An early warning system for self-injurious behavior in adolescents based on artificial intelligence

Abstract

The application discloses a kind of early warning system of self-injurious behavior of teenager based on artificial intelligence, specifically related to artificial intelligence technical field, including centralized behavior test division module, language content data acquisition module, language content data analysis module, physiological brain wave data acquisition module, physiological brain wave data analysis module, risk assessment module and early warning module.The application can not only be applied to school, family and other places where teenagers gather, but also can realize data sharing and linkage with mental health institutions, medical institutions and the like, to jointly build a prevention and control system for self-injurious behavior of teenagers.Through the application of the application, the occurrence of self-injurious behavior of teenagers can be effectively prevented and reduced, the physical and mental health of teenagers is protected, and the harmony and stability of society are promoted.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to an early warning system for self-harming behavior in adolescents based on artificial intelligence. Background Technology

[0002] Self-harm among teenagers is a growing social problem that not only affects their physical and mental health but may also lead to more serious consequences, such as suicide. Currently, early warning and intervention for self-harm among teenagers mainly rely on the observation and judgment of parents, teachers, and psychological counselors. However, this approach has obvious limitations, such as strong subjectivity, difficulty in quantification, and slow response speed, making it difficult to provide timely and accurate early warning and intervention.

[0003] The existing early warning system for adolescent self-harm behavior is a comprehensive process. First, the system collects a wide range of multi-source data on adolescents, including mental health assessments, daily behavior records, and social interactions, and performs data preprocessing to ensure the accuracy and consistency of the data. Then, the system extracts features related to self-harm behavior and uses these features to build and train a predictive model. The model will periodically or in real time assess the self-harm risk of adolescents and trigger an early warning mechanism when the risk exceeds a preset threshold, sending warning information to relevant personnel.

[0004] However, in practical use, it still has some shortcomings. For example, traditional early warning systems may lack accuracy in early warning, and cannot comprehensively assess the language content and physiological brainwave data of adolescents, resulting in the inability to accurately identify adolescents with a high risk of self-harm. Secondly, traditional early warning systems lack real-time and dynamic capabilities, and can only assess the behavior of adolescents at a certain moment or stage, but cannot adjust the early warning strategy in real time according to changes in adolescents' behavior. In addition, traditional early warning systems have limitations in data analysis and processing, and cannot make full use of artificial intelligence and big data technologies to deeply mine and analyze adolescents' behavioral data. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an early warning system for adolescent self-harm behavior based on artificial intelligence, which solves the problems mentioned in the background art through the following solutions.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an early warning system for self-harming behavior in adolescents based on artificial intelligence, comprising:

[0007] Concentrated Behavior Test Segmentation Module: This module is used to divide the target test subject's concentrated behavior test on preset self-harm words into various concentrated behavior tests according to the number of tests, and label them sequentially as 1, 2...n;

[0008] Language content data acquisition module: used to collect language feature data, communication behavior data, and risk statement data for each test, and transmit them to the language content data analysis module;

[0009] Language content data analysis module: includes language feature data analysis unit, communication behavior data analysis unit and risk statement data analysis unit, and transmits the analyzed data to risk assessment module;

[0010] Physiological brainwave data acquisition module: used to collect brainwave type data, brainwave activity data, and brainwave-behavior correlation data for each test, and transmit the collected data to the physiological brainwave data analysis module;

[0011] The physiological brainwave data analysis module includes a brainwave type data analysis unit, a brainwave activity data analysis unit, and a brainwave-behavior correlation data analysis unit. It then transmits the analyzed data to the risk assessment module.

[0012] Risk assessment module: By establishing a comprehensive risk assessment model, the language content analysis module and the physiological brainwave data analysis module are imported into the comprehensive risk assessment model to calculate the comprehensive risk index of the target test subject and transmit it to the early warning module.

[0013] Early warning module: Used to judge the comprehensive risk index of the target test subject based on the preset value of the comprehensive risk index, and issue an early warning signal based on the judgment result.

[0014] Preferably, the language content data acquisition module establishes a chatbot using big data and machine learning technologies, generates a language database of risk factors related to adolescent suicide and self-harm from previous research, and then puts it into the chatbot. The chatbot then conducts one-on-one interview-style chats with adolescents who have clearly exhibited suicidal or self-harm behaviors, and collects data from the chat content using natural language processing technology.

[0015] Preferably, the language feature data includes the number of positive words, the number of negative words, the number of self-harm related words, the proportion of sentences containing clauses or modifiers, the number of topic shifts, and the average time interval between topic shifts, labeled as Lp, La, Lf, Le, Ls, and Lt, respectively. The communication behavior data includes the average response time, the proportion of proactively initiated topics, the intensity score of emotional expression, and the positivity of self-description, labeled as Ct, Cp, Cs, and Cm, respectively. The risk statement data includes the number of times suicidal ideation was expressed, the number of times help-seeking intentions were expressed, the level of detail in self-harm behavior descriptions, and the number of times risk behavior plans were made, labeled as Rs, Rh, Rd, and Rp, respectively.

[0016] Preferably, the intensity score of emotional expression, the positivity of self-description, and the detail of self-harm behavior description are evaluated using natural language processing technology, with an evaluation range of 0-10.

[0017] Preferably, the language feature data analysis unit is used to establish a language feature data analysis model, import the language feature data transmitted by the language content data acquisition module into the language feature data analysis model, and calculate the language feature values ​​of each set of behavioral tests. The specific mathematical model is as follows: SL i Lp represents the language feature value of the i-th set behavior test. i Let La represent the number of positive words in the i-th concentrated behavior test. i Lf represents the number of negative words in the i-th concentration behavior test. i Let Le represent the number of self-injury-related words in the i-th concentrated behavioral test. i Ls represents the proportion of sentences containing clauses or modifiers in the i-th intensive behavior test. i Lt represents the number of topic shifts in the i-th concentrated behavior test. i Lt represents the average time interval for topic shifting during the i-th concentrated behavior test. min This represents the minimum average time interval for topic shifts in a concentrated behavior test.

[0018] Preferably, the communication behavior data analysis unit is used to establish a communication behavior data analysis model, import the communication behavior data transmitted by the language content data acquisition module into the communication behavior data analysis model, and calculate the communication behavior feature values ​​of each set of behavior tests. The specific mathematical model is as follows: SC i Let Ct represent the communication behavior feature value of the i-th concentrated behavior test. i Cp represents the average response time of the i-th concentrated behavioral test. i Cs represents the percentage of topics actively initiated during the i-th centralized behavior test. i Cm represents the intensity score of emotional expression in the i-th concentrated behavioral test. i This represents the degree of positivity in the self-report of the i-th concentrated behavior test.

[0019] Preferably, the risk statement data analysis unit is used to establish a risk statement data analysis model, import the risk statement data transmitted by the language content data acquisition module into the risk statement data analysis model, and calculate the risk statement feature values ​​for each set of behavioral tests. The specific mathematical model is as follows: SR i Rs represents the risk statement characteristic value of the i-th concentrated behavioral test. i Rh represents the number of suicidal ideation expressions during the i-th concentrated behavioral test.i Rd represents the number of times help-seeking intentions were expressed during the i-th concentrated behavior test. i Rp represents the level of detail in the description of self-injury behavior during the i-th concentrated behavioral test. i Rd represents the number of risk behavior plans in the i-th concentrated behavior test. max This represents the maximum level of detail in the description of self-harming behaviors in the concentrated behavior test.

[0020] Preferably, the EEG type data includes the average frequency of alpha waves, the average amplitude of alpha waves, the proportion of β wave activity time, the peak frequency of β waves, the number of occurrences of theta waves, and the average duration of theta waves, labeled as αf, αr, βt, βf, θf, and θt, respectively. The EEG activity data includes the maximum amplitude, the minimum amplitude, the similarity of corresponding wave segments between the left and right brain regions, and the standard deviation of EEG activity, labeled as Vb, Vs, Vl, and Vd, respectively. The EEG-behavioral association data includes the increase in alpha wave activity when positive words appear, the rate of change in β wave activity when response time is prolonged, the increase in the number of theta wave activities when suicidal thoughts are mentioned, and the decrease in delta wave activity when heart rate increases, labeled as Eα, Eβ, Eθ, and Eδ, respectively.

[0021] Preferably, the EEG type data analysis unit establishes an EEG type data analysis model, imports the EEG type data transmitted by the physiological EEG data acquisition module into the EEG type data analysis model, and calculates the EEG characteristic values ​​for each set of behavioral tests. The specific mathematical model is as follows: BK i Let αf represent the EEG characteristic value of the i-th concentration behavior test. i Let αr represent the average frequency of the α wave in the i-th concentrated behavior test. i Let βt represent the average amplitude of the α wave in the i-th concentration behavior test. i βf represents the percentage of β-wave activity time in the i-th concentrated behavioral test. i Let θf represent the peak frequency of the β wave in the i-th concentrated behavior test. i θt represents the number of occurrences of the θ wave in the i-th concentrated behavior test. i Let θf represent the average duration of the θ wave in the i-th concentrated behavior test. max This indicates the maximum number of occurrences of theta waves in the concentrated behavior test.

[0022] Preferably, the EEG activity data analysis unit establishes an EEG activity data analysis model, imports the EEG activity data transmitted by the physiological EEG data acquisition module into the EEG activity data analysis model, and calculates the EEG activity characteristic values ​​for each set of behavioral tests. The specific mathematical model is as follows: BV i Vb represents the characteristic value of EEG activity during the i-th concentration behavior test.i Vs represents the maximum amplitude of the i-th concentrated behavior test. i Vl represents the minimum amplitude of the i-th concentrated behavior test. i Vd represents the similarity of corresponding wavebands between the left and right hemispheres in the i-th concentration behavior test. i denoted as the standard deviation of EEG activity during the i-th concentration behavior test.

[0023] Preferably, the EEG and behavior correlation data analysis unit establishes an EEG and behavior correlation data analysis model, imports the EEG and behavior correlation data transmitted by the physiological EEG data acquisition module into the EEG and behavior correlation data analysis model, and calculates the neurobehavioral response rate for each set of behavioral tests. The specific mathematical model is as follows: BE i Eα represents the neurobehavioral response rate of the i-th concentration behavioral test. i Eβ represents the increase in alpha wave activity when positive words appear during the i-th concentration behavior test. i Eθ represents the rate of change of β-wave activity as the response time of the i-th concentration behavior test increases. i Eδ represents the number of increases in theta wave activity when suicidal ideation is mentioned during the i-th concentrated behavioral test. i Eα represents the percentage decrease in delta wave activity during the i-th concentration behavior test when the heart rate increases. min This indicates the maximum enhancement of alpha wave activity when positive words appear in the concentrated behavior test.

[0024] Preferably, the comprehensive risk assessment model is specifically expressed as follows: η represents the overall risk index of the target test subject, SL i SC represents the language feature value of the i-th concentrated behavior test. i SR represents the communication behavior characteristic value of the i-th concentrated behavior test. i Let BK represent the risk statement characteristic value of the i-th concentrated behavioral test. i BV represents the EEG characteristic value of the i-th concentration behavior test. i BE represents the characteristic value of EEG activity during the i-th concentration behavior test. i λ represents the neurobehavioral response rate of the i-th concentrated behavioral test, and λ represents other influencing factors of the comprehensive risk index.

[0025] Preferably, the preset value of the comprehensive risk index is represented as η. Def When η Def When η > η, it indicates that the self-harm risk of the target test subject is low, so the focused behavioral testing of the target test subject should continue. DefWhen the value is less than η, it indicates that the target test subject has a high risk of self-harm. In this case, an early warning signal is issued and the target test subject's concentrated behavioral test data is transmitted to the medical personnel's terminal.

[0026] The technical effects and advantages of this invention are as follows:

[0027] This invention, through the comprehensive application of artificial intelligence technology and big data analysis, achieves timely early warning and intervention for self-harm behavior in adolescents. Compared with traditional prevention and intervention methods, the early warning system of this invention has higher accuracy and efficiency, and can effectively reduce the incidence and risk of self-harm behavior in adolescents. This invention achieves comprehensive monitoring and analysis of adolescent self-harm behavior through the synergistic effect of multiple modules, including a centralized behavioral testing and segmentation module, a language content data collection module, and a language content data analysis module. This invention can accurately capture key information such as adolescents' language characteristics, communication behavior, and risk statements, and effectively process this information through intelligent analysis algorithms, thereby accurately identifying potential self-harm. To mitigate risks, this invention also employs advanced technologies such as natural language processing and machine learning to achieve intelligent and automated early warning systems for adolescent self-harm behavior. It can automatically perform data collection, analysis, and early warning operations, significantly reducing the burden of manual intervention and improving work efficiency. The early warning system of this invention also has broad application prospects and social value. This invention can not only be applied to places where adolescents gather, such as schools and homes, but also achieve data sharing and collaboration with mental health institutions and medical institutions to jointly build a prevention and control system for adolescent self-harm behavior. Through the application of this invention, the occurrence of adolescent self-harm behavior can be effectively prevented and reduced, protecting the physical and mental health of adolescents and promoting social harmony and stability. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] refer to Figure 1 The system shown is an AI-based early warning system for adolescent self-harm behavior, comprising a centralized behavior test segmentation module, a language content data collection module, a language content data analysis module, a physiological brainwave data collection module, a physiological brainwave data analysis module, a risk assessment module, and an early warning module.

[0031] The concentrated behavior test segmentation module is used to divide the concentrated behavior test of the target tester on the preset self-harm words into various concentrated behavior tests according to the number of tests, and label them as 1, 2...n in sequence.

[0032] The language content data acquisition module is used to collect language feature data, communication behavior data, and risk statement data for each test, and transmit them to the language content data analysis module.

[0033] The language content data acquisition module uses big data and machine learning technologies to build a chatbot, and generates a language database of risk factors related to adolescent suicide and self-harm from previous research, which is then placed into the chatbot. The chatbot then conducts one-on-one interview-style chats with adolescents who have a history of suicidal or self-harming behavior, and collects data from the chat content using natural language processing technology.

[0034] The language feature data includes the number of positive words, the number of negative words, the number of self-harm related words, the proportion of sentences containing clauses or modifiers, the number of topic shifts, and the average time interval between topic shifts, labeled as Lp, La, Lf, Le, Ls, and Lt, respectively. The communication behavior data includes the average response time, the proportion of proactively initiated topics, the intensity score of emotional expression, and the positivity of self-description, labeled as Ct, Cp, Cs, and Cm, respectively. The risk statement data includes the number of times suicidal ideation was expressed, the number of times help-seeking intentions were expressed, the level of detail in self-harm behavior descriptions, and the number of times risk behavior plans were made, labeled as Rs, Rh, Rd, and Rp, respectively.

[0035] The intensity of emotional expression, the positivity of self-description, and the level of detail in self-harm behavior description were assessed using natural language processing technology, with an assessment range of 0-10.

[0036] The language content data analysis module includes a language feature data analysis unit, a communication behavior data analysis unit, and a risk statement data analysis unit, and transmits the analyzed data to the risk assessment module.

[0037] The language feature data analysis unit is used to establish a language feature data analysis model. It imports the language feature data transmitted from the language content data acquisition module into the language feature data analysis model and calculates the language feature values ​​for each set of behavioral tests. The specific mathematical model is as follows: SL i Lp represents the language feature value of the i-th set behavior test. i Let La represent the number of positive words in the i-th concentrated behavior test. i Lf represents the number of negative words in the i-th concentration behavior test. i Let Le represent the number of self-injury-related words in the i-th concentrated behavioral test.i Ls represents the proportion of sentences containing clauses or modifiers in the i-th intensive behavior test. i Lt represents the number of topic shifts in the i-th concentrated behavior test. i Lt represents the average time interval for topic shifting during the i-th concentrated behavior test. min This represents the minimum average time interval for topic shifts in a concentrated behavior test.

[0038] The communication behavior data analysis unit is used to establish a communication behavior data analysis model. It imports the communication behavior data transmitted from the language content data acquisition module into the model and calculates the communication behavior feature values ​​for each set of behavioral tests. The specific mathematical model is as follows: SC i Let Ct represent the communication behavior feature value of the i-th concentrated behavior test. i Cp represents the average response time of the i-th concentrated behavioral test. i Cs represents the percentage of topics actively initiated during the i-th centralized behavior test. i Cm represents the intensity score of emotional expression in the i-th concentrated behavioral test. i This represents the degree of positivity in the self-report of the i-th concentrated behavior test.

[0039] The risk statement data analysis unit is used to establish a risk statement data analysis model. It imports the risk statement data transmitted from the language content data acquisition module into the risk statement data analysis model and calculates the risk statement feature values ​​for each set of behavioral tests. The specific mathematical model is as follows: SR i Rs represents the risk statement characteristic value of the i-th concentrated behavioral test. i Rh represents the number of suicidal ideation expressions during the i-th concentrated behavioral test. i Rd represents the number of times help-seeking intentions were expressed during the i-th concentrated behavior test. i Rp represents the level of detail in the description of self-injury behavior during the i-th concentrated behavioral test. i Rd represents the number of risk behavior plans in the i-th concentrated behavior test. max This represents the maximum level of detail in the description of self-harming behaviors in the concentrated behavior test.

[0040] The physiological brainwave data acquisition module is used to collect brainwave type data, brainwave activity data, and brainwave-behavior correlation data for each test, and transmits the collected data to the physiological brainwave data analysis module.

[0041] The EEG type data includes the average frequency of alpha waves, the average amplitude of alpha waves, the proportion of beta wave activity time, the peak frequency of beta waves, the number of occurrences of theta waves, and the average duration of theta waves, labeled as αf, αr, βt, βf, θf, and θt, respectively. The EEG activity data includes the maximum amplitude, the minimum amplitude, the similarity of corresponding wave segments between the left and right brain regions, and the standard deviation of EEG activity, labeled as Vb, Vs, Vl, and Vd, respectively. The EEG-behavioral association data includes the proportion of increased alpha wave activity when positive words are mentioned, the rate of change of beta wave activity when response time is prolonged, the number of times theta wave activity increases when suicidal thoughts are mentioned, and the proportion of decreased delta wave activity when heart rate increases, labeled as Eα, Eβ, Eθ, and Eδ, respectively.

[0042] Both the language content data acquisition module and the physiological brainwave data acquisition module employ data anonymization processing to protect the personal information of the target test subject; however, this embodiment will not provide specific details.

[0043] The physiological brainwave data analysis module includes a brainwave type data analysis unit, a brainwave activity data analysis unit, and a brainwave-behavior correlation data analysis unit, and transmits the analyzed data to the risk assessment module.

[0044] The EEG type data analysis unit establishes an EEG type data analysis model, imports the EEG type data transmitted by the physiological EEG data acquisition module into the EEG type data analysis model, and calculates the EEG characteristic values ​​for each set of behavioral tests. The specific mathematical model is as follows: BK i Let αf represent the EEG characteristic value of the i-th concentration behavior test. i Let αr represent the average frequency of the α wave in the i-th concentrated behavior test. i Let βt represent the average amplitude of the α wave in the i-th concentration behavior test. i βf represents the percentage of β-wave activity time in the i-th concentrated behavioral test. i Let θf represent the peak frequency of the β wave in the i-th concentrated behavior test. i θt represents the number of occurrences of the θ wave in the i-th concentrated behavior test. i Let θf represent the average duration of the θ wave in the i-th concentrated behavior test. max This indicates the maximum number of occurrences of theta waves in the concentrated behavior test.

[0045] The EEG activity data analysis unit establishes an EEG activity data analysis model, imports the EEG activity data transmitted by the physiological EEG data acquisition module into the EEG activity data analysis model, and calculates the EEG activity characteristic values ​​for each set of behavioral tests. The specific mathematical model is as follows: BV i Vb represents the characteristic value of EEG activity during the i-th concentration behavior test.i Vs represents the maximum amplitude of the i-th concentrated behavior test. i Vl represents the minimum amplitude of the i-th concentrated behavior test. i Vd represents the similarity of corresponding wavebands between the left and right hemispheres in the i-th concentration behavior test. i denoted as the standard deviation of EEG activity during the i-th concentration behavior test.

[0046] The EEG-behavior correlation data analysis unit establishes an EEG-behavior correlation data analysis model, imports the EEG-behavior correlation data transmitted by the physiological EEG data acquisition module into the EEG-behavior correlation data analysis model, and calculates the neurobehavioral response rate for each set of behavioral tests. The specific mathematical model is as follows: BE i Eα represents the neurobehavioral response rate of the i-th concentration behavioral test. i Eβ represents the increase in alpha wave activity when positive words appear during the i-th concentration behavior test. i Eθ represents the rate of change of β-wave activity as the response time of the i-th concentration behavior test increases. i Eδ represents the number of increases in theta wave activity when suicidal ideation is mentioned during the i-th concentrated behavioral test. i Eα represents the percentage decrease in delta wave activity during the i-th concentration behavior test when the heart rate increases. min This indicates the maximum enhancement of alpha wave activity when positive words appear in the concentrated behavior test.

[0047] The risk assessment module establishes a comprehensive risk assessment model, imports the language content analysis module and the physiological brainwave data analysis module into the comprehensive risk assessment model, calculates the comprehensive risk index of the target test subject, and transmits it to the early warning module.

[0048] The comprehensive risk assessment model is specifically expressed as follows: η represents the overall risk index of the target test subject, SL i SC represents the language feature value of the i-th concentrated behavior test. i SR represents the communication behavior characteristic value of the i-th concentrated behavior test. i Let BK represent the risk statement characteristic value of the i-th concentrated behavioral test. i BV represents the EEG characteristic value of the i-th concentration behavior test. i BE represents the characteristic value of EEG activity during the i-th concentration behavior test. i λ represents the neurobehavioral response rate of the i-th concentrated behavioral test, and λ represents other influencing factors of the comprehensive risk index.

[0049] The early warning module is used to judge the comprehensive risk index of the target test subject according to the preset value of the comprehensive risk index, and issue an early warning signal based on the judgment result.

[0050] The preset value of the comprehensive risk index is represented by η. Def When η Def When η > η, it indicates that the self-harm risk of the target test subject is low, so the focused behavioral testing of the target test subject should continue. Def When the value is less than η, it indicates that the target test subject has a high risk of self-harm. In this case, an early warning signal is issued and the target test subject's concentrated behavioral test data is transmitted to the medical personnel's terminal.

[0051] This invention, through the comprehensive application of artificial intelligence technology and big data analysis, achieves timely early warning and intervention for self-harm behavior in adolescents. Compared with traditional prevention and intervention methods, the early warning system of this invention has higher accuracy and efficiency, and can effectively reduce the incidence and risk of self-harm behavior in adolescents. This invention achieves comprehensive monitoring and analysis of adolescent self-harm behavior through the synergistic effect of multiple modules, including a centralized behavioral testing and segmentation module, a language content data collection module, and a language content data analysis module. This invention can accurately capture key information such as adolescents' language characteristics, communication behavior, and risk statements, and effectively process this information through intelligent analysis algorithms, thereby accurately identifying potential self-harm. To mitigate risks, this invention also employs advanced technologies such as natural language processing and machine learning to achieve intelligent and automated early warning systems for adolescent self-harm behavior. It can automatically perform data collection, analysis, and early warning operations, significantly reducing the burden of manual intervention and improving work efficiency. The early warning system of this invention also has broad application prospects and social value. This invention can not only be applied to places where adolescents gather, such as schools and homes, but also achieve data sharing and collaboration with mental health institutions and medical institutions to jointly build a prevention and control system for adolescent self-harm behavior. Through the application of this invention, the occurrence of adolescent self-harm behavior can be effectively prevented and reduced, protecting the physical and mental health of adolescents and promoting social harmony and stability.

[0052] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0053] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An early warning system for self-harming behavior in adolescents based on artificial intelligence, characterized in that, include: Concentrated Behavior Test Segmentation Module: This module is used to divide the target test subject's concentrated behavior test on preset self-harm words into various concentrated behavior tests according to the number of tests, and label them sequentially as 1, 2...n; Language content data acquisition module: used to collect language feature data, communication behavior data, and risk statement data for each test, and transmit them to the language content data analysis module; Language content data analysis module: includes language feature data analysis unit, communication behavior data analysis unit and risk statement data analysis unit, and transmits the analyzed data to risk assessment module; Physiological brainwave data acquisition module: used to collect brainwave type data, brainwave activity data, and brainwave-behavior correlation data for each test, and transmit the collected data to the physiological brainwave data analysis module; The physiological brainwave data analysis module includes a brainwave type data analysis unit, a brainwave activity data analysis unit, and a brainwave-behavior correlation data analysis unit. It then transmits the analyzed data to the risk assessment module. Risk assessment module: By establishing a comprehensive risk assessment model, the language content analysis module and the physiological brainwave data analysis module are imported into the comprehensive risk assessment model to calculate the comprehensive risk index of the target test subject and transmit it to the early warning module. Early warning module: Used to judge the comprehensive risk index of the target test subject based on the preset value of the comprehensive risk index, and issue an early warning signal based on the judgment result.

2. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The language feature data includes the number of positive words, the number of negative words, the number of self-harm related words, the proportion of sentences containing clauses or modifiers, the number of topic shifts, and the average time interval between topic shifts, labeled as Lp, La, Lf, Le, Ls, and Lt, respectively. The communication behavior data includes the average response time, the proportion of proactively initiated topics, the intensity score of emotional expression, and the positivity of self-description, labeled as Ct, Cp, Cs, and Cm, respectively. The risk statement data includes the number of times suicidal ideation was expressed, the number of times help-seeking intentions were expressed, the level of detail in self-harm behavior descriptions, and the number of times risk behavior plans were made, labeled as Rs, Rh, Rd, and Rp, respectively.

3. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The language feature data analysis unit is used to establish a language feature data analysis model. It imports the language feature data transmitted from the language content data acquisition module into the language feature data analysis model and calculates the language feature values ​​for each set of behavioral tests. The specific mathematical model is as follows: SL i Lp represents the language feature value of the i-th set behavior test. i Let La represent the number of positive words in the i-th concentrated behavior test. i Lf represents the number of negative words in the i-th concentration behavior test. i Let Le represent the number of self-injury-related words in the i-th concentrated behavioral test. i Ls represents the proportion of sentences containing clauses or modifiers in the i-th intensive behavior test. i Lt represents the number of topic shifts in the i-th concentrated behavior test. i Lt represents the average time interval for topic shifting during the i-th concentrated behavior test. min This represents the minimum average time interval for topic shifts in a concentrated behavior test.

4. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The communication behavior data analysis unit is used to establish a communication behavior data analysis model. It imports the communication behavior data transmitted from the language content data acquisition module into the model and calculates the communication behavior feature values ​​for each set of behavior tests. The specific mathematical model is as follows: SC i Let Ct represent the communication behavior feature value of the i-th concentrated behavior test. i Cp represents the average response time of the i-th concentrated behavioral test. i Cs represents the percentage of topics actively initiated during the i-th centralized behavior test. i Cm represents the intensity score of emotional expression in the i-th concentrated behavioral test. i This represents the degree of positivity in the self-report of the i-th concentrated behavior test.

5. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The risk statement data analysis unit is used to establish a risk statement data analysis model. It imports the risk statement data transmitted from the language content data acquisition module into the risk statement data analysis model and calculates the risk statement feature values ​​for each set of behavioral tests. The specific mathematical model is as follows: SR i Rs represents the risk statement characteristic value of the i-th concentrated behavioral test. i Rh represents the number of suicidal ideation expressions during the i-th concentrated behavioral test. i Rd represents the number of times help-seeking intentions were expressed during the i-th concentrated behavior test. i Rp represents the level of detail in the description of self-injury behavior during the i-th concentrated behavioral test. i Rd represents the number of risk behavior plans in the i-th concentrated behavior test. max This represents the maximum level of detail in the description of self-harming behaviors in the concentrated behavior test.

6. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The EEG type data includes the average frequency of alpha waves, the average amplitude of alpha waves, the proportion of beta wave activity time, the peak frequency of beta waves, the number of occurrences of theta waves, and the average duration of theta waves, labeled as αf, αr, βt, βf, θf, and θt, respectively. The EEG activity data includes the maximum amplitude, the minimum amplitude, the similarity of corresponding wave segments between the left and right brain regions, and the standard deviation of EEG activity, labeled as Vb, Vs, Vl, and Vd, respectively. The EEG-behavioral association data includes the proportion of increased alpha wave activity when positive words are mentioned, the rate of change of beta wave activity when response time is prolonged, the number of times theta wave activity increases when suicidal thoughts are mentioned, and the proportion of decreased delta wave activity when heart rate increases, labeled as Eα, Eβ, Eθ, and Eδ, respectively.

7. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The EEG type data analysis unit establishes an EEG type data analysis model, imports the EEG type data transmitted by the physiological EEG data acquisition module into the EEG type data analysis model, and calculates the EEG characteristic values ​​for each set of behavioral tests. The specific mathematical model is as follows: BK i Let αf represent the EEG characteristic value of the i-th concentration behavior test. i Let αr represent the average frequency of the α wave in the i-th concentrated behavior test. i Let βt represent the average amplitude of the α wave in the i-th concentration behavior test. i βf represents the percentage of β-wave activity time in the i-th concentrated behavioral test. i Let θf represent the peak frequency of the β wave in the i-th concentrated behavior test. i θt represents the number of occurrences of the θ wave in the i-th concentrated behavior test. i Let θf represent the average duration of the θ wave in the i-th concentrated behavior test. max This indicates the maximum number of occurrences of theta waves in the concentrated behavior test.

8. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The EEG-behavior correlation data analysis unit establishes an EEG-behavior correlation data analysis model, imports the EEG-behavior correlation data transmitted by the physiological EEG data acquisition module into the EEG-behavior correlation data analysis model, and calculates the neurobehavioral response rate for each set of behavioral tests. The specific mathematical model is as follows: BE i Eα represents the neurobehavioral response rate of the i-th concentration behavioral test. i Eβ represents the increase in alpha wave activity when positive words appear during the i-th concentration behavior test. i Eθ represents the rate of change of β-wave activity as the response time of the i-th concentration behavior test increases. i Eδ represents the number of increases in theta wave activity when suicidal ideation is mentioned during the i-th concentrated behavioral test. i Eα represents the percentage decrease in delta wave activity during the i-th concentration behavior test when the heart rate increases. min This indicates the maximum enhancement of alpha wave activity when positive words appear in the concentrated behavior test.

9. The early warning system for adolescent self-harm behavior based on artificial intelligence according to claim 1, characterized in that: The comprehensive risk assessment model is specifically expressed as follows: η represents the overall risk index of the target test subject, SL i SC represents the language feature value of the i-th concentrated behavior test. i SR represents the communication behavior characteristic value of the i-th concentrated behavior test. i Let BK represent the risk statement characteristic value of the i-th concentrated behavioral test. i BV represents the EEG characteristic value of the i-th concentration behavior test. i BE represents the characteristic value of EEG activity during the i-th concentration behavior test. i λ represents the neurobehavioral response rate of the i-th concentrated behavioral test, and λ represents other influencing factors of the comprehensive risk index.

Images