A multi-dimensional student mental health monitoring and early warning method and system based on AI

Abstract

The present invention discloses a multi-dimensional student mental health monitoring and early warning method and system based on AI, which relates to the field of mental health, including: S1, receiving multi-dimensional psychological interaction data of students. S2, monitoring the real-time operation and maintenance information of the data channel and adjusting it. S3, processing and analyzing to obtain the first verification label of the interaction quality performance data. S4, integrating and processing the interaction content data to obtain a classification prompt data collection and inputting it into the management terminal. The present invention provides a multi-dimensional student mental health monitoring and early warning method based on AI, screens the interaction quality of the interaction quality performance data uploaded by users to the student mental health platform and interprets the content of the interaction content data, analyzes and processes the interaction content data uploaded by users from multiple dimensions and obtains its comprehensive evaluation value, thereby realizing multi-dimensional analysis and management of student mental health data.

Description

Technical Field

[0001] The present invention relates to the field of mental health, and specifically to a multi-dimensional student mental health monitoring and early warning method and system based on AI. Background Art

[0002] In order to effectively monitor and protect students' mental health, many schools have adopted psychological assessment reports to evaluate students' mental state. However, this traditional method of student psychological monitoring can no longer meet the timeliness needs of psychological monitoring.

[0003] The prior art, such as the invention patent with announcement number: CN110477932B, provides a student psychological stress assessment method and system based on the Internet and cloud computing, wherein a student psychological stress assessment method based on the Internet and cloud computing includes: step 1: constructing a psychological stress database and uploading it to a cloud platform; step 2: constructing a psychological stress model set based on the database on the cloud platform; step 3: conducting a measured psychological stress assessment on the cloud platform according to the psychological stress model set. A student psychological stress assessment system based on the Internet and cloud computing is also disclosed, including: a psychological stress database construction module; a psychological stress model set construction module; and a psychological stress assessment module.

[0004] Existing technologies, such as the invention patent with announcement number: CN103919537B, provide an emotion recording analysis and guidance system and its implementation method, which mainly solves the problem that the existing technology lacks an emotion guidance system that can prevent and reduce the occurrence of mental illness in advance and cannot meet people's needs. Its implementation method includes: the emotion management platform reads the user's mood data and the physiological data collected by the emotional smart watch; transmits the user's mood data and physiological data to the data processing center; the data processing center analyzes the user information, obtains the user's current emotional state, and stores the analysis results; when the analysis shows that the user currently has a bad mood, the analysis results are fed back to the emotion management platform, and the emotion management platform automatically extracts funny pictures and / or jokes from the database and sends them to the user terminal.

[0005] Based on the above scheme, it can be seen that for the monitoring and early warning of mental health, the existing technology usually simply obtains the emotional data uploaded by the user and directly processes the data. However, in actual applications, due to the data collection process, the data uploaded by the user is usually missing, incomplete, and incomplete, which usually leads to quality defects in quality and integrity. At this time, if the collected data is processed directly, it will lead to unreasonable use of data processing resources. Therefore, there is a problem of insufficient data processing in the current student mental health monitoring. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a multi-dimensional student mental health monitoring and early warning method and system based on AI. To achieve the above purpose, the present invention is implemented through the following technical solutions: A multi-dimensional student mental health monitoring and early warning method and system based on AI, comprising:

[0007] The student mental health platform receives in real time the student multi-dimensional psychological interaction data transmitted by the edge channel port, wherein the student multi-dimensional psychological interaction data includes interaction quality performance data and interaction content data;

[0008] Synchronously monitor the real-time operation and maintenance information of the data channel of the student mental health platform for processing, and obtain the real-time bandwidth quota of the data channel of the student mental health platform for adjustment and processing;

[0009] Preprocessing and analyzing the received interaction quality performance data to obtain an initial quality inspection value of the interaction quality performance data, and generating a first verification tag based on the initial quality inspection value;

[0010] If the first verification tag indicates that the data verification is not compliant, the data quality abnormality information is transmitted to the edge channel port for prompting; if the first verification tag indicates that the data verification is compliant, the interactive content data is transmitted to the central processor;

[0011] The central processing unit receives the interactive content data for integrated processing, obtains a collection of classified prompt data of the interactive content data, and inputs the collection into a management terminal connected to the student mental health platform;

[0012] The interaction quality performance data and interaction content data specifically include:

[0013] The interaction quality performance data specifically includes text interaction quality performance data and video interaction quality performance data;

[0014] The interactive content data specifically includes text interactive content and video interactive content;

[0015] The preprocessing and analyzing of the received interaction quality performance data specifically includes:

[0016] The text interaction quality performance data specifically includes the total number of text bytes, the number of bytes of effective text content, and the length of time the text content is input;

[0017] The video interaction quality performance data specifically includes the total number of video bytes, the number of video valid content bytes, and the number of video continuous frames;

[0018] Based on the text interaction quality performance data and the video interaction quality performance data, the interaction quality performance verification data is extracted from the database, and the interaction quality performance verification data includes the ideal text valid byte ratio, the ideal video valid byte ratio and the ideal video continuous frame ratio. The initial quality inspection value of the interaction quality performance data is obtained by comparison processing. The initial quality inspection value of the interaction quality performance data is used to characterize the data quality level in the interaction stage. The specific processing method is as follows:

[0019] ;

[0020] in, is the initial quality inspection value of the interactive quality performance data, is the initial quality verification value of the text interaction quality performance data, It is the initial quality verification value of the video interaction quality performance data;

[0021] The first verification label is generated based on the initial quality inspection value, and the specific processing conditions are:

[0022] Based on the initial quality inspection value of the interaction quality performance data, an initial quality inspection value threshold is extracted from the student mental health platform. By comparison, if the initial quality inspection value is less than or equal to the initial quality inspection value threshold, the first verification label of the interaction quality performance data is defined as non-compliant; if the initial quality inspection value is greater than the initial quality inspection value threshold, the first verification label of the interaction quality performance data is defined as compliant;

[0023] The initial quality inspection value threshold is obtained by extracting from the database in the student mental health platform and is used to determine whether the interactive quality performance data in the student mental health platform meets the quality requirements;

[0024] The initial quality inspection value of the interaction quality performance data, the specific processing conditions are as follows:

[0025] ;

[0026] ;

[0027] in, is the initial quality verification value of the text interaction quality performance data, is the initial quality verification value of the video interaction quality performance data, is the total number of text bytes, is the number of bytes of text valid content, Enter the duration for the text content, Enter the duration for your ideal text content. is the number of bytes of the video's valid content. is the total number of video bytes, is the number of consecutive video frames, is the total number of video frames, is the ideal text effective byte ratio, is the ideal video effective byte ratio, is the ideal video continuous frame ratio, is the text byte weight factor, Enter the duration weight factor for the text content, is the video byte weight factor, is the video continuity weight factor.

[0028] As a preferred technical solution, the data channel real-time operation and maintenance information of the synchronous monitoring student mental health platform is processed, specifically including:

[0029] Real-time operation and maintenance information of the data channel of the student mental health platform, wherein the real-time operation and maintenance information of the data channel includes real-time delay of the data channel, real-time communication transmission rate of the data channel, real-time flow of the data channel and real-time packet loss rate of the data channel;

[0030] The real-time operation and maintenance information of the data channel of the student mental health platform is processed to obtain the real-time load index of the data channel of the student mental health platform. The specific processing method is as follows:

[0031] ;

[0032] in, Real-time load index of the data channel of the student mental health platform, Real-time communication transmission rate for student mental health platform, Real-time traffic of the data channel for the student mental health platform, Real-time delays for student mental health platforms, The real-time packet loss rate of the data channel of the student mental health platform, The ideal data channel communication transmission rate preset for the student mental health platform, Ideal data channel flow for student mental health platform, Ideal data channel delay for student mental health platform, The ideal data channel packet loss rate for the student mental health platform, is the real-time communication transmission rate weight factor, is the real-time delay weight factor, is the real-time traffic weight factor, is the real-time packet loss rate weight factor.

[0033] As a preferred technical solution, the real-time bandwidth quota of the data channel of the student mental health platform is obtained and adjusted, specifically including:

[0034] Based on the data channel real-time load index of the student mental health platform, the data channel real-time load index threshold is extracted from the student mental health platform. By comparison, if the data channel real-time load index is less than or equal to the data channel real-time load index threshold, the data channel real-time bandwidth quota is defined as zero. If the data channel real-time load index is greater than the data channel real-time load index threshold, the difference between the data channel real-time load index and the data channel real-time load index threshold is extracted and recorded as the data channel bandwidth adjustment reference value of the student mental health platform. The difference is mapped and matched with the bandwidth quota corresponding to each data channel bandwidth adjustment reference value interval set in the database to obtain the data channel real-time bandwidth quota of the student mental health platform, thereby performing bandwidth adjustment processing on the data channel of the student mental health platform according to the data channel real-time bandwidth quota;

[0035] The real-time load index threshold of the data channel is extracted from the database in the student mental health platform and is used to determine the critical state of the data channel load.

[0036] As a preferred technical solution, the central processor receives the interactive content data for integrated processing, specifically including:

[0037] Based on the positive keyword vocabulary set and the negative keyword vocabulary set in the database, the central processor performs keyword recognition on the text interaction content to obtain a number of keywords in the text interaction content, and maps and matches them one by one with the positive keyword vocabulary set and the negative keyword vocabulary set in the database, and counts the total number of positive keywords and the total number of negative keywords in the text interaction content;

[0038] The interactive video is segmented into a set number of frames to obtain each frame of the interactive video image, and by identifying whether there is a human face image in each frame of the interactive video image, if so, it is recorded as a valid frame image, thereby counting each valid frame image, and by identifying and counting the total number of positive facial micro-expressions and negative facial micro-expressions appearing in the valid frame image;

[0039] Extract the audio of the video interaction content for recognition and convert it into audio text interaction content, based on the positive keyword vocabulary set and negative keyword vocabulary set in the database, and perform keyword recognition on the audio text interaction content through the central processing unit to obtain several keywords in the audio text interaction content, and map and match them one by one with the positive keyword vocabulary set and negative keyword vocabulary set in the database, and count the total number of positive keywords and the total number of negative keywords in the audio text interaction content;

[0040] Based on the total number of positive keywords and negative keywords in the text interaction content, the total number of frames of each type of facial micro-expression in the valid frame images in the video interaction content, and the total number of positive keywords and negative keywords in the audio-text interaction content, a comprehensive evaluation index of the interaction content data is obtained through comprehensive processing.

[0041] As a preferred technical solution, the classification prompt data collection of the interactive content data obtained specifically includes:

[0042] Based on the interactive content data, a positive prompt data set and a negative prompt data set are counted therefrom, wherein the positive prompt data set includes the total number of positive keywords in the text interactive content, the total number of frames of positive facial micro-expressions in the video interactive content, and the total number of positive keywords in the audio-text interactive content; the negative prompt data set includes the total number of negative keywords in the text interactive content, the total number of frames of negative facial micro-expressions in the video interactive content, and the total number of negative keywords in the audio-text interactive content;

[0043] The positive prompt dataset and the negative prompt dataset are taken together as a classification prompt data collection of the interactive content data.

[0044] The present invention also provides a multi-dimensional student mental health monitoring and early warning system based on AI, which specifically includes:

[0045] A data information receiving module, used for the student mental health platform to receive in real time the student multi-dimensional psychological interaction data transmitted by the edge channel port, wherein the student multi-dimensional psychological interaction data includes interaction quality performance data and interaction content data;

[0046] The data channel processing module is used to synchronously monitor the real-time operation and maintenance information of the data channel of the student mental health platform for processing, and obtain the real-time bandwidth quota of the data channel of the student mental health platform for adjustment processing;

[0047] A data quality inspection module performs preprocessing and analysis on the received interaction quality performance data to obtain an initial quality inspection value of the interaction quality performance data, and generates a first verification tag based on the initial quality inspection value;

[0048] The data quality feedback module transmits data quality abnormality information to the edge channel port for prompting if the first verification tag indicates that the data verification is not compliant, and transmits the interactive content data to the central processor if the first verification tag indicates that the data verification is compliant;

[0049] In the data content processing module, the central processing unit receives the interactive content data for integrated processing, obtains a collection of classified prompt data of the interactive content data, and inputs it into the management terminal connected to the student mental health platform.

[0050] Compared with the prior art, the embodiments of the present invention have at least the following beneficial effects:

[0051] (1) The present invention provides an AI-based multi-dimensional student mental health monitoring and early warning method. By screening the interaction quality performance data uploaded by users to the student mental health platform and interpreting the content of the interaction content data, the interactive content data uploaded by users is analyzed and processed from multiple dimensions to obtain its comprehensive evaluation value.

[0052] (2) The present invention obtains the initial quality inspection value of the interaction quality performance data by analyzing the interaction quality performance data. The interaction quality performance data is first analyzed, and then it is determined whether to further analyze the interaction content data based on the quality analysis results, rather than directly processing the collected data. This further reduces the error phenomenon in the analysis results of the students' psychological data, promotes the rational use of data processing resources, and effectively solves the problem of insufficient data processing in the current monitoring of students' mental health.

[0053] (3) The present invention performs bandwidth adjustment processing on the data channel of the student mental health platform based on the real-time load index of the data channel of the student mental health platform, thereby avoiding delays and packet losses caused by network congestion during peak data transmission periods, thereby ensuring the timeliness and integrity of the data.

[0054] Of course, any product implementing the present invention does not necessarily need to achieve all of the above advantages at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the method flow of the present invention.

[0056] Figure 2 An example scatter diagram of comprehensive evaluation indexes of interactive content data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] In the description of the present invention, it should be understood that the terms "opening", "upper", "lower", "thickness", "top", "middle", "length", "inside", "all around" and the like indicating orientation or positional relationship are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the components or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0059] See also Figure 1 As shown, the embodiment of the present invention provides a multi-dimensional student mental health monitoring and early warning method based on AI, which specifically includes:

[0060] S1. The student mental health platform receives in real time the student multi-dimensional psychological interaction data transmitted by the edge channel port, wherein the student multi-dimensional psychological interaction data includes interaction quality performance data and interaction content data;

[0061] The edge channel port sends the collected multi-dimensional psychological interaction data of students to the message queue ApacheKafka, and the student mental health platform reads the multi-dimensional psychological interaction data of students from the message queue;

[0062] The interaction quality performance data specifically includes text interaction quality performance data and video interaction quality performance data;

[0063] The interactive content data specifically includes text interactive content data and video interactive content data.

[0064] S2, synchronously monitor the real-time operation and maintenance information of the data channel of the student mental health platform for processing, and obtain the real-time bandwidth quota of the data channel of the student mental health platform for adjustment processing;

[0065] The data channel real-time operation and maintenance information of the synchronous monitoring student mental health platform is processed, specifically including:

[0066] Use Zabbix monitoring platform to monitor server performance, database performance, operation logs and network device logs in real time.

[0067] Real-time operation and maintenance information of the data channel of the student mental health platform, wherein the real-time operation and maintenance information of the data channel includes real-time delay of the data channel, real-time communication transmission rate of the data channel, real-time flow of the data channel and real-time packet loss rate of the data channel;

[0068] The real-time operation and maintenance information of the data channel is obtained through the network monitoring tool Wireshark, including the real-time delay of the data channel, the real-time communication transmission rate of the data channel, the real-time traffic of the data channel and the real-time packet loss rate of the data channel.

[0069] The real-time operation and maintenance information of the data channel of the student mental health platform is processed to obtain the real-time load index of the data channel of the student mental health platform. The specific processing method is as follows:

[0070] ;

[0071] in, Real-time load index of the data channel of the student mental health platform, Real-time communication transmission rate for student mental health platform, Real-time traffic of the data channel for the student mental health platform, Real-time delays for student mental health platforms, The real-time packet loss rate of the data channel of the student mental health platform, The ideal data channel communication transmission rate preset for the student mental health platform, Ideal data channel flow for student mental health platform, Ideal data channel delay for student mental health platform, The ideal data channel packet loss rate for the student mental health platform, is the real-time communication transmission rate weight factor, is the real-time delay weight factor, is the real-time traffic weight factor, is the real-time packet loss rate weight factor.

[0072] It should be noted that the real-time communication transmission rate weight factor, real-time delay weight factor, real-time traffic weight factor and real-time packet loss rate weight factor all have a value range between 0 and 1 and satisfy The real-time communication transmission rate weight factor is an influence factor corresponding to the real-time load index of the data channel of the student mental health platform preset in the database, which indicates the numerical value of the influence degree of the real-time communication transmission rate on the real-time load index of the data channel of the student mental health platform. When used, the real-time communication transmission rate weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the real-time load index of the data channel of the student mental health platform and the real-time communication transmission rate weight factor corresponding to the real-time communication transmission rate form a mapping set, and the real-time communication transmission rate is input into the mapping set to obtain the real-time communication transmission rate weight factor, where the mapping relationship is one-to-one correspondence; the real-time delay weight factor is an influence factor corresponding to the real-time load index of the data channel of the student mental health platform preset in the database, which indicates the numerical value of the influence degree of the real-time delay on the real-time load index of the data channel of the student mental health platform. When used, the real-time delay weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the real-time load index of the data channel of the student mental health platform and the real-time delay weight factor corresponding to the real-time delay form a mapping set, and the real-time delay is input into the mapping set to obtain the real-time delay weight factor, where The mapping relationship is one-to-one; the real-time traffic weight factor is the influence factor corresponding to the real-time load index of the data channel of the student mental health platform preset in the database, which indicates the numerical value of the influence degree of real-time traffic on the real-time load index of the data channel of the student mental health platform. When used, the real-time traffic weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the real-time load index of the data channel of the student mental health platform and the real-time traffic weight factor corresponding to the real-time traffic form a mapping set, and the real-time traffic is input into the mapping set to obtain the real-time traffic weight factor, in which the mapping relationship is one-to-one; the real-time packet loss rate weight factor is the influence factor corresponding to the real-time load index of the data channel of the student mental health platform preset in the database, which indicates the numerical value of the influence degree of real-time packet loss rate on the real-time load index of the data channel of the student mental health platform. When used, the real-time packet loss rate weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the real-time load index of the data channel of the student mental health platform and the real-time packet loss rate weight factor corresponding to the real-time packet loss rate form a mapping set, and the real-time packet loss rate is input into the mapping set to obtain the real-time packet loss rate weight factor, in which the mapping relationship is one-to-one;

[0073] It should also be noted that there is a certain correlation between the real-time delay of the data channel, the real-time communication transmission rate of the data channel, the real-time traffic of the data channel and the real-time packet loss rate of the data channel. When the delay is high, it means that the data packet takes a longer time to transmit in the network, which will directly affect the transmission rate of real-time communication. High delay usually leads to a decrease in the data transmission rate because the data packet takes longer to reach the destination; on the contrary, low delay helps to increase the transmission rate of real-time communication because the data packet can be transmitted through the network faster, reducing waiting time.

[0074] Real-time traffic refers to the traffic generated by data packets during transmission in network communications. When the network traffic is large, it usually causes network congestion, which in turn increases the transmission delay of data packets. Therefore, the level of real-time traffic will indirectly affect the real-time delay of the data channel. In order to reduce the delay, it is necessary to reasonably control and manage network traffic to avoid congestion.

[0075] Packet loss rate refers to the ratio between the number of data packets lost during data transmission and the total number of data packets sent. High latency usually causes data packets to take too long to transmit in the network, increasing the risk of data packet loss and thus increasing the packet loss rate. Conversely, low latency helps reduce the packet loss rate because data packets can reach their destination faster, reducing the possibility of loss during transmission.

[0076] The real-time communication transmission rate is affected by both network bandwidth and real-time traffic. When the bandwidth is constant, an increase in real-time traffic will lead to a decrease in the transmission rate because the network needs to allocate more bandwidth to handle the increased traffic. Therefore, in order to maintain a high transmission rate, it is necessary to reasonably control the real-time traffic to avoid exceeding the network's carrying capacity.

[0077] The size of the real-time communication transmission rate will indirectly affect the packet loss rate. When the transmission rate is higher, data packets can be transmitted through the network faster, reducing the possibility of loss during transmission, thereby reducing the packet loss rate; however, if the transmission rate is too high and the network bandwidth or processing power is insufficient, it will also cause network congestion and increased packet loss rate.

[0078] The increase in real-time traffic will increase the density and complexity of data packet transmission in the network, thereby increasing the risk of data packet loss. When the network traffic exceeds its carrying capacity, data packets are usually discarded due to network congestion, resulting in an increase in packet loss rate. Therefore, in order to reduce the packet loss rate, it is necessary to reasonably control and manage real-time traffic to avoid network congestion.

[0079] The following table provides five examples of data channel real-time load indexes for the student mental health platform. In this example, the real-time communication transmission rate weight factor, real-time delay weight factor, real-time traffic weight factor, and real-time packet loss rate weight factor are set to 0.2, 0.3, 0.2, and 0.3, respectively. The specific output data channel real-time load index is as follows:

[0080] Table 1 Example of real-time load index of data channel of student mental health platform

[0081]

[0082] As shown in the above table, the greater the data channel real-time delay, the data channel real-time communication transmission rate, the data channel real-time traffic and the data channel real-time packet loss rate, the greater the data channel real-time load index of the student mental health platform. The greater the real-time load index of the data channel of the student mental health platform is positively correlated with these four parameters. Among them, the real-time load index of the data channel No. 4 is the largest. Compared with the other four groups of instances, the real-time delay and real-time packet loss rate of the data channel No. 4 are higher, indicating that when the real-time communication transmission rate and real-time traffic are at normal levels, the higher the real-time delay and the real-time packet loss rate, the greater the real-time load of the data channel.

[0083] The real-time bandwidth quota of the data channel of the student mental health platform is obtained and adjusted, specifically including:

[0084] Based on the real-time load index of the data channel of the student mental health platform, the data channel real-time load index threshold is extracted from the student mental health platform. By comparison, if the data channel real-time load index is less than or equal to the data channel real-time load index threshold, the data channel real-time bandwidth quota is defined as zero; if the data channel real-time load index is greater than the data channel real-time load index threshold, the difference between the data channel real-time load index and the data channel real-time load index threshold is extracted and recorded as the data channel bandwidth adjustment reference value of the student mental health platform. The difference is mapped and matched with the bandwidth quota corresponding to each data channel bandwidth adjustment reference value interval set in the database to obtain the data channel real-time bandwidth quota of the student mental health platform. The data channel of the student mental health platform is bandwidth adjusted according to the data channel real-time bandwidth quota.

[0085] The real-time load index threshold of the data channel is extracted from the database in the student mental health platform and is used to determine the critical state of the data channel load.

[0086] In a specific embodiment, the real-time load index threshold of the data channel is set directly in the database when the student mental health platform is developed. There are many methods for setting the real-time load index threshold of the data channel, such as obtaining it through statistical analysis, including statistics on the real-time operation and maintenance information of the channel of the student mental health platform under historical low load conditions (for example, the CPU and memory occupancy rates of the platform processor are below 80%), including historical data channel real-time delays, historical data channel real-time communication transmission rates, historical data channel real-time traffic, and historical data channel real-time packet loss rates. Multiple statistical processing is performed to obtain the corresponding data channel real-time load index, and the real-time load index of the data channel obtained multiple times is averaged to obtain the data channel real-time load index threshold, or experts judge the boundaries between peak and off-peak periods through corresponding historical data to set the threshold. In addition, user feedback surveys can also be used to collect opinions on usage experience. Satisfaction may decrease when the load is high, thereby adjusting the threshold.

[0087] The bandwidth adjustment process for the data channel of the student mental health platform specifically includes:

[0088] Log in to the router, check the existing interface configuration and its bandwidth limit, based on the real-time bandwidth quota of the data channel of the student mental health platform, use the congestion control algorithm in the TCP protocol to adjust the transmission rate of data packets, distribute traffic to different queues, set the bandwidth limit of the interface, and adjust the bandwidth of the data channel of the student mental health platform.

[0089] By adjusting the bandwidth of the data channel of the student mental health platform based on the real-time load index of the data channel of the student mental health platform, delays and packet losses caused by network congestion can be avoided during peak data transmission periods, ensuring the timeliness and integrity of the data. For mental health platforms that need to collect and analyze large amounts of data, effective bandwidth management can ensure the timely collection and transmission of data and support more efficient big data analysis.

[0090] S3, preprocessing and analyzing the received interaction quality performance data to obtain an initial quality inspection value of the interaction quality performance data, and generating a first verification tag based on the initial quality inspection value;

[0091] The preprocessing and analyzing of the received interaction quality performance data specifically includes:

[0092] Among them, the text interaction quality performance data is used to characterize the interactive data quality of users when uploading text data, including the total number of text bytes, the number of bytes of text valid content, and the length of time it takes to input text content;

[0093] The total number of text bytes reflects the length of the text uploaded by the user. If the total number of bytes is too large, it usually indicates that the user has uploaded detailed information. If the total number of bytes is too small, it usually means that the information is incomplete. A larger total number of text bytes usually indicates that the user has invested more time and effort in content generation.

[0094] Effective text content refers to text content that can be translated by encoding. In contrast, invalid text content refers to garbled content that cannot be translated by encoding. The ratio of effective text content bytes to total bytes can measure the actual value of user uploaded content. A high proportion of effective content indicates that the text data uploaded by users is more relevant and practical. A low number of effective content bytes usually means that the text contains a lot of irrelevant or repeated information, which will affect the quality of interaction. Using the correct encoding (such as ASCII, UTF-8, UTF-16) can accurately calculate the number of effective text content bytes.

[0095] The time it takes to input text content is used to represent the time it takes for users to think when generating content. If the input time is longer than the ideal input time set in the database, it usually indicates that the user has difficulty inputting text content or has made mistakes when using the platform. If the input time is shorter than the ideal input time set in the database, it usually indicates that the user did not think before generating content.

[0096] Video interaction quality performance data is used to characterize the interaction quality data when users upload video data, including the total number of video bytes, the number of bytes of effective video content, and the number of continuous video frames;

[0097] The total number of video bytes reflects the overall size of the video file. A larger total number of bytes usually means a longer video or a higher resolution. If the total number of video bytes is larger, it usually indicates that the user is willing to share more detailed or longer content, which usually increases the quality of the interaction.

[0098] The number of bytes of effective video content is the number of bytes with effective information in the video data uploaded by the user. It removes the empty shots and black screens in the video data, and reflects the effectiveness of the video data. The ratio relative to the total bytes can measure the relevance and practicality of the video content. A high ratio usually means that the video content is more compact and more focused. The higher the number of bytes of effective video content, the more valuable information there is in the video, and the less redundant and irrelevant content, which helps improve the quality of interaction. The number of bytes of effective video content can be extracted through FFmpeg.

[0099] The number of continuous video frames reflects the smoothness of video playback. A high continuous frame rate usually means that there is no frame skipping or lag during video playback. A large number of continuous video frames indicates that the video content is relatively complete and no important picture information is lost.

[0100] Based on the text interaction quality performance data and the video interaction quality performance data, the interaction quality performance verification data is extracted from the database, and the interaction quality performance verification data includes the ideal text valid byte ratio, the ideal video valid byte ratio and the ideal video continuous frame ratio. The initial quality inspection value of the interaction quality performance data is obtained by comparison processing. The initial quality inspection value of the interaction quality performance data is used to characterize the data quality level in the interaction stage. The specific processing method is as follows:

[0101] ;

[0102] ;

[0103] ;

[0104] in, is the initial quality inspection value of the interactive quality performance data, is the initial quality verification value of the text interaction quality performance data, is the initial quality verification value of the video interaction quality performance data, is the total number of text bytes, is the number of bytes of text valid content, Enter the duration for the text content, Enter the duration for your ideal text content. is the number of bytes of the video's valid content. is the total number of video bytes, is the number of consecutive video frames, is the total number of video frames, is the ideal text effective byte ratio, is the ideal video effective byte ratio, is the ideal video continuous frame ratio, is the text byte weight factor, Enter the duration weight factor for the text content, is the video byte weight factor, is the video continuity weight factor.

[0105] It should be noted that the text byte weight factor, text content input duration weight factor, video byte weight factor and video continuity weight factor all have a value range between 0 and 1 and satisfy , , the text byte weight factor is an influence factor corresponding to the initial quality inspection value of the interactive quality performance data preset in the database, indicating the numerical value of the influence of the number of text bytes on the initial quality inspection value of the interactive quality performance data. When used, the text byte weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the initial quality inspection value of the interactive quality performance data and the text byte weight factor corresponding to the number of text bytes form a mapping set, and the real-time text byte number is input into the mapping set to obtain the text byte weight factor, wherein the mapping relationship is one-to-one correspondence; the text content input duration weight factor is an influence factor corresponding to the initial quality inspection value of the interactive quality performance data preset in the database, indicating the numerical value of the influence of the text content input duration on the initial quality inspection value of the interactive quality performance data. When used, the text content input duration weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the initial quality inspection value of the interactive quality performance data and the text content input duration weight factor corresponding to the text content input duration form a mapping set, and the real-time text content input duration is input into the mapping set to obtain the text content input duration weight factor, wherein the text content input duration weight factor is one-to-one correspondence. The mapping relationship in is one-to-one; the video byte weight factor is an influence factor corresponding to the initial quality inspection value of the interactive quality performance data preset in the database, indicating the numerical value of the influence of the number of video bytes on the initial quality inspection value of the interactive quality performance data. When used, the video byte weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the initial quality inspection value of the interactive quality performance data and the video byte weight factor corresponding to the number of video bytes form a mapping set, and the real-time video byte number is input into the mapping set to obtain the video byte weight factor, wherein the mapping relationship is one-to-one; the video continuity weight factor is an influence factor corresponding to the initial quality inspection value of the interactive quality performance data preset in the database, indicating the numerical value of the influence of the number of continuous video frames on the initial quality inspection value of the interactive quality performance data. When used, the video continuity weight factor can be directly obtained from the database, and the corresponding relationship is a preset mapping relationship, for example: the initial quality inspection value of the interactive quality performance data and the video continuity weight factor corresponding to the number of continuous video frames form a mapping set, and the real-time video continuous frame number is input into the mapping set to obtain the video continuity weight factor, wherein the mapping relationship is one-to-one;

[0106] It should be noted that there is a certain correlation between the total number of text bytes, the number of text effective content bytes and the length of time it takes to enter text content. Generally, when the total number of text bytes is high, the number of effective content bytes is also usually high, because longer texts usually contain more information. However, there are usually differences between the two. A very long text usually contains a lot of formatting information, irrelevant filler content or repeated information, which will cause the number of effective content bytes to be not proportional to the total number of text bytes. The longer the text entered by the user, the longer it will take. Therefore, the total number of text bytes is usually proportional to the length of time it takes to enter text content. The input time usually does not increase linearly with the increase in text length, because the user's typing speed, thinking time and editing process usually affect the input time. If the user spends more time and thinking on content generation, more effective content will usually be generated. Therefore, the number of effective content bytes is usually proportional to the input time. Different users usually have different input times when generating the same amount of effective content, depending on their typing speed, thinking process and expression ability.

[0107] In addition, there is a certain correlation between the total number of video bytes, the number of video effective content bytes, and the number of video continuous frames. The number of video effective content bytes is part of the total number of video bytes. The total number of bytes includes all video data, while the effective content bytes only refer to the video content that is valuable to the viewer. The ratio of effective content bytes to the total number of bytes can reflect the quality of the video content. If the proportion of effective content is high, it means that there is a lot of useful information in the video. Otherwise, it usually contains a lot of non-essential data. High total number of video bytes and high effective content bytes usually mean that the video content is rich and most of the data is valid. If the number of continuous frames is also high, the video is usually long and coherent.

[0108] The first verification label is generated based on the initial quality inspection value, and the specific processing conditions are:

[0109] Based on the initial quality inspection value of the interaction quality performance data, the initial quality inspection value threshold is extracted from the student mental health platform. Through comparison, if the initial quality inspection value is less than or equal to the initial quality inspection value threshold, the first verification label of the interaction quality performance data is defined as non-compliant; if the initial quality inspection value is greater than the initial quality inspection value threshold, the first verification label of the interaction quality performance data is defined as compliant.

[0110] The initial quality inspection value threshold is extracted from the database in the student mental health platform and is used to determine whether the interaction quality performance data in the student mental health platform meets the quality requirements.

[0111] In a specific embodiment, the initial quality inspection value is set directly in the database when the student mental health platform is developed. There are many methods for setting the initial quality inspection value threshold. For example, during development, multiple groups of sample interaction quality performance data that have been verified by personnel and confirmed to be of good quality are uploaded, and the multiple groups of sample interaction quality performance data are analyzed to obtain the initial quality inspection value of the multiple groups of interaction quality performance data. The initial quality inspection values ​​of the multiple groups of interaction quality performance data are then averaged to obtain the initial quality inspection value threshold. In addition, the threshold can also be continuously adjusted according to user feedback and verification results during platform operation. For example, an initial value is set and then gradually adjusted according to the actual situation of user feedback.

[0112] If the first verification tag indicates that the data verification is not compliant, the data quality abnormality information is transmitted to the edge channel port for prompting; if the first verification tag indicates that the data verification is compliant, the interactive content data is transmitted to the central processor;

[0113] S4. The central processing unit receives the interactive content data for integrated processing, obtains a classification prompt data collection of the interactive content data, and inputs it into a management terminal connected to the student mental health platform.

[0114] The central processor receives the interactive content data for integrated processing, specifically including:

[0115] Based on the positive keyword vocabulary set and the negative keyword vocabulary set in the database, and through the central processor, the natural language processing (NLP) technology is used to perform keyword recognition on the text interaction content, and a number of keywords in the text interaction content are obtained, and they are mapped and matched one by one with the positive keyword vocabulary set and the negative keyword vocabulary set in the database, and the total number of positive keywords and the total number of negative keywords in the text interaction content are counted;

[0116] Use video processing software PR to segment the interactive video into set frames to obtain each frame of the interactive video. Use the face detection algorithm Haar feature classifier to detect each frame to determine whether there is a human face image. If so, it is recorded as a valid frame image. Thus, each valid frame image is counted, and the total number of positive and negative facial micro-expressions appearing in the valid frame images is counted by identifying and counting.

[0117] The total number of frames of each type of facial micro-expression in the valid frame image is identified and counted, specifically including:

[0118] Active Shape Models (ASM) are used to extract key facial feature points, such as eyes, eyebrows, nose and mouth, and the geometric relationships between feature points, including distance and angle, are calculated. The movement patterns of facial muscles are analyzed, small changes in facial expressions are detected, and the start and end frames of micro-expressions are determined. The Affdex sentiment analysis tool is used to classify micro-expressions based on the standards of the Facial Action Coding System (FACS). The micro-expressions in valid frame images are divided into positive facial micro-expressions and negative facial micro-expressions, and the total number of frames of positive facial micro-expressions and negative facial micro-expressions are counted respectively.

[0119] Using automatic speech recognition (ASR) technology, extract the audio of the video interaction content for recognition and convert it into audio text. Based on the positive keyword vocabulary set and negative keyword vocabulary set in the database, the central processor performs keyword recognition on the audio text interaction content to obtain several keywords in the audio text interaction content, and maps and matches them one by one with the positive keyword vocabulary set and negative keyword vocabulary set in the database, and counts the total number of positive keywords and negative keywords in the audio text interaction content.

[0120] The obtained classification prompt data collection of interactive content data specifically includes:

[0121] Based on the interactive content data, a positive prompt data set and a negative prompt data set are counted from it, wherein the positive prompt data set includes the total number of positive keywords in the text interactive content, the total number of frames of positive facial micro-expressions in the video interactive content, and the total number of positive keywords in the audio text interactive content; the negative prompt data set includes the total number of negative keywords in the text interactive content, the total number of frames of negative facial micro-expressions in the video interactive content, and the total number of negative keywords in the audio text interactive content.

[0122] The positive prompt dataset and the negative prompt dataset are taken together as a classification prompt data collection of the interactive content data.

[0123] The positive keyword vocabulary set and the negative keyword vocabulary set in the database are automatically updated according to the popular online vocabulary. In this embodiment, the positive keyword vocabulary set includes keywords expressing positivity, happiness and joy, such as "haha", "happy", etc., and the negative keyword vocabulary set includes keywords expressing negativity, sadness and depression, such as "very annoying", "sad", etc.;

[0124] In another embodiment, the central processor receives the interactive content data for integrated processing, and further includes the following analysis steps:

[0125] Based on the total number of positive and negative keywords in text interaction content, the total number of frames of each type of facial micro-expression in valid frame images in video interaction content, and the total number of positive and negative keywords in audio-text interaction content, a comprehensive evaluation index of the interaction content data is obtained.

[0126] The obtaining of the comprehensive evaluation index of the interactive content data specifically includes:

[0127] ;

[0128] in, is the comprehensive evaluation index of interactive content data, is the total number of negative keywords in text interaction content, is the total number of positive keywords in the text interaction content, is the total number of negative keywords annotated in the audio interaction content, is the total number of positive keywords in the audio-text interaction content, is the total number of frames with negative facial micro-expressions in the valid frame images in the video interaction content, The total number of frames with positive facial micro-expressions in the effective frame images in the video interaction content, is the threshold of the positive keyword ratio of the text interaction content extracted from the database, is the threshold of the proportion of positive facial micro-expressions in the video interaction content extracted from the database, is the threshold of the positive keyword ratio of the audio-text interaction content extracted from the database, is the text interaction content weight factor, is the video interaction content weight factor, is the audio-text interaction content weight factor.

[0129] It should be noted that the text interaction content weight factor, video interaction content weight factor and audio text interaction content weight factor all have a value range between 0 and 1 and satisfy The text interaction content weight factor is an influence factor corresponding to the comprehensive evaluation index of the interactive content data preset in the database, which indicates the numerical value of the influence degree of the evaluation index of the text interaction content on the comprehensive evaluation index of the interactive content data. When used, the text interaction content weight factor can be directly obtained from the database, and the corresponding relationship is a pre-set mapping relationship. For example, the comprehensive evaluation index of the interactive content data and the text interaction content weight factor corresponding to the evaluation index of the text interaction content form a mapping set, and the evaluation index of the real-time text interaction content is input into the mapping set to obtain the text interaction content weight factor, wherein the mapping relationship is one-to-one correspondence; the video interaction content weight factor is an influence factor corresponding to the comprehensive evaluation index of the interactive content data preset in the database, which indicates the numerical value of the influence degree of the evaluation index of the video interaction content on the comprehensive evaluation index of the interactive content data. When used, the video interaction content weight factor can be directly obtained from the database, and the corresponding relationship is pre-set. For example, the comprehensive evaluation index of the interactive content data and the video interactive content weight factor corresponding to the evaluation index of the video interactive content form a mapping set, and the evaluation index of the real-time video interactive content is input into the mapping set to obtain the video interactive content weight factor, wherein the mapping relationship is one-to-one correspondence; the audio-text interactive content weight factor is an influence factor corresponding to the comprehensive evaluation index of the interactive content data preset in the database, and represents a numerical value of the influence degree of the evaluation index of the audio-text interactive content on the comprehensive evaluation index of the interactive content data. When used, the audio-text interactive content weight factor can be directly obtained from the database, and its corresponding relationship is a pre-set mapping relationship, for example, the comprehensive evaluation index of the interactive content data and the audio-text interactive content weight factor corresponding to the evaluation index of the audio-text interactive content form a mapping set, and the evaluation index of the real-time audio-text interactive content is input into the mapping set to obtain the audio-text interactive content weight factor, wherein the mapping relationship is one-to-one correspondence;

[0130] In another embodiment, the comprehensive evaluation index of the interactive content data is processed to obtain an integrated processing classification prompt data collection of the interactive content data, specifically including:

[0131] Based on the comprehensive evaluation index of the interactive content data, the comprehensive evaluation index threshold of the interactive content data is extracted from the student mental health platform. Through comparison, if the comprehensive evaluation index is less than or equal to the comprehensive evaluation index threshold, the classified prompt data collection of the interactive content data is marked as requiring warning; if the comprehensive evaluation index is greater than the comprehensive evaluation index threshold, the classified prompt data collection of the interactive content data is marked as not requiring warning.

[0132] The comprehensive evaluation index threshold of the interactive content data is obtained by analyzing the historical verification comprehensive evaluation data on the student mental health platform, and is used to determine whether the interactive content data in the student mental health platform meets the comprehensive evaluation requirements. The specific analysis method includes:

[0133] In a specific embodiment, the comprehensive evaluation index threshold of the interactive content data is set directly in the database when the student mental health platform is developed. There are many methods for setting the comprehensive evaluation index threshold of the interactive content data, such as analyzing it through statistical methods, including statistical interactive content data in a historical high comprehensive evaluation index state (for example, the total number of positive keywords in the historical text interactive content accounts for more than 60%), including the total number of positive keywords and the total number of negative keywords in the historical text interactive content, the total number of frames of each type of facial micro-expression in the valid frame image in the historical video interactive content, and the total number of positive keywords and the total number of negative keywords in the audio text interactive content, multiple statistical processing to obtain the corresponding comprehensive evaluation index of the interactive content data, and the comprehensive evaluation index of the interactive content data obtained multiple times is averaged to obtain the comprehensive evaluation index threshold of the interactive content data, or experts judge that the interactive content data is at the boundary of the need for warning and no need for warning through the corresponding historical data to set it. In addition, opinions on the usage experience can also be collected through user feedback surveys. When the judgment is wrong, the satisfaction may decrease, thereby adjusting the threshold.

[0134] like Figure 2 Shown is a scatter plot example of a comprehensive evaluation index of interactive content data involved in another embodiment of the present invention. The example graph shows comprehensive evaluation indexes of ten different groups of interactive content data. In this embodiment, the horizontal axis is the instance number of the interactive content data, and the vertical axis is the comprehensive evaluation index of the interactive content data. The higher the comprehensive evaluation index in the graph, the higher the positive tendency of the interactive content data. The comprehensive evaluation index of 100,000 is used as the threshold of the comprehensive evaluation index of the interactive content data. In the example graph, the dotted line is used as the baseline. In the example graph, the classified prompt data collection of the interactive content data below the baseline is marked as a demand warning, and the interactive content data above the baseline is marked as no demand warning.

[0135] In this embodiment, the present invention provides a multi-dimensional student mental health monitoring and early warning system based on AI, which specifically includes:

[0136] A data information receiving module, used for the student mental health platform to receive in real time the student multi-dimensional psychological interaction data transmitted by the edge channel port, wherein the student multi-dimensional psychological interaction data includes interaction quality performance data and interaction content data;

[0137] The data channel processing module is used to synchronously monitor the real-time operation and maintenance information of the data channel of the student mental health platform for processing, and obtain the real-time bandwidth quota of the data channel of the student mental health platform for adjustment processing;

[0138] A data quality inspection module performs preprocessing and analysis on the received interaction quality performance data to obtain an initial quality inspection value of the interaction quality performance data, and generates a first verification tag based on the initial quality inspection value;

[0139] The data quality feedback module transmits data quality abnormality information to the edge channel port for prompting if the first verification tag indicates that the data verification is not compliant, and transmits the interactive content data to the central processor if the first verification tag indicates that the data verification is compliant;

[0140] In the data content processing module, the central processing unit receives the interactive content data for integrated processing, obtains a collection of classified prompt data of the interactive content data, and inputs it into the management terminal connected to the student mental health platform.

[0141] The present invention provides an AI-based multi-dimensional student mental health monitoring and early warning method, which analyzes and processes the interactive content data uploaded by users to the student mental health platform from multiple dimensions and obtains its comprehensive evaluation value by screening the interactive quality performance data of the interactive quality uploaded by users to the student mental health platform and interpreting the interactive content data.

[0142] The present invention analyzes and obtains the initial quality inspection value of the interaction quality performance data. The interaction quality performance data is first analyzed, and then it is determined whether to further analyze the interaction content data based on the quality analysis result, rather than directly processing the collected data. This further reduces the error phenomenon in the analysis results of the students' psychological data, promotes the rational use of data processing resources, and effectively solves the problem of insufficient data processing in the current monitoring of students' mental health.

[0143] The present invention performs bandwidth adjustment processing on the data channel of the student mental health platform based on the real-time load index of the data channel of the student mental health platform, which can avoid delays and packet losses caused by network congestion during data transmission peak periods and ensure the timeliness and integrity of data.

[0144] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0145] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that technicians in the relevant technical field can understand and use the present invention well. As long as they do not deviate from the structure of the present invention or exceed the scope defined by the present invention, they should all belong to the protection scope of the present invention.

Claims

1. A multi-dimensional student mental health monitoring and early warning method based on AI, characterized in that: include: The student mental health platform receives in real time the student multi-dimensional psychological interaction data transmitted by the edge channel port, wherein the student multi-dimensional psychological interaction data includes interaction quality performance data and interaction content data; Synchronously monitor the real-time operation and maintenance information of the data channel of the student mental health platform for processing, and obtain the real-time bandwidth quota of the data channel of the student mental health platform for adjustment and processing; Preprocessing and analyzing the received interaction quality performance data to obtain an initial quality inspection value of the interaction quality performance data, and generating a first verification tag based on the initial quality inspection value; If the first verification tag indicates that the data verification is not compliant, the data quality abnormality information is transmitted to the edge channel port for prompting; if the first verification tag indicates that the data verification is compliant, the interactive content data is transmitted to the central processor; The central processing unit receives the interactive content data for integrated processing, obtains a collection of classified prompt data of the interactive content data, and inputs the collection into a management terminal connected to the student mental health platform; The interaction quality performance data and interaction content data specifically include: The interaction quality performance data specifically includes text interaction quality performance data and video interaction quality performance data; The interactive content data specifically includes text interactive content and video interactive content; The preprocessing and analyzing of the received interaction quality performance data specifically includes: The text interaction quality performance data specifically includes the total number of text bytes, the number of bytes of effective text content, and the length of time the text content is input; The video interaction quality performance data specifically includes the total number of video bytes, the number of video valid content bytes, and the number of video continuous frames; Based on the text interaction quality performance data and the video interaction quality performance data, the interaction quality performance verification data is extracted from the database, and the interaction quality performance verification data includes the ideal text valid byte ratio, the ideal video valid byte ratio and the ideal video continuous frame ratio. The initial quality inspection value of the interaction quality performance data is obtained by comparison processing. The initial quality inspection value of the interaction quality performance data is used to characterize the data quality level in the interaction stage. The specific processing method is as follows: ; in, is the initial quality inspection value of the interactive quality performance data, is the initial quality verification value of the text interaction quality performance data, It is the initial quality verification value of the video interaction quality performance data; The first verification label is generated based on the initial quality inspection value, and the specific processing conditions are: Based on the initial quality inspection value of the interaction quality performance data, an initial quality inspection value threshold is extracted from the student mental health platform. By comparison, if the initial quality inspection value is less than or equal to the initial quality inspection value threshold, the first verification label of the interaction quality performance data is defined as non-compliant; if the initial quality inspection value is greater than the initial quality inspection value threshold, the first verification label of the interaction quality performance data is defined as compliant; The initial quality inspection value threshold is obtained by extracting from the database in the student mental health platform and is used to determine whether the interactive quality performance data in the student mental health platform meets the quality requirements; The initial quality inspection value of the interaction quality performance data, the specific processing conditions are as follows: ; ; in, is the initial quality verification value of the text interaction quality performance data, is the initial quality verification value of the video interaction quality performance data, is the total number of text bytes, is the number of bytes of text valid content, Enter the duration for the text content, Enter the duration for your ideal text content. is the number of bytes of the video's valid content. is the total number of video bytes, is the number of consecutive video frames, is the total number of video frames, is the ideal text effective byte ratio, is the ideal video effective byte ratio, is the ideal video continuous frame ratio, is the text byte weight factor, Enter the duration weight factor for the text content, is the video byte weight factor, is the video continuity weight factor.

2. According to claim 1, a multi-dimensional student mental health monitoring and early warning method based on AI is characterized by: The data channel real-time operation and maintenance information of the synchronous monitoring student mental health platform is processed, specifically including: Real-time operation and maintenance information of the data channel of the student mental health platform, wherein the real-time operation and maintenance information of the data channel includes real-time delay of the data channel, real-time communication transmission rate of the data channel, real-time flow of the data channel and real-time packet loss rate of the data channel; The real-time operation and maintenance information of the data channel of the student mental health platform is processed to obtain the real-time load index of the data channel of the student mental health platform. The specific processing method is as follows: ； in, Real-time load index of the data channel of the student mental health platform, Real-time communication transmission rate for student mental health platform, Real-time traffic of the data channel for the student mental health platform, Real-time delays for student mental health platforms, The real-time packet loss rate of the data channel of the student mental health platform, The ideal data channel communication transmission rate preset for the student mental health platform, Ideal data channel flow for student mental health platform, Ideal data channel delay for student mental health platform, The ideal data channel packet loss rate for the student mental health platform, is the real-time communication transmission rate weight factor, is the real-time delay weight factor, is the real-time traffic weight factor, is the real-time packet loss rate weight factor.

3. According to claim 2, a multi-dimensional student mental health monitoring and early warning method based on AI is characterized by: The real-time bandwidth quota of the data channel of the student mental health platform is obtained and adjusted, specifically including: Based on the data channel real-time load index of the student mental health platform, the data channel real-time load index threshold is extracted from the student mental health platform. By comparison, if the data channel real-time load index is less than or equal to the data channel real-time load index threshold, the data channel real-time bandwidth quota is defined as zero. If the data channel real-time load index is greater than the data channel real-time load index threshold, the difference between the data channel real-time load index and the data channel real-time load index threshold is extracted and recorded as the data channel bandwidth adjustment reference value of the student mental health platform. The difference is mapped and matched with the bandwidth quota corresponding to each data channel bandwidth adjustment reference value interval set in the database to obtain the data channel real-time bandwidth quota of the student mental health platform, thereby performing bandwidth adjustment processing on the data channel of the student mental health platform according to the data channel real-time bandwidth quota; The real-time load index threshold of the data channel is extracted from the database in the student mental health platform and is used to determine the critical state of the data channel load.

4. According to claim 1, a multi-dimensional student mental health monitoring and early warning method based on AI is characterized by: The central processor receives the interactive content data for integrated processing, specifically including: Based on the positive keyword vocabulary set and the negative keyword vocabulary set in the database, the central processor performs keyword recognition on the text interaction content to obtain a number of keywords in the text interaction content, and maps and matches them one by one with the positive keyword vocabulary set and the negative keyword vocabulary set in the database, and counts the total number of positive keywords and the total number of negative keywords in the text interaction content; The interactive video is segmented into a set number of frames to obtain each frame of the interactive video image, and by identifying whether there is a human face image in each frame of the interactive video image, if so, it is recorded as a valid frame image, thereby counting each valid frame image, and by identifying and counting the total number of positive facial micro-expressions and negative facial micro-expressions appearing in the valid frame image; Extract the audio of the video interaction content for recognition and convert it into audio text interaction content, based on the positive keyword vocabulary set and negative keyword vocabulary set in the database, and perform keyword recognition on the audio text interaction content through the central processing unit to obtain several keywords in the audio text interaction content, and map and match them one by one with the positive keyword vocabulary set and negative keyword vocabulary set in the database, and count the total number of positive keywords and the total number of negative keywords in the audio text interaction content; Based on the total number of positive keywords and negative keywords in the text interaction content, the total number of frames of each type of facial micro-expression in the valid frame images in the video interaction content, and the total number of positive keywords and negative keywords in the audio-text interaction content, a comprehensive evaluation index of the interaction content data is obtained through comprehensive processing.

5. According to claim 4, a multi-dimensional student mental health monitoring and early warning method based on AI is characterized by: The obtained classification prompt data collection of interactive content data specifically includes: Based on the interactive content data, a positive prompt data set and a negative prompt data set are counted therefrom, wherein the positive prompt data set includes the total number of positive keywords in the text interactive content, the total number of frames of positive facial micro-expressions in the video interactive content, and the total number of positive keywords in the audio-text interactive content; the negative prompt data set includes the total number of negative keywords in the text interactive content, the total number of frames of negative facial micro-expressions in the video interactive content, and the total number of negative keywords in the audio-text interactive content; The positive prompt dataset and the negative prompt dataset are taken together as a classification prompt data collection of the interactive content data.

6. A system using an AI-based multi-dimensional student mental health monitoring and early warning method as described in any one of claims 1 to 5, characterized in that: Includes the following modules: A data information receiving module, used for the student mental health platform to receive in real time the student multi-dimensional psychological interaction data transmitted by the edge channel port, wherein the student multi-dimensional psychological interaction data includes interaction quality performance data and interaction content data; The data channel processing module is used to synchronously monitor the real-time operation and maintenance information of the data channel of the student mental health platform for processing, and obtain the real-time bandwidth quota of the data channel of the student mental health platform for adjustment processing; A data quality inspection module performs preprocessing and analysis on the received interaction quality performance data to obtain an initial quality inspection value of the interaction quality performance data, and generates a first verification tag based on the initial quality inspection value; The data quality feedback module transmits data quality abnormality information to the edge channel port for prompting if the first verification tag indicates that the data verification is not compliant, and transmits the interactive content data to the central processor if the first verification tag indicates that the data verification is compliant; In the data content processing module, the central processing unit receives the interactive content data for integrated processing, obtains a collection of classified prompt data of the interactive content data, and inputs it into the management terminal connected to the student mental health platform.

Images