A Multimodal Non-Contact Screening System for Adolescents and its Implementation Method

Abstract

This invention discloses a non-intrusive multimodal screening system and its implementation method for adolescents, relating to the field of adolescent mental health assessment technology. It includes a data collection terminal layer, a core algorithm layer, and a platform service layer. The data collection terminal layer is used to collect adolescent multimodal behavioral data non-contactly and without physical contact, based on the MSSMHS scale's collection requirements. This non-intrusive multimodal screening system and its implementation method for adolescents effectively overcomes the shortcomings of traditional subjective assessments, which are easily influenced by concealment and emotional interference, by establishing a standardized mapping between scale dimensions and objective behavior and a dedicated algorithm. This significantly improves the objectivity and clinical relevance of the test results. Simultaneously, relying on non-intrusive data collection and local encryption processing mechanisms, it eliminates the psychological defenses of test subjects, increases their willingness to cooperate, and fundamentally protects the security of personal privacy data, achieving accurate identification and compliant management of mental states.

Description

Technical Field

[0001] This invention relates to the field of adolescent mental health assessment technology, specifically to an adolescent multimodal non-intrusive screening system and its implementation method. Background Technology

[0002] Currently, the detection rate of psychological and emotional problems among adolescents aged 12-18 in my country has reached 17.5%. Mental health issues have become a core public health problem affecting the healthy growth of adolescents. The Mental Health Scale for Middle School Students (MSSMHS) is the most widely used and reliable national standard psychological assessment tool for adolescents in China, and it is also the core legal basis for mental health screening in primary and secondary schools.

[0003] However, the traditional MSSMHS scale, which relies primarily on paper-and-pencil self-assessment, faces the following bottlenecks in its routine application on campuses: Adolescents are prone to concealing their true psychological state due to social expectations, defensive mentality, and stigma, leading to distorted test results; moreover, a single paper-and-pencil test takes more than 20 minutes, and screening all students and staff across the entire school requires a large number of dedicated psychological teachers; in addition, straightforward test questions can easily trigger rebellious psychology, and avoidance behavior, especially among high-risk groups, can easily lead to missed screenings. Summary of the Invention

[0004] The purpose of this invention is to provide a multimodal non-intrusive screening system and implementation method for adolescents, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a multimodal non-intrusive screening system for adolescents, comprising a data acquisition terminal layer, a core algorithm layer, and a platform service layer; The data collection terminal layer is used to collect multimodal behavioral data of adolescents in a non-contact and non-invasive manner based on the collection requirements of the MSSMHS scale. The core algorithm layer includes an MSSMHS three-level mapping module, a multimodal weighted calibration algorithm module, and a risk level intelligent determination module. The MSSMHS three-level mapping module establishes a three-level standardized mapping relationship of "MSSMHS core dimensions - clinical behavioral factors - multimodal collection indicators" to achieve the matching of abstract dimensions of the scale to objective collection indicators. The multimodal weighted calibration algorithm module, based on the MSSMHS scale weight system and age-specific norms, achieves accurate conversion of collected data into standardized scale scores. The risk level intelligent determination module, based on MSSMHS norm thresholds, completes the standardized determination of psychological risk levels. The platform service layer is used to implement functions such as class management, abnormal warning, targeted intervention matching, home-school collaboration, and model iteration based on the MSSMHS scale assessment results.

[0006] Furthermore, in the MSSMHS three-level mapping module, the four core dimensions of the MSSMHS scale include emotion regulation ability, social adaptation ability, focus and executive ability, and sensory adaptation; the weight percentages are 30%, 30%, 25%, and 15% respectively. Each core dimension is broken down into three secondary clinical behavioral factors, and the secondary factors are equally weighted according to the corresponding dimension.

[0007] Furthermore, the formula for the secondary factor basic score of the multimodal weighted calibration algorithm module is as follows: in These are measured values. The standard values ​​for age-specific norms of MSSMHS are as follows. For factor weights, For calibration coefficients, This is the conversion factor for fractions.

[0008] Furthermore, the multimodal weighted calibration algorithm module employs a multi-source data fusion formula: in Teachers will be given feedback scores based on their observations.

[0009] Furthermore, the intelligent risk level determination module is set with four risk levels. The high warning determination condition is: scores of two or more primary dimensions < 60 points, or core factors are lower than 50% of the norm.

[0010] Furthermore, the core algorithm layer also includes a local encryption and blockchain evidence storage module, which is used to destroy the original audio and video data after local processing, encrypt and store only the feature values ​​and scoring results, and generate a unique traceability ID through the blockchain to achieve full-chain traceability.

[0011] One implementation method, which applies the above-mentioned multimodal non-intrusive screening system for adolescents, includes the following steps: S1: Construct a three-level precise mapping system for the MSSMHS scale, establish a standardized transformation path of "MSSMHS core dimensions - clinical behavior factors - multimodal collection indicators", and clarify the age-specific norm standard values ​​and collection rules for each factor; S2: Using a wearable multimodal acquisition terminal, multimodal data corresponding to 12 clinical behavioral factors of adolescents are collected seamlessly based on a three-level mapping system. S3: The collected data is processed using the MSSMHS proprietary multimodal weighted calibration algorithm to calculate the standardized dimension scores of 0-100 points, which are aligned with the national standard scale. This step also includes a three-level data verification mechanism: outlier removal, consistency verification, and missing value supplementation; to ensure the reliability of the scoring results. S4: Based on the national norm threshold of the MSSMHS scale, automatically determine the psychological risk level of adolescents and match the corresponding targeted intervention plan. S5: Generates encrypted assessment reports, completes end-to-end tamper-proof notarization through blockchain, and pushes them to teachers' and parents' ends; S6: Based on clinical feedback and retest data, the three-level mapping system and algorithm model are optimized through incremental learning to form a closed loop across the entire chain; The model iteration cycle consists of monthly optimization of calibration coefficients and quarterly updates to the mapping system and weight allocation to ensure continuous improvement in detection accuracy.

[0012] Furthermore, in step S1, the three-level mapping system dynamically adjusts the norm standard values ​​according to the three educational stages of primary school, junior high school, and senior high school to adapt to the age-specific norm characteristics of the MSSMHS scale.

[0013] Furthermore, in step S3, the multimodal weighted calibration algorithm module also includes a dynamic correction mechanism for environmental factors, the specific steps of which are as follows: Real-time acquisition of light intensity values ​​in the current testing environment Environmental decibel levels and population density value Construct a comprehensive environmental disturbance index The calculation formula is: in, The preset optimal light threshold; The preset ambient background noise threshold; These are the environmental factor weighting coefficients; based on the aforementioned comprehensive environmental disturbance index. Weights of secondary factors Dynamically adjust and generate environment adaptation weights. : in, As an environmental sensitivity modulator, For environmental disturbance nonlinear correction function; when When the preset threshold is exceeded, the system automatically increases the weight of the "perceptual adaptation" dimension and reduces the original score weight of the "focus" dimension, which is affected by environmental interference.

[0014] This invention provides a multimodal non-intrusive screening system and implementation method for adolescents, which has the following beneficial effects: 1. This invention, by establishing a standardized mapping between scale dimensions and objective behavior and a proprietary algorithm, effectively overcomes the shortcomings of traditional subjective assessments, which are easily subject to concealment and emotional interference. It significantly improves the objectivity and clinical relevance of test results. At the same time, relying on non-intrusive data collection and local encryption processing mechanisms, it eliminates the psychological defenses of test subjects and increases their willingness to cooperate, while fundamentally ensuring the security of personal privacy data and achieving accurate identification and compliant management of psychological states.

[0015] 2. This invention enhances the system's adaptability to the external physical environment by introducing an environmental perception and dynamic weight adjustment mechanism. It can automatically balance the scoring weights of different psychological dimensions according to the real-time environmental interference level, thereby effectively filtering out abnormal data interference caused by external factors such as light and noise. This ensures that the evaluation results can still truly reflect the individual's psychological traits even in non-ideal environments, and greatly improves the system's stability and applicability. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of a multimodal non-intrusive screening system and implementation method for adolescents according to the present invention. Figure 2 This is a schematic diagram of the weight adjustment process for a multimodal non-intrusive screening system and implementation method for adolescents according to the present invention. Detailed Implementation

[0017] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0018] like Figures 1-2 As shown, a multimodal non-intrusive screening system for adolescents and its implementation method are disclosed. The system includes a data acquisition terminal layer, a core algorithm layer, and a platform service layer. The data collection terminal layer is used to collect multimodal behavioral data of adolescents in a non-contact and non-invasive manner based on the collection requirements of the MSSMHS scale. The core algorithm layer includes an MSSMHS three-level mapping module, a multimodal weighted calibration algorithm module, and a risk level intelligent determination module. The MSSMHS three-level mapping module establishes a three-level standardized mapping relationship of "MSSMHS core dimensions - clinical behavioral factors - multimodal collection indicators" to achieve the matching of abstract dimensions of the scale to objective collection indicators. The multimodal weighted calibration algorithm module, based on the MSSMHS scale weight system and age-specific norms, achieves accurate conversion of collected data into standardized scale scores. The risk level intelligent determination module, based on MSSMHS norm thresholds, completes the standardized determination of psychological risk levels. The platform service layer is used to realize functions such as class management, abnormal warning, targeted intervention matching, home-school collaboration and model iteration based on the MSSMHS scale assessment results; Data collection terminal layer: Deployed in campus scenarios, it includes visual, haptic, and audio acquisition devices and edge computing units, used to collect data such as facial micro-expressions and body movements of teenagers in interactive tasks, to achieve contactless, non-intrusive, non-wearable, non-question-answering, non-disclosure of assessment, and single session duration ≤5 minutes of multimodal behavioral data collection of teenagers; MSSMHS Level 3 Mapping Module: Establishes a mapping between "core dimensions - clinical behavioral factors - collected indicators", with the specific mapping relationship shown in the table below: Mapping rules: Each collected indicator strictly corresponds to the factor content of the MSSMHS scale to ensure the clinical validity of the test results. The norm standard values ​​are dynamically adjusted according to the three educational stages of primary school, junior high school, and senior high school to match the age-specific norm characteristics of the MSSMHS scale. All collected indicators are repeatable and stable quantitative indicators with no subjective judgment items. Moreover, each scoring result can be traced back to the corresponding collected indicator and scale factor, achieving full-link traceability. Multimodal weighted calibration algorithm module: Based entirely on the MSSMHS scale's weighting system and age-specific norm design, it achieves accurate conversion of collected data into standardized scale scores. The core algorithm logic is as follows: Based on the national norms of the MSSMHS scale, a specific calculation formula was designed for each secondary clinical behavior factor: in: : No. The first-level dimension The base scores of each secondary factor; : Measured values ​​of factors acquired through multimodal sampling; Age-specific norm standard values ​​for the corresponding factors of the MSSMHS scale; Factor weights: The secondary factors are evenly distributed and correspond to the fixed weights of the primary dimensions (e.g., the emotion regulation dimension has a weight of 30%, and the three secondary factors each account for 10%). Factor calibration coefficient, ranging from 0.95 to 1.05, was obtained through training with ≥10,000 clinical samples of MSSMHS and was used to correct for systematic bias. The score conversion factor is fixed at 100 to ensure that the final score is uniformly converted to 0-100 points, which is fully aligned with the MSSMHS scale scoring system. By integrating objectively collected data with teacher observations and feedback, and avoiding bias from single data sources, the integration formula is as follows: in: Final score of secondary factors; The weight of objective data is fixed at 0.7. Teacher feedback weight is fixed at 0.3; That is, the above fusion formula can be converted into: in, Teachers observe and provide feedback, scoring on a four-level scale: "Excellent (100 points), Good (80 points), Average (60 points), Needs Improvement (40 points)". If no feedback is provided, the score is [not specified]. ; Final score calculation for the first dimension: The dimensional synthesis logic is completely consistent with that of the MSSMHS scale; the final score of the first-level dimension is the arithmetic mean of the final scores of its corresponding second-level factors. in, For the first The final score for each primary dimension, The number of secondary factors under this dimension is used to output the standardized scores of 0-100 for the four dimensions. To comply with psychometric standards and ensure the reliability of the scores, a three-level data verification mechanism is used to automatically remove absolute outliers, and interpolation of valid data from the same scenario is used to complete the data with an interpolation error of ≤5%. When the difference between data from multiple scenarios for the same factor is >30%, the average value is taken as the measured value. Furthermore, when data is missing, the mean of the norm of the same grade and gender group is used to supplement the data.

[0019] Intelligent Risk Level Determination Module: Based on the national norm thresholds of the MSSMHS scale and combined with clinical diagnostic criteria, a four-level risk level determination system is constructed, with the following rules: Targeted intervention matching rules: Based on the MSSMHS scale assessment results, for the dimensions and factors that have not met the standards, the corresponding intervention scenarios are precisely matched to achieve targeted intervention that "fills in the gaps"; for example, the weakness in emotion regulation ability is matched with "emotion management theme interaction scenario", and the weakness in social adaptation ability is matched with "social collaboration theme interaction scenario". Model Iteration Mechanism: Establish an incremental learning iteration framework to reverse-accumulate retest data, clinical diagnosis results, and intervention effect data, optimize the algorithm calibration coefficient monthly, and update the indicator thresholds and weight allocation of the three-level mapping system quarterly to ensure that the system's detection accuracy continues to improve as the sample size increases; Based on local privacy protection requirements, the original image and audio data are destroyed immediately after being processed locally on the terminal. No original audio or video data that can identify an individual is uploaded. Only the feature values ​​and scoring results are encrypted and stored. At the same time, blockchain evidence storage is required. By generating a unique blockchain traceability ID for each assessment report, the scoring results, mapping parameters, algorithm version, collection time and other tamper-proof information are stored on the blockchain, supporting auditing and traceability by education departments and parents. Access control is also required. By adopting the RBAC hierarchical access control model, teachers can only view data for their own class, and parents can only access their own child's reports. All operations are audited and recorded.

[0020] Example 1: Student information: 10 years old, male, 5th grade of primary school, no history of psychological problems; Data collection: Students completed a 5-minute interactive assessment through a data collection terminal. The system, based on a three-level mapping system, collected data showing a focus duration of 110 seconds and a target achievement rate of 65%. Algorithm calculation: Substituting the MSSMHS exclusive algorithm formula, the device-collected score for the focus and executive function dimension is 68 points, the teacher feedback score is 70 points, and the final score for multi-source fusion is 68.6 points; Risk Assessment: The system automatically classified it as "Attention" level and prompted that "the duration of attention is slightly lower than the MSSMHS upper elementary school norm value"; Intervention and retest: Match the "Focus Improvement Theme Intervention Scenario", it is recommended to participate twice a week, and retest after 1 month. The student's focus duration increased to 135 seconds, the dimension score increased to 78 points, and the student returned to the "normal" level. Example 2: Student information: 13 years old, female, second year of junior high school, homeroom teacher feedback: introverted personality, unwilling to participate in group activities; Data collection: The terminal collected 1 instance of social initiation, 70% cooperation rate, and met the boundary awareness and adaptability standards; Algorithm calculation: The final score for the social adaptability dimension was 62 points, and the score for the emotion regulation dimension was 68 points; Risk Assessment: The system automatically assesses the risk as "early warning," triggering alerts from the homeroom teacher and school counselor. Intervention implementation: Matching the "social collaboration theme intervention scenario", the psychological teacher developed a personalized intervention plan. After 2 weeks, the retest showed that the number of times the student initiated social interaction increased to 3 times, and the score of the social adaptation dimension increased to 78 points, returning to the "normal" level. Example 3: The terminal layer integrates an ambient light sensor and a high-precision microphone array to collect the current ambient light intensity in real time before and during the evaluation. (Unit: lux) Ambient noise decibels (Unit: dB) and crowd density obtained through visual recognition ; The system calls the environmental interference assessment submodule to calculate the comprehensive environmental interference index according to a preset formula. The formula is: In the formula, Used to measure the degree to which illumination deviates from the optimal value; Used to measure the increase in noise level beyond the background noise level; Directly reflects the degree of space crowding; The preset optimal light threshold; The preset ambient background noise threshold; The environmental factor weighting coefficients are set to 0.3, 0.5, and 0.2 respectively, indicating that noise has the highest interference weight on psychological assessments; based on the aforementioned comprehensive environmental interference index... Weights of secondary factors Dynamically adjust and generate environment adaptation weights. : when At the (slight interference threshold), the system maintains the original weights. If the environmental factors remain unchanged, their impact on the evaluation is relatively small, and the data collection is relatively stable. when When the (severe interference threshold) is reached, the system triggers the weight correction function. In this situation, the noise and crowding levels are high, which may have a significant impact on the evaluation data. Specific correction logic: Since noise and crowded environments can objectively cause deviations in "focus" test data (such as eye focus), the system will automatically adjust the weighting of the "focus and execution" dimension. Multiply by an attenuation factor To eliminate false "decrease in concentration" caused by environmental noise; at the same time, in order to capture the real stimuli of the environment to the senses, the system will increase the weight of the "perceptual adaptation" dimension accordingly. Adjusted environment adaptation weights Substitute into the basic scoring formula The final calculation is performed; by introducing environmental factor correction, this system can still maintain extremely high evaluation accuracy in non-standard laboratory environments (such as ordinary classrooms and activity centers), solving the problem of large score fluctuations caused by environmental variable interference in existing technologies.

[0021] In summary, this multimodal non-intrusive screening system and its implementation method for adolescents effectively overcome the shortcomings of traditional subjective assessments, which are easily subject to concealment and emotional interference, by establishing a standardized mapping between scale dimensions and objective behaviors and a dedicated algorithm. This significantly improves the objectivity and clinical relevance of the test results. At the same time, relying on non-intrusive data collection and local encryption processing mechanisms, it eliminates the psychological defenses of test subjects and increases their willingness to cooperate, while fundamentally ensuring the security of personal privacy data and achieving accurate identification and compliant management of psychological states. By introducing an environmental perception and dynamic weight adjustment mechanism, the system's adaptability to the external physical environment is enhanced. It can automatically balance the scoring weights of different psychological dimensions according to the real-time environmental interference level, thereby effectively filtering out abnormal data interference caused by external factors such as light and noise. This ensures that the evaluation results can still truly reflect the individual's psychological traits even in non-ideal environments, greatly improving the system's stability and applicability.

[0022] The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and to design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A multimodal, non-intrusive screening system for adolescents, characterized in that, It includes the data acquisition terminal layer, the core algorithm layer, and the platform service layer; The data collection terminal layer is used to collect multimodal behavioral data of adolescents in a non-contact and non-invasive manner based on the collection requirements of the MSSMHS scale. The core algorithm layer includes a MSSMHS three-level mapping module, a multimodal weighted calibration algorithm module, and a risk level intelligent determination module; The MSSMHS three-level mapping module establishes a three-level standardized mapping relationship of "MSSMHS core dimensions - clinical behavioral factors - multimodal collection indicators" to achieve the matching of abstract dimensions of the scale to objective collection indicators; the multimodal weighted calibration algorithm module, based on the MSSMHS scale weight system and age-specific norms, achieves accurate conversion of collected data into standardized scale scores; and the risk level intelligent judgment module, based on MSSMHS norm thresholds, completes the standardized judgment of psychological risk level. The platform service layer is used to implement functions such as class management, abnormal warning, targeted intervention matching, home-school collaboration, and model iteration based on the MSSMHS scale assessment results.

2. The multimodal non-intrusive screening system for adolescents according to claim 1, characterized in that, In the MSSMHS three-level mapping module, the four core dimensions of the MSSMHS scale include emotion regulation ability, social adaptation ability, focus and executive ability, and sensory adaptation; the weight percentages are 30%, 30%, 25%, and 15% respectively. Each core dimension is broken down into three secondary clinical behavioral factors, and the secondary factors are equally weighted according to the corresponding dimension.

3. The multimodal non-intrusive screening system for adolescents according to claim 1, characterized in that, The formula for the second-level factor base score of the multimodal weighted calibration algorithm module is as follows: in These are measured values. The standard values ​​for age-specific norms of MSSMHS are as follows. For factor weights, For calibration coefficients, This is the conversion factor for fractions.

4. The multimodal non-intrusive screening system for adolescents according to claim 3, characterized in that, The multimodal weighted calibration algorithm module employs a multi-source data fusion formula: in Teachers will be given feedback scores based on their observations.

5. The multimodal non-intrusive screening system for adolescents according to claim 1, characterized in that, The intelligent risk level determination module is set with four risk levels. The high warning determination condition is: scores of two or more primary dimensions < 60 points, or core factors are lower than 50% of the norm.

6. The multimodal non-intrusive screening system for adolescents according to claim 1, characterized in that, The core algorithm layer also includes a local encryption and blockchain evidence storage module, which is used to destroy the original audio and video data after local processing, encrypt and store only the feature values ​​and scoring results, and generate a unique traceability ID through the blockchain to achieve full-chain traceability.

7. A method of implementation, wherein the system employs the multimodal non-intrusive screening system for adolescents as described in any one of claims 1-6, characterized in that, Includes the following steps: S1: Construct a three-level precise mapping system for the MSSMHS scale, establish a standardized transformation path of "MSSMHS core dimensions - clinical behavior factors - multimodal collection indicators", and clarify the age-specific norm standard values ​​and collection rules for each factor; S2: Using a wearable multimodal acquisition terminal, multimodal data corresponding to 12 clinical behavioral factors of adolescents are collected seamlessly based on a three-level mapping system. S3: The collected data is processed using the MSSMHS proprietary multimodal weighted calibration algorithm to calculate the standardized dimension scores of 0-100 points, which are aligned with the national standard scale. This step also includes a three-level data validation mechanism: outlier removal, consistency verification, and missing value supplementation; to ensure the reliability of the scoring results. S4: Based on the national norm threshold of the MSSMHS scale, automatically determine the psychological risk level of adolescents and match the corresponding targeted intervention plan. S5: Generates encrypted assessment reports, completes end-to-end tamper-proof notarization through blockchain, and pushes them to teachers' and parents' ends; S6: Based on clinical feedback and retest data, the three-level mapping system and algorithm model are optimized through incremental learning to form a closed loop across the entire chain; The model iteration cycle involves monthly optimization of calibration coefficients and quarterly updates to the mapping system and weight allocation to ensure continuous improvement in detection accuracy.

8. The implementation method according to claim 7, characterized in that, In step S1, the three-level mapping system dynamically adjusts the norm standard values ​​according to the three educational stages of primary school, junior high school, and senior high school to adapt to the age-specific norm characteristics of the MSSMHS scale.

9. The implementation method according to claim 7, characterized in that, In step S3, the multimodal weighted calibration algorithm module also includes a dynamic correction mechanism for environmental factors, the specific steps of which are as follows: Real-time acquisition of light intensity values ​​in the current testing environment Environmental decibel levels and population density value Construct a comprehensive environmental disturbance index The calculation formula is: in, The preset optimal light threshold; The preset ambient background noise threshold; These are the environmental factor weighting coefficients; based on the aforementioned comprehensive environmental disturbance index. Weights of secondary factors Dynamically adjust and generate environment adaptation weights. : in, As an environmental sensitivity modulator, For nonlinear correction functions of environmental disturbances; when When the preset threshold is exceeded, the system automatically increases the weight of the "perceptual adaptation" dimension and reduces the original score weight of the "focus" dimension, which is affected by environmental interference.

Images