation. For example, they might use a web or mobile application to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device then sends this data to the server as an HTTP request.

[1010] Analyzing received data

[1011] The server analyzes the received consultation content using natural language processing technology. It obtains text data and performs tokenization and entity recognition using NLTK and SpaCy. This allows it to understand the intent of the consultation content.

[1012] Response Generation

[1013] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. For example, a prompt such as "I've been feeling more stressed at work lately, so I'd like some advice" is input into the model, and the generated text data is retrieved. The response generated is, "To reduce stress, it's effective to try daily relaxation techniques. Examples include meditation, deep breathing, and light exercise."

[1014] Response Display

[1015] The device displays the response received from the server to the user, and displays the response message in a text area on a web page or in an application, or in the chatbot's UI, allowing the user to receive appropriate advice.

[1016] Specific examples

[1017] Example 1: Stress consultation

[1018] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[1019] The terminal sends this input to the server.

[1020] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[1021] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1022] The terminal displays the generated response to the user, and the user can receive specific advice.

[1023] Example 2: Relationship advice

[1024] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[1025] The terminal sends this input to the server.

[1026] The server receives the consultation content, analyzes it using natural language processing technology, and prepares the input data to generate an appropriate response.

[1027] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[1028] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[1029] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[1030] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1031] Step 1:

[1032] The server uses API calls to obtain the information to be collected. Specifically, it obtains data including past consultation cases from health support centers and peer supporters from the API endpoint. The input is the API endpoint URL and authentication information, and the output is data in JSON format. This data is stored in a database such as MongoDB or PostgreSQL.

[1033] Step 2:

[1034] The server cleanses the collected data and converts it into a unified format. Specifically, it uses Python scripts to remove unnecessary characters and noise, and performs word segmentation and tokenization using natural language processing libraries such as NLTK and SpaCy. The input is the collected JSON data, and the output is structured data in CSV or Parquet format.

[1035] Step 3:

[1036] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). Specifically, the data is divided into a training set and a test set, and the model is trained using a deep learning framework such as TensorFlow or PyTorch. The input is the preprocessed CSV or Parquet data, and the output is a trained generative AI model. This model is stored on the server.

[1037] Step 4:

[1038] The user uses a terminal to input and send the content of their consultation. Specifically, they enter something like "I've been feeling more stressed at work lately, so I'd like some advice" into a form on a web or mobile application, and then press the send button. The input is text data entered by the user, and the output is sent to the server as an HTTP request.

[1039] Step 5:

[1040] The server analyzes the consultation content received from the device. Specifically, it takes the received text data and performs tokenization and entity recognition using natural language processing libraries such as NLTK and SpaCy. The input is the received text data, and the output is the tokenized analyzed data. This analysis allows the intent and gist of the consultation content to be understood.

[1041] Step 6:

[1042] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. Specifically, the analyzed data is input into the model as a prompt sentence, and the generated text data is obtained. The input is the tokenized analyzed data and the prompt sentence, and the output is the generated response text. In this example, a response such as "To reduce stress, it is effective to try daily relaxation methods. Examples include meditation, deep breathing, and light exercise" is generated.

[1043] Step 7:

[1044] The terminal displays the response from the server to the user. Specifically, it displays the received response text in a text area on a web page or in an application, or in the chatbot's UI. The input is the response text received from the server, and the output is a text display that the user can check on the screen.

[1045] (Application example 1)

[1046] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1047] Mental health problems among employees are a serious problem that can lead to reduced work efficiency and a worsening work environment. Particularly in brick-and-mortar stores, where employees often have direct contact with customers, accumulating stress and interpersonal problems can negatively impact how they treat customers. The present invention aims to provide an environment where employees working in brick-and-mortar stores can easily receive mental care, thereby improving the work environment and the mental health of employees.

[1048] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1049] In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing consultation content from a user, a means for generating a response using the generative AI model based on the analyzed consultation content, and a means for displaying the generated response on a terminal used by an employee at a physical store. This enables employees working at physical stores to easily seek mental health consultation and receive appropriate advice.

[1050] "Data collection methods" are methods for collecting information provided by wellness centers and past consultation cases of peer supporters across the country.

[1051] The "preprocessing means" is a means for cleansing the collected data, removing unnecessary characters and noise, and converting the data into a unified format.

[1052] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to employee inquiries.

[1053] The "means for receiving the consultation content from the user" is a means for transmitting the consultation content input by the employee to the terminal to the server and receiving it.

[1054] The "means for analyzing" refers to a means for analyzing the received consultation content using natural language processing technology and understanding the meaning and intent of the content.

[1055] The "means for generating a response" is a means for generating an appropriate response using a generative AI model based on the analyzed consultation content.

[1056] The "means for displaying the generated response" is a means for displaying the response sent from the server on the terminal used by the employee.

[1057] "Brick and mortar store" refers to a physical sales or service location where employees interact directly with customers.

[1058] This invention relates to an AI peer support system to prevent employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers. This system is designed especially for use by employees working in brick-and-mortar stores.

[1059] The system is configured as follows:

[1060] Hardware and Software

[1061] The system's main hardware consists of a smartphone or tablet connected to a server. The server is responsible for data collection, preprocessing, training the generative AI model, and generating responses. The smartphone or tablet acts as an interface where users input their inquiries and view responses from the server. The software primarily uses OpenAI's API, Python scripts, and the HTTP request library.

[1062] Data collection

[1063] The server collects information provided by the wellness center and past consultation cases of peer supporters nationwide through API calls or database queries. This data is used as the basis for user consultations.

[1064] Data Preprocessing

[1065] The collected data is cleansed by the server to remove unnecessary characters and noise, and then converted into a structured format using techniques such as tokenization and word segmentation. This preprocessing prepares the data in a format that can be analyzed.

[1066] Model learning

[1067] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[1068] Receiving and analyzing inquiries from users

[1069] Users input and submit their consultation details using a device such as a smartphone or tablet. The device then sends this input data to a server, which then analyzes the received consultation details using natural language processing technology.

[1070] Response generation and display

[1071] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is sent to the device and displayed to the user, allowing the user to obtain appropriate countermeasures and advice.

[1072] Specific examples

[1073] Example 1: Stress consultation

[1074] User: I'm under a lot of pressure at work, how can I reduce it?

[1075] Prompt: User wants to know: I'm under a lot of pressure at work. How can I alleviate it?\nGive appropriate advice:

[1076] Example 2: Relationship Advice

[1077] User: I'm having some disagreements with a coworker and it's been a strained relationship. Is there anything I can do to improve this?

[1078] Prompt: User's question: I'm having disagreements with a colleague and it's been a strained relationship. Is there a way to improve this?\nPlease provide appropriate advice:

[1079] In this way, the AI ​​peer supporter system of the present invention allows store employees to easily seek mental health advice and receive appropriate advice in real time, enabling them to smoothly carry out their daily work and improve the quality of customer service.

[1080] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1081] Step 1:

[1082] The server collects information provided by wellness centers and past consultation cases of peer supporters nationwide using API calls or database queries. The input is the API endpoints for wellness centers and peer support data, and the output is the collected data.

[1083] Step 2:

[1084] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format using tokenization, word segmentation, etc. The input is the collected data, and the output is the preprocessed data.

[1085] Step 3:

[1086] The server uses the preprocessed data to learn and train a generative AI model (e.g., GPT-4), which improves the model's accuracy. The input is the preprocessed data, and the output is a trained generative AI model.

[1087] Step 4:

[1088] The user inputs the content of the consultation using a device such as a smartphone or tablet and sends the input to the server. The input is the content of the user's consultation, and the output is the data sent to the server.

[1089] Step 5:

[1090] The server analyzes the received consultation content using natural language processing technology (e.g., NLP technology) to understand the meaning and intent of the content. The input is the consultation content from the user, and the output is the analyzed consultation content.

[1091] Step 6:

[1092] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. The input is the analyzed consultation content and the generative AI model, and the output is the generated response.

[1093] Step 7:

[1094] The terminal displays the response sent by the server to the user. The input is the response generated, and the output is the advice or action displayed to the user.

[1095] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1096] This invention relates to an AI peer support system for preventing mental health problems among employees. This system combines generative AI and an emotion engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[1097] The program of this system performs the following processing.

[1098] 1. Data Collection

[1099] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country using API calls and database queries.

[1100] 2. Data Preprocessing

[1101] The server cleanses the collected data, removing unnecessary characters and noise, and then converts the data into a structured format through segmentation and tokenization. This preprocessing prepares the data in a form that can be analyzed.

[1102] 3. Model Training

[1103] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[1104] 4. Receiving and analyzing consultation content

[1105] The user inputs the content of their consultation into the terminal and sends it. The terminal then sends the input data to the server. The server then analyzes the received content using natural language processing technology to understand the meaning and intent of the content.

[1106] 5. Emotion analysis

[1107] The server uses an emotion engine to analyze the emotion from the user's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response based on the analysis.

[1108] 6. Response Generation

[1109] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. This response is generated by inputting the consultation content and emotional data into the model and obtaining the generated text data.

[1110] 7. Response Display

[1111] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[1112] Specific examples

[1113] Example 1: Stress consultation

[1114] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice."

[1115] The terminal sends this input to the server.

[1116] The server receives the consultation content and analyzes it using natural language processing technology.

[1117] The server uses an emotion engine to analyze the user's emotions and extracts emotion data indicating "increasing stress."

[1118] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1119] The terminal displays the generated response to the user, and the user can receive specific advice.

[1120] Example 2: Relationship advice

[1121] The user uses the terminal to input, "I'm having trouble with my relationships at work. What should I do?"

[1122] The terminal sends this input to the server.

[1123] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning and intent.

[1124] The server uses an emotion engine to analyze the user's emotions and extracts data indicating that "anxiety" is strongly felt.

[1125] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[1126] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[1127] In this way, the AI ​​peer supporter system of the present invention, combined with an emotion engine, provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice. This makes it possible to detect signs of mental illness early and provide appropriate countermeasures, with the aim of realizing a workplace environment that is healthy both physically and mentally.

[1128] The processing flow will be explained below.

[1129] Processing Steps

[1130] Step 1:

[1131] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[1132] How it works: The server uses API calls and database queries to gather the necessary data, which is then stored locally or in cloud storage.

[1133] Step 2:

[1134] The server pre-processes the collected data.

[1135] How it works: The server cleanses the text data, removing noise and unnecessary characters, and then segments and tokenizes the data to make it parseable.

[1136] Step 3:

[1137] The server formats the preprocessed data into a training dataset.

[1138] What it does: Splits the dataset into training and validation sets, and performs sampling and data augmentation as needed to keep the data balanced.

[1139] Step 4:

[1140] The server trains a generative AI model (e.g., GPT-4).

[1141] How it works: The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model using the preprocessed dataset, applying adaptive learning rates and other optimization techniques to improve the model's accuracy.

[1142] Step 5:

[1143] The server prepares to receive the consultation content from the user.

[1144] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[1145] Step 6:

[1146] The user uses the terminal to input and transmit the consultation contents.

[1147] Operation: The user enters the content of the consultation in the text field and clicks the "Send" button. The device sends the consultation data to the server.

[1148] Step 7:

[1149] The terminal transmits the user's input data to the server.

[1150] Operation: The text data of the consultation is sent to the server using an HTTP request.

[1151] Step 8:

[1152] The server analyzes the received consultation content.

[1153] How it works: The server uses natural language processing technology to analyze the meaning and intent of the consultation, extracting important keywords and context.

[1154] Step 9:

[1155] The server uses an emotion engine to analyze the emotion from the content of the user's consultation.

[1156] How it works: The emotion engine detects emotion types (e.g., joy, sadness, fear) and their intensity from text data. This emotion data is fed into a generative AI model.

[1157] Step 10:

[1158] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data.

[1159] How it works: The server inputs the consultation content and emotional data into the generative AI model, receives the generated text response, and adjusts the tone and content of the response based on the emotional data.

[1160] Step 11:

[1161] The server sends the generated response to the terminal.

[1162] Operation: The generated response data is sent back to the terminal via an HTTP response.

[1163] Step 12:

[1164] The terminal displays the response received from the server to the user.

[1165] Behavior: Displays the received response text on the user's screen.

[1166] Specific examples

[1167] Example 1: Stress consultation

[1168] Step 1:

[1169] The server collects information provided by the wellness center and past consultation cases of peer supporters.

[1170] Step 2:

[1171] The server cleanses the collected data and puts it into an analyzable format.

[1172] Step 3:

[1173] The server formats the preprocessed data into a training dataset and trains a generative AI model.

[1174] Step 4:

[1175] The user uses the terminal to input "I've been feeling more and more stressed at work recently, so I'd like some advice," and then sends it.

[1176] Step 5:

[1177] The terminal transmits the user's input data to the server.

[1178] Step 6:

[1179] The server receives the consultation content and analyzes it using natural language processing technology.

[1180] Step 7:

[1181] The server uses an emotion engine to extract emotion data such as "stress is increasing."

[1182] Step 8:

[1183] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1184] Step 9:

[1185] The server sends the generated response to the terminal.

[1186] Step 10:

[1187] The terminal displays the response to the user, and the user can receive specific advice.

[1188] The above is the specific processing flow of a system that combines an emotion engine.

[1189] Example 2

[1190] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1191] To prevent employees from developing mental health problems, it is necessary to provide an environment where employees can easily seek advice. However, conventional systems lack the data necessary to generate appropriate responses and perform insufficient emotion analysis, making it difficult to provide personalized responses to users. As a result, employees may not receive the support they need, which could worsen their mental health.

[1192] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing the consultation content from the user, a sentiment analysis means for analyzing the sentiment from the user's consultation content and generating sentiment data for response generation, a means for generating a response using the generative AI model based on the analyzed consultation content and sentiment data, and a means for displaying the generated response to the user. This makes it possible to provide a personalized response tailored to the user's emotional state and realize an environment where employees can easily seek consultation.

[1193] "Data collection methods" are methods for collecting necessary data from past consultation cases of wellness centers and supporters.

[1194] "Preprocessing means" refers to means for cleansing collected data, removing unnecessary data and noise, and converting data into a unified format.

[1195] A "means for training using a generative AI model" is a means for training a generative AI model using preprocessed data.

[1196] The "means for receiving and analyzing the consultation content" is a means for receiving the consultation content from the user and analyzing the content using natural language processing technology.

[1197] The "emotion analysis means" is a means for analyzing emotions from the content of the user's consultation and generating emotion data.

[1198] The "means for generating a response" refers to a means for generating an appropriate response using a generative AI model based on the analyzed consultation content and emotional data.

[1199] The "means for displaying a response" is a means for displaying the generated response to the user.

[1200] A "wellness center" is a specialized institution that provides information on maintaining employee health and mental care.

[1201] A "support person" is a person or expert who serves as a source of advice for employees.

[1202] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system combines generative AI and an emotion analysis engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by specialist institutions.

[1203] The system's program is composed of the following hardware and software: The server uses Python libraries (e.g., requests and SQLAlchemy) to collect the necessary information from the wellness center and supporter databases. Additionally, as a preprocessing method, the pandas and nltk libraries are used to cleanse the collected data and convert it into a unified format.

[1204] The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to learn and train data using a generative AI model (e.g., GPT-4). It receives the user's consultation and analyzes it using natural language processing techniques (e.g., spaCy or BERT). Furthermore, it uses a sentiment analysis engine (e.g., TextBlob or VADER) to analyze the user's sentiment and generate sentiment data for response generation.

[1205] Let's explain with a concrete example. A user uses a device to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device sends this input to the server. The server receives the consultation content and performs natural language processing using spaCy to analyze the meaning and intent of the sentence. Next, the server performs sentiment analysis using the VADER library and extracts the emotional data "stress is increasing." The server uses a GPT-4 model to generate an appropriate response based on the following prompt sentence:

[1206] "The user is asking for advice about work stress. Please provide appropriate advice."

[1207] The terminal then receives the generated response and displays it to the user.

[1208] Similarly, if a user types "I'm having trouble with relationships at work. What should I do?" into their device, the same process will occur. The server will analyze the consultation content using spaCy and perform sentiment analysis using TextBlob. If "anxiety" is strongly detected, the server will use the GPT-4 model to input the following prompt:

[1209] "The user is asking for advice about relationships at work. Please provide advice on how to improve them."

[1210] The generated response is then received by the terminal and displayed to the user.

[1211] This system provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice, making it possible to catch early signs of mental health problems and provide appropriate countermeasures.

[1212] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1213] Step 1: Data collection

[1214] The server collects information from the wellness center and supporter databases. It uses Python's requests library to send API requests and retrieves data in JSON format. It also extracts supporter past consultation cases from the database using SQL queries. The input is the wellness center API and database queries, and the output is the collected raw data.

[1215] Step 2: Data Preprocessing

[1216] The server cleanses the collected data to remove noise and unnecessary characters. Specifically, it creates a data frame using Python's pandas library and filters out unnecessary data using regular expressions. It then uses nltk to tokenize the data and convert it into a unified format (e.g., JSON). The input is the collected raw data, and the output is the preprocessed data.

[1217] Step 3: Training the generative AI model

[1218] The server trains a generative AI model (e.g., GPT-4) using the preprocessed data. Specifically, it builds a model using a deep learning framework (e.g., TensorFlow or PyTorch) and runs training using the preprocessed data. The input is the preprocessed data, and the output is a trained generative AI model.

[1219] Step 4: Receiving consultation details

[1220] The user enters the content of their consultation into an input field on the terminal. The terminal then sends the entered content to the server. Specifically, an HTTP POST request is used. The input is the user's consultation content, and the output is the data to be sent to the server.

[1221] Step 5: Analysis of consultation content

[1222] The server analyzes the received consultation content and understands the meaning and intent of the text using natural language processing technology (e.g., spaCy or BERT). The input is the received consultation content, and the output is the analyzed meaning and intent of the text.

[1223] Step 6: Sentiment Analysis

[1224] The server uses an emotion analysis engine (e.g., TextBlob or VADER) to extract emotion data from the analyzed consultation content. It measures the type of emotion (joy, sadness, anger, etc.) and its intensity. The input is the analyzed meaning and intent, and the output is emotion data.

[1225] Step 7: Response Generation

[1226] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data. Specifically, it inputs a prompt sentence (e.g., "The user is asking for advice about work stress. Please provide appropriate advice.") into the generative AI model and obtains the generated response. The input is the analyzed consultation content and emotional data, and the output is the generated response.

[1227] Step 8: Display the response

[1228] The terminal receives the response sent by the server and displays it to the user, typically using a text view or a dialog box to display the response. The input is the response generated, and the output is what is displayed to the user.

[1229] (Application example 2)

[1230] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1231] There is a need for systems that can detect mental stress and mental disorders among employees early and deal with them promptly and appropriately. However, existing employee support systems lack effective emotion analysis and personalized responses, and do not provide an environment where employees can easily seek advice. Furthermore, there are many situations where an appropriate response in real time is required, so technology to solve this issue is needed.

[1232] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a preprocessing means for cleansing collected data and converting it into a unified format; a means for learning the preprocessed data using a generative AI model and training the model; a means for receiving and analyzing consultation content from a user; a means for generating a response using the generative AI model based on the analyzed consultation content; a means for displaying the generated response to the user; a means for an employee to directly input the consultation content using a smart device; an emotion analysis means for analyzing emotions based on the input consultation content; and a means for generating the emotion analysis result as a prompt sentence for the generative AI model. This enables employees to easily receive mental support and enables early detection and appropriate treatment of mental disorders.

[1233] "Data collection means" refers to the means used to collect information such as past consultation cases of employees and information provided by the wellness center.

[1234] "Preprocessing means" refers to means for cleansing collected data and converting it into a unified format.

[1235] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to user inquiries.

[1236] A "training means" is a means for training preprocessed data using a generative AI model.

[1237] The "means for receiving consultation content" is a means for receiving consultation content from employees.

[1238] The "analysis means" is a means for analyzing the received consultation content and understanding its meaning and intent.

[1239] The "response generation means" is a means for generating a personalized response using a generative AI model based on the analyzed consultation content.

[1240] A "display means" is a means for displaying the generated response on the employee's device.

[1241] "Smart devices" are electronic devices such as smartphones and tablets that employees use to input consultation details.

[1242] The "emotion analysis means" is a means for analyzing emotions based on the content of an employee's consultation and obtaining the results.

[1243] The "prompt generation means" is a means for generating prompt sentences for the generative AI model based on the results of emotion analysis.

[1244] This invention relates to an AI peer support system for preventing employee mental health problems, and provides a consultation environment by combining generative AI models and emotion analysis technology. This system creates an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers.

[1245] This system is configured as follows:

[1246] Hardware and Software

[1247] 1. Hardware: smartphones, tablets, servers.

[1248] 2. Software: Python program, OpenAI GPT-4, TextBlob.

[1249] Explanation of main processes

[1250] 1. Data collection methods:

[1251] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[1252] 2. Pretreatment methods:

[1253] The server cleanses the collected data, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format, making it parseable.

[1254] 3. Training methods:

[1255] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4) using a deep learning framework, which performs iterative learning to improve the model's accuracy.

[1256] 4. How to receive consultation details:

[1257] The user inputs and sends the consultation details using a smart device, which then sends the input to the server.

[1258] 5. Analysis methods:

[1259] The server analyzes the received consultation content using natural language processing technology to understand the meaning and intent of the consultation content.

[1260] 6. Emotion analysis means:

[1261] The server uses an emotion engine to analyze emotions from the employee's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response.

[1262] 7. Prompt Generation Method:

[1263] Based on the results of the sentiment analysis, a prompt sentence is generated for the generative AI model. For example, the following prompt sentence is generated:

[1264] text

[1265] User's question: I've been so busy at work lately that I'm feeling stressed.

[1266] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[1267] Provide specific advice to this user.

[1268] 8. Response Generation Methods:

[1269] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. A personalized response is created by inputting a prompt sentence to the generative AI model and obtaining the generated text data.

[1270] 9. Display means:

[1271] The smart device displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate countermeasures or advice.

[1272] Specific examples

[1273] Example 1: Stress consultation

[1274] The user uses the terminal to input, "I've been so busy at work recently that I'm stressed out."

[1275] The smart device sends this input to the server.

[1276] The server receives the consultation content and analyzes it using natural language processing technology.

[1277] The server uses an emotion engine to analyze emotions and extract data indicating "increasing stress."

[1278] The server uses a generative AI model to generate a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1279] The smart device displays the generated response to the user, who can then receive specific advice.

[1280] As a result, the AI ​​peer supporter system of the present invention combines an emotion engine and a generative AI model to provide personalized responses tailored to the user's own emotional state, creating an environment where employees can feel comfortable seeking advice.

[1281] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1282] Step 1:

[1283] Data collection methods

[1284] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[1285] Input: Wellness center and peer supporter data.

[1286] Output: A dataset of collected information.

[1287] Step 2:

[1288] Pretreatment means

[1289] The server cleanses the data collected in step 1, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format.

[1290] Input: Collected information dataset.

[1291] Output: A cleansed and tokenized dataset.

[1292] Step 3:

[1293] Training methods

[1294] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4), using a deep learning framework to iteratively train the model to improve its accuracy.

[1295] Input: The preprocessed dataset.

[1296] Output: A trained generative AI model.

[1297] Step 4:

[1298] Means of receiving consultation details

[1299] The user inputs the consultation details using the smart device, and the terminal transmits this input to the server.

[1300] Input: User's consultation content.

[1301] Output: Consultation content data sent to the server.

[1302] Step 5:

[1303] Analysis means

[1304] The server analyzes the consultation content received in step 4 using natural language processing technology to understand the meaning and intent of the consultation content.

[1305] Input: Received consultation data.

[1306] Output: Parsed consultation data.

[1307] Step 6:

[1308] Emotion analysis means

[1309] The server analyzes emotions using an emotion engine based on the analyzed consultation content, and detects the type and intensity of the emotion.

[1310] Input: Parsed consultation content data.

[1311] Output: Sentiment analysis data.

[1312] Step 7:

[1313] Prompt Generation Method

[1314] The server generates prompt sentences for the generative AI model based on the emotion analysis data.

[1315] Input: Sentiment analysis data.

[1316] Output: The generated prompt statement.

[1317] Examples:

[1318] text

[1319] User's question: I've been so busy at work lately that I'm feeling stressed.

[1320] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[1321] Provide specific advice to this user.

[1322] Step 8:

[1323] Response Generation Method

[1324] The server inputs the prompt sentence into a generative AI model, which then generates an appropriate response.

[1325] Input: The generated prompt statement.

[1326] Output: The generated response text.

[1327] Step 9:

[1328] Display means

[1329] The terminal displays the response sent from the server to the user, allowing the user to receive specific advice and solutions.

[1330] Input: The generated response text.

[1331] Output: The response text displayed to the user.

[1332] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1333] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1334] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1335] [Fourth embodiment]

[1336] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1337] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1338] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1339] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1340] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1341] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1342] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1343] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1344] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1345] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1346] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1347] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1348] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1349] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[1350] The program of this system performs the following processing.

[1351] 1. Data Collection

[1352] The server acquires the data to be collected, i.e., information provided by wellness centers and past consultation cases of peer supporters nationwide. This collection is performed using API calls and database queries.

[1353] 2. Data Preprocessing

[1354] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format by segmenting and tokenizing it. This preprocessing prepares the data in a format that can be analyzed.

[1355] 3. Model Training

[1356] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[1357] 4. Receiving inquiries from users

[1358] The user inputs the details of the consultation into the terminal and transmits them. The terminal transmits the input data to the server.

[1359] 5. Data Analysis

[1360] The server uses natural language processing technology to analyze the received consultation content and understand its meaning and intent. This analysis prepares input data for generating an appropriate response.

[1361] 6. Response Generation

[1362] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is generated by inputting the consultation content into the model and obtaining the generated text data.

[1363] 7. Response Display

[1364] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[1365] Specific examples

[1366] Example 1: Stress consultation

[1367] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[1368] The terminal sends this input to the server.

[1369] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[1370] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1371] The terminal displays the generated response to the user, and the user can receive specific advice.

[1372] Example 2: Relationship advice

[1373] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[1374] The terminal sends this input to the server.

[1375] The server receives the consultation content, analyzes it using natural language processing, and prepares the input data to generate an appropriate response.

[1376] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[1377] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[1378] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[1379] The processing flow will be explained below.

[1380] Processing Steps

[1381] Step 1:

[1382] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[1383] How it works: It uses API calls to retrieve information from the wellness center database, and also uses database queries to gather peer supporter consultation cases stored locally or in cloud storage.

[1384] Step 2:

[1385] The server pre-processes the collected data.

[1386] How it works: Text data is cleansed to remove noise and unnecessary characters, then natural language processing techniques are used to segment and tokenize it, converting it into a structured format.

[1387] Step 3:

[1388] The server formats the preprocessed data into a training dataset.

[1389] How it works: Split the dataset into training and validation sets, and format them into JSON or CSV formats. Consider the balance of the data, and perform sampling or data augmentation if there is an imbalance.

[1390] Step 4:

[1391] The server trains a generative AI model (e.g., GPT-4).

[1392] How it works: The preprocessed dataset is fed into a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model. Adaptive learning rates and other optimization techniques are applied to improve the model's accuracy.

[1393] Step 5:

[1394] The server prepares to receive the consultation content from the user.

[1395] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[1396] Step 6:

[1397] The user uses the terminal to input and transmit the consultation contents.

[1398] Operation: Enter the content of your inquiry into the text field and click the "Send" button. This will cause the device to send the content of your inquiry to the server.

[1399] Step 7:

[1400] The terminal transmits the user's input data to the server.

[1401] Operation: The text data of the consultation is sent to the server via an HTTP request.

[1402] Step 8:

[1403] The server analyzes the received consultation content.

[1404] How it works: Using natural language processing technology, the text of the consultation is semantically analyzed and important keywords and context are extracted.

[1405] Step 9:

[1406] The server generates a response using a generative AI model based on the analyzed data.

[1407] What it does: It feeds the analysis results into a generative AI model to generate an appropriate response text, which is then formatted appropriately and prepared to be sent back to the user.

[1408] Step 10:

[1409] The server sends the generated response to the terminal.

[1410] Behavior: Sends the generated response to the device as an HTTP response.

[1411] Step 11:

[1412] The terminal displays the response received from the server to the user.

[1413] Behavior: Displays the received response text on the user interface.

[1414] Specific examples

[1415] Example 1: Consultation regarding stress

[1416] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice" (step 6).

[1417] The terminal transmits the consultation content to the server (step 7).

[1418] The server receives the consultation content and analyzes it using natural language processing (step 8).

[1419] The server generates a response using a generative AI model. "To reduce stress, try daily relaxation techniques, such as meditation, deep breathing, and light exercise" (Step 9).

[1420] The server sends the generated response to the terminal (step 10).

[1421] The terminal displays the response to the user (step 11).

[1422] The above is a detailed flow of processing in the system.

[1423] Example 1

[1424] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1425] It is important to detect and appropriately address mental health issues among employees early. However, due to a lack of environments where employees can easily seek advice, problems are often discovered late. Furthermore, conventional systems require a great deal of time and effort for data collection, preprocessing, and model training, making it difficult to respond in real time. For this reason, there is a need for a system that can provide effective and prompt mental health support.

[1426] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1427] In this invention, the server includes means for acquiring information to be collected, means for cleansing the collected information and converting it into a unified format, means for learning a generative AI model using the preprocessed data and training the model, means for receiving and transmitting consultation content from a user, means for analyzing the received consultation content, means for generating a response using the generative AI model based on the analyzed consultation content, and means for displaying the generated response to the user. This makes it possible to effectively utilize the collected data and provide appropriate responses in real time in response to employee consultations.

[1428] "Information to be collected" refers to the information source that the system is set up to acquire, and in this case, this refers to data such as information provided by the health support center and past consultation cases of peer supporters.

[1429] "Cleansing" is a preprocessing process to remove unnecessary characters and noise from collected data and ensure data quality.

[1430] A "uniform format" is a standardization procedure for converting data into a format suitable for analysis and learning, and is a data format that ensures consistency.

[1431] A "generative AI model" is an artificial intelligence model that learns from collected and preprocessed data and generates appropriate responses based on the user's inquiry.

[1432] "Learning" is the process by which a generative AI model uses training data to improve the accuracy of response generation.

[1433] "Training" is the process of repeatedly learning a generative AI model using collected data.

[1434] "User consultation content" is text data that system users input and send regarding mental health issues and questions.

[1435] "Analysis" refers to the process of analyzing the meaning of the received user's consultation content using natural language processing technology and understanding the intent of the consultation content.

[1436] A "response" is a text message containing advice or information generated by the generative AI model based on the analyzed content of the user's consultation.

[1437] "Display" refers to an operation or function for visually presenting the generated response on the user's terminal.

[1438] This invention relates to an AI peer support system for preventing mental health problems among employees. This system uses a generative AI model to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by health support centers.

[1439] Program processing explanation

[1440] The main processing of this system is carried out by three elements: the server, the terminal, and the user.

[1441] Data collection

[1442] The server acquires the information to be collected. Specifically, it uses API calls to collect information provided by the health support center and past consultation cases by peer supporters. This data is stored in a database such as MongoDB or PostgreSQL.

[1443] Data Preprocessing

[1444] The server cleanses the collected data and converts it into a unified format, using Python scripts to remove unnecessary data, and natural language processing libraries such as NLTK and SpaCy for segmentation and tokenization. Finally, the data is converted into formats such as CSV and Parquet.

[1445] Model learning

[1446] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). The data is split into a training dataset and a test dataset, and training is performed using a deep learning framework such as TensorFlow or PyTorch. Once training is complete, the model is saved.

[1447] Accepting inquiries from users

[1448] The user uses the device to input and send the content of their consultation. For example, they might use a web or mobile application to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device then sends this data to the server as an HTTP request.

[1449] Analyzing received data

[1450] The server analyzes the received consultation content using natural language processing technology. It obtains text data and performs tokenization and entity recognition using NLTK and SpaCy. This allows it to understand the intent of the consultation content.

[1451] Response Generation

[1452] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. For example, a prompt such as "I've been feeling more stressed at work lately, so I'd like some advice" is input into the model, and the generated text data is retrieved. The response generated is, "To reduce stress, it's effective to try daily relaxation techniques. Examples include meditation, deep breathing, and light exercise."

[1453] Response Display

[1454] The device displays the response received from the server to the user, and displays the response message in a text area on a web page or in an application, or in the chatbot's UI, allowing the user to receive appropriate advice.

[1455] Specific examples

[1456] Example 1: Stress consultation

[1457] The user uses the device to type, "I've been feeling more and more stressed at work lately, so I'd like some advice," and then sends it.

[1458] The terminal sends this input to the server.

[1459] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning of the consultation content.

[1460] Using a generative AI model, the server generates a response such as, "To reduce stress, it is effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1461] The terminal displays the generated response to the user, and the user can receive specific advice.

[1462] Example 2: Relationship advice

[1463] The user uses the terminal to type, "I'm having trouble with my relationships at work. What should I do?" and submits the message.

[1464] The terminal sends this input to the server.

[1465] The server receives the consultation content, analyzes it using natural language processing technology, and prepares the input data to generate an appropriate response.

[1466] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[1467] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[1468] In this way, the AI ​​peer supporter system of the present invention aims to prevent mental health problems by providing an environment where employees can easily seek advice. This system allows employees to receive appropriate advice, realizing a workplace environment that is healthy both physically and mentally.

[1469] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1470] Step 1:

[1471] The server uses API calls to obtain the information to be collected. Specifically, it obtains data including past consultation cases from health support centers and peer supporters from the API endpoint. The input is the API endpoint URL and authentication information, and the output is data in JSON format. This data is stored in a database such as MongoDB or PostgreSQL.

[1472] Step 2:

[1473] The server cleanses the collected data and converts it into a unified format. Specifically, it uses Python scripts to remove unnecessary characters and noise, and performs word segmentation and tokenization using natural language processing libraries such as NLTK and SpaCy. The input is the collected JSON data, and the output is structured data in CSV or Parquet format.

[1474] Step 3:

[1475] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). Specifically, the data is divided into a training set and a test set, and the model is trained using a deep learning framework such as TensorFlow or PyTorch. The input is the preprocessed CSV or Parquet data, and the output is a trained generative AI model. This model is stored on the server.

[1476] Step 4:

[1477] The user uses a terminal to input and send the content of their consultation. Specifically, they enter something like "I've been feeling more stressed at work lately, so I'd like some advice" into a form on a web or mobile application, and then press the send button. The input is text data entered by the user, and the output is sent to the server as an HTTP request.

[1478] Step 5:

[1479] The server analyzes the consultation content received from the device. Specifically, it takes the received text data and performs tokenization and entity recognition using natural language processing libraries such as NLTK and SpaCy. The input is the received text data, and the output is the tokenized analyzed data. This analysis allows the intent and gist of the consultation content to be understood.

[1480] Step 6:

[1481] The server uses a generative AI model to generate an appropriate response based on the analyzed consultation content. Specifically, the analyzed data is input into the model as a prompt sentence, and the generated text data is obtained. The input is the tokenized analyzed data and the prompt sentence, and the output is the generated response text. In this example, a response such as "To reduce stress, it is effective to try daily relaxation methods. Examples include meditation, deep breathing, and light exercise" is generated.

[1482] Step 7:

[1483] The terminal displays the response from the server to the user. Specifically, it displays the received response text in a text area on a web page or in an application, or in the chatbot's UI. The input is the response text received from the server, and the output is a text display that the user can check on the screen.

[1484] (Application example 1)

[1485] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1486] Mental health problems among employees are a serious problem that can lead to reduced work efficiency and a worsening work environment. Particularly in brick-and-mortar stores, where employees often have direct contact with customers, accumulating stress and interpersonal problems can negatively impact how they treat customers. The present invention aims to provide an environment where employees working in brick-and-mortar stores can easily receive mental care, thereby improving the work environment and the mental health of employees.

[1487] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1488] In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing consultation content from a user, a means for generating a response using the generative AI model based on the analyzed consultation content, and a means for displaying the generated response on a terminal used by an employee at a physical store. This enables employees working at physical stores to easily seek mental health consultation and receive appropriate advice.

[1489] "Data collection methods" are methods for collecting information provided by wellness centers and past consultation cases of peer supporters across the country.

[1490] The "preprocessing means" is a means for cleansing the collected data, removing unnecessary characters and noise, and converting the data into a unified format.

[1491] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to employee inquiries.

[1492] The "means for receiving the consultation content from the user" is a means for transmitting the consultation content input by the employee to the terminal to the server and receiving it.

[1493] The "means for analyzing" refers to a means for analyzing the received consultation content using natural language processing technology and understanding the meaning and intent of the content.

[1494] The "means for generating a response" is a means for generating an appropriate response using a generative AI model based on the analyzed consultation content.

[1495] The "means for displaying the generated response" is a means for displaying the response sent from the server on the terminal used by the employee.

[1496] "Brick and mortar store" refers to a physical sales or service location where employees interact directly with customers.

[1497] This invention relates to an AI peer support system to prevent employees from developing mental health problems. This system uses generative AI to provide an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers. This system is designed especially for use by employees working in brick-and-mortar stores.

[1498] The system is configured as follows:

[1499] Hardware and Software

[1500] The system's main hardware consists of a smartphone or tablet connected to a server. The server is responsible for data collection, preprocessing, training the generative AI model, and generating responses. The smartphone or tablet acts as an interface where users input their inquiries and view responses from the server. The software primarily uses OpenAI's API, Python scripts, and the HTTP request library.

[1501] Data collection

[1502] The server collects information provided by the wellness center and past consultation cases of peer supporters nationwide through API calls or database queries. This data is used as the basis for user consultations.

[1503] Data Preprocessing

[1504] The collected data is cleansed by the server to remove unnecessary characters and noise, and then converted into a structured format using techniques such as tokenization and word segmentation. This preprocessing prepares the data in a format that can be analyzed.

[1505] Model learning

[1506] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[1507] Receiving and analyzing inquiries from users

[1508] Users input and submit their consultation details using a device such as a smartphone or tablet. The device then sends this input data to a server, which then analyzes the received consultation details using natural language processing technology.

[1509] Response generation and display

[1510] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. This response is sent to the device and displayed to the user, allowing the user to obtain appropriate countermeasures and advice.

[1511] Specific examples

[1512] Example 1: Stress consultation

[1513] User: I'm under a lot of pressure at work, how can I reduce it?

[1514] Prompt: User wants to know: I'm under a lot of pressure at work. How can I alleviate it?\nGive appropriate advice:

[1515] Example 2: Relationship Advice

[1516] User: I'm having some disagreements with a coworker and it's been a strained relationship. Is there anything I can do to improve this?

[1517] Prompt: User's question: I'm having disagreements with a colleague and it's been a strained relationship. Is there a way to improve this?\nPlease provide appropriate advice:

[1518] In this way, the AI ​​peer supporter system of the present invention allows store employees to easily seek mental health advice and receive appropriate advice in real time, enabling them to smoothly carry out their daily work and improve the quality of customer service.

[1519] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1520] Step 1:

[1521] The server collects information provided by wellness centers and past consultation cases of peer supporters nationwide using API calls or database queries. The input is the API endpoints for wellness centers and peer support data, and the output is the collected data.

[1522] Step 2:

[1523] The server cleanses the collected data, removing unnecessary characters and noise, and converts the data into a structured format using tokenization, word segmentation, etc. The input is the collected data, and the output is the preprocessed data.

[1524] Step 3:

[1525] The server uses the preprocessed data to learn and train a generative AI model (e.g., GPT-4), which improves the model's accuracy. The input is the preprocessed data, and the output is a trained generative AI model.

[1526] Step 4:

[1527] The user inputs the content of the consultation using a device such as a smartphone or tablet and sends the input to the server. The input is the content of the user's consultation, and the output is the data sent to the server.

[1528] Step 5:

[1529] The server analyzes the received consultation content using natural language processing technology (e.g., NLP technology) to understand the meaning and intent of the content. The input is the consultation content from the user, and the output is the analyzed consultation content.

[1530] Step 6:

[1531] The server generates an appropriate response based on the analyzed consultation content using a trained generative AI model. The input is the analyzed consultation content and the generative AI model, and the output is the generated response.

[1532] Step 7:

[1533] The terminal displays the response sent by the server to the user. The input is the response generated, and the output is the advice or action displayed to the user.

[1534] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1535] This invention relates to an AI peer support system for preventing mental health problems among employees. This system combines generative AI and an emotion engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by wellness centers.

[1536] The program of this system performs the following processing.

[1537] 1. Data Collection

[1538] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country using API calls and database queries.

[1539] 2. Data Preprocessing

[1540] The server cleanses the collected data, removing unnecessary characters and noise, and then converts the data into a structured format through segmentation and tokenization. This preprocessing prepares the data in a form that can be analyzed.

[1541] 3. Model Training

[1542] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4). This training process uses a deep learning framework and involves repeated training to improve the model's accuracy.

[1543] 4. Receiving and analyzing consultation content

[1544] The user inputs the content of their consultation into the terminal and sends it. The terminal then sends the input data to the server. The server then analyzes the received content using natural language processing technology to understand the meaning and intent of the content.

[1545] 5. Emotion analysis

[1546] The server uses an emotion engine to analyze the emotion from the user's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response based on the analysis.

[1547] 6. Response Generation

[1548] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. This response is generated by inputting the consultation content and emotional data into the model and obtaining the generated text data.

[1549] 7. Response Display

[1550] The terminal displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate measures or advice.

[1551] Specific examples

[1552] Example 1: Stress consultation

[1553] The user uses the terminal to input, "I've been feeling more and more stressed at work lately, so I'd like some advice."

[1554] The terminal sends this input to the server.

[1555] The server receives the consultation content and analyzes it using natural language processing technology.

[1556] The server uses an emotion engine to analyze the user's emotions and extracts emotion data indicating "increasing stress."

[1557] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1558] The terminal displays the generated response to the user, and the user can receive specific advice.

[1559] Example 2: Relationship advice

[1560] The user uses the terminal to input, "I'm having trouble with my relationships at work. What should I do?"

[1561] The terminal sends this input to the server.

[1562] The server receives the consultation content, analyzes it using natural language processing technology, and understands the meaning and intent.

[1563] The server uses an emotion engine to analyze the user's emotions and extracts data indicating that "anxiety" is strongly felt.

[1564] Using a generative AI model, the server generates a response that reads, "To improve human relationships, it's important to first try to understand the other person's position. It's effective to have repeated dialogue and have a forum where you can share your opinions with each other."

[1565] The terminal displays the generated responses to the user, who can then receive specific advice on how to improve relationships at work.

[1566] In this way, the AI ​​peer supporter system of the present invention, combined with an emotion engine, provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice. This makes it possible to detect signs of mental illness early and provide appropriate countermeasures, with the aim of realizing a workplace environment that is healthy both physically and mentally.

[1567] The processing flow will be explained below.

[1568] Processing Steps

[1569] Step 1:

[1570] The server collects information provided by wellness centers and past consultation cases of peer supporters across the country.

[1571] How it works: The server uses API calls and database queries to gather the necessary data, which is then stored locally or in cloud storage.

[1572] Step 2:

[1573] The server pre-processes the collected data.

[1574] How it works: The server cleanses the text data, removing noise and unnecessary characters, and then segments and tokenizes the data to make it parseable.

[1575] Step 3:

[1576] The server formats the preprocessed data into a training dataset.

[1577] What it does: Splits the dataset into training and validation sets, and performs sampling and data augmentation as needed to keep the data balanced.

[1578] Step 4:

[1579] The server trains a generative AI model (e.g., GPT-4).

[1580] How it works: The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to train a generative AI model using the preprocessed dataset, applying adaptive learning rates and other optimization techniques to improve the model's accuracy.

[1581] Step 5:

[1582] The server prepares to receive the consultation content from the user.

[1583] How it works: Set up a RESTful API or WebSocket to receive data sent from the user's device in real time.

[1584] Step 6:

[1585] The user uses the terminal to input and transmit the consultation contents.

[1586] Operation: The user enters the content of the consultation in the text field and clicks the "Send" button. The device sends the consultation data to the server.

[1587] Step 7:

[1588] The terminal transmits the user's input data to the server.

[1589] Operation: The text data of the consultation is sent to the server using an HTTP request.

[1590] Step 8:

[1591] The server analyzes the received consultation content.

[1592] How it works: The server uses natural language processing technology to analyze the meaning and intent of the consultation, extracting important keywords and context.

[1593] Step 9:

[1594] The server uses an emotion engine to analyze the emotion from the content of the user's consultation.

[1595] How it works: The emotion engine detects emotion types (e.g., joy, sadness, fear) and their intensity from text data. This emotion data is fed into a generative AI model.

[1596] Step 10:

[1597] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data.

[1598] How it works: The server inputs the consultation content and emotional data into the generative AI model, receives the generated text response, and adjusts the tone and content of the response based on the emotional data.

[1599] Step 11:

[1600] The server sends the generated response to the terminal.

[1601] Operation: The generated response data is sent back to the terminal via an HTTP response.

[1602] Step 12:

[1603] The terminal displays the response received from the server to the user.

[1604] Behavior: Displays the received response text on the user's screen.

[1605] Specific examples

[1606] Example 1: Stress consultation

[1607] Step 1:

[1608] The server collects information provided by the wellness center and past consultation cases of peer supporters.

[1609] Step 2:

[1610] The server cleanses the collected data and puts it into an analyzable format.

[1611] Step 3:

[1612] The server formats the preprocessed data into a training dataset and trains a generative AI model.

[1613] Step 4:

[1614] The user uses the terminal to input "I've been feeling more and more stressed at work recently, so I'd like some advice," and then sends it.

[1615] Step 5:

[1616] The terminal transmits the user's input data to the server.

[1617] Step 6:

[1618] The server receives the consultation content and analyzes it using natural language processing technology.

[1619] Step 7:

[1620] The server uses an emotion engine to extract emotion data such as "stress is increasing."

[1621] Step 8:

[1622] Using a generative AI model, the server generates a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1623] Step 9:

[1624] The server sends the generated response to the terminal.

[1625] Step 10:

[1626] The terminal displays the response to the user, and the user can receive specific advice.

[1627] The above is the specific processing flow of a system that combines an emotion engine.

[1628] Example 2

[1629] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1630] To prevent employees from developing mental health problems, it is necessary to provide an environment where employees can easily seek advice. However, conventional systems lack the data necessary to generate appropriate responses and perform insufficient emotion analysis, making it difficult to provide personalized responses to users. As a result, employees may not receive the support they need, which could worsen their mental health.

[1631] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a data collection means, a preprocessing means for cleansing the collected data and converting it into a unified format, a means for learning the preprocessed data using a generative AI model and training the model, a means for receiving and analyzing the consultation content from the user, a sentiment analysis means for analyzing the sentiment from the user's consultation content and generating sentiment data for response generation, a means for generating a response using the generative AI model based on the analyzed consultation content and sentiment data, and a means for displaying the generated response to the user. This makes it possible to provide a personalized response tailored to the user's emotional state and realize an environment where employees can easily seek consultation.

[1632] "Data collection methods" are methods for collecting necessary data from past consultation cases of wellness centers and supporters.

[1633] "Preprocessing means" refers to means for cleansing collected data, removing unnecessary data and noise, and converting data into a unified format.

[1634] A "means for training using a generative AI model" is a means for training a generative AI model using preprocessed data.

[1635] The "means for receiving and analyzing the consultation content" is a means for receiving the consultation content from the user and analyzing the content using natural language processing technology.

[1636] The "emotion analysis means" is a means for analyzing emotions from the content of the user's consultation and generating emotion data.

[1637] The "means for generating a response" refers to a means for generating an appropriate response using a generative AI model based on the analyzed consultation content and emotional data.

[1638] The "means for displaying a response" is a means for displaying the generated response to the user.

[1639] A "wellness center" is a specialized institution that provides information on maintaining employee health and mental care.

[1640] A "support person" is a person or expert who serves as a source of advice for employees.

[1641] This invention relates to an AI peer support system for preventing employees from developing mental health problems. This system combines generative AI and an emotion analysis engine to provide an environment where employees can easily seek advice, and has the function of generating appropriate responses based on past consultation cases and information provided by specialist institutions.

[1642] The system's program is composed of the following hardware and software: The server uses Python libraries (e.g., requests and SQLAlchemy) to collect the necessary information from the wellness center and supporter databases. Additionally, as a preprocessing method, the pandas and nltk libraries are used to cleanse the collected data and convert it into a unified format.

[1643] The server uses a deep learning framework (e.g., TensorFlow or PyTorch) to learn and train data using a generative AI model (e.g., GPT-4). It receives the user's consultation and analyzes it using natural language processing techniques (e.g., spaCy or BERT). Furthermore, it uses a sentiment analysis engine (e.g., TextBlob or VADER) to analyze the user's sentiment and generate sentiment data for response generation.

[1644] Let's explain with a concrete example. A user uses a device to input, "I've been feeling more stressed at work recently, so I'd like some advice." The device sends this input to the server. The server receives the consultation content and performs natural language processing using spaCy to analyze the meaning and intent of the sentence. Next, the server performs sentiment analysis using the VADER library and extracts the emotional data "stress is increasing." The server uses a GPT-4 model to generate an appropriate response based on the following prompt sentence:

[1645] "The user is asking for advice about work stress. Please provide appropriate advice."

[1646] The terminal then receives the generated response and displays it to the user.

[1647] Similarly, if a user types "I'm having trouble with relationships at work. What should I do?" into their device, the same process will occur. The server will analyze the consultation content using spaCy and perform sentiment analysis using TextBlob. If "anxiety" is strongly detected, the server will use the GPT-4 model to input the following prompt:

[1648] "The user is asking for advice about relationships at work. Please provide advice on how to improve them."

[1649] The generated response is then received by the terminal and displayed to the user.

[1650] This system provides personalized responses tailored to the user's emotional state, creating an environment where employees can easily seek advice, making it possible to catch early signs of mental health problems and provide appropriate countermeasures.

[1651] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1652] Step 1: Data collection

[1653] The server collects information from the wellness center and supporter databases. It uses Python's requests library to send API requests and retrieves data in JSON format. It also extracts supporter past consultation cases from the database using SQL queries. The input is the wellness center API and database queries, and the output is the collected raw data.

[1654] Step 2: Data Preprocessing

[1655] The server cleanses the collected data to remove noise and unnecessary characters. Specifically, it creates a data frame using Python's pandas library and filters out unnecessary data using regular expressions. It then uses nltk to tokenize the data and convert it into a unified format (e.g., JSON). The input is the collected raw data, and the output is the preprocessed data.

[1656] Step 3: Training the generative AI model

[1657] The server trains a generative AI model (e.g., GPT-4) using the preprocessed data. Specifically, it builds a model using a deep learning framework (e.g., TensorFlow or PyTorch) and runs training using the preprocessed data. The input is the preprocessed data, and the output is a trained generative AI model.

[1658] Step 4: Receiving consultation details

[1659] The user enters the content of their consultation into an input field on the terminal. The terminal then sends the entered content to the server. Specifically, an HTTP POST request is used. The input is the user's consultation content, and the output is the data to be sent to the server.

[1660] Step 5: Analysis of consultation content

[1661] The server analyzes the received consultation content and understands the meaning and intent of the text using natural language processing technology (e.g., spaCy or BERT). The input is the received consultation content, and the output is the analyzed meaning and intent of the text.

[1662] Step 6: Sentiment Analysis

[1663] The server uses an emotion analysis engine (e.g., TextBlob or VADER) to extract emotion data from the analyzed consultation content. It measures the type of emotion (joy, sadness, anger, etc.) and its intensity. The input is the analyzed meaning and intent, and the output is emotion data.

[1664] Step 7: Response Generation

[1665] The server generates a response using a generative AI model based on the analyzed consultation content and emotional data. Specifically, it inputs a prompt sentence (e.g., "The user is asking for advice about work stress. Please provide appropriate advice.") into the generative AI model and obtains the generated response. The input is the analyzed consultation content and emotional data, and the output is the generated response.

[1666] Step 8: Display the response

[1667] The terminal receives the response sent by the server and displays it to the user, typically using a text view or a dialog box to display the response. The input is the response generated, and the output is what is displayed to the user.

[1668] (Application example 2)

[1669] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1670] There is a need for systems that can detect mental stress and mental disorders among employees early and deal with them promptly and appropriately. However, existing employee support systems lack effective emotion analysis and personalized responses, and do not provide an environment where employees can easily seek advice. Furthermore, there are many situations where an appropriate response in real time is required, so technology to solve this issue is needed.

[1671] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a preprocessing means for cleansing collected data and converting it into a unified format; a means for learning the preprocessed data using a generative AI model and training the model; a means for receiving and analyzing consultation content from a user; a means for generating a response using the generative AI model based on the analyzed consultation content; a means for displaying the generated response to the user; a means for an employee to directly input the consultation content using a smart device; an emotion analysis means for analyzing emotions based on the input consultation content; and a means for generating the emotion analysis result as a prompt sentence for the generative AI model. This enables employees to easily receive mental support and enables early detection and appropriate treatment of mental disorders.

[1672] "Data collection means" refers to the means used to collect information such as past consultation cases of employees and information provided by the wellness center.

[1673] "Preprocessing means" refers to means for cleansing collected data and converting it into a unified format.

[1674] A "generative AI model" is an artificial intelligence model that learns from preprocessed data and generates appropriate responses to user inquiries.

[1675] A "training means" is a means for training preprocessed data using a generative AI model.

[1676] The "means for receiving consultation content" is a means for receiving consultation content from employees.

[1677] The "analysis means" is a means for analyzing the received consultation content and understanding its meaning and intent.

[1678] The "response generation means" is a means for generating a personalized response using a generative AI model based on the analyzed consultation content.

[1679] A "display means" is a means for displaying the generated response on the employee's device.

[1680] "Smart devices" are electronic devices such as smartphones and tablets that employees use to input consultation details.

[1681] The "emotion analysis means" is a means for analyzing emotions based on the content of an employee's consultation and obtaining the results.

[1682] The "prompt generation means" is a means for generating prompt sentences for the generative AI model based on the results of emotion analysis.

[1683] This invention relates to an AI peer support system for preventing employee mental health problems, and provides a consultation environment by combining generative AI models and emotion analysis technology. This system creates an environment where employees can easily seek advice, and generates appropriate responses based on past consultation cases and information provided by wellness centers.

[1684] This system is configured as follows:

[1685] Hardware and Software

[1686] 1. Hardware: smartphones, tablets, servers.

[1687] 2. Software: Python program, OpenAI GPT-4, TextBlob.

[1688] Explanation of main processes

[1689] 1. Data collection methods:

[1690] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[1691] 2. Pretreatment methods:

[1692] The server cleanses the collected data, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format, making it parseable.

[1693] 3. Training methods:

[1694] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4) using a deep learning framework, which performs iterative learning to improve the model's accuracy.

[1695] 4. How to receive consultation details:

[1696] The user inputs and sends the consultation details using a smart device, which then sends the input to the server.

[1697] 5. Analysis methods:

[1698] The server analyzes the received consultation content using natural language processing technology to understand the meaning and intent of the consultation content.

[1699] 6. Emotion analysis means:

[1700] The server uses an emotion engine to analyze emotions from the employee's consultation content, detect the type and intensity of the emotion, and generate data to adjust the content and tone of the response.

[1701] 7. Prompt Generation Method:

[1702] Based on the results of the sentiment analysis, a prompt sentence is generated for the generative AI model. For example, the following prompt sentence is generated:

[1703] text

[1704] User's question: I've been so busy at work lately that I'm feeling stressed.

[1705] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[1706] Provide specific advice to this user.

[1707] 8. Response Generation Methods:

[1708] The server generates an appropriate response using a generative AI model based on the analyzed consultation content and emotional data. A personalized response is created by inputting a prompt sentence to the generative AI model and obtaining the generated text data.

[1709] 9. Display means:

[1710] The smart device displays the response sent from the server to the user, allowing the user to check the generated response and obtain appropriate countermeasures or advice.

[1711] Specific examples

[1712] Example 1: Stress consultation

[1713] The user uses the terminal to input, "I've been so busy at work recently that I'm stressed out."

[1714] The smart device sends this input to the server.

[1715] The server receives the consultation content and analyzes it using natural language processing technology.

[1716] The server uses an emotion engine to analyze emotions and extract data indicating "increasing stress."

[1717] The server uses a generative AI model to generate a response such as, "To reduce stress, it's effective to try daily relaxation techniques, such as meditation, deep breathing, and light exercise."

[1718] The smart device displays the generated response to the user, who can then receive specific advice.

[1719] As a result, the AI ​​peer supporter system of the present invention combines an emotion engine and a generative AI model to provide personalized responses tailored to the user's own emotional state, creating an environment where employees can feel comfortable seeking advice.

[1720] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1721] Step 1:

[1722] Data collection methods

[1723] The server collects information provided by wellness centers and peer supporters across the country using API calls and database queries.

[1724] Input: Wellness center and peer supporter data.

[1725] Output: A dataset of collected information.

[1726] Step 2:

[1727] Pretreatment means

[1728] The server cleanses the data collected in step 1, removing unnecessary characters and noise, then tokenizes the data and converts it into a structured format.

[1729] Input: Collected information dataset.

[1730] Output: A cleansed and tokenized dataset.

[1731] Step 3:

[1732] Training methods

[1733] The server uses the preprocessed data to train a generative AI model (e.g., GPT-4), using a deep learning framework to iteratively train the model to improve its accuracy.

[1734] Input: The preprocessed dataset.

[1735] Output: A trained generative AI model.

[1736] Step 4:

[1737] Means of receiving consultation details

[1738] The user inputs the consultation details using the smart device, and the terminal transmits this input to the server.

[1739] Input: User's consultation content.

[1740] Output: Consultation content data sent to the server.

[1741] Step 5:

[1742] Analysis means

[1743] The server analyzes the consultation content received in step 4 using natural language processing technology to understand the meaning and intent of the consultation content.

[1744] Input: Received consultation data.

[1745] Output: Parsed consultation data.

[1746] Step 6:

[1747] Emotion analysis means

[1748] The server analyzes emotions using an emotion engine based on the analyzed consultation content, and detects the type and intensity of the emotion.

[1749] Input: Parsed consultation content data.

[1750] Output: Sentiment analysis data.

[1751] Step 7:

[1752] Prompt Generation Method

[1753] The server generates prompt sentences for the generative AI model based on the emotion analysis data.

[1754] Input: Sentiment analysis data.

[1755] Output: The generated prompt statement.

[1756] Examples:

[1757] text

[1758] User's question: I've been so busy at work lately that I'm feeling stressed.

[1759] Emotion analysis results: Polarity: -0.5, Subjectivity: 0.6

[1760] Provide specific advice to this user.

[1761] Step 8:

[1762] Response Generation Method

[1763] The server inputs the prompt sentence into a generative AI model, which then generates an appropriate response.

[1764] Input: The generated prompt statement.

[1765] Output: The generated response text.

[1766] Step 9:

[1767] Display means

[1768] The terminal displays the response sent from the server to the user, allowing the user to receive specific advice and solutions.

[1769] Input: The generated response text.

[1770] Output: The response text displayed to the user.

[1771] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1772] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1773] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1774] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1775] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1776] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1777] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1778] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1779] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1780] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1781] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1782] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1783] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1784] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1785] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1786] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1787] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1788] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1789] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1790] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1791] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1792] The following is further disclosed regarding the above embodiment.

[1793] (Claim 1)

[1794] data collection means;

[1795] As a pre-processing means, a means for cleansing the collected data and converting it into a unified format;

[1796] A means for using a generative AI model to learn from the preprocessed data and train the model;

[1797] A means for receiving and analyzing the content of a consultation from a user;

[1798] A means for generating a response using a generative AI model based on the analyzed consultation content;

[1799] means for displaying the generated response to a user;

[1800] A system including:

[1801] (Claim 2)

[1802] The system of claim 1, wherein the trained generative AI model generates a response in real time to the user's inquiry.

[1803] (Claim 3)

[1804] The system of claim 1, wherein the collected data includes information provided by wellness centers and past consultation cases of peer supporters nationwide.

[1805] "Example 1"

[1806] (Claim 1)

[1807] A means for obtaining the information to be collected;

[1808] A means of cleansing the collected information and converting it into a unified format;

[1809] a means for learning a generative AI model using the preprocessed data and training the model; and

[1810] means for receiving and transmitting consultation contents from users;

[1811] A means for analyzing the received consultation content;

[1812] A means for generating a response using a generative AI model based on the analyzed consultation content;

[1813] means for displaying the generated response to a user;

[1814] A system including:

[1815] (Claim 2)

[1816] The system of claim 1, wherein the trained generative AI model generates a response in real time to the user's inquiry.

[1817] (Claim 3)

[1818] 2. The system according to claim 1, wherein the information to be collected includes information provided by the health support center and past consultation cases of peer supporters.

[1819] "Application Example 1"

[1820] (Claim 1)

[1821] data collection means;

[1822] As a pre-processing means, a means for cleansing the collected data and converting it into a unified format;

[1823] A means for using a generative AI model to learn from the preprocessed data and train the model;

[1824] A means for receiving and analyzing the content of a consultation from a user;

[1825] A means for generating a response using a generative AI model based on the analyzed consultation content;

[1826] a means for displaying the generated response on a terminal used by an employee at the physical store;

[1827] A system including:

[1828] (Claim 2)

[1829] The system of claim 1, in which the trained generative AI model generates responses in real time to inquiries from users, and is used by employees in physical stores.

[1830] (Claim 3)

[1831] The system of claim 1, wherein the collected data includes information provided by health centers and past consultation cases of supporters nationwide.

[1832] "Example 2: Combining Emotion Engines"

[1833] (Claim 1)

[1834] data collection means;

[1835] As a pre-processing means, a means for cleansing the collected data and converting it into a unified format;

[1836] A means for using a generative AI model to learn from the preprocessed data and train the model;

[1837] A means for receiving and analyzing the content of a consultation from a user;

[1838] As an emotion analysis means, a means for analyzing emotions from the content of a user's consultation and generating emotion data for generating a response;

[1839] A means for generating a response using a generative AI model based on the analyzed consultation content and emotion data;

[1840] means for displaying the generated response to a user;

[1841] A system including:

[1842] (Claim 2)

[1843] The system of claim 1, wherein the trained generative AI model generates a response in real time to the consultation content and emotional data from the user.

[1844] (Claim 3)

[1845] The system of claim 1, wherein the collected data includes information provided by specialized institutions and past consultation cases of domestic supporters.

[1846] "Application example 2 when combining emotion engines"

[1847] (Claim 1)

[1848] data collection means;

[1849] As a pre-processing means, a means for cleansing the collected data and converting it into a unified format;

[1850] A means for using a generative AI model to learn from the preprocessed data and train the model;

[1851] A means for receiving and analyzing the content of a consultation from a user;

[1852] A means for generating a response using a generative AI model based on the analyzed consultation content;

[1853] means for displaying the generated response to a user;

[1854] A method for employees to directly input the details of their consultation using a smart device,

[1855] emotion analysis means for analyzing emotions based on the input consultation content;

[1856] A means for generating the emotion analysis result as a prompt sentence for a generative AI model;

[1857] A system including:

[1858] (Claim 2)

[1859] The system of claim 1, wherein the trained generative AI model generates a response in real time to the user's inquiry.

[1860] (Claim 3)

[1861] The system of claim 1, wherein the collected data includes information provided by wellness centers and past consultation cases of peer supporters nationwide. [Explanation of symbols]

[1862] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. data collection means; As a pre-processing means, a means for cleansing the collected data and converting it into a unified format; A means for using a generative AI model to learn from the preprocessed data and train the model; A means for receiving and analyzing the content of a consultation from a user; A means for generating a response using a generative AI model based on the analyzed consultation content; means for displaying the generated response to a user; A system including:

2. The system of claim 1, wherein the trained generative AI model generates a response in real time to a user's inquiry.

3. The system according to claim 1, wherein the collected data includes information provided by wellness centers and past consultation cases of peer supporters nationwide.

Images