Depression tendency early warning method and device, electronic equipment and storage medium

Through the evaluation of the dialogue content scoring model, deep learning models are used to warn the depressive tendency, which solves the problems of low efficiency and insufficient accuracy of traditional methods, and achieves rapid and accurate warning and evaluation of the depressive tendency.

CN118969278APending Publication Date: 2024-11-15SHENZHEN UNITED INFORMATION TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411085050.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

Traditional early warning methods for depression tendency are inefficient and inaccurate enough to accurately identify depression tendencies, especially in adolescent groups, and are unable to effectively utilize natural conditions and behaviors to assess social hierarchy.

Method used

By obtaining the conversation content, selecting specific conversation information and calling the trained scoring model for scoring, sending alarm information based on the scoring results, using deep learning models such as BERT and RoBERTa for text classification, and combining the context information of the conversation for model training to improve the accuracy of evaluation sex.

Benefits of technology

It realizes rapid and accurate warning of depression tendencies, automatic prediction process, fast response, and adjusts warning thresholds according to needs, improving the accuracy and efficiency of depression tendency evaluation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118969278A_ABST
    Figure CN118969278A_ABST
Patent Text Reader

Abstract

The invention discloses a depression tendency early warning method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining dialogue content of a dialogue target; the dialogue content comprises N pieces of dialogue information; selecting M pieces of dialogue information from the N pieces of dialogue information, and calling a trained scoring model to independently score each piece of dialogue information in the M pieces of dialogue information; the score is used for indicating the depression tendency degree of the dialogue target; and when the total score of the M pieces of dialogue information exceeds a set threshold value, sending alarm information. According to the depression tendency early warning method, depression tendency evaluation is realized based on the dialogue text and by using the recognition intelligent model, and early warning is performed based on the evaluation result. The prediction process is fully automatic, the response is fast, the accuracy is high, and the early warning threshold value can be adjusted according to specific requirements.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the field of artificial intelligence technology, and in particular, relates to a method, device, electronic device and storage medium for early warning of depressive tendency. Background Art

[0002] In the current field of mental health, especially in research focusing on adolescent depression, a persistent challenge is how to identify and predict depressive tendencies in a timely and accurate manner. Traditional methods rely on face-to-face clinical assessments, self-report questionnaires, and observations by parents or teachers. Although these methods are effective in certain situations, they also have certain limitations. For example, self-report questionnaires may be affected by the subject's current emotional state, self-perception ability, and the authenticity of the answers, while the accessibility and frequency of clinical assessments are limited by resource availability.

[0003] With the explosive growth of digital information, people are faced with the challenge of extracting valuable information from massive text data. As an effective solution, text classification technology can automatically classify text data to help users quickly locate and understand information. In the background technology of text classification, machine learning algorithms play a core role. Traditional text classification methods are mainly based on the combination of feature engineering and machine learning algorithms. Feature engineering involves extracting meaningful feature representations from text, such as term frequency-inverse document frequency (TF-IDF), n-gram, etc. These features are then input into machine learning models (such as naive Bayes, support vector machines, decision trees, etc.) for training to learn how to map text data to corresponding categories.

[0004] In this context, it becomes feasible and necessary to develop a scoring model for depressive tendency based on conversation texts to predict adolescents' depressive tendency. Summary of the invention

[0005] The present application provides a depression tendency warning method, device, electronic device and storage medium, aiming to solve the problems that traditional depression tendency warning methods have limitations, low efficiency and inaccurate evaluation results; and the inability to accurately identify the identity of animals and effectively evaluate social levels based on animal behavior in methods of estimating social levels under natural conditions.

[0006] In a first aspect, the present application provides a method for early warning of depression tendency, the method comprising:

[0007] Acquire the conversation content of the conversation target; the conversation content includes N pieces of conversation information;

[0008] Selecting M dialogue messages from the N dialogue messages, and calling the trained scoring model to score each of the M dialogue messages individually; the score is used to indicate the degree of depression tendency of the dialogue target;

[0009] When the total score of the M dialogue messages exceeds a set threshold, an alarm message is sent.

[0010] In a second aspect, the present application provides a depression tendency early warning device, the device comprising:

[0011] An acquisition module, used to acquire the conversation content of the conversation target; the conversation content includes N pieces of conversation information;

[0012] A processing module, configured to select M dialogue messages from the N dialogue messages, and call a trained scoring model to score each of the M dialogue messages individually; the score is used to indicate the degree of depression tendency of the dialogue target;

[0013] The early warning module is used to send an alarm message when the total score of the M dialogue messages exceeds a set threshold.

[0014] In a third aspect, the present application provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned depression tendency warning method when executing the computer program.

[0015] In a fourth aspect, a storage medium stores computer-readable instructions thereon, wherein the computer-readable instructions are executed by one or more processors to implement the depressive tendency early warning method as described above.

[0016] In a fifth aspect, a computer program product is provided, which includes computer-readable instructions stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the depressive tendency warning method as described above.

[0017] Compared with the prior art, the depression tendency early warning method of the present application is based on the conversation text and uses the cognitive intelligence model to realize the evaluation of depression tendency, and issues early warning based on the evaluation results. Its prediction process is fully automated, with fast response and high accuracy, and the threshold of early warning can be adjusted according to specific needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of the main process of a depression tendency early warning method provided in an embodiment of the present application;

[0020] Figure 2 A structural block diagram of an electronic depression tendency warning device provided in an embodiment of the present application;

[0021] Figure 3 A structural block diagram of an electronic device provided for an embodiment of the present application. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0024] See also Figure 1 , Figure 1 Schematic diagram of the main process of the depression tendency early warning method shown in the embodiment of the present application. The depression tendency early warning method shown in the embodiment of the present application includes the following steps S100 to S300:

[0025] S100, obtaining the conversation content of the conversation target; the conversation content includes N pieces of conversation information; N is a positive integer.

[0026] The historical conversation content of the conversation target is obtained or the conversation content of the conversation target is obtained in real time. Usually, the conversation content includes multiple conversation information.

[0027] In a specific embodiment, a conversation with a dialogue target can be conducted through a trained language model, that is, a trained scoring model is integrated into the conversation system of a trained large language model. The large language model can complete a conversation with the dialogue target and record the conversation text information of the dialogue target, and then further input the conversation text information into the scoring model.

[0028] The process of completing the dialogue with the dialogue target through the trained language model can better restore the psychological state of the dialogue target, and the obtained dialogue content is more realistic, so that the final scoring effect is better.

[0029] S200, selecting M dialogue messages from the N dialogue messages, and calling the trained scoring model to score each of the M dialogue messages individually; the score is used to indicate the degree of depression tendency of the dialogue target; M is a positive integer.

[0030] A certain number of dialogue messages are randomly selected or selected in a set manner from the multiple dialogue messages obtained, and the scoring model can score each selected dialogue message separately, and the score is used to evaluate the degree of depression tendency of the dialogue target. Among them, the scoring model is used to process the classification task, which is used to classify the obtained dialogue messages, and the corresponding score is used as the classification label. For example, the score can be divided into 0 to 3 points, where 0 points means no depression tendency and 3 points means severe depression tendency.

[0031] In a specific embodiment, the conversation information may be arranged in chronological order, and the last 100 conversation information in chronological order may be selected for scoring, that is, the most recent conversation information may be selected for scoring.

[0032] S300: When the total score of the M dialogue messages exceeds a set threshold, an alarm message is sent.

[0033] According to the M conversation messages obtained, the total score is recorded and calculated. In one embodiment, the most recent 100 messages are selected. If the total score of the most recent 100 conversation messages exceeds 100 points, an early warning is triggered, otherwise it is not triggered. The specific number of conversation messages and the threshold value of the score can be set according to the actual situation.

[0034] In an exemplary embodiment, based on the above-mentioned depression tendency early warning method, the training method for the scoring model in the method may include the following steps:

[0035] S10, obtaining training samples; the training samples include a number of dialogue information samples in different scenarios and scoring labels annotated on the dialogue information samples;

[0036] In order to collect diverse data, the conversation content of the language model in multiple usage scenarios can be collected, including but not limited to online learning platforms, mental health counseling systems, etc., to capture the conversation content of users in different situations.

[0037] For example, in a mental health consultation system scenario, conversation information about health consultation can be collected, such as:

[0038] “User: I feel very depressed and unhappy recently.

[0039] Big Model: What happened that made you unhappy?

[0040] User: The new semester has begun and I find it difficult to adapt to the new environment.

[0041] Big Model: It does take some time to adapt to the new environment. What specific aspects make you feel uncomfortable?

[0042] User: It’s mainly the course content and the relationships with classmates. The new course is difficult and I don’t know any of my new classmates. ”…

[0043] Due to the lack of diverse data, it is difficult for the scoring model to score (classify) noise and data; therefore, by collecting conversation content from various scenarios and combining the contextual information of the conversation, the scoring model is trained to improve the performance of the model and thus more accurately assess the user's depressive tendencies.

[0044] In order to accurately assess the user's depressive tendencies, the collected conversation data will be manually annotated by psychologists with rich experience and expertise. The annotation team will undergo rigorous training to ensure the consistency and accuracy of the annotation results. During the annotation process, the annotation team will comprehensively judge the user's depressive tendency based on the content of the conversation, combined with factors such as the user's age, gender, and cultural background. The annotation results will serve as an important reference for model training.

[0045] S20: Input the training samples into a pre-built scoring model for training to obtain the trained scoring model.

[0046] In a specific embodiment, step S20 may include the following steps:

[0047] S21, inputting part of the training samples as a training set into the pre-built scoring model to output a first prediction scoring result;

[0048] S22: training the scoring model based on the comparison result between the scoring result and the scoring label to obtain a trained scoring model;

[0049] S23, inputting part of the training samples as a test set into the trained scoring model to output a second prediction scoring result;

[0050] S24: Evaluate the performance of the trained scoring model based on the comparison result between the second predicted scoring result and the scoring label. If the performance does not meet the set conditions, retrain the scoring model.

[0051] In the model building and training phase, we will use labeled data to train deep learning models, such as BERT, RoBERTa, etc. These models have strong text representation and context understanding capabilities, and can automatically learn complex features and patterns in text. These models have acquired rich language knowledge and semantic understanding capabilities by pre-training on large-scale corpora. When applying pre-trained language models to text classification tasks, they can quickly adapt to specific data sets and classification requirements through fine-tuning, thereby achieving better performance.

[0052] The model is trained using supervised learning, and the labeled data is divided into a training set and a test set. During the training process, the parameters and structure of the model will be continuously adjusted to optimize the performance of the model. At the same time, cross-validation and other techniques will be used to evaluate the generalization ability and robustness of the model.

[0053] In an exemplary embodiment, step S10 further includes step S11:

[0054] S11. Preprocess the dialogue information sample to obtain a processed dialogue information sample.

[0055] Specifically, the purpose of preprocessing sample data is to remove irrelevant information in the sample data to reduce interference with model training.

[0056] In one implementation, the dialogue information samples may be cleaned in detail, such as removing noise text, emoticons, special characters, etc., and the dialogue information samples may be processed by word segmentation and stop word removal, so that the model can better capture the key information in the text.

[0057] In one implementation, the dialogue information samples may also be normalized to ensure that the model can handle text inputs of different lengths during training.

[0058] The present application also provides a depressive tendency warning device, which includes an acquisition module, a processing module and an early warning module. The acquisition module is used to acquire the conversation content of the conversation target; the conversation content includes N conversation messages; the processing module is used to select M conversation messages from the N conversation messages, and call the trained scoring model to score each of the M conversation messages separately; the score is used to indicate the degree of depression tendency of the conversation target; the early warning module is used to send an alarm message when the total score of the M conversation messages exceeds a set threshold.

[0059] like Figure 2 As shown, an embodiment of the present application provides a depressive tendency warning device, which includes an acquisition module, a processing module and an early warning module; wherein the acquisition module is used to obtain the dialogue content of the dialogue target; the dialogue content includes N dialogue messages; the processing module is used to select M dialogue messages from the N dialogue messages, and call the trained scoring model to score each of the M dialogue messages separately; the score is used to indicate the degree of depressive tendency of the dialogue target; the early warning module is used to send an alarm message when the total score of the M dialogue messages exceeds a set threshold.

[0060] In an exemplary embodiment, the acquisition module includes a dialogue unit, and the dialogue unit is used to conduct a dialogue with a dialogue target through a trained language model to obtain the dialogue content of the dialogue target.

[0061] In an exemplary embodiment, the depressive tendency warning device also includes a training module, which is used to obtain training samples; the training samples include a number of dialogue information samples in different scenarios and scoring labels annotated on the dialogue information samples; and the training samples are input into a pre-built scoring model for training to obtain the trained scoring model.

[0062] In an exemplary embodiment, the training module further includes a preprocessing unit, which is used to preprocess the dialogue information sample to obtain a processed dialogue information sample.

[0063] In an exemplary embodiment, the preprocessing unit includes a preprocessing subunit, and the first preprocessing subunit is used to perform denoising and / or normalization processing on the dialogue information sample to obtain a processed dialogue information sample.

[0064] See also Figure 3 In an embodiment of the present application, an electronic device 4000 is provided, and the electronic device 4000 includes at least one processor 4001 and at least one memory 4003.

[0065] The data interaction between the processor 4001 and the memory 4003 can be realized through at least one communication bus 4002. The communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0066] Optionally, the electronic device 4000 may further include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0067] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0068] The memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program instructions or codes in the form of instructions or data structures and can be accessed by the electronic device 4000, but is not limited to this.

[0069] Computer-readable instructions are stored in the memory 4003 , and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002 .

[0070] The computer-readable instructions are executed by one or more processors 4001 to implement the depression tendency warning method in the above-mentioned embodiments.

[0071] In addition, a storage medium is provided in an embodiment of the present application, on which computer-readable instructions are stored. The computer-readable instructions are executed by one or more processors to implement the above-mentioned depression tendency warning method.

[0072] In an embodiment of the present application, a computer program product is provided. The computer program product includes computer-readable instructions. The computer-readable instructions are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the above-mentioned depression tendency warning method.

[0073] Compared with the related art, the depressive tendency warning method of the present application is based on the dialogue text and uses the cognitive intelligent model to realize the assessment of depressive tendency, and warns based on the assessment results. Its prediction process is fully automated, fast in response, high in accuracy, and the threshold of the warning can be adjusted according to specific needs. In addition, in the training stage of the model, by collecting the dialogue content of various scenes and combining the contextual information of the dialogue, the scoring model is trained to improve the performance of the model, so as to more accurately assess the user's depressive tendency. And by cleaning the training data, irrelevant information such as noise text, emoticons, special characters, etc. is removed to reduce the interference of noise on model training, thereby further improving the performance of the model. In terms of model construction, pre-trained language models (such as BERT, GPT, etc.) are used. These models have obtained rich language knowledge and semantic understanding capabilities by pre-training on large-scale corpora. When the pre-trained language model is applied to text classification tasks, it can be quickly adapted to specific data sets and classification requirements by fine-tuning, thereby achieving better performance.

[0074] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0075] The present application is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present application. In addition, under the teachings of the present application, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present application. Therefore, the present application is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of the present application are within the scope protected by the present application.

Claims

1. A method for early warning of depression tendency, characterized in that: The method comprises: Acquire the conversation content of the conversation target; the conversation content includes N pieces of conversation information; Selecting M dialogue messages from the N dialogue messages, and calling the trained scoring model to score each of the M dialogue messages individually; the score is used to indicate the degree of depression tendency of the dialogue target; When the total score of the M dialogue messages exceeds a set threshold, an alarm message is sent.

2. The method according to claim 1, characterized in that The acquisition of the dialogue content of the dialogue target specifically includes: The trained language model is used to conduct a dialogue with the dialogue target to obtain the dialogue content of the dialogue target.

3. The method according to claim 1, characterized in that The training method of the scoring model includes: Acquire training samples; the training samples include a number of dialogue information samples in different scenarios and score labels annotated on the dialogue information samples; The training samples are input into a pre-built scoring model for training to obtain the trained scoring model.

4. The method according to claim 3, characterized in that After obtaining the training sample, the method further includes: The dialogue information sample is preprocessed to obtain a processed dialogue information sample.

5. The method according to claim 4, characterized in that The preprocessing of the dialogue information sample to obtain a processed dialogue information sample includes: The dialogue information sample is subjected to denoising and / or normalization processing to obtain a processed dialogue information sample.

6. The method according to claim 3, characterized in that The step of inputting the training sample into a pre-built scoring model for training to obtain the trained scoring model includes: Inputting part of the training samples as a training set into the pre-built scoring model to output a first prediction scoring result; Training the scoring model based on the comparison result between the scoring result and the scoring label to obtain a trained scoring model; Inputting part of the training samples as a test set into the trained scoring model to output a second prediction scoring result; The performance of the trained scoring model is evaluated based on the comparison result between the second predicted scoring result and the scoring label. If the set conditions are not met, the scoring model is retrained.

7. The method according to any one of claims 1 to 6, characterized in that: The scoring model is constructed using a pre-trained BERT model or a RoBERTa model.

8. A depressive tendency early warning device, characterized in that: The device comprises: An acquisition module, used to acquire the conversation content of the conversation target; the conversation content includes N pieces of conversation information; A processing module, configured to select M dialogue messages from the N dialogue messages, and call a trained scoring model to score each of the M dialogue messages individually; the score is used to indicate the degree of depression tendency of the dialogue target; The early warning module is used to send an alarm message when the total score of the M dialogue messages exceeds a set threshold.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the method for early warning of depression tendency according to any one of claims 1 to 7 is implemented.

10. A storage medium having computer-readable instructions stored thereon, characterized in that: The computer-readable instructions are executed by one or more processors to implement the depression tendency early warning method according to any one of claims 1 to 7.