Continuous concentration evaluation system and method based on multi-dimensional data

Abstract

The invention provides a continuous concentration assessment system and method based on multi-dimensional data, and belongs to the technical field of attention assessment. The system comprises a multi-dimensional data acquisition module used for acquiring multi-dimensional test data; the curve construction module is used for constructing a baseline and generating an individual curve with the same characteristics as the baseline by using a sliding window technology; the analysis module is used for carrying out abnormity judgment when the individual curve is generated; and the control module is used for extracting an attenuation characteristic value of the individual curve through a symmetric rotation equivalent convolution technology in an early warning observation period, and starting an evaluation termination mechanism when the early warning observation period is ended and the attenuation characteristic value still shows abnormity. According to the invention, through multi-modal data fusion and individual baseline construction, high-precision dynamic identification of the attention attenuation process is realized, and based on evaluation of the starting and ending time and output of the continuous concentration duration as an evaluation index, the accuracy and personalized adaptive capacity of concentration evaluation are significantly improved.

Description

Technical Field

[0001] The present invention relates to the field of attention assessment technology, and in particular to a continuous concentration assessment system and method based on multi-dimensional data. Background Art

[0002] During a child's growth process, attention is a key factor affecting their learning and quality of life. In a school environment, children with focused attention can absorb knowledge efficiently and have significantly higher learning efficiency than children with inattention; children with inattention are more likely to miss key knowledge points, which in turn affects their academic performance. In the medical field, accurately assessing a child's attention status is of great significance for the diagnosis of diseases such as Attention Deficit Hyperactivity Disorder (ADHD) and is an important basis for formulating scientific treatment plans. Educational training institutions also need to optimize teaching content and methods based on the students' attention status to improve teaching effectiveness.

[0003] However, existing technologies have many shortcomings. The commonly used concentration test methods internationally only assess attention based on the speed and accuracy of ordinary answering questions, while children's attention is affected by many factors such as emotions and environmental interference. These factors cannot be reflected in simple answering data, resulting in the evaluation results being difficult to fully and truly reflect the children's attention level. Some studies use the average EEG value within a fixed time to judge concentration, but EEG signals are extremely weak and require high-multiple amplification during collection and processing, which will amplify environmental noise and the device's own interference signals, resulting in larger errors. In addition, existing testing methods generally have the problem of fixed test duration. Young children have limited endurance and attention span. Faced with tests of 15-30 minutes or even longer, they are easily distracted by emotions such as fatigue and boredom, resulting in the inability to complete the test or inaccurate results, which cannot truly reflect their attention level. Summary of the Invention

[0004] The present invention provides a continuous concentration evaluation system and method based on multi-dimensional data, which are used to solve the problems of low test efficiency and poor accuracy in the prior art.

[0005] To achieve the above-mentioned objectives, an embodiment of the present invention provides a continuous concentration assessment system based on multidimensional data, the continuous concentration assessment system comprising: a multidimensional data acquisition module for collecting multidimensional test data according to preset time intervals and sequences, the multidimensional test data comprising question-answering response data, EEG activity data, and eye movement data; a curve construction module for constructing a baseline comprising prefrontal EEG signal activity, the proportion of time spent looking at the screen, and parameters related to question-answering response based on a portion of the collected multidimensional test data, and processing the collected multidimensional test data using a sliding window technique to generate an individual curve having the same characteristics as the baseline; an analysis module for performing an abnormality warning judgment using a preset abnormality judgment rule when generating the individual curve; and a control module for extracting an attenuation characteristic value of the individual curve using a symmetric rotation equivariant convolution technique during a warning observation period. When the warning observation period ends and the attenuation characteristic value still shows an abnormality, an assessment termination mechanism is initiated, and the duration from the start to the end of the test is used as an evaluation indicator for the continuous concentration duration.

[0006] Optionally, the construction includes a baseline of the activity of the prefrontal EEG signal, the proportion of time the eyes look at the screen, and the parameters of the response to questions, including: after normalizing and correcting some of the collected multidimensional test data, extracting the features of the activity of the prefrontal EEG signal, the proportion of time the eyes look at the screen, and the parameters related to the response to questions to obtain multimodal features; and using a weighted Gaussian mixture model to fuse the multimodal features to obtain the baseline.

[0007] Optionally, the sliding window technology is used to process the collected multidimensional test data and generate an individual curve with the same characteristics as the baseline, including: initializing the sliding window and its parameter configuration according to a preset step size; extracting the characteristics of the prefrontal EEG signal activity, the proportion of time the eyes look at the screen, and parameters related to the answering response within the current sliding window to obtain multimodal features; fusing the multimodal features, and performing time axis alignment processing and missing value processing to generate an individual curve of the current sliding window; and executing the window closing process.

[0008] Optionally, when generating the individual curve, a preset abnormality judgment rule is used to perform abnormality judgment, including: if the parameter characteristics related to the answering reaction show that the dynamic average accuracy rate within the first preset time is less than the corresponding characteristics in the baseline, and the variance of the answering reaction time is greater than multiple times the dynamic variance of the dimension data within the first preset time, and the duration exceeds the preset abnormal time, then the indicator of the question answering reaction state dimension is abnormal; if the characteristics of the proportion of time the eyes look at the screen show that the visual concentration is lower than the preset value, then the indicator of the eye movement state dimension is abnormal; if the characteristics of the activity of the prefrontal EEG signal show that it is lower than the corresponding characteristics in the baseline by a preset degree, then the indicator of the EEG activity state dimension is abnormal.

[0009] Optionally, when generating the individual curve, a preset abnormality judgment rule is used to perform abnormality judgment, including: comparing and analyzing the data of each dimension in the individual curve with the characteristics of the baseline, and combining the abnormality judgment rule to judge the abnormality of the data of each dimension; when indicators of more than two dimensions are judged to be abnormal, it is determined that the attention state is attenuated and marked as an observation period.

[0010] Optionally, the attenuation characteristic value of the individual curve is extracted by the symmetric rotation equivariant convolution technology, including: constructing a model for extracting the attenuation characteristic value of the individual curve, wherein a symmetric rotation equivariant convolution layer is configured to extract the basic features of the rotation equivariance of the individual curve, a multi-scale feature pyramid module is configured to capture the attenuation pattern of different time scales, and an attenuation feature regression module is configured to output the attenuation characteristic value; based on the training data set, and using a three-stage training method, the model is trained to extract the attenuation characteristic value of the individual curve using the trained model.

[0011] Optionally, the control module is further configured to: continue the evaluation process when, during the early warning observation period, the attenuation characteristic value returns to a normal fluctuation range of the previous attenuation characteristic value.

[0012] Optionally, the continuous concentration evaluation system further includes a result presentation module, which is used to: generate a visual chart based on the concentration evaluation result to display the attenuation change trend and the final determined continuous concentration duration value; and generate the concentration evaluation result.

[0013] An embodiment of the present invention further provides a method for evaluating continuous concentration, the method comprising: collecting multidimensional test data according to a preset time interval and sequence, the multidimensional test data comprising answer response data, EEG activity data, and eye movement data;

[0014] Based on part of the collected multidimensional test data, a baseline is constructed, including the activity of prefrontal EEG signals, the proportion of time spent looking at the screen, and parameters related to answering responses. The collected multidimensional test data is processed using sliding window technology, and an individual curve with the same characteristics as the baseline is generated. When generating the individual curve, an abnormal warning judgment is performed using preset abnormality judgment rules. After the abnormal warning is triggered, a warning observation period begins, and the attenuation characteristic value of the individual curve is extracted through the symmetric rotation and equivariant convolution technology. When the warning observation period ends and the attenuation characteristic value still shows an abnormality, the evaluation termination mechanism is activated, and the time from the start to the end of the test is used as an evaluation indicator for the continuous concentration time.

[0015] An embodiment of the present invention further provides a machine-readable storage medium having stored thereon instructions, which enable a machine to execute the above-mentioned continuous concentration evaluation method.

[0016] The present invention provides a continuous concentration evaluation system based on multidimensional data, including: a multi-source data acquisition module, which is used to collect multidimensional test data according to preset time intervals and sequences, and the multidimensional test data includes question-answering response data, EEG activity data and eye movement data; a curve construction module, which is used to construct a baseline including the activity of prefrontal EEG signals, the proportion of time spent looking at the screen and parameters related to question-answering response based on part of the collected multidimensional test data, and use sliding window technology to process the collected multidimensional test data and generate individual curves with the same characteristics as the baseline; an analysis module, which is used to perform abnormality judgment using preset abnormality judgment rules when generating the individual curve; a control module, which is used to extract the attenuation characteristic value of the individual curve through symmetric rotation and equivariant convolution technology during the early warning observation period, and when the early warning observation period ends and the attenuation characteristic value still shows an abnormality, the evaluation termination mechanism is activated, and the duration from the start to the end of the evaluation is used as an evaluation indicator for the continuous concentration duration. The embodiments of the present invention achieve high-precision dynamic extraction of concentration decay characteristics through multimodal data fusion (EEG signals, eye tracking and answer response data) and sliding window technology; based on the dynamically constructed individual baseline and multi-dimensional abnormality judgment rules, it can monitor attention fluctuations in real time and trigger an early warning mechanism, significantly improving the accuracy, anti-interference and personalized adaptability of concentration assessment, and can be widely used in the quantitative analysis of attention status in education, medical care and human-computer interaction scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the present invention or the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0018] Figure 1 2 is a schematic diagram of the structure of a continuous concentration evaluation system provided by an embodiment of the present invention;

[0019] Figure 2 2 is a schematic structural diagram of a continuous concentration evaluation system provided by a preferred embodiment of the present invention;

[0020] Figure 3 is a workflow diagram of an example continuous concentration assessment system;

[0021] Figure 4 This is a flowchart of the example baseline construction process;

[0022] Figure 5 is a schematic diagram of an example continuous concentration assessment;

[0023] Figure 6 This is a flowchart of the example symmetric rotation equivariant convolution technique. DETAILED DESCRIPTION

[0024] The following describes the specific implementation of the embodiment of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiment of the present invention and is not used to limit the embodiment of the present invention.

[0025] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, certain software, components, models, and other existing solutions in the industry may be mentioned. These should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use such solutions.

[0026] During a child's development, attention, as a core cognitive ability, directly impacts learning efficiency, knowledge absorption, and overall behavior. In education, children with focused attention are more likely to achieve their learning goals, while those with inattention face learning difficulties and academic lags. Therefore, developing a more accurate and efficient attention assessment system is crucial.

[0027] In response to the above problems, the present invention provides a continuous concentration assessment system and method based on multi-dimensional data. By integrating multi-dimensional data of EEG, eye movements and behavioral responses, combined with deep learning to construct dynamic attenuation baselines and individual baselines, and using symmetric rotation and equivariant convolution to accurately capture attention attenuation characteristics, real-time monitoring and early warning across scenarios are achieved. Its adaptive test duration adjustment mechanism and anti-interference data processing technology significantly improve the assessment accuracy, which can meet the precise needs of educational diagnosis, medical rehabilitation and other fields for quantitative analysis of individualized concentration.

[0028] Please refer to Figure 1-Figure 3 The present invention provides a continuous concentration assessment system based on multidimensional data. The continuous concentration assessment system includes: a multidimensional data acquisition module for collecting multidimensional test data according to preset time intervals and sequences, the multidimensional test data including question-answering response data, EEG activity data, and eye movement data; a curve construction module for constructing a baseline including prefrontal EEG signal activity, the proportion of time spent looking at the screen, and parameters related to question-answering response based on a portion of the collected multidimensional test data, and processing the collected multidimensional test data using a sliding window technique to generate an individual curve with the same characteristics as the baseline; an analysis module for performing an abnormality warning judgment using a preset abnormality judgment rule when generating the individual curve; and a control module for entering an early warning observation period after the abnormality warning is triggered, extracting the attenuation characteristic value of the individual curve using the symmetric rotation equivariant convolution technique, and activating an assessment termination mechanism when the early warning observation period ends and the attenuation characteristic value still shows an abnormality, and using the duration from the start to the end of the test as an evaluation indicator for the continuous concentration duration.

[0029] The parameters related to the response to the questions may include the accuracy of the questions and / or the variability of the response time to the questions.

[0030] The embodiment of the present invention realizes high-precision dynamic extraction of concentration decay characteristics through multimodal data fusion (electroencephalogram signals, eye tracking and answer response data) and sliding window technology; based on the dynamically constructed individual baseline and multi-dimensional abnormality judgment rules, it can monitor attention fluctuations in real time and trigger an early warning mechanism, significantly improving the accuracy, anti-interference and personalized adaptability of concentration assessment, and can be widely used in quantitative analysis of attention status in education, medical care and human-computer interaction scenarios. In the embodiment of the invention, the (individual) baseline and individual curve (including the baseline) are constructed based on the individual's multi-dimensional test data, and the decay characteristic value is performed for the individual, making the monitoring of the continuous concentration assessment system more objective.

[0031] Please refer to Figure 2For example, after collecting multidimensional test data, embodiments of the present invention preferably utilize an improved deep singular value decomposition (DSVD) denoising algorithm to effectively improve the quality of multidimensional test data. For example, when collecting multidimensional information such as children's EEG signals and eye movement data, noise such as power frequency interference, electromyographic artifacts, and ambient light fluctuations can easily be mixed in, making it difficult for traditional methods to accurately separate signal from noise. The DSVD denoising algorithm, however, constructs a deep neural network to automatically learn noise distribution characteristics and utilizes a learnable nonlinear transformation layer to adaptively separate the signal subspace from the noise subspace. This algorithm can fully preserve subtle features in EEG signals, such as changes in the power of the prefrontal alpha, beta, theta, and gamma waves that reflect attentional states, while efficiently removing various types of noise interference, significantly improving the signal-to-noise ratio of the EEG signals. It also effectively purifies interference information in the eye movement data, resulting in a purer multidimensional data. This lays a solid foundation for the subsequent precise extraction of attenuation feature values ​​using symmetric rotation and equivariant convolution techniques, and for accurate anomaly detection and concentration assessment based on the attenuation curve baseline, effectively ensuring the reliability and accuracy of the results of the continuous concentration assessment system.

[0032] Furthermore, image data in the multidimensional test data are randomly selected and downsampled; the downsampled image data are spliced ​​in a grid form to generate a hybrid image so that the image retains multi-scale information; the labels of all categories in the generated hybrid image are merged to generate a hybrid image label in a multi-hot encoding form to improve label diversity.

[0033] The system further optimizes the image data in the multidimensional test data using data augmentation techniques. By randomly selecting and downsampling images, the downsampled images are then spliced ​​together in a grid to create a hybrid image. Labels are then merged to form multi-hot encoded hybrid image labels. This series of operations effectively creates a rich and diverse "learning material" for the model. On the one hand, the hybrid image preserves multi-scale information, enabling the model to learn image features at different levels of detail and enhance its understanding of complex image structures. On the other hand, the multi-hot encoded labels enrich the label diversity, better meeting the requirements of multi-label image classification training. This allows the model to access more comprehensive sample information during training, thereby improving its generalization and robustness. These data processing methods work closely together, from unified integration of the temporal dimension of the data, to data quality purification and enhancement, to enrichment and enhancement of data features, comprehensively optimizing the data processing process. This effectively reduces data errors and interference, improves data availability, and provides solid and reliable data support for the subsequent system's use of symmetric rotation and equivariant convolution techniques to extract data features and accurately assess continuous focus, significantly improving the accuracy and stability of the entire evaluation system.

[0034] In a preferred embodiment of the present invention, the construction of a baseline including the activity of prefrontal EEG signals, the proportion of time spent looking at the screen, and the variability of reaction time to answer questions may include: normalizing and correcting part of the collected multidimensional test data, extracting the characteristics of the activity of prefrontal EEG signals, the proportion of time spent looking at the screen, and the variability of reaction time to answer questions to obtain multimodal features; and using a weighted Gaussian mixture model to fuse the multimodal features to obtain the baseline.

[0035] Among them, part of the multidimensional test data can be the multidimensional test data collected at the beginning of the preset time period of the tester. After extracting the characteristics of the activity of the prefrontal EEG signal, the proportion of time the eyes look at the screen, and the variability of the reaction time to answer questions, each feature can also be expanded. For example, the alpha wave envelope slope (i.e., the time-varying characteristics of the prefrontal EEG signal) can be added to the characteristics of the activity of the prefrontal EEG signal, the gaze hotspot transfer entropy (i.e., the eye movement characteristics) can be added to the characteristics of the proportion of time the eyes look at the screen, and the reaction acceleration ratio (the rate of change of reaction time in the later trials, i.e., the reaction pattern) can be added to the characteristics of the variability of the reaction time to answer questions.

[0036] In an embodiment of the present invention, a weighted Gaussian mixture model is used to fuse the multimodal features. The weighted Gaussian mixture model can be constructed using the following formula:

[0037]

[0038] Among them, X is part of the multidimensional test data after normalization and correction, K is the preset number of clusters, and w k can be the weight set according to expert experience, u k is the mean vector of the k-th feature, Σ k is the diagonal covariance matrix.

[0039] When using a weighted Gaussian mixture model to fuse the multimodal features and obtain a baseline, drift detection can also be performed. When significant drift is detected, a learning engine can be added to the weighted Gaussian mixture model to perform a credibility assessment. When the credibility exceeds a threshold, the baseline is corrected. For example, the test subject can be prompted to close their eyes for three minutes before preparing for the test.

[0040] Please refer to Figure 4 For example, an embodiment of the present invention uses WGMM to capture the nonlinear relationship between features to achieve multimodal joint probability modeling; uses a static statistical baseline combined with a dynamic probability baseline to achieve a dual-track baseline; and based on the combination of drift detection and incremental learning, it adaptively updates in real time and provides an accurate baseline.

[0041] In a preferred embodiment of the present invention, the sliding window technology is used to process the collected multidimensional test data and generate an individual curve with the same characteristics as the baseline, including: initializing the sliding window and its parameter configuration according to a preset step size; extracting the characteristics of the activity of the frontal lobe EEG signal, the proportion of time the eyes look at the screen, and the variability of the reaction time to answer questions within the current sliding window to obtain multimodal features; fusing the multimodal features and performing time axis alignment processing and missing value processing to generate an individual curve of the current sliding window; and executing the window closing process.

[0042] In an embodiment of the present invention, the sliding window and its parameter configuration are initialized according to a preset step size (detection interval, for example, data is collected once every 1s after the test starts). For example, parameter configurations such as the time-driven window, window length, sliding step size, minimum proportion of valid data, and time-triggered window are defined. Within the current sliding window, the above-mentioned weighted Gaussian mixture model can be reused to extract the characteristics of the activity of the frontal lobe EEG signal, the proportion of time the eyes look at the screen, and the variability of the response time to answer questions to obtain multimodal features, and after time axis alignment processing and missing value processing, the individual curve of the current sliding window is generated. In a preferred embodiment of the present invention, a multi-level window architecture can be utilized, which can include a time-driven window (preset step size) and an event-driven window (events such as task start, task end, and curve presentation). Furthermore, the window closing process can be executed when the end time of the window is reached. For example, the window mark can be invalidated and the buffer memory can be released for resource recovery.

[0043] In a preferred embodiment of the present invention, window parameters can also be dynamically adjusted based on data quality. For example, when the EEG activity data quality is poor, the window time can be extended; when the eye movement loss rate is high, an event-driven window can be switched; when a high-frequency response period occurs, the window time can be shortened, etc.

[0044] The embodiment of the present invention achieves the generation of high-quality individual curves through a carefully designed sliding window mechanism and multimodal feature calculation, providing reliable data support for continuous concentration assessment.

[0045] In a preferred embodiment of the present invention, when generating the individual curve, a preset abnormality judgment rule is used to perform abnormality judgment, including: if the parameter characteristics related to the answering reaction show that the dynamic average accuracy within the first preset time is less than the corresponding characteristics in the baseline, and the variance of the answering reaction time is greater than multiple times the dynamic variance of the dimension data within the first preset time, and the duration exceeds the preset abnormal time, then the indicator of the answering reaction state dimension is abnormal; if the characteristics of the proportion of time the eyes look at the screen show that the visual concentration is lower than the preset value, then the indicator of the eye movement state dimension is abnormal; if the characteristics of the activity of the prefrontal EEG signal show that it is lower than the corresponding characteristics in the baseline by a preset degree, then the indicator of the EEG activity state dimension is abnormal.

[0046] In a preferred embodiment of the present invention, for each dimension in the test data, the extracted feature values ​​are compared and analyzed in real time with the data of the corresponding time and dimension on the pre-built individual baseline. Through careful comparison, subtle changes in the data can be keenly captured. Combined with the preset abnormality judgment rules, this module can accurately judge the abnormality of the data in each dimension. When the data of more than one dimension is judged to be abnormal and there is no reverse trend in the remaining dimensions, the system will determine that the attention state has been significantly reduced. This judgment method that integrates multi-dimensional data and considers the trend relationship between data avoids misjudgment caused by single-dimensional abnormalities, is more in line with the complex actual situation of children's attention changes, and significantly improves the accuracy and reliability of judgment. Once the attention state is judged to be significantly reduced, or the test time reaches the test time corresponding to the subject, the system will immediately stop the multi-source data acquisition module for data collection, promptly terminate the invalid or achieved data collection process, and improve evaluation efficiency. Based on the abnormality of the data in each dimension and the judgment results, combined with the individual baseline, a comprehensive and integrated assessment of the child's continuous concentration is performed, and the continuous concentration duration is determined. This evaluation method is no longer limited to simple data comparison, but fully considers the changing trends, abnormal situations and baseline standards of the data. It can more realistically and accurately reflect the children's concentration state during the test, and provide extremely valuable reference for education, medical care and other fields related to children's attention, which will help to formulate more scientific and reasonable intervention measures and teaching plans.

[0047] In a preferred embodiment of the present invention, when generating the individual curve, a preset abnormality judgment rule is used to perform abnormality judgment, including: comparing and analyzing the data of each dimension in the individual curve with the characteristics of the baseline, and combining the abnormality judgment rule to judge the abnormality of the data of each dimension; when the indicators of more than two dimensions are judged to be abnormal, it is determined that the attention state is attenuated and marked as an observation period.

[0048] In a preferred embodiment of the present invention, a scientific and flexible early warning mechanism is added to the continuous concentration assessment system, which significantly improves the safety and effectiveness of the assessment process. When the system detects that two dimensional data are judged to be abnormal, it immediately starts the high-frequency acquisition and analysis mode to conduct real-time and intensive tracking and monitoring of the abnormal dimensional data. During the observation period, if the abnormal dimensional data is restored to a state consistent with the subject's own previous normal fluctuation range, it indicates that the subject may have only a short-term fluctuation in attention or is subject to instantaneous interference. The system will automatically restore the normal test process to avoid unnecessary test interruptions and ensure the continuity of the assessment process. Once the preset time period is exceeded and the abnormal data still fails to return to normal, it means that there is a persistent problem with the subject's attention state. The system will immediately trigger the stop test instruction and terminate the test process in time, which can not only avoid the generation of invalid data, but also prevent the subject from being affected by excessive fatigue or poor condition. The accuracy of the test results is affected, providing a dynamic and reliable risk management guarantee for the entire assessment process.

[0049] In an embodiment of the present invention, accurate and targeted abnormality judgment rules are set for each dimension of data: for answering response data, by comparing the dynamic average accuracy rate within the first preset time with the corresponding value in the individual baseline, and combining the relationship between the total average and the dynamic average and standard deviation, while considering the duration of the abnormal state, it can effectively eliminate accidental fluctuations and accurately identify abnormal answering responses caused by decreased attention; eye movement data uses the preset value of visual concentration as the measurement standard. Once it is lower than this value, it is determined that the child has a distracted state; EEG activity data is compared with the individual baseline by the difference ratio. When the EEG activity data is significantly lower than the EEG activity data recorded in the individual baseline, it can be perceived as abnormal brain neural activity and judged as abnormal attention state. These rules start from multiple dimensions such as cognitive behavior, visual attention and brain neural activity, and comprehensively cover data related to children's attention, providing a reliable basis for accurately judging children's attention abnormalities, and significantly improving the accuracy and credibility of the evaluation results.

[0050] Please refer to Figure 5For example, consider a 7-year-old participant undergoing a continuous attention assessment. After the test begins, multi-dimensional data acquisition modules operate synchronously: During the visual response phase, participants identify and click on changing graphics on a computer screen, with the system recording their response data in real time. An EEG headband collects EEG signals from the Fp1 and Fp2 locations of their prefrontal cortex. A 3D camera also captures their eye movements. Four minutes into the test, the system detects the following abnormal signals: The dynamic accuracy rate for the response data over the past minute (corresponding to the first preset time period) is 76%, significantly lower than the 90% average for the corresponding time period in the individual's baseline. Furthermore, the response variance during this period is more than 2.5 times the baseline variance, and this state persists for 31 seconds, exceeding the set trigger threshold (20 seconds), meeting the criteria for abnormal response data. Eye movement data analysis reveals that the participant spent 50% of their time gazing at the center of the screen in the last minute, below the preset lower limit and thus identified as abnormal eye movement. EEG activity data indicates that the participant's signal strength in the target band is only 8.5 μV. 2 / Hz, which is significantly lower than the corresponding value in its individual baseline, is also considered abnormal. Since two or more data dimensions are judged to be abnormal in a row, the system automatically enters the early warning observation period. During this stage, the system continuously monitors whether various indicators show a recovery trend. If at the end of the early warning observation period, the comprehensive attenuation value still does not meet the recovery judgment conditions, the system automatically terminates the assessment. In the end, the system assessed that the participant's effective continuous concentration time was approximately 4 minutes and 30 seconds, and generated visual charts and individualized assessment reports based on multi-dimensional data to provide a quantitative basis for subsequent education or medical intervention.

[0051] In a preferred embodiment of the present invention, the continuous concentration evaluation system also includes a result presentation module, which is used to: generate a visual chart based on the concentration evaluation result to show the attenuation change trend and the final determined continuous concentration duration value; generate and display the concentration evaluation result.

[0052] Among them, the charts intuitively show the attenuation trend and changes in focus time. The report presents the evaluation results in detail and gives targeted corrective suggestions, making the evaluation results easier to understand and apply, and improving the practical value of the system.

[0053] In the embodiment of the present invention, the baseline construction module is further used to compare the individual baseline with the attenuation curve baseline. When the difference between the data recorded on the individual baseline and the data recorded on the attenuation curve baseline is too large, the test is terminated.

[0054] In a preferred embodiment of the present invention, extracting the attenuation characteristic values ​​of the individual curves using symmetric rotational equivariant convolution technology includes: constructing a model for extracting the attenuation characteristic values ​​of the individual curves. A symmetric rotational equivariant convolution layer is configured to extract basic features of the rotational equivariance of the individual curves; a multi-scale feature pyramid module is configured to capture attenuation patterns at different time scales; and an attenuation feature regression module is configured to output the attenuation characteristic values. The model is trained based on a training dataset and using a three-stage training method, so that the trained model can be used to extract the attenuation characteristic values ​​of the individual curves.

[0055] Symmetric rotation equivariant convolution is a convolutional neural network technology that maintains equivariance under symmetric transformations such as rotation and reflection. For attenuation feature extraction in time series, it has the advantages of being insensitive to curve phase shifts, automatically capturing multi-scale attenuation patterns, and reducing the need for data enhancement. Please refer to Figure 6 For example, a model constructed for extracting the attenuation feature value of the individual curve may include an input layer (the number of channels is, for example, 3); the symmetric rotation equivariant convolution layer may be configured with a rotation number of, for example, 12, and a convolution kernel size of, for example, 15; the multi-scale feature pyramid module may be configured with a scale number of, for example, 3, and an expansion rate of [1, 3, 5]; the attenuation feature regression module may include an attenuation extractor may include branch channels (for example, 32) and an output layer, and the feature dimension may be, for example, 3.

[0056] Among them, the symmetric rotation equivariant convolution layer can be configured by the following formula:

[0057]

[0058] Among them, K is the rotational symmetry order, x represents the input curve, k represents the rotation index, rot() represents the rotation transformation operator, and w k represents the science department convolution kernel, Indicates the rotation angle.

[0059] Furthermore, the multi-objective loss function can be expressed as follows:

[0060] L=λ1L rate +λ2L stability +λ3L shape

[0061] Among them, λ1, λ2, λ3 represent weights, L rate Represents the attenuation rate loss, L stability Indicates the stability loss, L shape represents the morphological similarity loss.

[0062] In a preferred embodiment of the present invention, the model is trained based on a training dataset and using a three-stage training method, so that the attenuation characteristic values ​​of the individual curves can be extracted using the trained model. For example, in the first stage (pre-training stage), the attenuation curve can be synthesized using a physical model; in the second stage (fine-tuning stage), the individual curves can be fine-tuned and gradually unfrozen, and the unfreezing can be performed from the output layer to the input layer; in the third stage (adaptive stage), domain adversarial training is performed based on unlabeled individual curves. Using the three-stage training method, the accuracy of the model can be improved.

[0063] In a preferred embodiment of the present invention, the attenuation characteristic value of each dimension can include multiple indicators. For example, the attenuation characteristic value of the prefrontal EEG signal activity can include α indicating fatigue warning, β indicating attention stability, and γ indicating morphological abnormality.

[0064] The present invention utilizes symmetric rotational equivariant convolution technology to significantly improve the feature extraction capabilities of multidimensional data. By maintaining invariance to data transformations such as rotation and translation, it effectively addresses changes in EEG signal acquisition position caused by slight head movements during testing, stably extracting neural activity features and reducing the impact of environmental interference. By designing rotational equivariant filters of different scales, it achieves the fusion capture of multi-scale features such as microscopic gaze patterns and macroscopic saccade paths in eye movement data, comprehensively reflecting the distribution of visual attention. By constructing a multi-channel network architecture, it automatically learns the inherent correlations between different dimensions, such as response, EEG signals, and eye movement data, enabling collaborative analysis of multidimensional data. Furthermore, it leverages the data's symmetry prior to reduce the number of model parameters, reducing them by over 40% while ensuring assessment accuracy. This significantly improves real-time processing capabilities and meets the needs of dynamic assessment. This technology accurately captures attention decay characteristics, adapts to individual differences, and improves assessment efficiency. It is a key technology for achieving high-precision and robust continuous attention assessment, effectively ensuring the accuracy and reliability of the system's continuous attention assessment of children.

[0065] In a preferred embodiment of the present invention, the analysis module is further configured to: for the data of each dimension in the individual curve, compare and analyze the extracted attenuation characteristic value with the characteristics of the baseline, and judge the abnormality of the data of each dimension in combination with the abnormality judgment rule; when the data of two or more dimensions are judged to be abnormal and there is no reverse trend in the remaining dimensions, the attention state is judged to be attenuated; when the attention state is judged to be attenuated or the test time reaches the test time corresponding to the subject, stop data collection.

[0066] An embodiment of the present invention further provides a method for evaluating continuous concentration, which may include steps S210-S240:

[0067] Step S210: Collect multi-dimensional test data according to a preset time interval and sequence, wherein the multi-dimensional test data includes question response data, EEG activity data, and eye movement data.

[0068] Step S220: Based on the collected part of the multidimensional test data, a baseline is constructed including the activity of the frontal lobe EEG signal, the proportion of time the eyes look at the screen, and parameters related to the answering response, and the collected multidimensional test data is processed using the sliding window technology to generate an individual curve with the same characteristics as the baseline.

[0069] Step S230: When generating the individual curve, use the preset abnormality judgment rules to perform abnormality warning judgment.

[0070] Step S240: After the abnormal warning is triggered, the warning observation period begins. The attenuation characteristic value of the individual curve is extracted through the symmetrical rotation equivariant convolution technology. When the warning observation period ends and the attenuation characteristic value still shows an abnormality, the evaluation termination mechanism is activated, and the time from the start to the end of the test is used as an evaluation indicator for the continuous concentration time.

[0071] An embodiment of the present invention further provides a machine-readable storage medium having stored thereon instructions, which enable a machine to execute the above-mentioned continuous concentration evaluation method.

[0072] Additionally, the terms "system" and "network" are often used interchangeably. The term "and / or" is simply used to describe a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " generally indicates an "or" relationship between the related objects.

[0073] It should be understood that in the embodiments of the present invention, "B corresponding to A" means that B is associated with A and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B based solely on A; B can also be determined based on A and / or other information.

[0074] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0075] Those skilled in the art will clearly understand that, for the sake of convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0076] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or can be electrical, mechanical or other forms of connection.

[0077] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the objectives of the embodiments of the present invention.

[0078] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0079] Through the description of the above embodiments, it will be clear to those skilled in the art that the present invention can be implemented in hardware, firmware, or a combination thereof. When implemented using software, the above functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media includes any medium that facilitates the transmission of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a computer. By way of example and not limitation, computer-readable media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer. In addition, any connection can appropriately become a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. As used herein, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, wherein disks usually reproduce data magnetically, while discs use lasers to reproduce data optically. The above combinations should also be included in the scope of protection of computer-readable media.

[0080] In short, the above description is only a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A continuous concentration evaluation system based on multi-dimensional data, characterized in that: The continuous concentration evaluation system includes: A multi-dimensional data acquisition module is used to collect multi-dimensional test data according to a preset time interval and sequence, wherein the multi-dimensional test data includes answer response data, EEG activity data, and eye movement data; A curve construction module is used to construct a baseline including the activity of the frontal lobe EEG signal, the proportion of time the eyes look at the screen, and parameters related to the response to the test questions based on the collected part of the multidimensional test data, and use the sliding window technology to process the collected multidimensional test data and generate an individual curve with the same characteristics as the baseline; An analysis module, configured to perform abnormal warning judgment using preset abnormal judgment rules when generating the individual curve; The control module is used to enter the warning observation period after the abnormal warning is triggered, and extract the attenuation characteristic value of the individual curve through the symmetric rotation equivariant convolution technology. When the warning observation period ends and the attenuation characteristic value still shows an abnormality, the evaluation termination mechanism is activated, and the time from the start to the end of the test is used as an evaluation indicator of the continuous concentration time.

2. The continuous concentration evaluation system according to claim 1, characterized in that: The construction includes a baseline of prefrontal EEG signal activity, the proportion of time eyes look at the screen, and response parameters, including: After normalizing and correcting some of the collected multidimensional test data, extract the characteristics of the prefrontal EEG signal activity, the proportion of time spent looking at the screen, and parameters related to the response to the questions to obtain multimodal features; and A weighted Gaussian mixture model is used to fuse the multimodal features to obtain the baseline.

3. The continuous concentration evaluation system according to claim 1, characterized in that: The method of processing the collected multidimensional test data using a sliding window technique and generating an individual curve having the same characteristics as the baseline comprises: Initialize the sliding window and its parameter configuration according to the preset step size; Within the current sliding window, extract the features of the prefrontal EEG signal activity, the proportion of time the eyes gaze at the screen, and parameters related to the answering response to obtain multimodal features; After fusing the multimodal features, performing time axis alignment processing and missing value processing, an individual curve of the current sliding window is generated; Execute the window closing process.

4. The continuous concentration evaluation system according to claim 1, wherein: When generating the individual curve, performing abnormality judgment using a preset abnormality judgment rule includes: If the parameter characteristics related to the answering reaction show that the dynamic average accuracy rate within the first preset time is lower than the corresponding characteristics in the baseline, and the variance of the answering reaction time is greater than multiple times the dynamic variance of the dimension data within the first preset time, and the duration exceeds the preset abnormal time, then the indicator of the answering reaction status dimension is abnormal; If the time proportion of eyes looking at the screen shows that the visual concentration is lower than the preset value, the indicator of the eye movement state dimension is abnormal; If the prefrontal EEG signal activity feature shows a preset degree lower than the corresponding feature in the baseline, the indicator of the EEG activity state dimension is abnormal.

5. The continuous concentration evaluation system according to claim 1, characterized in that: When generating the individual curve, performing abnormality judgment using a preset abnormality judgment rule includes: Comparing and analyzing the data of each dimension in the individual curve with the characteristics of the baseline, and combining the abnormality judgment rules to judge the abnormality of the data of each dimension; When indicators in more than two dimensions are judged to be abnormal, the attention state is determined to be attenuated and marked as a warning observation period.

6. The continuous concentration evaluation system according to claim 1, characterized in that: The attenuation characteristic value of the individual curve is extracted by using the symmetric rotation equivariant convolution technology, including: constructing a model for extracting the attenuation characteristic value of the individual curve, Among them, a symmetric rotation equivariant convolution layer is configured to extract the basic features of the rotation equivariance of the individual curve. Configure a multi-scale feature pyramid module to capture decay patterns at different time scales. configuring an attenuation feature regression module to output the attenuation feature value; The model is trained based on a training data set and using a three-stage training method, so as to extract the attenuation characteristic value of the individual curve using the trained model.

7. The continuous concentration evaluation system according to claim 1, characterized in that: The control module is further configured to: During the early warning observation period, when the attenuation characteristic value shows a return to the normal fluctuation range of the previous attenuation characteristic value, the evaluation process continues.

8. The continuous concentration evaluation system according to claim 1, characterized in that: The continuous concentration evaluation system further includes a result presentation module, which is configured to: Based on the concentration evaluation results, a visual chart is generated to show the attenuation trend and the final determined continuous concentration duration. Generate focus assessment results.

9. A method for evaluating continuous concentration, characterized in that: The continuous concentration assessment method includes: Collect multi-dimensional test data according to a preset time interval and sequence, wherein the multi-dimensional test data includes response data, EEG activity data, and eye movement data; Based on the collected multidimensional test data, a baseline is constructed, including the activity of the frontal lobe EEG signal, the proportion of time the eyes are looking at the screen, and parameters related to the response to the test questions. The collected multidimensional test data is processed using a sliding window technique to generate an individual curve with the same characteristics as the baseline. When generating the individual curve, using the preset abnormality judgment rules, abnormality warning judgment is performed; After the abnormal warning is triggered, the warning observation period begins. The attenuation characteristic value of the individual curve is extracted through the symmetric rotation equivariant convolution technology. When the warning observation period ends and the attenuation characteristic value still shows an abnormality, the evaluation termination mechanism is activated, and the time from the start to the end of the test is used as an evaluation indicator of the continuous concentration time.

10. A machine-readable storage medium, characterized in that The machine-readable storage medium stores instructions, which enable the machine to execute the continuous concentration evaluation method according to claim 9.

Images