Method and system for providing cognitive behavioral therapy program based on large language model

Abstract

A method for providing a cognitive behavioral therapy program based on a large language model according to the present invention may comprise the steps of: providing, to a user terminal, at least one questionnaire related to a cognitive behavioral therapy; and by using response data to the questionnaire received from the user terminal, generating a prompt to be input to a large language model.

Description

Method and System for Providing Cognitive Behavioral Therapy Programs Based on Large Language Models

[0001] The present invention relates to a method and system for providing a Cognitive Behavioral Therapy (CBT) program based on a large language model.

[0002] With the rapid advancement of artificial intelligence (AI) technology in recent years, generative AI models capable of natural conversation with humans (e.g., ChatGPT) have emerged. In particular, unlike existing chatbots, Large Language Models (LLMs) provide conversational quality similar to humans based on their ability to understand various contexts and generate natural language, and are being rapidly adopted in various industrial fields such as medical, education, and healthcare.

[0003] In particular, along with the advancement of artificial intelligence (AI) technology, the utilization of AI in the medical industry is rapidly expanding. For example, there is growing interest in AI technology that uses large language models to provide customized cognitive behavioral therapy programs remotely to patients who require such therapy.

[0004] Cognitive Behavioral Therapy (CBT) can be understood as a psychotherapy method based on the premise that a person's thoughts, emotions, and behaviors are closely interconnected. It involves patients recognizing their own emotions, identifying faulty thoughts, and changing problematic behaviors during the treatment process.

[0005] Recently, as the effectiveness of cognitive behavioral therapy has been scientifically proven, the use of the therapy for mental health problems such as depression, anxiety, insomnia, and pain is gradually increasing, leading to active research on methods to provide the therapy to patients.

[0006] However, since such cognitive behavioral therapy requires continuous treatment, the fact that patients must visit the hospital frequently poses a significant burden. Consequently, there is growing interest in artificial intelligence technology that provides customized cognitive behavioral therapy programs remotely to patients who require such therapy.

[0007] In such non-face-to-face cognitive behavioral therapy, it is essential to provide a customized program that takes into account the user's current condition. If a treatment program is provided uniformly without considering the user's current state and medical history, it may hinder therapeutic effectiveness or even worsen symptoms. Furthermore, users may be unable to engage with content irrelevant to their condition and are more likely to drop out early, which can significantly lower treatment retention rates and overall satisfaction.

[0008] There is a need to address these issues with non-face-to-face cognitive behavioral therapy and to provide user-customized cognitive behavioral therapy programs.

[0009] The present invention is intended to provide a method and system capable of providing a user-customized cognitive behavioral therapy program.

[0010] Specifically, the present invention aims to provide a method and system for providing a large language model-based cognitive behavioral therapy program that can generate feedback on response data to a survey provided to a user terminal using a large language model, and update the cognitive behavioral therapy program based on the generated feedback.

[0011] More specifically, the present invention aims to provide a method and system for providing a large language model-based cognitive behavioral therapy program capable of generating prompts containing feedback of different emotion types through emotion classification of natural language response data included in survey response data.

[0012] Furthermore, the present invention aims to provide a method and system for providing a large language model-based cognitive behavioral therapy program capable of generating update information for updating the cognitive behavioral therapy program using a large language model according to feedback of different emotion types.

[0013] Furthermore, the present invention aims to provide a method and system for providing a large language model-based cognitive behavioral therapy program that can predict a user's emotional state using user information and response data to a survey, and provide notification information related to the predicted emotional state to a user terminal.

[0014] To solve the problem described above, the present invention proposes a method for providing a user-customized cognitive behavioral therapy program that can interact with a user remotely and perform cognitive behavioral therapy using a large language model. The method for providing a cognitive behavioral therapy program based on a large language model according to the present invention may include the steps of: providing at least one survey related to cognitive behavioral therapy to a user terminal; generating a prompt to be input into a large language model using response data for the survey received from the user terminal; inputting the prompt into the large language model and generating feedback related to a treatment program for cognitive behavioral therapy based on natural language response data included in the response data; updating the prompt based on the generated feedback, inputting it into the large language model, and updating the treatment program through the large language model; and providing the updated treatment program to the user terminal.

[0015] Furthermore, the step of generating the above prompt may include the step of extracting a plurality of natural language response data from the response data for the above survey; the step of analyzing the emotion type corresponding to each of the plurality of natural language response data to group the natural language response data corresponding to the same emotion type among the plurality of natural language response data; and the step of generating a plurality of different prompts to be input into the large language model based on each of the groups of the plurality of natural language response data corresponding to different emotion types.

[0016] Furthermore, the plurality of natural language response data groups may include at least one of a first type natural language response data group comprising at least one natural language response data corresponding to a first emotion type among the plurality of natural language response data, and at least one second type natural language response data group comprising at least one natural language response data corresponding to a second emotion type among the plurality of natural language response data.

[0017] Furthermore, the step of generating the above-mentioned different multiple prompts involves generating multiple prompts corresponding to each of the multiple natural language response data groups corresponding to the above-mentioned different emotion types, and the multiple prompts may include at least one of a first type prompt associated with the first type natural language response data group and a second type prompt associated with the second type natural language response data group.

[0018] Furthermore, a plurality of treatment programs corresponding to different topics are stored in the database, and a plurality of treatment contents constituting the specific treatment program are sequentially provided to the user terminal according to the treatment week set for the specific treatment program related to the user's cognitive behavioral therapy among the plurality of treatment programs, and the step of generating the feedback can be performed by inputting at least one of the first type prompt and the second type prompt into the large language model to generate the feedback for the specific treatment program through the large language model.

[0019] Furthermore, the feedback includes at least one of a first type feedback corresponding to the first type prompt and a second type feedback corresponding to the second type prompt, and each of the first type feedback and the second type feedback may include different types of update information for the specific treatment program.

[0020] Furthermore, the step of updating the treatment program may update information related to at least one of the treatment weeks in which each of the plurality of treatment contents is provided, based on the different types of update information for the specific treatment program, a specific topic matched to the specific treatment program, the configuration of each of the plurality of treatment contents included in the specific treatment program, and the treatment week in which each of the plurality of treatment contents is provided.

[0021] Furthermore, the above-mentioned at least one survey may include at least one objective survey consisting of at least one objective question item related to cognitive behavioral therapy and a plurality of selection items corresponding to the objective question item, and at least one subjective survey capable of receiving natural language input for at least one subjective question item from the user terminal related to cognitive behavioral therapy.

[0022] Furthermore, the above response data may include at least one of the objective survey response data for the objective survey and the subjective survey response data for the subjective survey, and the objective survey response data may include the natural language response data corresponding to a specific selection item selected by user input among the plurality of selection items, and the subjective survey response data may include the natural language input for the subjective question item and the natural language response data corresponding to at least one of the natural language answers obtained through the large language model by processing the natural language input as an input to the large language model.

[0023] Furthermore, the method may further include the steps of predicting the emotional pattern of a user subject to cognitive behavioral therapy using the large language model, and providing notification information containing prediction information corresponding to the emotional pattern to the user terminal according to the predicted emotional pattern, wherein the step of predicting the emotional pattern of the user may include the step of generating a prediction prompt for predicting the emotional pattern using at least one of the response data to the survey, the feedback, and user information collected from the database, and the step of processing the prediction prompt as input to the large language model to generate the prediction information of a type corresponding to the emotional pattern through the large language model.

[0024] Meanwhile, the system for providing a cognitive behavioral therapy program based on a large language model according to the present invention includes a communication unit that provides at least one survey related to cognitive behavioral therapy to a user terminal, and a control unit that generates a prompt to be input into a large language model using response data for the survey received from the user terminal. The control unit inputs the prompt into the large language model, generates feedback related to a treatment program for cognitive behavioral therapy based on natural language response data included in the response data, updates the prompt based on the generated feedback, inputs it into the large language model, updates the treatment program through the large language model, and provides the updated treatment program to the user terminal.

[0025] Meanwhile, the program is executed by one or more processes in an electronic device and is stored on a computer-readable recording medium, and the program may include instructions for performing the steps of: providing at least one survey related to cognitive behavioral therapy to a user terminal; generating a prompt to be input into a large language model using response data for the survey received from the user terminal; inputting the prompt into the large language model and generating feedback related to a treatment program for cognitive behavioral therapy based on natural language response data included in the response data; updating the prompt based on the generated feedback and inputting it into the large language model and updating the treatment program through the large language model; and providing the updated treatment program to the user terminal.

[0026] The method and system for providing a large language model-based cognitive behavioral therapy program according to the present invention can improve the efficiency of cognitive behavioral therapy and user engagement by providing a user-customized cognitive behavioral therapy program based on user response data to a survey and user information.

[0027] Furthermore, the method and system for providing a large language model-based cognitive behavioral therapy program according to the present invention can dynamically adjust the therapy program by updating the therapy program provided to the user based on feedback according to the user's current state, thereby responding sensitively to changes in the user's state and ensuring the continuity of long-term therapy.

[0028] Furthermore, the method and system for providing a large language model-based cognitive behavioral therapy program according to the present invention can enable an early intervention and prevention-centered treatment approach for cognitive behavioral therapy by predicting a user's emotional state in advance and preemptively providing notification information based on the predicted emotional state.

[0029] FIG. 1 is a conceptual diagram illustrating a large language model-based cognitive behavioral therapy program providing system according to the present invention.

[0030] FIG. 2 is a flowchart for explaining, in general, a method for providing a large language model-based cognitive behavioral therapy program according to the present invention.

[0031] FIGS. 3a and FIGS. 3b are conceptual diagrams for specifically explaining a survey provided to a user terminal according to the present invention.

[0032] FIG. 4a is a conceptual diagram illustrating the process of extracting natural language response data from response data to a survey according to the present invention.

[0033] FIG. 4b is a conceptual diagram for explaining the process of generating a natural language response data group based on the sentiment type corresponding to the natural language response data according to the present invention.

[0034] FIG. 4c is a conceptual diagram illustrating the process of generating a prompt to be input into a large language model using natural language response data and user information according to the present invention.

[0035] FIG. 5a is a conceptual diagram illustrating a cognitive behavioral therapy program according to the present invention.

[0036] FIG. 5b is a conceptual diagram illustrating the process of generating feedback related to cognitive behavioral therapy using a large language model according to the present invention.

[0037] FIGS. 6a and 6b are conceptual diagrams for explaining the process of updating a cognitive behavioral therapy program based on feedback according to the present invention and providing the updated therapy program to a user terminal.

[0038] FIG. 7 is a conceptual diagram illustrating the process of predicting a user's emotional pattern according to the present invention and providing notification information corresponding to the predicted emotional pattern to a user terminal.

[0039] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components are assigned the same reference number regardless of the drawing symbols, and redundant descriptions thereof will be omitted. The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably solely for the ease of drafting the specification and do not have distinct meanings or roles in themselves. Furthermore, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art could obscure the essence of the embodiments disclosed in this specification, such detailed description will be omitted. Additionally, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification; the technical concept disclosed in this specification is not limited by the attached drawings, and it should be understood that they include all modifications, equivalents, and substitutions that fall within the spirit and technical scope of the present invention.

[0040] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but said components are not limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.

[0041] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0042] A singular expression includes a plural expression unless the context clearly indicates otherwise.

[0043] In this application, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0044] The present invention relates to a method and system for providing a cognitive behavioral therapy program for a user based on a large language model. More specifically, the present invention relates to a method and system capable of generating feedback based on the user's current state and updating the cognitive behavioral therapy program based on the generated feedback. Here, the term "Large Language Model" may refer to an artificial intelligence model capable of understanding and generating natural language by learning from a vast amount of data.

[0045] The “Cognitive Behavioral Therapy (CBT)” according to the present invention may refer to a psychotherapy method in which, by understanding and regulating the interaction between thought (cognition) and behavior, the patient recognizes their own emotions and identifies erroneous thoughts during the treatment process, thereby changing problematic behaviors.

[0046] In the present invention, a questionnaire related to cognitive behavioral therapy is provided, and a user-customized cognitive behavioral therapy program (hereinafter referred to as the "therapy program") can be generated using response data to the questionnaire. Here, the questionnaire according to the present invention may include questions related to cognitive behavioral therapy for checking at least one of the user's pain level, degree of cognitive distortion, duration of pain, severity of pain, mood at the time of experiencing pain, degree of negative emotion regarding pain, degree of stress regarding pain, physical health status, and mental health status.

[0047] Here, “pain level” can be understood as a measure indicating the intensity or severity of pain experienced by an individual. Pain is a subjective and personal experience; as people may perceive and interpret pain differently, it can involve physical discomfort, emotional distress, and suffering. Various methods can be used to assess this pain level. For example, a numerical scale (Numerical Rating Scale, NRS) can be provided to the patient to evaluate the intensity of the pain they perceive (e.g., no pain, mild pain, severe pain, etc.).

[0048] Furthermore, the “pain patient” described in the present invention may refer to a patient experiencing pain due to an indication, and in the present invention, the term “pain patient” may be used interchangeably with “patient” or “user.”

[0049] A treatment program according to the present invention may be configured to include at least one treatment content (or treatment module). Here, “treatment content” may refer to content related to a detailed category (or sub-topic) for cognitive behavioral therapy regarding a specific topic. Such treatment content may be understood as a constituent unit of a treatment program matched to a specific topic. For example, “treatment content” may include worksheets, educational materials, assignments, activities, etc., composed of at least one of text, voice, image, or video related to a specific topic. Furthermore, a plurality of treatment contents constituting the treatment program may be provided sequentially to a user terminal according to a treatment week pre-set in the treatment program.

[0050] The “treatment week” described in the present invention may be understood as the sequence (or period) of providing (or activating) treatment content through a user terminal (10) so that the user performs cognitive behavioral therapy according to the treatment program. In the present invention, a “total number of treatment weeks” and a “weekly treatment period” corresponding to each treatment week may be pre-set. Also, in the present invention, it may be understood that a “total treatment period” is pre-defined. The total treatment period may be determined by the product of the pre-set total number and the pre-set weekly treatment period. For example, if the total number is “8” and the weekly treatment period is 1 week (7 days), the pre-set total treatment period may be “weeks.”

[0051] In the present invention, a large language model is used to generate feedback based on response data to a survey provided to a user terminal (10), and a treatment program can be updated based on said feedback. For example, in the present invention, information related to a topic matched to the treatment program, the configuration of each of the plurality of treatment contents included in the treatment program, and at least one of the treatment weeks in which each of the plurality of treatment contents is provided can be updated.

[0052] More specifically, the present invention classifies (or analyzes) the emotion types of response data to generate different types of feedback, and can update a treatment program based on the emotion type corresponding to the feedback. Furthermore, by providing it to a user terminal, the treatment program updated according to the current user's emotional state can be provided to the user terminal.

[0053] Furthermore, the present invention can predict a user's emotional pattern by analyzing natural language response data, including survey response data received from a user terminal, and generate notification information regarding warnings, advice, or emotional support corresponding to the predicted emotional pattern. Moreover, the present invention can support the user in recognizing their emotional and psychological state in advance and taking necessary measures by providing the generated notification information to the user terminal.

[0054] For the sake of convenience of explanation, the present invention focuses on cognitive behavioral therapy for patients with pain due to musculoskeletal indications, but is not necessarily limited thereto. Here, "indication" refers to a symptom or clinical situation requiring specific treatment or examination, and can be understood as the user's indications or symptoms. Furthermore, an indication may refer to symptoms regarding a specific part of the user that has been physically injured.

[0055] As an example, the treatment program described in the present invention may be related to cognitive behavioral therapy for a user due to various indications (e.g., cancer, diabetes, hypertension, etc.). Additionally, the present invention may generate a program necessary for mental health in daily life, rather than a treatment program for therapeutic purposes related to the user's indications.

[0056] In the foregoing, a method for providing a large language model-based cognitive behavioral therapy program according to the present invention has been generally described, and this can be implemented by a large language model-based cognitive behavioral therapy program providing system described below. Below, with reference to FIG. 1, a large language model-based cognitive behavioral therapy program providing system according to the present invention will be described in detail. FIG. 1 is a conceptual diagram illustrating a large language model-based cognitive behavioral therapy program providing system according to the present invention.

[0057] As illustrated in FIG. 1, a large language model-based cognitive behavioral therapy program providing system according to the present invention (hereinafter referred to as the “therapy program providing system,” 100) may include at least one of a communication unit (110), a storage unit (120), and a control unit (130). At this time, the therapy program providing system (100) according to the present invention is not limited to the components described above and may further include components that perform the same or similar roles as described in the description of this specification.

[0058] Additionally, the treatment program providing system (100) according to the present invention may be configured as a device that performs learning and inference as described in the present invention. For example, the treatment program providing system (100) according to the present invention may be a machine learning model(s) that is trained on learning data by one or more machine learning algorithms. Furthermore, the machine learning model(s) trained in the inference stage of the treatment program providing system (100) may receive input data including one or more inference / prediction requests.

[0059] In this case, the machine learning algorithm or machine learning model(s) according to the present invention may include a deep learning algorithm or model utilizing a neural network. Furthermore, the trained machine learning model(s) according to the present invention may provide one or more inferences and / or prediction(s) as outputs in response to an inference / prediction request. Accordingly, the trained machine learning model(s) according to the present invention may include one or more models of one or more machine learning algorithms. In one embodiment, the trained machine learning model(s) according to the present invention may use the output inference(s) and / or prediction(s) as input feedback. Furthermore, the trained machine learning model(s) may use past inference(s) as inputs for generating new inference(s).

[0060] Meanwhile, the treatment program providing system (100) according to the present invention may be implemented as an application or software. The treatment program providing system (100) implemented as software in this manner may be downloaded via a program (e.g., Play Store) that allows downloading applications on a user terminal (10), or implemented via an initial installation program on the user terminal (10). In this case, the communication unit (110), storage unit (120), and control unit (130) according to the present invention may be utilized as components of the user terminal (10). In the present invention, the user terminal (10) can be understood as referring to an application installed on the user terminal (10). Such an application (or software) can be understood as a component of the treatment program providing system (100) according to the present invention.

[0061] Meanwhile, the treatment program providing system may exist inside a server (hereinafter referred to as the server) established to perform a specific purpose (e.g., a function related to providing a treatment program for cognitive behavioral therapy) separately from the user terminal, or it may exist as a system separate from said server. When the treatment program providing system (100) exists inside the server, it may provide various services related to the present invention (e.g., a treatment program for cognitive behavioral therapy, notification information for cognitive behavioral therapy, etc.) through at least one component located inside the server, or through configuration modules that perform functions similar to said components. In this case, the application may provide various services related to the present invention on the user terminal where the application is installed through communication with the server.

[0062] In addition, the treatment program providing system (100) may be linked with a central server or an external server to implement the cognitive behavioral therapy program providing method according to the present invention and provide the treatment program to a user terminal. Meanwhile, the patient may diagnose the degree of pain and the degree of cognitive distortion regarding their own pain through an application or webpage provided by the treatment program providing system (100) according to the present invention, and receive a treatment program according to the degree of pain and the degree of cognitive distortion. In addition, the patient may receive various services related to cognitive behavioral therapy (e.g., notification information for cognitive behavioral therapy) and manage at least one of the mental health and physical health conditions.

[0063] At this time, the patient may have a user account registered in the treatment program providing system (100). For convenience of explanation in this specification, the account of the patient user is referred to as the “patient account” or “user account.”

[0064] The “account” described above can be created through a page linked to the treatment program providing system (100) examined above. Alternatively, the “account” can also be created in at least one other system linked to the treatment program providing system (100).

[0065] Accordingly, in this specification, the system in which the account was issued is not distinguished separately, and all accounts based on the treatment program providing system (100) according to the present invention are referred to as “accounts already registered in the treatment program providing system (100) according to the present invention.”

[0066] For convenience of explanation, the present specification describes the treatment program providing system (100) according to the present invention as an embodiment implemented as an application on a user terminal (10).

[0067] In the present invention, the user terminal (10) may also be named a 'mobile terminal' or 'electronic device', and the user terminal (10) described in this specification may include a mobile phone, a smartphone, a smart TV, a laptop computer, a digital broadcasting terminal, a PDA (personal digital assistants), a PMP (portable multimedia player), a navigation device, a slate PC, a tablet PC, an ultrabook, a wearable device (e.g., a smartwatch, a smart glass, a head-mounted display), etc.

[0068] More specifically, the user terminal (10) according to the present invention is not limited to an electronic device with an application activated, but may refer to an electronic device connected to an electronic device with an application activated. As an example, based on the fact that the user terminal (10) with an application activated according to the present invention is a smartphone, the user terminal (10) may refer to a smart TV connected to the smartphone. That is, the user terminal (10) according to the present invention may refer to at least one of an electronic device owned by a user (or logged in with a user account), an electronic device owned by a medical professional (or logged in with a medical professional account), and an electronic device used in common.

[0069] Meanwhile, the treatment program providing system (100) may exist inside a server (hereinafter referred to as the server) established to perform a specific purpose (e.g., providing a cognitive behavioral therapy program), or it may exist as a separate device from the server. When the treatment program providing system (100) exists inside the server, the treatment program providing system (100) according to the present invention may update the treatment program and provide the updated treatment program to a user terminal through at least one component among a communication unit (110), a storage unit (120), and a control unit (130) located inside the server, or through a module that performs a function similar to each of the above components. In this case, the application may provide the treatment program to the user terminal (10) where the application is installed through communication with the server. Furthermore, the treatment program providing system (100) according to the present invention may provide the treatment program according to the present invention to the user terminal (10) by linking with a plurality of different external servers.

[0070] A user (or patient, U) and a medical staff member (or doctor) according to the present invention may possess an account registered in the treatment program providing system (100) according to the present invention. For convenience of explanation, the account of the user (or patient) in this specification is referred to as the "user account (or patient account)." The "account" described above may be created through a page linked to the treatment program providing system (100). Alternatively, the "account" may be created on at least one other server (e.g., a medical staff server) linked to the treatment program providing system (100) according to the present invention. Accordingly, in this specification, without distinguishing the server where the account was issued, all accounts based on the treatment program providing system (100) according to the present invention are referred to as "accounts already registered in the treatment program providing system (100) according to the present invention."

[0071] Meanwhile, the communication unit (110) according to the present invention may be connected to a user terminal (10), an LLM server (140), a central server, a device, and at least one network via a wireless or wired network, and may be configured to receive or transmit overall data and information necessary for the operation of the treatment program providing system (100) according to the present invention.

[0072] Specifically, the communication unit (110) may transmit at least one survey related to cognitive behavioral therapy for a user to a user terminal (10). For example, the communication unit (110) may transmit at least one survey related to cognitive behavioral therapy to a user terminal (10) where a user account is logged in. Furthermore, the communication unit (110) may receive response data (or survey response data) for at least one survey provided to the user terminal (10). Here, “receiving response data” may mean receiving an input signal (or selection signal) corresponding to a user input entered through the user terminal (10). According to the present invention, the response data may include at least one of text, an image (or video), and voice. In this case, the treatment program providing system (100) may further include a module that converts voice into text.

[0073] The communication unit (110) can receive user information from the database (200). Furthermore, the communication unit (110) can receive a plurality of treatment contents stored in a database (or DB, 200) that includes a treatment program DB. The communication unit (110) may include at least one communication module capable of wireless communication and wired communication between the treatment program providing system (100) and the communication target. Additionally, the communication unit (110) may include a communication module that connects the treatment program providing system (100) to at least one network.

[0074] Meanwhile, the communication unit (110) can support various communication methods depending on the communication standard of the communicating device. For example, the communication unit (110) may be configured to perform communication using at least one of the following technologies: WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), HSDPA (High Speed ​​Downlink Packet Access), HSUPA (High Speed ​​Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced), 5G (5th Generation Mobile Telecommunication), Bluetooth (Bluetooth™ Frequency Identification), Infrared Communication (Infrared Data Association; IrDA), UWB (Ultra-Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus).

[0075] Next, the storage unit (120) may be configured to store various information related to the present invention. In the present invention, the storage unit (120) may be provided in the treatment program providing system (100) itself, or alternatively, at least a part of the storage unit (120) may mean a database (Database: DB, 200).

[0076] The storage unit (120) may include one or more non-transient computer-readable storage media that can be read and / or accessed by at least one processor. One or more computer-readable storage media may include volatile and / or non-volatile storage components such as optical, magnetic, organic, or other memory or disk storage devices. In some examples, the storage unit (120) may be implemented using a single physical device (e.g., one optical, magnetic, organic, or other memory or disk storage device), whereas in other examples, the storage unit (120) may be implemented using multiple physical devices. The storage unit (120) may include computer-readable instructions and additional data.

[0077] The storage unit (120) may include a storage necessary to perform at least some of the methods and techniques described herein and / or at least some of the functions of the device and network. That is, the storage unit (120) is sufficient as a space in which information necessary to generate a treatment program for cognitive behavioral therapy according to the present invention and to update the generated treatment program is stored, and the storage unit (120) may be understood as not being restricted by physical space.

[0078] Data and commands necessary for the operation of the treatment program providing system (100) according to the present invention may be stored in the storage unit (120). More specifically, data and commands necessary for the operation of the emotion analysis module (132) may be stored in the storage unit (120). Here, the emotion analysis module (132) may refer to a module that performs the function of analyzing linguistic expressions included in natural language response data among response data to a survey provided to a user terminal, and classifying the user's emotion type (e.g., positive or negative) based thereon.

[0079] For example, the sentiment analysis module (132) may include at least one artificial intelligence model among BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), KoBERT, KorBERT, DistilBERT, ELECTRA, and ALBERT. Additionally, the sentiment analysis module (132) may include at least one classifier among LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit), or Transformer-based classifiers (e.g., BERT, Classification head), and may improve precision by ensembling with a CNN-based text classifier if necessary. Furthermore, to account for subtle differences in emotional expression, the sentiment analysis module (132) may calculate Emotion Intensity scores in parallel or derive probability values ​​for each emotion using Softmax or Sigmoid-based activation functions.

[0080] The emotion analysis module (132) according to the present invention may be pre-trained or fine-tuned based on a large-scale emotion labeling dataset (KoNLPy, AIHub Sentimental Conversation Corpus, etc.) or actual clinical-based emotion response data. Furthermore, a domain-specific learning strategy may be applied to the emotion analysis module (132) to reflect differences in emotional expression according to the user's age and cultural expression characteristics.

[0081] The method for classifying (or analyzing) the emotion type (or class) corresponding to the natural language response data according to the present invention may be very diverse, and the present specification does not limit the method for classifying the emotion type corresponding to the natural language response data. In the present invention, the emotion analysis module (132) is not limited in its type and method as long as it is a module capable of classifying the emotion type corresponding to the natural language response data. The storage unit (120) according to the present invention may further store a command for performing an operation of at least one of a module and an algorithm that perform the same function as the emotion analysis module (132).

[0082] The storage unit (120) may store user authentication information included in user account information. Here, “user authentication information” may refer to information used in a user authentication process performed to log in to a user account on a user terminal. As an example, user authentication information may be various, such as i) ID, ii) password, iii) password pattern, iv) user’s fingerprint authentication information, v) face authentication information, vi) voice authentication information, vii) iris authentication information, viii) vein authentication information, etc., set by the user.

[0083] Meanwhile, the database (DB, 200) may be configured to store various information related to the provision of a treatment program for the user's cognitive behavioral therapy. Specifically, the database (200) may store multiple treatment contents (e.g., worksheets, assignments, educational materials, etc.) matched to specific topics related to the user's cognitive behavioral therapy.

[0084] Additionally, user information for each of multiple different users may be stored in the database (200). For example, the database (200) may store user information including at least one of the user's i) name, ii) date of birth, iii) user's cognitive behavioral therapy history, iv) medical records, v) surgical records, vi) past medical history, vii) prescription information, and viii) past response data.

[0085] In the present invention, the storage unit (120) is described as existing separately from the database (200), but is not limited thereto, and the storage unit (120) may include the database (200).

[0086] Next, the control unit (130) may be configured to control the overall operation of the treatment program providing system (100) related to the present invention. The control unit (130) may include at least one of a prompt generation unit (131) and an emotion analysis module (132). Furthermore, the control unit (130) may process signals, data, information, etc. that are input or output through the components described above, or provide or process appropriate information and functions to the user.

[0087] The control unit (130) can control the output of a service page for cognitive behavioral therapy through a display unit (or touchscreen) provided in the user terminal (10). Such a service page may be output on the user terminal (10) through an application or web page installed on the user terminal (10). The service page may be a page linked to the treatment program providing system (100) according to the present invention and may be configured to be controlled by the treatment program providing system (100) according to the present invention.

[0088] Furthermore, if the service page is provided in the form of an application, the service page may be controlled by the CPU (Central Processing Unit) of the user terminal (10) on which the application is installed. In this case, the CPU of the user terminal (10) may generate a treatment program and update the generated treatment program based on information provided by the treatment program providing system (100) according to the present invention.

[0089] Meanwhile, the control unit (130) may provide at least one survey related to cognitive behavioral therapy for the user on a service page provided to the user terminal (10). Furthermore, the control unit (130) may extract natural language response data from the response data for at least one survey received through the user terminal (10).

[0090] Furthermore, the control unit (130) can analyze (or classify) the sentiment type (or sentiment class) corresponding to each of the multiple natural language response data using the sentiment analysis module (132). Specifically, the sentiment analysis module (132) can preprocess the natural language response data (e.g., stop word removal, sentence-unit tokenization, etc.) and convert the preprocessed text data into a semantic vector by embedding it. Furthermore, the sentiment analysis module (132) can classify the sentiment type (or class) of the embedded semantic vector using at least one classifier. At this time, it can be understood that both multi-class classification and multi-label classification methods can be applied to at least one classifier. For example, an emotion type (or emotion class) according to the present invention may be defined as various types such as 'positive', 'negative', 'depressed', 'anxious', 'anger', 'sadness', 'lethargy', 'joy', 'calm', 'confusion', etc., and the types of emotion types are not limited in this specification. Hereinafter, prompts to be input into a large language model are described based on two different emotion types, but this is not limited thereto, and additional prompts corresponding to each emotion type may be generated according to various emotion type classifications.

[0091] The control unit (130) can group natural language response data corresponding to the same emotion type (or class) based on the analysis (or classification) results of each of the multiple natural language response data. Furthermore, the control unit (130) can generate prompts to be input into a large language model by using user information collected from the database (200) and the groups of natural language response data corresponding to each of the different emotion types. Specifically, the control unit (130) can generate different types of prompts corresponding to each emotion type associated with the multiple groups of natural language response data.

[0092] The LLM server (140) according to the present invention may include at least one large language model. For example, the LLM server (140) may include at least one large language model based on at least one of GPT (Generative Pre-trained Transformer), T5 (Text-to-Text Transfer Transformer), Vision-Language Models (e.g., CLIP, Flamingo), Multimodal Generative Models (GPT-4 Multimodal, PaLM-E), BERT (Bidirectional Encoder Representations from Transformers), and LaMDA (Language Model for Dialogue Applications).

[0093] In the present invention, the LLM server (140) is described as existing separately from the treatment program providing system (100), but is not limited thereto, and the treatment program providing system (100) may be configured to include the LLM server (140). That is, the treatment program providing system (100) and the LLM server (140) according to the present invention may exist separately, or the LLM server (140) may be included in the treatment program providing system (100). For convenience of explanation, the LLM server (140) and the large language model are used interchangeably below, and the use of the large language model by the control unit (130) can be understood as using at least one large language model included in the LLM server (140).

[0094] Additionally, although the present invention describes the control unit (130) as including a prompt generation unit (131), it is not limited thereto, and the control unit (130) and the prompt generation unit (131) may exist separately. In this case, the control unit (130) can generate a prompt in conjunction with the prompt generation unit (131).

[0095] The control unit (130) according to the present invention can generate feedback for cognitive behavioral therapy of a user according to a plurality of prompts corresponding to different input types through a large language model (140). Specifically, the large language model (140) can generate each of different types of feedback based on the type corresponding to the prompt.

[0096] Furthermore, the control unit (130) can update the prompt to be input into the large language model (140) using natural language response data, user information, and generated feedback. Based on the updated prompt, the large language model can update the treatment program for the user's cognitive behavioral therapy.

[0097] In the present invention, it is described that the control unit (130) generates feedback corresponding to natural language response data in conjunction with the large language model (140), and updates the prompt according to the generated feedback to generate and update the treatment program. However, this is not limited thereto, and the control unit (130) itself may generate and update the treatment program by including the function of the large language model (140) or performing the same function. Therefore, in this specification, the control unit (130) generating and updating the treatment program and the large language model (140) generating and updating the treatment program may be expressed interchangeably.

[0098] Meanwhile, the treatment program providing system (100) may include one or more processors, and such processors may include one or more general-purpose processors and / or one or more special-purpose processors (e.g., digital signal processors, tensor processing units (TPUs), graphics processing units (GPUs), neural network processing units (NPUs), application-specific integrated circuits, application-specific semiconductors (ASICs), etc.). One or more processors may be configured to execute instructions, computer-readable instructions, and / or other instructions described herein that are stored (or included) in the storage unit (120). The treatment program providing system (100) may perform data processing described below in cooperation with memory and at least one processor. The processor may perform a series of operations and data processing using data and information stored in memory. Here, “memory” may be a component of the storage unit (120), and “processor” may be used interchangeably with the control unit (130).

[0099] In the above description, the treatment program providing system (100) of the present invention has been described, and it can be implemented based on the large language model-based cognitive behavioral treatment program providing method described below.

[0100] Hereinafter, with reference to FIG. 2 together with FIG. 3a, FIG. 3b, FIG. 4a to FIG. 4c, FIG. 5a, FIG. 5b, FIG. 6a, and FIG. 6b, a method for providing a cognitive behavioral therapy program based on a large language model according to the present invention will be described in more detail. FIG. 2 is a flowchart for generally explaining a method for providing a cognitive behavioral therapy program based on a large language model according to the present invention, and FIG. 3a and FIG. 3b are conceptual diagrams for specifically explaining a survey provided to a user terminal according to the present invention. FIG. 4a is a conceptual diagram for explaining the process of extracting natural language response data from response data to a survey according to the present invention, FIG. 4b is a conceptual diagram for explaining the process of generating a natural language response data group based on an emotion type corresponding to the natural language response data according to the present invention, and FIG. 4c is a conceptual diagram for explaining the process of generating a prompt to be input into a large language model using natural language response data and user information according to the present invention. FIG. 5a is a conceptual diagram illustrating a cognitive behavioral therapy program according to the present invention, and FIG. 5b is a conceptual diagram illustrating a process of generating feedback related to cognitive behavioral therapy using a large language model according to the present invention. FIG. 6a and FIG. 6b are conceptual diagrams illustrating a process of updating a cognitive behavioral therapy program based on feedback according to the present invention and providing the updated therapy program to a user terminal.

[0101] In the present invention, a process of providing at least one questionnaire related to cognitive behavioral therapy to a user terminal may be carried out (S210, see FIG. 2).

[0102] The control unit (130) may provide at least one survey related to cognitive behavioral therapy on a service page (1000) displayed on a user terminal (10). Here, the at least one survey may include a multiple-choice survey consisting of at least one multiple-choice question item related to cognitive behavioral therapy and a plurality of selection items corresponding to the multiple-choice question item.

[0103] As illustrated in FIG. 3a, the control unit (130) may provide a survey page (or service page, 1000) containing a plurality of surveys on a user terminal (10). In this case, the page containing the plurality of surveys may be output through the touch screen (or display) of the user terminal (10). Here, the survey (or question, or problem, or item, or test) provided on the page may include a survey for relative evaluation of pain-related factors. In the present invention, pain-related factors may be diverse. For example, pain-related factors may include various elements for evaluating the patient's condition related to pain, such as pain level, duration of pain, mental health, and physical health. However, the above-described factors are merely examples, and it is sufficient for the pain-related factors described in the present invention to include all elements for evaluating the patient's condition.

[0104] These pain-related factors may include credible questionnaires actually used in psychiatry to diagnose the patient's condition. Additionally, the control unit (130) can periodically update the questionnaire by additionally collecting questionnaires to diagnose the patient's condition through a central server, an external server, or a website.

[0105] More specifically, the control unit (130) may provide a plurality of multiple-choice questionnaires corresponding to attributes of at least one pain-related factor on a service page (1000) output to a user terminal (10). For example, the multiple-choice questionnaire may include at least one of a first multiple-choice questionnaire corresponding to a first attribute (e.g., pain) and a second multiple-choice question corresponding to a second attribute (e.g., competence) relative to the first attribute. At this time, the attributes of the multiple-choice questionnaire provided to the user terminal (10) in the present invention are not limited to the described examples and may be very diverse.

[0106] As illustrated in FIG. 3a (a) and FIG. 3b, the control unit (130) provides a first multiple-choice survey (310) corresponding to a first attribute and a second multiple-choice survey corresponding to a second attribute to a service page (1000), and can receive responses to at least one multiple-choice question item (311, 312) related to the first attribute and responses to at least one multiple-choice question item (321, 322) related to the second attribute. Specifically, the multiple-choice survey may be composed of a multiple-choice question item related to cognitive behavioral therapy and a plurality of selection items corresponding to each of the plurality of different responses to the multiple-choice question item, forming pairs. At this time, the format of at least one multiple-choice question item constituting the multiple-choice survey according to the present invention may be very diverse.

[0107] For example, a first type of query item (311, 321) may be composed of pairs of multiple selection items corresponding to each of the questions related to cognitive behavioral therapy (e.g., “How is your pain today?”) and multiple different responses to said question (e.g., “no pain,” “almost minor pain,” “moderate,” “slightly severe pain,” “terrible pain”), wherein different scores related to cognitive behavioral therapy may be matched to each of the multiple different responses. For example, among the multiple different responses (e.g., “no pain,” “almost minor pain,” “moderate,” “slightly severe pain,” “terrible pain”) composed of the first type of query item (311, 321), a first score related to cognitive behavioral therapy (e.g., “points”) is matched to the first response (e.g., “no pain”), and a fifth score related to cognitive behavioral therapy (e.g., “points”) is matched to the fifth response (e.g., “almost minor pain”). It can exist.

[0108] As another example, the second type of query item (312, 322) may be composed of a pair of query items related to cognitive behavioral therapy (e.g., “Please select all symptoms of the current pain area?”) and multiple selection items corresponding to each of the multiple different responses to the query item (e.g., “stiffness in movement,” “pain during specific movements,” “weakness in muscle strength,” etc.). In this case, the multiple different responses may include selection items expressed in different natural languages ​​related to cognitive behavioral therapy.

[0109] Accordingly, as a response to at least one question item constituting a multiple-choice survey, the user may select a specific option that is considered to be the most suitable (or appropriate or accurate) for the user's condition among multiple option items corresponding to each of multiple different responses.

[0110] Meanwhile, the control unit (130) may provide at least one open-ended survey related to cognitive behavioral therapy on a service page (1000) displayed on a user terminal (10). Specifically, the at least one survey provided on the service page may further include an open-ended survey capable of receiving natural language input for at least one open-ended question item from the user terminal in relation to the cognitive behavioral therapy.

[0111] As illustrated in FIG. 3b (a), the control unit (130) may provide at least one open-ended survey (331, 334, 335) corresponding to different attributes to the user terminal (10). For example, the control unit (130) may provide a first open-ended survey (331) corresponding to a first attribute (e.g., pain) to the user terminal (10). At this time, the first open-ended survey (331) may provide an open-ended question item related to the first attribute (e.g., “Please freely describe the pattern or characteristics of the pain you are currently feeling in your own words,” 332) and an input area for inputting a natural language response to the open-ended question item. Here, the input area may refer to an interface capable of receiving a natural language response from at least one of text and voice.

[0112] The control unit (130) according to the present invention may further include a module that converts voice input into an input area into text. For example, the control unit (130) may input voice data received from a user terminal (10) into a voice conversion module. Specifically, the voice conversion module may include at least one voice recognition model based on a Speech-to-Text (STT) algorithm that converts voice data into text. As an example, the voice conversion module may include at least one voice recognition model among a Hidden Markov Model (HMM), a Hidden Markov Model-Gaussian Mixture Model (HMM-GMM), a Connectionist Temporal Classification (CTC) based model, a Beam Search based model, a Deep Neural Network (DNN) based model, a Recurrent Neural Network (RNN) based model, a Sequence-to-Sequence (Seq2Seq) model, and a Transformer based model.

[0113] The control unit (130) can convert voice (or voice data) into text based on a STT (Speech-to-Text) algorithm and process the converted text as response data for subjective question items.

[0114] As illustrated in (b) of FIG. 3b, the control unit (130) can conduct an interactive open-ended survey (340). Here, the “interactive open-ended survey” can be performed by using at least one large language model (140) and at least one chatbot to conduct a conversation related to cognitive behavioral therapy and to collect response data input through the conversation process. Here, at least one large language model (140) may refer to a large language model (140) based on at least one of GPT (Generative Pre-trained Transformer), T5 (Text-to-Text Transfer Transformer), Vision-Language Models (e.g., CLIP, Flamingo), Multimodal Generative Models (GPT-4 Multimodal, PaLM-E), BERT (Bidirectional Encoder Representations from Transformers), and LaMDA (Language Model for Dialogue Applications).

[0115] In addition, the “chatbot” according to the present invention may be implemented to receive information that is difficult to obtain from multiple-choice questionnaires in relation to cognitive behavioral therapy. For example, the “chatbot” according to the present invention may include at least one of a rule-based chatbot, an artificial intelligence chatbot, a natural language processing chatbot, and a hybrid chatbot.

[0116] As illustrated in (b) of FIG. 3b, the control unit (130) may conduct an interactive open-ended survey (340) to collect information related to cognitive behavioral therapy. For example, the control unit (130) may generate and provide an interactive open-ended question item (341) related to cognitive behavioral therapy on a service page (1000) provided to a user terminal (10). In this case, the control unit (130) may provide the interactive open-ended survey (340) through at least one of a large language model (140) and a chatbot so that the user can freely respond to the interactive open-ended question item (341). Furthermore, the control unit (130) may receive the user's response (or natural language input, 342) to the interactive open-ended question item (341) through the service page (1000). In this case, the user's response may refer to natural language input entered as at least one of text and voice for a subjective question item.

[0117] Meanwhile, at least one questionnaire according to the present invention is not limited to the examples of questionnaires (or questions) described above, and may include at least one questionnaire among multiple-choice questionnaires and open-ended questionnaires, and the at least one questionnaire may include various questionnaires (or questions) related to cognitive behavioral therapy.

[0118] In the present invention, the control unit (130) can receive response data for at least one survey provided to the user terminal (10). Specifically, the control unit (130) can receive objective survey response data corresponding to objective surveys included in at least one survey.

[0119] The control unit (130) may receive response data for a multiple-choice survey (or multiple-choice survey response data) from a user terminal (10), which includes a response matched to an item selected by user input among a plurality of selection items. At this time, the multiple-choice survey response data may include natural language response data corresponding to a specific selection item selected by user input among a plurality of selection items.

[0120] For example, as illustrated in (a) and (b) of FIG. 3a, the control unit (130) can receive a natural language response (e.g., recognized the importance of health through pain) as a choice response data corresponding to a choice selected by user input among a plurality of choice items for a question related to the user’s indication (e.g., “What is your thought regarding the impact of current pain on life?”) included in the multiple-choice survey (310, 320).

[0121] Furthermore, the control unit (130) may receive subjective survey response data for a subjective survey included in at least one survey. At this time, the subjective survey may include natural language input for a subjective question item and natural language response data corresponding to at least one of the natural language answers obtained through a large language model by processing the natural language input as input to a large language model. As an example, as illustrated in (b) of FIG. 3b, the control unit (130) may receive natural language input (e.g., “Daily life is difficult. I can’t do anything, so I’m depressed”, 342) for a subjective question item (or, conversational subjective question item, 341) as subjective survey response data.

[0122] Although the above description explains how to receive subjective survey response data using an interactive subjective survey (340) as an example, the present invention may also receive natural language input entered in an input area where a response to a subjective question item (332) provided in the subjective survey (330) can be entered as subjective survey response data.

[0123] Furthermore, the control unit (130) may provide a preset number of open-ended survey questions (or interactive open-ended survey questions) to the user terminal (10) and receive open-ended survey response data for each open-ended survey question. At this time, the control unit (130) may provide sample answers to the open-ended survey questions along with the open-ended survey questions.

[0124] In the present invention, a process of generating a prompt to be input into a large language model using response data to a survey received from a user terminal can be performed (S220, see FIG. 2).

[0125] As illustrated in FIG. 4a, the control unit (130) can extract a plurality of natural language response data (430) from response data (400) for a survey received from a user terminal. Specifically, the control unit (130) can extract natural language response data (430) from at least one of the objective survey response data (410) and subjective survey response data (420) included in the response data. For example, the control unit (130) can extract subjective survey response data (e.g., daily life is difficult, I can't do anything, I'm depressed) received from the user terminal (10) as the first natural language response data (431) for a subjective question item included in the subjective survey (e.g., Please freely explain regarding the pain you are currently feeling).

[0126] For another example, the control unit (130) can extract a multiple-choice survey response data (e.g., I realized the importance of health through pain) corresponding to a multiple-choice question item included in the multiple-choice survey (e.g., What is your thought closest to the impact of current pain on your life?) selected through user input among multiple options as a second natural language response data (432).

[0127] Furthermore, the control unit (130) can perform sentiment analysis (or classification) on a plurality of natural language response data (400) extracted from response data (400) using a sentiment analysis module (132). Specifically, the sentiment analysis module (132) can identify a sentiment type (or class) corresponding to each of the extracted plurality of natural language response data. Here, the sentiment type can be identified as at least one of 'positive' and 'negative'.

[0128] The method for specifying the emotion type (or class) corresponding to each of the extracted plurality of natural language response data according to the present invention may be very diverse, and the present specification does not limit the method for specifying the emotion type (or class) corresponding to each of the extracted plurality of natural language response data. In the present invention, the emotion analysis module (132) is not limited in type and method as long as it is a module capable of specifying the emotion type (or class) corresponding to each of the extracted plurality of natural language response data. The control unit (130) according to the present invention may further include at least one of a module and an algorithm that perform the same function as the emotion analysis module (132). At this time, the emotion analysis module can specify the emotion type (or class) corresponding to each of the extracted plurality of natural language response data through at least one algorithm.

[0129] For example, the sentiment analysis module (132) can perform a preprocessing process for sentiment analysis of natural language response data. Specifically, the sentiment analysis module (132) may consist of word segmentation using a morphological analyzer (e.g., KoNLPy-based Okt, Mecab, etc.), part-of-speech tagging, stop word removal, and spelling and spacing refinement. Through this process, a token sequence suitable for sentiment analysis can be constructed.

[0130] Furthermore, the sentiment analysis module (132) can input preprocessed natural language response data into an artificial intelligence-based sentiment classification model. At this time, the sentiment classification model may be configured based on a pre-trained large language model (140). For example, the sentiment classification model may use at least one of a Korean-specific BERT model (KoBERT), a sentiment analysis-tuned KLUE-BERT, or a GPT-family language model.

[0131] The sentiment analysis module (132) can embed preprocessed natural language response data using a sentiment classification model and generate a semantic vector that reflects the contextual meaning of the natural language response data. Furthermore, the sentiment analysis module (132) can input the semantic vector into a classifier that performs sentiment type classification corresponding to the semantic vector. Here, the classifier may refer to at least one of an artificial intelligence model and an algorithm capable of identifying the sentiment type (or sentiment class) of each of the multiple natural language response data using a semantic vector corresponding to each of the multiple natural language response data. For example, the classifier may include at least one of machine learning-based classifiers such as Nave Bayes, Support Vector Machine (SVM), Random Forest, and Logistic Regression; deep learning-based classifiers such as LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit), and CNN (for text classification); Korean-specific models such as BERT (Bidirectional Encoder Representations from Transformers), RoBERTa, KoBERT, and KLUE-BERT; and large language model-based classifiers such as GPT and T5. The classifier according to the present invention is not limited to the algorithms or models described above, and may further include at least one of models and algorithms that perform the same function as the classifier according to the present invention.

[0132] As an example, the classifier according to the present invention inputs a semantic vector into at least one of a multilayer perceptron (MLP) and a recurrent neural network (RNN) family structure (e.g., BiLSTM), and in the final output layer, can calculate a probability distribution for each emotion type through a softmax function. The emotion types may consist of, for example, 'positive' and 'negative', and the type (or class) with the highest probability can be identified as the emotion type corresponding to the natural language response data.

[0133] As illustrated in FIG. 4b, the control unit (130) can analyze the emotion type corresponding to each of the plurality of natural language response data and group the natural language response data corresponding to the same emotion type among the plurality of natural language response data. Specifically, the control unit (130) can generate a first type natural language response data group (440) that includes at least one natural language response data corresponding to the first emotion type among the plurality of natural language response data. For example, the control unit (130) can generate a first type natural language response data group (440) that includes a first natural language response data (e.g., “Daily life is difficult. I feel depressed because I can’t do anything”, 431) and a fourth natural language response data (e.g., “I felt lethargic and wanted to give up”, 434) that correspond to the first emotion type (e.g., negative) among the plurality of natural language response data.

[0134] Alternatively, the control unit (130) may generate a second type natural language response data group (450) that includes at least one natural language response data corresponding to a second emotion type among a plurality of natural language response data. For example, the control unit (130) may generate a second type natural language response data group (450) that includes a second natural language response data (e.g., “I actually realized the importance of health through the pain”, 432) and a third natural language response data (e.g., “I had some pain, but I finished everything I had to do today”, 433) that correspond to a second emotion type (e.g., positive) different from a first emotion type among a plurality of natural language response data.

[0135] Furthermore, the control unit (130) can generate multiple different prompts to be input into a large language model based on each of the multiple natural language response data groups corresponding to different emotion types. As illustrated in FIG. 4c, the control unit (130) can generate multiple prompts (470) corresponding to each of the multiple natural language response data groups corresponding to different emotion types by using the prompt generation unit (131). Specifically, the prompt generation unit (131) can generate multiple prompts (470) of different types by using user information (460) collected from the database (200) and natural language response data including at least one of the first type natural language response data group (440) and the second type natural language response data group (450). Here, “user information (460)” may include at least one of the user’s i) name, ii) date of birth, iii) user’s cognitive behavioral therapy history, iv) medical records, v) surgical records, vi) past medical history, vii) prescription information and viii) past response data.

[0136] For example, the prompt generation unit (131) can generate a plurality of different prompts (470) including at least one of a first type prompt associated with a first type natural language response data group and a second type prompt associated with a second type natural language response data group.

[0137] Specifically, the prompt generation unit (131) can generate a first type prompt (480) by using at least one of the natural language response data and user information (460) corresponding to the first sentiment type included in the first type natural language response data group (440). Alternatively, the prompt generation unit (131) can generate a second type prompt (490) by using at least one of the natural language response data and user information (460) corresponding to the second sentiment type included in the second type natural language response data group (450).

[0138] In the present invention, a process may be carried out in which a prompt is input into a large language model to generate feedback related to a treatment program for cognitive behavioral therapy based on natural language response data included in the response data (S230, see FIG. 2).

[0139] Meanwhile, a plurality of treatment programs corresponding to different topics may be stored in the database (200) according to the present invention. The control unit (130) may sequentially provide a plurality of treatment contents constituting a specific treatment program to the user terminal (10) according to the treatment week set for a specific treatment program related to the user's cognitive behavioral therapy among the plurality of treatment programs. Here, "treatment contents" may refer to contents related to a detailed category (or sub-topic) for cognitive behavioral therapy of a pain patient regarding a specific topic, and may include worksheets, educational materials, assignments, etc. composed of at least one of text, voice, image, or video related to the specific topic.

[0140] The “treatment week” described in the present invention can be understood as the sequence (or period) of providing (or activating) treatment content through a user terminal (10) so that the user performs cognitive behavioral therapy according to the treatment program. In the present invention, a “total number of treatment weeks” and a “weekly treatment period” corresponding to each treatment week may be pre-set. Additionally, in the present invention, it can be understood that a “total treatment period” is pre-defined.

[0141] For convenience of explanation, the following description will be given as an example where different treatment weeks arrive “times” during each one-week treatment period. That is, in the present invention, the first treatment week arrives during the first treatment session, and the second treatment week arrives during the second treatment session. Accordingly, in the present invention, “treatment week” may be used interchangeably with “treatment session,” “treatment round,” “treatment period,” and “treatment sequence.”

[0142] As illustrated in FIG. 5a, the present invention may have a predefined total number of times (e.g., “8”) for a specific treatment program (500). Here, assuming that the specific treatment program includes “first treatment content to eighth treatment content,” in this case, any one of the first treatment week to eighth treatment week may be set and exist in each of the first treatment content to eighth treatment content (503).

[0143] Furthermore, for each of the multiple therapeutic contents (e.g., first therapeutic content to eighth therapeutic content), multiple topics (502) may correspond (or be matched) to each. For example, the first therapeutic content may refer to content for training or treatment related to the first topic (e.g., motivational reinforcement), and the second therapeutic content may refer to content for training or treatment related to the second topic (e.g., emotion identification).

[0144] In the present invention, a plurality of treatment contents constituting a specific treatment program (500) can be provided sequentially to a user terminal (10) in accordance with the treatment week set in the specific treatment program. For example, the control unit (130) can provide a first treatment content among a plurality of treatment contents (503) constituting the specific treatment program (500) to the user terminal (10) sequentially during the first treatment week.

[0145] As illustrated in FIG. 5b, the control unit (130) can use a large language model (140) to generate feedback (510) for cognitive behavioral therapy corresponding to a prompt (470). Specifically, the control unit (130) can use multiple prompts of different types to generate multiple feedbacks of different types corresponding to each of the multiple prompts. For example, the control unit (130) can input at least one of a first type prompt (480) and a second type prompt (490) into the large language model (140) to generate feedback for a specific treatment program through the large language model (140).

[0146] Specifically, the control unit (130) can use a large language model (140) to generate a first type feedback (520) corresponding to a first type prompt (480) and generate a second type feedback (530) corresponding to a second type prompt (490). Here, the first type feedback (520) may include first type update information (521) for a specific treatment program provided to the user terminal (10) according to the first type prompt (480), which includes first type natural language response data corresponding to a first emotion type.

[0147] For example, when the first emotion type corresponds to the 'negative' type, the large language model (140) can analyze the first type natural language response data to generate a first analysis result (e.g., “The current user’s response shows recurring helplessness, loss of meaning, and skepticism about change. This is a reaction commonly seen in users with high levels of cognitive exhaustion and depression.”). Furthermore, based on the first analysis result, the large language model (140) can generate first type feedback (520) including first type update information (521) for updating a specific treatment program.

[0148] Alternatively, the second type feedback (530) may include second type update information (532) for a specific treatment program provided to the user terminal (10) according to a second type prompt (490) associated with a second type natural language response data group corresponding to the second emotion type. As an example, when the second emotion type corresponds to the 'positive' type, the large language model (140) may analyze the second type natural language response data to generate a second analysis result (e.g., “The user is showing autonomy and continuity in behavioral activation and emotion regulation during the recovery process, which is a very positive indicator. In particular, a reduction in emotional variability, recovery of daily rhythm, and improvement in self-awareness are observed.”). Furthermore, the large language model (140) may generate second type feedback (530) including second type update information (531) for updating the specific treatment program based on the second analysis result.

[0149] That is, in the present invention, each of the first type feedback (520) and the second type feedback (530) may include different types of update information for a specific treatment program. Through this, the control unit (130) can perform different types of updates for a specific treatment program provided to the user terminal (10) based on different types of update information.

[0150] In the present invention, based on the generated feedback, a process may be carried out to update a prompt, input it into a large language model, and update a treatment program through the large language model (S240, see FIG. 2).

[0151] As illustrated in FIG. 6a, the control unit (130) can update a prompt (600) to be input into a large language model (140) using at least one of natural language response data (430) including groups of different types of natural language response data (440, 450), user information (460), and feedback (510) for cognitive behavioral therapy including multiple types of different types of feedback (520, 530).

[0152] Specifically, the control unit (130) can update the prompt based on different types of update information regarding a specific treatment program included in the feedback (510). There may be many different methods for updating the prompt based on different types of update information according to the present invention, and the present specification does not limit the method of updating the prompt based on different types of update information. In the present invention, the prompt generation unit (131) is not limited in type and method as long as it is a module (or unit) capable of updating the prompt based on different types of update information. The control unit (130) according to the present invention may further include at least one of a module and an algorithm that perform the same function as the prompt generation unit (131). At this time, the prompt generation unit (131) can update the prompt based on different types of update information through at least one algorithm.

[0153] For example, the control unit (130) can analyze update information included in user feedback (510) for a specific treatment program and update a prompt to repeatedly provide or omit at least one treatment content in which the satisfaction score of the treatment content constituting the specific treatment program satisfies a pre-set satisfaction standard. Here, the “satisfaction score” may be included in the user’s response data regarding the treatment content provided in the previous week, based on the treatment history information included in the user information. Specifically, when the control unit (130) performs the ‘3rd session’ of a specific treatment program, it may provide a survey related to the satisfaction of the treatment content performed in the ‘2nd session’ and receive information related to the satisfaction score from the user terminal (10) as response data. At this time, the “pre-set satisfaction standard” according to the present invention may be set in various ways, and the present invention is not limited to the pre-set satisfaction standard.

[0154] As another example, when the control unit (130) receives natural language response data of a specific type (e.g., positive type, etc.) regarding specific therapeutic content (e.g., meditation, walking, attention diversion techniques) included in a specific therapeutic program, the control unit (130) can update a prompt to add a list of therapeutic content similar to the specific therapeutic content to the specific therapeutic program.

[0155] Furthermore, the control unit (130) can update a prompt for updating a specific treatment program based on at least one of natural language response data (430), user information (460), and feedback (510) for cognitive behavioral therapy. In this way, the control unit (130) can generate an updated prompt (600) that updates the structure and response method of the treatment program for cognitive behavioral therapy based on changes in the user's state and feedback.

[0156] The control unit (130) can process the updated prompt (600) as input to the large language model (140) to update the treatment program for the user's cognitive behavioral therapy. For example, based on the updated prompt (600), the control unit (130) can update information related to a specific topic matched to a specific treatment program, the configuration of each of the plurality of treatment contents included in the specific treatment program, and at least one of the treatment weeks in which each of the plurality of treatment contents is provided.

[0157] As illustrated in FIG. 6b, the control unit (130) can update a specific treatment program for cognitive behavioral therapy using a large language model (140). Specifically, the control unit (130) can update the specific treatment program based on at least one of natural language response data included in the updated prompt (600), multiple feedbacks of different types, and user information. For example, the control unit (130) can change specific treatment content provided to the user terminal (10) sequentially according to the specific treatment program based on specific type of feedback included in the updated prompt. At this time, the control unit (130) can receive the specific treatment content and the modified treatment content to be changed from among a plurality of treatment contents already stored in the database.

[0158] For example, the control unit (130) can update a specific treatment program (610) by changing specific treatment content in a specific treatment session to first treatment content according to first type feedback (e.g., “Reconfiguring behavior to start small changes”, 620) based on first type feedback included in the updated prompt. For another example, the control unit (130) can update a specific treatment program (610) by changing specific treatment content in a specific treatment session to second treatment content according to second type feedback (e.g., “Recovery strategy design exercises”, 630) based on second type feedback included in the updated prompt.

[0159] That is, the control unit (130) may update a specific topic matched to the specific treatment program, the configuration of each of the plurality of treatment contents, and the treatment week in which each of the plurality of treatment contents is provided, in order to provide a specific treatment program including a first treatment content and a second treatment content. For example, if the first type feedback includes a content that the first treatment content must be provided in a specific treatment week (e.g., 3 sessions), the control unit (130) may update (or change) the treatment week in which the plurality of treatment contents constituting the specific treatment program are provided sequentially in order to provide the first treatment content in a specific treatment week.

[0160] Furthermore, the control unit (130) can update the delivery method of multiple therapeutic contents constituting a specific therapeutic program (e.g., method of presenting examples, emotional tone of feedback, degree of empathy, etc.). For example, the control unit (130) can update the specific therapeutic program based on the first type of feedback to provide a response related to the first emotional type (e.g., how about taking a walk for just 5 minutes or drinking a glass of water today?) along with specific therapeutic content related to the first emotional type (e.g., negative type).

[0161] In the present invention, a process of providing an updated treatment program to the user terminal may be carried out (S250, see FIG. 2).

[0162] As previously described, the control unit (130) can update a specific treatment program to be provided to the user terminal (10) in a user-customized manner by using a large language model (140). Furthermore, the control unit (130) can provide the updated specific treatment program to the user terminal (10). The control unit (130) can provide a user environment that enables a user undergoing cognitive behavioral therapy to continuously perform cognitive behavioral therapy. For example, the control unit (130) can provide a GUI (Graphical User Interface) that allows the user to access the specific treatment program along with recommendation information for the updated specific treatment program. The user can select the GUI (e.g., “Proceed Now”) to access a screen where the updated specific treatment program related to the recommendation information can be performed, and then proceed with cognitive behavioral therapy through the updated specific treatment program. Based on the selection of the GUI by the user terminal (10), the control unit (130) can provide a page where the specific treatment program can be performed. The control unit (130) can provide a page containing the updated specific treatment program on the user terminal (10).

[0163] In the foregoing, a method for providing a cognitive behavioral therapy program based on a large language model according to the present invention has been described in detail. Below, a method for predicting a user's emotional pattern and providing prediction information corresponding to the user's emotional pattern to a user terminal (10) using the method for providing a cognitive behavioral therapy program based on a large language model is described. FIG. 7 is a conceptual diagram for explaining the process of predicting a user's emotional pattern according to the present invention and providing notification information corresponding to the predicted emotional pattern to a user terminal.

[0164] Meanwhile, the control unit (130) can use a large language model to predict the emotional pattern of a user who is the subject of cognitive behavioral therapy and provide notification information corresponding to the predicted emotional pattern to the user terminal.

[0165] As illustrated in FIG. 7, the control unit (130) can generate a prediction prompt for predicting the emotion pattern using at least one of the response data (400) for a survey, feedback (510) for cognitive behavioral therapy, and user information (460) collected from the database (200). As previously described, the response data (400) for a survey includes at least one of multiple-choice survey response data (410) and open-ended survey response data (420), and the control unit (130) can extract natural language response data from at least one of the multiple-choice survey response data (410) and open-ended survey response data (420).

[0166] The control unit (130) can generate a prediction prompt for predicting a user's emotion pattern based on a plurality of natural language response data and different types of feedback (520, 530) corresponding to different emotion types included in at least one of the multiple-choice survey response data (410) and the open-ended survey response data (420).

[0167] Additionally, the control unit (130) can infer an emotional pattern (or psychological pattern, behavioral tendency, emotional state, etc.) exhibited by a user based on at least one of natural language response data, user information (460), and feedback (510) for cognitive behavioral therapy. For example, the control unit (130) can analyze each of different types of natural language response data to extract repeatedly mentioned keywords (or sentences (e.g., “I always fail,” “Nobody likes me”)). Specifically, the control unit (130) can use an emotion analysis module (132) to extract repeated keywords and sentences corresponding to different emotional types.

[0168] The method for extracting repeat keywords and sentences corresponding to different emotion types according to the present invention may be very diverse, and the present specification does not limit the method for extracting repeat keywords and sentences corresponding to different emotion types. In the present invention, the emotion analysis module (132) is not limited in type and method as long as it is a module capable of extracting repeat keywords and sentences corresponding to different emotion types. The control unit (130) according to the present invention may further include at least one of a module and an algorithm that perform the same function as the emotion analysis module (132). At this time, the emotion analysis module (132) can extract repeat keywords and sentences corresponding to different emotion types through at least one algorithm.

[0169] For example, the sentiment analysis module (132) may include at least one keyword extraction algorithm. Specifically, the sentiment analysis module (132) can extract repeated keywords and sentences corresponding to different sentiment types through at least one of TF-IDF, Count Vectorizer-based frequency-based keyword extraction, TextRank, KeyBERT-based key expression extraction, and BERT-based attention weight analysis.

[0170] Additionally, the control unit (130) can infer emotional patterns (e.g., social avoidance, helplessness, excessive self-blame) by reflecting at least one of the age, gender, underlying indication information, and treatment history included in the user information (460). For example, the control unit (130) can predict behavioral tendencies based on individual user characteristics by using a pre-learned behavioral pattern model related to the user information. Here, the pre-learned behavioral pattern model may refer to a prediction model constructed by learning the statistical correlations of thoughts, emotions, and behavioral patterns that are repeatedly observed according to a specific user group (age, gender, diagnosis, treatment history, etc.).

[0171] Specifically, the pre-trained behavioral pattern model can be trained based on training data including large-scale clinical records, survey response data, and datasets collected during cognitive behavioral therapy. As an example, the pre-trained behavioral pattern model can be implemented using the aforementioned training data as a supervised learning-based classifier, a clustering-based type classification model, or a time-series analysis-based prediction model. For instance, 'age' information may be associated with avoidant regression behavior or a tendency toward reduced self-efficacy in adolescents; 'gender' information may be linked to the suppression of emotional expression or increased expression of anger in male users; and 'underlying indication information'—specifically, the presence of a history of depression, panic disorder, or PTSD—can be understood to be highly associated with specific emotional and behavioral patterns such as lethargy, insomnia, and excessive self-blame.

[0172] The control unit (130) can generate a prediction prompt that generates notification information including prediction information corresponding to the predicted emotion pattern by using at least one of response data (400) for a survey, feedback (510) for cognitive behavioral therapy, and user information (460) collected from a database (200).

[0173] Furthermore, the control unit (130) can process the prediction prompt as input to the large language model and generate prediction information of a type corresponding to the emotion pattern through the large language model. For example, the control unit (130) can process the prediction prompt as input to the large language model (140) and generate prediction information corresponding to the predicted emotion pattern (e.g., possibility of future emotional deterioration, state of accumulated fatigue, recurring negative emotions, etc.) through the large language model (140).

[0174] For a specific example, the large language model (140) can generate first type prediction information corresponding to the first emotion pattern (e.g., “I don’t want to do anything” and “I have no energy”) based on a prediction prompt that generates prediction information corresponding to the first emotion pattern associated with the first emotion type (e.g., “Responses such as “I don’t want to do anything” and “I have no energy” have been repeated twice in a row recently. If lethargy continues, it may lead to reduced activity and worsening depression. How about taking a walk for just 5 minutes or drinking a glass of water today?”, 720).

[0175] Alternatively, the large language model (140) can generate second type prediction information corresponding to the second emotion pattern (e.g., “Feeling better”, “I’ve started walking again”) based on a prediction prompt that causes the large language model (140) to generate prediction information corresponding to the second emotion pattern associated with the second emotion type (e.g., “Positive” type) and the second type prediction information (e.g., “In your recent responses, the content of ‘feeling better’ and ‘starting walking again’ is repeating. This is clearly a sign of recovery. These small changes you have made are giving you great strength. Please continue to observe your emotions and take time to take care of yourself as you are now. You are doing great!”, 730). That is, the control unit (130) can generate notification information (710) containing different types of prediction information (720, 730) based on the type of the predicted emotion pattern.

[0176] Furthermore, the control unit (130) can provide notification information generated through a large language model (140) to a user terminal. Specifically, the control unit (130) can provide notification information including prediction information corresponding to the emotion pattern to the user terminal (10) according to the predicted emotion pattern.

[0177] Additionally, the control unit (130) may provide notification information containing prediction information corresponding to the emotion pattern to the user terminal (10). Here, the notification information may be provided to the user terminal (10) in the form of a pop-up, a notification window, an interactive message, etc.

[0178] As described above, the method and system for providing a large language model-based cognitive behavioral therapy program according to the present invention can improve the efficiency of cognitive behavioral therapy and user engagement by providing a user-customized cognitive behavioral therapy program based on user survey response data and user information.

[0179] Furthermore, the method and system for providing a large language model-based cognitive behavioral therapy program according to the present invention can dynamically adjust the therapy program by updating the therapy program provided to the user based on feedback according to the user's current state, thereby responding sensitively to changes in the user's state and ensuring the continuity of long-term therapy.

[0180] Furthermore, the method and system for providing a large language model-based cognitive behavioral therapy program according to the present invention can enable an early intervention and prevention-centered treatment approach for cognitive behavioral therapy by predicting a user's emotional state in advance and preemptively providing notification information based on the predicted emotional state.

[0181] Meanwhile, computer-readable media include all types of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SSD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

[0182] Furthermore, the computer-readable medium may be a server or cloud storage that includes a storage and is accessible to an electronic device via communication. In this case, the computer may download the program according to the present invention from the server or cloud storage via wired or wireless communication.

[0183] Furthermore, in the present invention, the computer described above is an electronic device equipped with a processor, namely a CPU (Central Processing Unit), and no special limitations are placed on its type.

[0184] Meanwhile, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention.

Claims

1. A step of providing at least one questionnaire related to cognitive behavioral therapy to a user terminal; A step of generating a prompt to be input into a large language model using response data to the survey received from the user terminal; A step of inputting the above prompt into the above large language model to generate feedback related to the treatment program for cognitive behavioral therapy based on natural language response data included in the above response data; A step of updating the prompt based on the generated feedback, inputting it into the large language model, and updating the treatment program through the large language model; and A method for providing a large language model-based cognitive behavioral therapy program characterized by including the step of providing the updated treatment program to the user terminal.

2. In Paragraph 1, The step of generating the above prompt is, A step of extracting multiple natural language response data from the response data for the above survey; A step of analyzing the emotion type corresponding to each of the plurality of natural language response data and grouping the natural language response data corresponding to the same emotion type among the plurality of natural language response data; and A method for providing a large language model-based cognitive behavioral therapy program, characterized by including the step of generating multiple different prompts to be input into the large language model based on each of a multiple natural language response data group corresponding to different emotion types.

3. In Paragraph 2, The above plurality of natural language response data groups are, A first type natural language response data group comprising at least one natural language response data corresponding to a first sentiment type among the plurality of natural language response data above, and A method for providing a large language model-based cognitive behavioral therapy program characterized by including at least one of at least one group of second type natural language response data, which includes at least one natural language response data corresponding to a second emotion type among the plurality of natural language response data.

4. In Paragraph 3, The step of generating the above-mentioned multiple different prompts is, Generate multiple prompts corresponding to each of the multiple natural language response data groups corresponding to the above different emotion types, and The above plurality of prompts are, A method for providing a large language model-based cognitive behavioral therapy program characterized by including at least one of a first type prompt associated with the first type natural language response data group and a second type prompt associated with the second type natural language response data group.

5. In Paragraph 4, Multiple treatment programs corresponding to different topics are matched and stored in the database, and In the above user terminal, According to the treatment week set in the specific treatment program related to the user's cognitive behavioral therapy among the plurality of treatment programs mentioned above, a plurality of treatment contents constituting the specific treatment program are provided sequentially, and The step of generating the above feedback is, A method for providing a large language model-based cognitive behavioral therapy program characterized by inputting at least one of the first type prompt and the second type prompt into the large language model to generate feedback for the specific therapy program through the large language model.

6. In Paragraph 5, The above feedback is, It includes at least one of a first type feedback corresponding to the first type prompt and a second type feedback corresponding to the second type prompt. Each of the above-mentioned first type feedback and the above-mentioned second type feedback is, A method for providing a large language model-based cognitive behavioral therapy program characterized by including different types of update information for the specific treatment program mentioned above.

7. In Paragraph 6, The step of updating the above treatment program is, A method for providing a large language model-based cognitive behavioral therapy program, characterized by updating information related to at least one of the treatment weeks in which each of the plurality of treatment contents is provided, based on the different types of update information for the specific treatment program.

8. In Paragraph 1, The above at least one survey is, A multiple-choice questionnaire consisting of at least one multiple-choice question item related to the above-mentioned cognitive behavioral therapy and a plurality of selection items corresponding to the above-mentioned multiple-choice question item, and A method for providing a large language model-based cognitive behavioral therapy program, characterized by including at least one open-ended questionnaire capable of receiving natural language input for at least one open-ended question item from the user terminal in relation to the above cognitive behavioral therapy.

9. In Paragraph 8, The above response data is, It includes at least one of the objective survey response data for the above objective survey and the subjective survey response data for the above subjective survey, and The above multiple-choice survey response data is, It includes the natural language response data corresponding to a specific selection item selected by user input among the plurality of selection items above, and The above open-ended survey response data is, The above natural language input and for the above subjective query items A method for providing a large language model-based cognitive behavioral therapy program, characterized by processing the above natural language input as an input to the large language model and including the above natural language response data corresponding to at least one of the natural language answers obtained through the large language model.

10. In Paragraph 1, A step of predicting the emotional patterns of a user subject to cognitive behavioral therapy using the above-mentioned large language model, and The method further includes the step of providing notification information to the user terminal that includes prediction information corresponding to the emotion pattern according to the predicted emotion pattern. The step of predicting the emotional pattern of the user mentioned above is, A step of generating a prediction prompt for predicting the emotion pattern using at least one of the response data to the above survey, the above feedback, and user information collected from the database; and A method for providing a large language model-based cognitive behavioral therapy program, characterized by including the step of processing the above prediction prompt as input to the large language model and generating the above prediction information of a type corresponding to the above emotion pattern through the large language model.

11. A communication unit that provides at least one questionnaire related to cognitive behavioral therapy to a user terminal, and a control unit that generates a prompt to be input into a large language model using response data for the questionnaire received from the user terminal, The above control unit is, Input the above prompt into the above large language model to generate feedback related to the treatment program for cognitive behavioral therapy based on the natural language response data included in the above response data, and Based on the generated feedback, the prompt is updated and input into the large language model, and the treatment program is updated through the large language model, and A large language model-based cognitive behavioral therapy program providing system characterized by providing the updated treatment program to the user terminal.

12. Executed by one or more processes on an electronic device and can be read by a computer As a program stored on an existing recording medium, The above program is, A step of providing at least one questionnaire related to cognitive behavioral therapy to a user terminal; A step of generating a prompt to be input into a large language model using response data to the survey received from the user terminal; A step of inputting the above prompt into the above large language model to generate feedback related to the treatment program for cognitive behavioral therapy based on natural language response data included in the above response data; A step of updating the prompt based on the generated feedback, inputting it into the large language model, and updating the treatment program through the large language model; and A program stored on a computer-readable recording medium characterized by including instructions that perform the step of providing the updated treatment program to the user terminal.

Images