KR20240027185A - Content providing platform for digital psychological schema therapy and method thereof - Google Patents

Abstract

An embodiment of the present invention includes an AI-based depression diagnosis unit that transmits and receives a message to a user terminal, performs a depression diagnosis based on the transmitted and received message, and calculates a depression analysis result; And collect and crawl content published on the web, process and store the crawled content, match the content based on the depression analysis results and provide it to the user terminal, and determine content preference for the provided content. It provides a content providing platform for digital psychological schema therapy that includes a content recommendation unit that collects data from the user terminal and recommends the content based on the content preference and the depression analysis result.

Description

Content provision platform and method for digital psychological schema therapy {CONTENT PROVIDING PLATFORM FOR DIGITAL PSYCHOLOGICAL SCHEMA THERAPY AND METHOD THEREOF}

The present invention relates to a content provision platform and method for digital psychological schema therapy. More specifically, the present invention relates to a content provision platform and method for treating digital psychological schemas. More specifically, the present invention relates to a chatbot using natural language processing to determine the mental health status of a patient through daily conversations with the patient and perform an accurate diagnosis of the patient. It relates to a content provision platform and method that matches and provides mental illness treatment content based on diagnosis results.

Psychological schema therapy is a psychotherapy method that expands traditional cognitive behavioral therapy and is mainly used to improve depression, anxiety, PTSD (post-traumatic stress disorder), and chronic personality disorders. Here, because cognitive behavioral therapy focuses on solving problems that appear in the patient's current life, it is difficult to apply it to patients with chronic and difficult-to-treat psychological disorders. However, psychological schema therapy explores childhood and adolescence to understand the origins of a patient's psychological problems and helps solve the problem by finding ways to satisfy emotional needs.

Recently, as the demand for non-face-to-face treatment has exploded, research on psychological schema therapy based on digital therapy is actively underway. Digital therapy is a technology that uses advanced software (mobile applications, games, virtual reality, chatbots, etc.) to prevent, manage, and treat diseases or disorders. In order to take advantage of these digital therapeutic technologies, conventional technologies have been developed to diagnose and treat patients using software.

For example, when diagnosing depression, the doctor determines the degree of the patient's mental illness through patient information (language characteristics, speaking habits, etc.). Depression treatment is based on the patient's responses to the doctor's specific questions (past mental illness, lifestyle habits, environment, etc.). Outpatient treatment for many depressed patients who are not hospitalized is provided at intervals of more than one month, and the doctor in charge checks the mental health status of the depressed patient based on questions for depression diagnosis and the patient's answers at each repeated outpatient treatment. .

Therefore, for patients with depression, technology is needed to collect the mental health status of the patient's daily life through various methods after treatment and before the patient's visit to the hospital, and to provide the collected results to the patient's doctor.

Meanwhile, many patients diagnosed with mental illness find it difficult to disclose their medical history to those around them. Additionally, some patients diagnosed with mental illness refuse to seek medical help. In addition, some patients diagnosed with mental illness attempt to self-diagnose their mental health condition in their daily lives and receive treatment periodically.

If mental illness diagnosis and treatment can be done non-face-to-face, even patients who do not receive ongoing treatment can participate in diagnosis and treatment, allowing more patients to receive treatment than when using existing diagnosis and treatment methods.

Additionally, digital therapy services related to the treatment of mental illness have been actively developed recently. However, digital treatment services currently being developed only provide standardized mental illness treatment content and information. This digital treatment service has the problem of low satisfaction because it matches the patient's attributes and content and simply provides content with similar attributes.

Therefore, there is a need to develop technology that can conduct non-face-to-face diagnosis for patients diagnosed with mental illness and provide customized content and information to patients based on the diagnosis results.

The technical task to be achieved by the present invention is to classify depression keywords using a chatbot applying natural language processing, generate questions based on a depression evaluation scale, and provide a content provision platform and method that can objectively determine mental health status. It is done.

In addition, the technical task to be achieved by the present invention is to provide a content provision platform and method that can identify the mental health status of the patient, perform an accurate diagnosis of the patient, and generate data for appropriate treatment without separate management personnel. .

In addition, the technical task to be achieved by the present invention is to provide a content provision platform and method that can classify and store content related to mental health by crawling content on the web and construct data related to psychological schema therapy.

In addition, the technical task to be achieved by the present invention is to provide a content provision platform and method that can match and provide content based on the results of a patient's mental health analysis.

In addition, the technical problem to be achieved by the present invention is to collect and analyze the user's content preference for the provided content, and to provide a content provision platform that can recommend content based on the analyzed content preference and the patient's mental health analysis results; and That method is provided.

In order to achieve the above technical problem, an embodiment of the present invention includes an AI-based depression diagnosis unit that transmits and receives messages to a user terminal, performs a depression diagnosis based on the transmitted and received messages, and calculates a depression analysis result; And collect and crawl content published on the web, process and store the crawled content, match the content based on the depression analysis results and provide it to the user terminal, and determine content preference for the provided content. It provides a content providing platform for digital psychological schema therapy that includes a content recommendation unit that collects data from the user terminal and recommends the content based on the content preference and the depression analysis result.

The content recommendation unit includes a content collection unit that collects text content and video content published on the web; a content crawling unit that crawls and classifies the collected text content and video content; a text content processing unit that processes the crawled text content, divides it into cognitive elements, emotional elements, and physical elements, and gives a score to each element; Collect and extract the metadata of the crawled video content, filter the regular video content based on the metadata, classify it according to the cognitive element, the emotional element, and the physical element, and give a score for each element. a video content processing unit; a content database that stores metadata of the separated text content and video content; a content matching unit that matches the stored text content and the video content based on the depression analysis results; a content provider providing the matched text content and video content to the user terminal; a user preference analysis unit that collects content preferences for the text content and the video content from the user terminal and calculates a content preference score; a user preference database storing the content preferences; and a content recommendation unit that recommends the text content and the video content based on the content preference and the depression analysis result.

The AI-based depression diagnosis unit includes a depression keyword classification unit that classifies depression keywords from the text of a received message received through the chatbot; If the received message is classified as a depression keyword, a depression question generator that generates a question message about a depression questionnaire related to the depression keyword; a depression score calculation unit that calculates a depression score for the depression questionnaire from the text of the response message to the question message; A depression analysis engine that provides depression analysis results based on the calculated depression score; And a database including first artificial intelligence model data learned so that the depression keyword classification unit can classify the depression keyword, and second artificial intelligence model data learned so that the depression score calculation unit can calculate a depression score; It can be included.

The content matching unit may compare and match the depression score included in the depression analysis result with the score for each element included in the text content and the video content.

The content preference may include content usage frequency, content usage time, and content satisfaction.

The metadata may include the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content.

In addition, in order to achieve the above technical problem, an embodiment of the present invention includes the steps of: a) transmitting and receiving messages with a user terminal using a chatbot, performing a depression diagnosis based on the transmitted and received messages, and calculating a depression analysis result; b) collecting and crawling content published on the web, processing and storing the crawled content, matching the content with the content based on the depression analysis results, and providing the content to the user terminal; c) collecting content preferences for the provided content from the user terminal, and recommending the content based on the content preferences and the depression analysis results.

Step a) includes activating the chatbot; Receiving a message from a user terminal by the chatbot; classifying depression keywords in the text of the received message; Generating a question message about depression questionnaire items related to the classified depression keywords; Transmitting the generated question message to a user terminal; calculating a depression score from the text of a response message to the question message; And a step of outputting a depression analysis result of the user based on the calculated depression score, wherein the depression keyword classification step uses first artificial intelligence model data learned to classify the depression keyword from the message. In the depression score calculation step, learned second artificial intelligence model data can be used to calculate the depression score from the message.

Step b) includes collecting text content and video content published on the web; Crawling and classifying the collected text content and video content; Processing the crawled text content, dividing it into cognitive elements, emotional elements, and physical elements, assigning a score to each element, and storing it; Collect and extract the metadata of the crawled video content, filter the regular video content based on the metadata, classify it according to the cognitive element, the emotional element, and the physical element, and give a score for each element. and saving; Matching the stored text content and the video content based on the depression analysis results; Providing the matched text content and the video content to the user terminal; Collecting content preferences for the text content and the video content from the user terminal to calculate a content preference score; And recommending the text content and the video content based on the content preference and the depression analysis result. May include ;.

According to an embodiment of the present invention, it is possible to accurately diagnose and treat individual depression patients using a chatbot in daily life with a relatively simple configuration.

Additionally, according to an embodiment of the present invention, not only depression patients but also people who are unaware of depression symptoms can receive mental health care.

Additionally, according to an embodiment of the present invention, it is possible to crawl and store content related to mental health existing on the web and to construct unique data. Additionally, according to an embodiment of the present invention, by summarizing text content including expert psychological analysis information, users can be encouraged to understand easily and intuitively.

Additionally, according to an embodiment of the present invention, metadata of video content can be collected and extracted. Additionally, it is possible to filter based on metadata. Since metadata includes location information of video content, the location of video content can be determined without separately storing large video content.

Additionally, according to an embodiment of the present invention, content can be matched and provided based on the results of the patient's mental health analysis. In addition, users' content preferences for provided content can be collected and analyzed, and content can be recommended based on the analyzed content preferences and the patient's mental health analysis results. Therefore, more satisfactory content can be provided. Additionally, it is possible to induce users to increase the time they spend on the content provision platform.

1 is a diagram illustrating the operation of a content providing platform according to an embodiment of the present invention.
Figure 2 is a diagram showing the configuration of an AI-based depression diagnosis unit according to an embodiment of the present invention.
Figure 3 is a diagram showing the configuration of a database according to an embodiment of the present invention.
Figure 4 is a diagram illustrating a depression keyword classification process by a depression keyword classification unit according to an embodiment of the present invention.
Figure 5 is a diagram illustrating a chatbot message transmission process by the depression question generator according to an embodiment of the present invention.
Figure 6 is a diagram illustrating a depression score calculation process by a depression score calculation unit according to an embodiment of the present invention.
Figure 7 is a diagram showing the configuration of a content recommendation unit according to an embodiment of the present invention.
Figure 8 is a diagram illustrating a process for providing content by diagnosing user depression according to an embodiment of the present invention.
Figure 9 is a diagram showing a method of diagnosing depression using a chatbot according to an embodiment of the present invention.
Figure 10 is a diagram illustrating a process of matching content based on a user depression analysis result according to an embodiment of the present invention.
Figure 11 is a diagram illustrating text content displayed on a user terminal according to an embodiment of the present invention.
Figure 12 is a diagram illustrating a process for recommending content based on a user's content preference according to an embodiment of the present invention.
Figure 13 is a diagram illustrating content recommendation by reflecting a user's content preference according to an embodiment of the present invention.

Hereinafter, the present invention will be described with reference to the attached drawings. However, the present invention may be implemented in various different forms and, therefore, is not limited to the embodiments described herein. In order to clearly explain the present invention in the drawings, parts that are not related to the description are omitted, and similar parts are given similar reference numerals throughout the specification.

Throughout the specification, when a part is said to be "connected (connected, contacted, combined)" with another part, this means not only "directly connected" but also "indirectly connected" with another member in between. "Includes cases where it is. Additionally, when a part is said to “include” a certain component, this does not mean that other components are excluded, but that other components can be added, unless specifically stated to the contrary.

The terms used herein are only used to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as “comprise” or “have” are intended to indicate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but are not intended to indicate the presence of one or more other features. It should be understood that this does not exclude in advance the possibility of the existence or addition of elements, numbers, steps, operations, components, parts, or combinations thereof.

Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

1 is a diagram illustrating the operation of a content providing platform according to an embodiment of the present invention.

Referring to Figure 1, the content provision platform may include an AI-based depression diagnosis unit 100 and a content recommendation unit 200.

Here, the AI-based depression diagnosis unit 100 transmits and receives messages with the user terminal 10, and may perform a depression diagnosis based on the transmitted and received messages. The AI-based depression diagnosis unit 100 can be implemented in server or cloud form and is equipped with or connected to an artificial intelligence model, database, natural language processing engine, etc.

Additionally, the user terminal 10 can utilize various methods such as messenger programs, virtual characters, and voice call programs. As an example, the AI-based depression diagnosis unit 100 can induce a conversation for depression diagnosis based on a chatbot. When a user carries out a daily conversation with a chatbot using text or voice using the user terminal 10, the chatbot can perform natural language processing to determine the user's conversation intention and induce a conversation for diagnosing depression. At this time, a depression score is calculated from the conversation as an objective diagnostic index for depression, and this is used as data for depression diagnosis and treatment. Hereinafter, the configuration of the AI-based depression diagnosis unit 100 will be described in detail with reference to FIG. 2.

Meanwhile, the content recommendation unit 200 may receive the depression analysis results determined from the AI-based depression diagnosis unit 100 and provide content based on the received depression analysis results.

The content recommendation unit 200 may collect and crawl content related to psychological schema therapy. Here, the content may include text content and video content published on the web. The content recommendation unit 200 may collect and classify content based on keywords related to preset psychological schema therapy. Crawling is a data collection method using a search engine robot. It collects documents published on the web such as SNS, news, web information, etc., classifies and stores search targets. At this time, the content recommendation unit 200 may process the crawled content. The content recommendation unit 200 can process and store text content. As an example, the content recommendation unit 200 can summarize the expert's psychological analysis information and provide it to the user more intuitively.

Additionally, the content recommendation unit 200 may extract and collect metadata of video content. Metadata of video content is data given to the content in order to efficiently find the information you are looking for among a large amount of information. In other words, metadata is an index of information for finding content faster. As an example, metadata may include the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content. Additionally, the content recommendation unit 200 may perform filtering based on metadata of the collected video content. Therefore, the content recommendation unit 200 can only store metadata of video content related to psychological schema therapy.

In addition, the content recommendation unit 200 can distinguish the depth of text content and video content according to cognitive elements, emotional elements, and physical elements that can be used in psychological schema therapy. As an example, the content recommendation unit 200 may classify content into cognitive elements, emotional elements, and physical elements, and assign a score to each element. Additionally, the content recommendation unit 200 may classify content to be provided according to the score given to each element. In other words, the content recommendation unit 200 can provide content according to the depth of a specific element. Here, the depth of a specific element can be determined based on information included in the content.

Additionally, the content recommendation unit 200 may match text content and video content based on depression analysis results. Here, the content recommendation unit 200 may match content according to the depression score and provide the matched content to the user terminal 10.

Additionally, the content recommendation unit 200 may collect and analyze the user's content preferences for content provided from the user terminal 10. Content preference may include content usage frequency, content usage time, and content satisfaction. At this time, the content recommendation unit 200 may recommend content based on the analyzed content preferences and depression analysis results. The content recommendation unit 200 can extend the time spent on the content provision platform and provide more satisfying content by reflecting the user's content preferences.

Figure 2 is a diagram showing the configuration of an AI-based depression diagnosis unit according to an embodiment of the present invention.

As shown in Figure 2, the AI-based depression diagnosis unit 100 includes a chatbot 110, a depression keyword classification unit 120, a natural language response generation unit 130, a depression question generation unit 140, and a depression score calculation unit. (150), it may include a depression analysis engine (160) and a database (170).

The AI-based depression diagnosis unit 100 transmits and receives messages to the user terminal 10 through a chatbot 110, which is a messenger app or virtual character. The AI-based depression diagnosis unit 100 may be implemented as a server accessible to multiple users, a stand-alone device providing dedicated access, or a cloud system.

The user terminal 10 may be a device equipped with a processor, memory, communication, and voice recognition functions, such as a smartphone, tablet PC, laptop PC, or desktop PC. Alternatively, it can be a dedicated device, kiosk, or clinical point of care (POC) device with display, input, and voice recognition capabilities.

Text and voice can be input to the user terminal 10. The voice input to the user terminal 10 may be converted into text data encoded in the Unicode standard by a voice recognition function included in the user terminal 10.

The received message received from the user terminal 10 to the chatbot may be text. The received message may be received from the user terminal 10 to the chatbot 110. Additionally, the transmission message may be transmitted from the chatbot 110 to the user terminal 10.

The chatbot 110 may deliver the received message to the depression keyword classification unit 120 and the depression score calculation unit 150. Additionally, the transmitted message may be transmitted from the natural language response generator 130 and the depression question generator 140 through the chatbot 110. Additionally, the chatbot 110 may store the received and transmitted messages in the database 170.

The depression keyword classification unit 120 may be implemented as a natural language processing (NLP) model that classifies received messages. Specifically, the depression keyword classification unit 120 performs natural language processing to classify text in the received message into text related to depression (text containing a depression keyword, depression questionnaire or response item) and text not related to depression. It can be implemented as a model. Depression keywords can be classified using the first artificial intelligence model, which will be described later.

Additionally, if the received message is classified as being related to depression, the depression keyword classification unit 120 may forward the received message to the depression question generating unit 140. Additionally, if the received message is classified as not related to depression by the depression keyword classification unit 120, the received message may be delivered to the natural language response generating unit 130.

The natural language response generator 130 may be implemented with a natural language processing model that generates a transmitted message corresponding to the received message delivered to the natural language response generator 130 and delivers it to the chatbot 110. Specifically, the transmitted message generated by the natural language response generator 130 may be a response selected from predetermined data that elicits everyday conversation and emotional empathy.

The depression question generator 140 may be implemented as a natural language processing model that generates questions corresponding to depression-related received messages delivered to the depression question generator 140 and delivers them to the chatbot 110. Specifically, the question message generated by the depression question generator 140 is received based on questions extracted from depression questionnaire items of depression assessment scales (BDI, HAM-D, PHQ-9, etc.) and information in the received message. You can include questions that best fit your message.

The depression score calculation unit 150 may be implemented as a natural language processing model that calculates a depression score corresponding to the received message delivered to the depression score calculation unit 150. Specifically, the depression rating scale corresponding to the questionnaire response items of the depression rating scale (BDI, HAM-D, PHQ-9, etc.) is classified from the response message to the question, and the depression score is calculated based on the depression rating scale. .

The depression analysis engine 160 may determine the depression state of the received message by combining the calculated depression score delivered to the depression analysis engine 160 and the message from the chatbot.

Specifically, the depression analysis engine 160 diagnoses the level of depression revealed in the verbal expression of the received message and outputs the depression analysis result. For example, using the characteristic that depressed people use more 'first person singular pronouns' during linguistic expressions than healthy people, the depression analysis engine 160 analyzes the 'first person singular pronoun' used in the received message. Measure the level of depression by calculating the frequency. In this way, the depression analysis engine 160 can be used to derive depression analysis results by combining the calculated depression score and the characteristics of the verbal expression.

Figure 3 is a diagram showing the configuration of a database according to an embodiment of the present invention.

Referring to FIG. 3, the database 170 may include natural language processing model data 171, depression questionnaire list data 172, and depression analysis engine data 173.

The natural language processing model data 171 may include natural language response data 171a, first artificial intelligence model data 171b, depression question generation data 171c, and second artificial intelligence model data 171d.

The natural language response data 171a may include data that the natural language response generator 130 of FIG. 2 reads for natural language processing. Specifically, the natural language response data 171a may include everyday conversation data and response data with emotional sympathy.

The first artificial intelligence model data 171b is data about the first artificial intelligence model learned so that the depression keyword classification unit 120 of FIG. 2 can classify keywords related to depression in the received message. The first artificial intelligence model data (171b) can be learned through depression questionnaire data, recording data with depression patients, academic materials related to depression, etc., based on the relationship between keywords included in received messages and the preceding and following context. It is possible to quantify the relationship with depression and provide it.

The depression question generating data 171c may include data that the depression question generating unit 140 of FIG. 2 reads for natural language processing. Specifically, the depression question generating data 171c may include questionnaire questions or questionnaire items of a depression diagnostic scale associated with the depression keyword of the received message.

The second artificial intelligence model data 171d is data about an artificial intelligence model learned so that the depression score calculation unit 150 of FIG. 2 can calculate a score on a depression diagnosis scale from a received message. The second artificial intelligence model data 171d can be learned through a list of answers to depression questionnaire items, diagnosis results of pre-diagnosed depression patients, recorded data with depression patients, depression scores, and academic materials related to statistics. The depression score calculation unit 150 reads the second artificial intelligence model data 170d for natural language processing of the received message, extracts terms, keywords, and phrases related to the depression score from the received message, and provides questionnaire items for measuring depression scale. By deriving the correlation, an accurate depression score is calculated.

The depression questionnaire list data 172 may store accumulated data on depression questionnaires for each user account accessed through the chatbot. More specifically, the depression questionnaire list data 172 can store the chatbot's sent and received messages, calculated depression scores, and depression analysis results for each user.

Depression analysis engine data 173 is data that the depression analysis engine 160 of FIG. 2 reads to output depression analysis results, and may store the calculated depression score and data of verbal expressions related to depression. For example, the depression analysis engine data 173 may store depression scale data matched with the frequency of use of 'first person singular pronoun' in received messages.

Figure 4 is a diagram illustrating a depression keyword classification process by a depression keyword classification unit according to an embodiment of the present invention.

As shown in FIG. 4, the depression keyword classification unit 120 can view the first artificial intelligence model data 171b during the classification process of the received message. Additionally, when the depression keyword classification unit 120 classifies the received message as being related to depression, it can extract the psychological type of the received message.

For example, as shown in FIG. 4, the depression keyword classification unit 120 may view the depression keywords stored in the first artificial intelligence model data 171b and classify the received message as being related to depression. Specifically, the depression keyword classification unit 120 may classify the received message as a message related to depression related to a change in sleep pattern.

Figure 5 is a diagram illustrating a chatbot message transmission process by the depression question generator according to an embodiment of the present invention.

As shown in FIG. 5 , the depression question generator 140 may view the depression question creation data 171c during the process of creating a transmission message. For example, when a depression keyword related to a change in sleep pattern is detected in the received message, as shown in FIG. 5, the depression question generator 140 selects the sleep pattern from the depression questionnaire items stored in the depression question creation data 171c. Questionnaire questions related to change can be extracted. The depression questionnaire items correspond to questions for the depression questionnaire, and the chatbot's question message can be implemented in the form of questions corresponding to the depression questionnaire items.

Figure 6 is a diagram illustrating a depression score calculation process by a depression score calculation unit according to an embodiment of the present invention.

As shown in FIG. 6, the depression score calculation unit 150 can view the second artificial intelligence model data 170d in the process of calculating the depression score. Additionally, the depression score calculation unit 150 may calculate the depression score of the received message corresponding to the response to the transmitted message. The depression score calculation unit 150 can classify positive, medium, weak negative, and strong negative from the response message using the second artificial intelligence model data 171d and match it with the score of the depression questionnaire item.

For example, as shown in FIG. 6, the depression score calculation unit 150 says ‘That’s right. The received message 'I can't sleep so much that I'm going to die' is changed to '3: I sleep almost all day/I sleep more than before' in the BDI sleep pattern change included in the depression questionnaire items stored in the second artificial intelligence model data (171d). It can be categorized as ‘waking up 1-2 hours early and having difficulty falling back to sleep’ and a depression score of 3 can be calculated.

Figure 7 is a diagram showing the configuration of a content recommendation unit according to an embodiment of the present invention.

Referring to Figure 7, the content recommendation unit 200 includes a content collection unit 210, a content crawling unit 220, a content processing unit 230, a content database 240, a content matching unit 250, and a content providing unit ( 260), a user preference analysis unit 270, a user preference database 280, and a content recommendation unit 290.

First, the content collection unit 210 can collect text content and video content published on the web. As an example, the content collection unit 210 may collect content published on related sites such as psychotherapy, medicine, and health.

The content crawling unit 220 can crawl the collected content. Crawling is a data collection method using a search engine robot. It collects documents published on the web such as SNS, news, web information, etc., classifies and stores search targets. Here, the content crawling unit 220 can classify and store the collected content based on preset keywords. As an example, keywords may consist of depression questionnaire items including depression assessment scales (BDI, HAM-D, PHQ-9, etc.).

Meanwhile, the content processing unit 230 may include a text content processing unit 231 and a video content processing unit 232.

The text content processing unit 231 can process crawled text content. As an example, the text content processing unit 231 may analyze and summarize text content. As a result, the text content processing unit 231 can provide psychological analysis information from mental health-related experts in an easy-to-understand manner.

The text content processing unit 231 may classify text content based on preset keywords. Additionally, the text content processing unit 231 can classify the depth of text content according to cognitive elements, emotional elements, and physical elements that can be used in psychological schema therapy. As an example, the text content processing unit 231 may classify text content into cognitive elements, emotional elements, and physical elements, and assign a score to each element. Accordingly, the text content processing unit 231 can classify the text content according to the score assigned to each element. As will be described later, based on this, the content recommendation platform 200 can provide text content corresponding to the user depression analysis result.

Meanwhile, the video content processing unit 232 can extract and collect meta data of crawled video content. Metadata of video content is data given to the content in order to efficiently find the information you are looking for among a large amount of information. Metadata may include the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content. The video content processing unit 232 may filter video content based on metadata. As an example, the video content processing unit 232 may determine whether metadata includes a preset keyword. If metadata does not include a keyword, the video content processing unit 232 may exclude the corresponding video content.

Additionally, the video content processing unit 232 may distinguish filtered video content based on preset keywords. As an example, the video content processing unit 232 may classify video content into cognitive elements, emotional elements, and physical elements, and assign a score to each element. Accordingly, the video content processing unit 232 can classify the video content according to the score assigned to each element.

The content database 240 can store separated text content and video content. Here, the content database 240 may store metadata of video content. Even if the video content is not stored in the content database 240, the meta data includes location information of the video content, so the location of the video content can be determined based on the meta data. Therefore, it is possible to efficiently manage the capacity of the content database 240. Additionally, the content database 240 is capable of constructing data related to psychological schema therapy. In addition, the content database 240 can be updated in real time. Accordingly, the content database 240 can store the latest content, thereby securing accurate data related to psychological schema therapy.

The content matching unit 250 may match text content and video content based on the user depression analysis result. First, the content matching unit 250 may receive the user depression analysis result from the database 170. The content matching unit 250 may receive content stored in the content database 240 based on the user depression analysis result. Additionally, the content matching unit 250 may determine text content and video content corresponding to the user depression analysis result. As an example, the content matching unit 250 may compare the depression score included in the user depression analysis result with the score for each element included in the content. That is, the content matching unit 250 can match content based on content attributes and user attributes. Here, the matched content may be provided to the user terminal 10 through the content provider 260.

The content provider 260 may provide matched content to the user terminal 10. Additionally, the content provider 260 may provide scores given to cognitive elements, emotional elements, and physical elements in a graph. As an example, the content provider 260 may display scores given to each element according to time/date.

The user preference analysis unit 270 may collect the user's content preference for content provided from the user terminal 10. Content preference may include content usage frequency, content usage time, and content satisfaction. For example, the user preference analysis unit 270 may determine that the user's content preference is higher as the content is used more often, the usage time is longer, and the level of satisfaction is higher. Additionally, the user preference analysis unit 270 may calculate a content preference score by assigning a score to each content preference element.

The user preference database 280 may store analyzed content preferences. Content preferences stored in the user preference database 280 can be used by the content matching unit 250 to match content based on the user depression analysis results. That is, the content matching unit 250 can match content based on the user's content preference and the user's depression analysis result. Here, the content matching unit 250 can provide content with a higher level of satisfaction by matching content by reflecting content preferences. Additionally, it is possible to encourage users to increase the time they spend on the content provision platform.

The content recommendation unit 290 may provide recommended content to the user terminal 10. As an example, the content recommendation unit 290 may recommend content that reflects the content preferences of users with similar depression analysis results. Here, the content recommendation unit 290 may recommend content in the order of content preference. As a result, customized content and information can be provided by recommending content that reflects the user's preferences.

Figure 8 is a diagram illustrating a process for providing content by diagnosing user depression according to an embodiment of the present invention.

In step S100, the content providing platform may use a chatbot to diagnose the user's depression. Here, the content providing platform can transmit and receive messages to and from the user terminal and perform a depression diagnosis based on the transmitted and received messages. As an example, a content provision platform can induce conversations for diagnosing depression based on chatbots. When a user carries out a daily conversation with a chatbot using text or voice using the user terminal, the chatbot can perform natural language processing to determine the user's conversation intention and induce a conversation for diagnosing depression. At this time, a depression score is calculated from the conversation as an objective diagnostic index for depression, and this is used as data for depression diagnosis and treatment. In addition, the content provision platform can determine the depression state of the message received from the user terminal by combining the calculated depression score and the chat message. As an example, a content provision platform can diagnose the level of depression revealed in the verbal expression of a received message and output a depression analysis result.

In step S200, the content providing platform may match content based on the depression analysis results. The content provision platform can collect and crawl content related to psychological schema therapy. Here, the content may include text content and video content published on the web. The content provision platform can classify and store the content collected by crawling. Additionally, the content provision platform can process content. As an example, a content provision platform can process and store text content. The content provision platform provides a summary of text content, allowing users to more easily understand the expert's psychological analysis information. Additionally, the content providing platform can collect metadata of video content and filter based on the collected metadata. Metadata of video content is data given to the content in order to efficiently find the information you are looking for among a large amount of information. As an example, metadata may include the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content. As a result, the content provision platform can only store metadata of video content related to psychological schema therapy.

Additionally, the content provision platform can classify the depth of content according to cognitive elements, emotional elements, and physical elements that can be used in psychological schema therapy. As an example, a content provision platform can classify content into each element and give scores. Additionally, the content provision platform can classify content according to preset conditions for the score of each element.

Additionally, the content provision platform can match and provide content based on the depression analysis results. As an example, a content provision platform can match content according to depression scores. At this time, the content provision platform can compare the depression score from the depression analysis result with the score for each element of the content. The content provision platform can match content containing scores for each element corresponding to the depression score.

In step S300, the content providing platform may recommend content based on the user's content preference. The content providing platform may collect the user's content preference for content from the user terminal. Content preference may include content usage frequency, content usage time, and content satisfaction. At this time, the content providing platform can recommend content based on content preference and depression analysis results. Here, the content provision platform can provide more satisfying content by reflecting the user's content preferences.

Figure 9 is a diagram showing a method of diagnosing depression using a chatbot according to an embodiment of the present invention.

In step S101, the content providing platform activates the chatbot in response to the request from the user terminal. Chatbots are activated by applications on the user's terminal or various types of programs, and personal information is identified through the user's account information.

When the chatbot is activated, the content provision platform maintains a state in which it can receive messages from the user terminal. The content provision platform can be implemented as a server accessible to multiple users, a stand-alone device that provides dedicated access, or a cloud system. Chatbots are connected to wired and wireless networks. The content provision platform sends and receives messages to the user terminal through a chatbot. Text and voice can be input to the user terminal. The voice input to the user terminal can be converted into text data encoded in the Unicode standard method by the voice recognition function included in the user terminal.

In step S102, the chatbot receives a message from the user terminal. The received message received from the user terminal to the chatbot may be text. The messages you receive may be depression-related or non-depression-related. If the received message is not related to depression, the depression diagnosis system uses natural language response data to induce continued conversation. Additionally, the chatbot delivers received messages to the depression keyword classification unit and depression score calculation unit.

In step S103, the depression keyword classification unit within the content providing platform classifies messages and keywords related to depression in the text of the received message. As described above, depression keywords can be classified using a pre-trained first artificial intelligence model. If the received message is classified as related to depression by the depression keyword classification unit, the depression keyword classification unit may transfer the received message and the keyword to the depression question generation unit. The depression keyword classification unit can be implemented as a natural language processing model that classifies received messages related to depression among received messages.

In step S104, the depression question generator within the content providing platform generates a question message corresponding to the depression-related received message delivered to the depression question generator. The generated question message is delivered to the chatbot by the depression question generator. The question message generated by the depression question generator includes questions extracted from depression questionnaire items of depression assessment scales (BDI, HAM-D, PHQ-9, etc.) and questions for diagnosing depression based on information in the received message. can do. Specifically, in the process of generating the transmission message of the depression question generator, questionnaire questions or questions extracted from questionnaire items for measuring a plurality of depression rating scales pre-stored in the database may be used.

In step S105, the question message generated by the depression question generator in the content providing platform is transmitted from the depression question generator to the chatbot, and the transmitted question message is transmitted to the user terminal through the chatbot.

In step S106, the chatbot stores the received and transmitted messages in the database. Specifically, message data sent to the user terminal through the chatbot and message data received from the user terminal to the chatbot are stored in the depression questionnaire list data of the database.

In step S107, the depression score calculation unit within the content providing platform calculates the depression score using the response message to the question generated by the depression question generator. The depression score calculation unit can be implemented as a natural language processing model, and uses a second artificial intelligence model that learns the depression scale and relevance from the processed natural language.

Specifically, the depression score calculation unit matches the questionnaire items of the depression assessment scale (BDI, HAM-D, PHQ-9, etc.) with the processing results of the received message, and calculates the depression score based on the depression assessment scale of the matched questionnaire items. do.

In step S108, the depression analysis engine within the content provision platform synthesizes the calculated depression score delivered to the depression analysis engine, the received message of the chatbot, and the transmitted message, and outputs a depression analysis result of the received message.

The depression analysis engine diagnoses the level of depression revealed in the verbal expression of the received message and further reflects it in the depression analysis results. For example, by using the characteristic that depressed people use more 'first person singular pronouns' during linguistic expressions than healthy people, the depression analysis engine calculates the frequency of 'first person singular pronouns' used in incoming messages. The level of depression can be calculated and reflected in the depression analysis results.

In step S109, the depression analysis engine within the content providing platform stores the depression analysis results in the database. Additionally, the depression score calculation unit stores the calculated depression score in the database. Additionally, received and transmitted messages between the chatbot and the user terminal can be stored in the database. Specifically, the database may store depression analysis results, calculated depression scores, received messages, and sent messages for each user account.

Additionally, the database may include natural language processing model data, depression questionnaire list data, and depression analysis engine data. Natural language processing model data in the database may store natural language response data, first artificial intelligence model data, data for generating depression questions, and second artificial intelligence model data.

Figure 10 is a diagram illustrating a process of matching content based on a user depression analysis result according to an embodiment of the present invention.

In step S201, the content providing platform may collect text content and video content published on the web. As an example, a content provision platform may collect content published on sites related to psychological schema therapy, such as psychotherapy, medicine, and health.

In step S202, the content providing platform may crawl the collected content. Crawling is a data collection method using a search engine robot. It collects documents published on the web such as SNS, news, web information, etc., classifies and stores search targets. Here, the content provision platform can classify and store the collected content based on preset keywords. As an example, keywords may consist of depression questionnaire items including depression assessment scales (BDI, HAM-D, PHQ-9, etc.).

In step S203, the content providing platform may process the crawled text content. As an example, a content provision platform can analyze and summarize text content. As a result, the content provision platform can provide easy-to-understand psychological analysis information from mental health experts.

In step S204, the content providing platform may collect and filter meta data of the crawled video content. Metadata may include the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content. Since metadata includes location information of video content, the location of video content can be determined based on metadata. Therefore, the content providing platform does not need to store video content. As a result, the content provision platform can manage capacity efficiently. Additionally, the content provision platform can filter video content based on metadata. As an example, a content provision platform may determine whether metadata includes preset keywords. The content provision platform may exclude the video content if the metadata does not include keywords. Keywords may consist of depression questionnaire items including depression assessment scales (BDI, HAM-D, PHQ-9, etc.).

In step S205, the content providing platform may separate and store the content. As an example, a content provision platform may classify content based on preset keywords. In addition, content provision platforms can be classified according to cognitive elements, emotional elements, and physical elements, which are elements that can be used in psychological schema therapy. At this time, the content provision platform can give scores for each element. The content provision platform can classify content according to the score given to each element. Additionally, the content provision platform stores classified content, making it possible to construct data related to psychological schema therapy.

In step S206, the content providing platform may match text content and video content based on the user's depression analysis results. First, the content provision platform can determine text content and video content that correspond to the depression analysis results. As an example, the content provision platform can compare the depression score included in the depression analysis result with the score for each element included in the content. As a result, it is possible for the content providing platform to match content attributes with user attributes.

In step S207, the content provision platform may provide matched text content and video content to the linked user terminal.

Figure 11 is a diagram illustrating text content displayed on a user terminal according to an embodiment of the present invention.

Referring to Figure 11, processed text content (T1) and health index (T2) may be displayed. The processed text content (T1) is text content matched based on the user's depression analysis results. Here, the processed text content T1 may display preset text. The preset text may be determined according to the results of the user's depression analysis. As an example, the preset text may correspond to a depression score in the user's depression analysis result. Additionally, the preset text may be displayed in the form of simple phrases that intuitively indicate the results of the user's depression analysis.

Additionally, the processed text content T1 may be a summary of the text content. Users can easily understand psychological analysis information from mental health experts through processed text content (T1).

The health index (T2) is a graphical representation of the scores given to the cognitive, emotional and physical components. The health index (T2) can be displayed in the form of a graph of the score given to each element of text content matched according to time/date. By displaying the health index (T2) on the user terminal, the user can easily check changes in the depression analysis results.

Figure 12 is a diagram illustrating a process for recommending content based on a user's content preference according to an embodiment of the present invention.

In step S301, the content providing platform may collect and analyze the user's content preferences for content provided from the user terminal. Content preference may include content usage frequency, content usage time, and content satisfaction. For example, the content provision platform may determine that the user's content preference is higher as the content is used more often, the usage time is longer, and the level of satisfaction is higher. Additionally, the content provision platform can calculate a content preference score by assigning scores for each content preference element.

In step S302, the content providing platform may recommend or provide content based on the analyzed user's content preferences and depression analysis results. The content provision platform can match content based on the user's content preferences and depression analysis results. Here, the content provision platform can provide more satisfying content by reflecting the user's content preferences.

Additionally, the content provision platform can recommend content that reflects the content preferences of users with similar depression analysis results. In other words, the content provision platform can provide content that reflects the content preferences of other users. As an example, a content provision platform may recommend content in order of content preference. As a result, the content provision platform can encourage users to increase their stay time and improve dependency.

Figure 13 is a diagram illustrating content recommendation by reflecting a user's content preference according to an embodiment of the present invention.

Content preference can be calculated based on the number of times content is used, content use time, and content satisfaction. Referring to Figure 13, content preference can be calculated based on content satisfaction (V1) and content use time (V2). Content satisfaction (V1) can be determined by whether or not the content is recommended. For example, content satisfaction (V1) may be judged as high when a heart-shaped icon is selected, and low when an icon with a finger pointing down is selected. Additionally, the content usage time (V2) may be determined based on the actual content usage time.

Content preference information (d1) may include content preference determined based on content satisfaction (V1) and content use time (V2). Here, content satisfaction (V1) can determine content preference based on whether or not it is recommended. Additionally, in the content preference information (d1), the content satisfaction (V1) may be displayed as True if it is recommended, as null if it is not recommended, and as False if it is not recommended. Content usage time (V2) can be granted according to preset standards. Additionally, content satisfaction (V1) and content usage time (V2) may be prioritized according to preset conditions. For example, when higher weight is placed on content satisfaction (V1), video content UID1 may be given higher priority than video content UID2. In this case, it may be determined that video content UID1 has a higher content preference than video content UID2.

Recommended content (V3) may display video content with high content preference. Additionally, similar content (V4) may be displayed in order of content preference. For example, recommended content (V3) may display video content UID1, and similar content (V4) may display video content UID3 and video content UID2 in that order. As a result, customized content and information can be provided by recommending content that reflects the user's preferences.

The description of the present invention described above is for illustrative purposes, and those skilled in the art will understand that the present invention can be easily changed to another specific form without changing the technical idea or essential features of the present invention. will be. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. For example, each component described as unitary may be implemented in a distributed manner, and similarly, components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims described below, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention.

100: AI-based depression diagnosis unit
200: Content recommendation section

Claims (12)

An AI-based depression diagnosis unit that transmits and receives messages to and from the user terminal, performs a depression diagnosis based on the transmitted and received messages, and calculates depression analysis results; and
Collect and crawl content published on the web, process and store the crawled content, match it with the content based on the depression analysis results, and provide it to the user terminal,
Content for digital psychological schema therapy comprising a content recommendation unit that collects content preferences for the provided content from the user terminal and recommends the content based on the content preferences and the depression analysis results. Delivery platform.
According to paragraph 1,
The content recommendation section
A content collection unit that collects text content and video content published on the web;
a content crawling unit that crawls and classifies the collected text content and video content;
a text content processing unit that processes the crawled text content, divides it into cognitive elements, emotional elements, and physical elements, and gives a score to each element;
Collect and extract the metadata of the crawled video content, filter the regular video content based on the metadata, classify it according to the cognitive element, the emotional element, and the physical element, and give a score for each element. a video content processing unit;
a content database that stores metadata of the separated text content and video content;
a content matching unit that matches the stored text content and the video content based on the depression analysis results;
a content provider providing the matched text content and video content to the user terminal;
a user preference analysis unit that collects content preferences for the text content and the video content from the user terminal and calculates a content preference score;
a user preference database storing the content preferences; and
A content providing platform for digital psychological schema therapy that includes a content recommendation unit that recommends the text content and the video content based on the content preference and the depression analysis results.
According to clause 1,
The AI-based depression diagnosis unit
a depression keyword classification unit that classifies depression keywords from the text of the received message received through the chatbot;
If the received message is classified as a depression keyword, a depression question generator that generates a question message about a depression questionnaire related to the depression keyword;
a depression score calculation unit that calculates a depression score for the depression questionnaire from the text of the response message to the question message;
A depression analysis engine that provides depression analysis results based on the calculated depression score; and
A database including first artificial intelligence model data learned so that the depression keyword classification unit can classify the depression keyword, and second artificial intelligence model data learned so that the depression score calculation unit can calculate a depression score. A content provision platform for digital psychological schema therapy.
According to clause 2,
The content matching unit compares and matches the depression score included in the depression analysis result with the score for each element included in the text content and the video content.
According to paragraph 1,
The content provision platform for digital psychological schema therapy, wherein the content preference includes the number of content uses, content use time, and content satisfaction.
According to clause 2,
The metadata includes the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content. A content provision platform for digital psychological schema therapy.
a) Sending and receiving messages with the user terminal using a chatbot, performing a depression diagnosis based on the sent and received messages, and calculating depression analysis results;
b) collecting and crawling content published on the web, processing and storing the crawled content, matching the content with the content based on the depression analysis results, and providing the content to the user terminal;
c) collecting content preferences for the provided content from the user terminal, and recommending the content based on the content preferences and the depression analysis results; a content providing method for digital psychological schema therapy comprising a. .
The method of claim 7, wherein step a) is
activating the chatbot;
Receiving a message from a user terminal by the chatbot;
classifying depression keywords in the text of the received message;
Generating a question message about depression questionnaire items related to the classified depression keywords;
Transmitting the generated question message to a user terminal;
calculating a depression score from the text of a response message to the question message; and
A step of outputting a depression analysis result of the user based on the calculated depression score,
The depression keyword classification step uses first artificial intelligence model data learned to classify the depression keyword from the message,
The depression score calculation step is a method of providing content for digital psychological schema therapy, wherein the learned second artificial intelligence model data is used to calculate the depression score from the message.
The method of claim 7, wherein step b) is:
collecting text content and video content published on the web;
Crawling and classifying the collected text content and video content;
Processing the crawled text content, dividing it into cognitive elements, emotional elements, and physical elements, assigning a score to each element, and storing it;
Collect and extract the metadata of the crawled video content, filter the regular video content based on the metadata, classify it according to the cognitive element, the emotional element, and the physical element, and give a score for each element. and saving;
Matching the stored text content and the video content based on the depression analysis results;
Providing the matched text content and the video content to the user terminal;
Collecting content preferences for the text content and the video content from the user terminal to calculate a content preference score; And
A content providing method for digital psychological schema therapy comprising a content step of recommending the text content and the video content based on the content preference and the depression analysis result.
According to clause 9,
Content for digital psychological schema treatment that compares and matches the depression score included in the depression analysis result with the scores given to the cognitive element, the emotional element, and the physical element included in the text content and the video content. How to provide.
In clause 7,
A method of providing content for digital psychological schema therapy, wherein the content preference includes the number of times of content use, content use time, and content satisfaction.
According to clause 9,
The metadata includes the location, content, comments, captions, hashtags, title, author information, terms of use, and usage history of the video content. A method of providing content for digital psychological schema therapy.

Images