Student emotion and concentration degree monitoring and feedback system based on fNIRS

Abstract

The invention relates to the technical field of emotion and concentration monitoring, and discloses an fNIRS-based student emotion and concentration monitoring and feedback system, which comprises an fNIRS sensing module, which adopts a low-power-consumption wireless head-mounted device and continuously monitors blood oxygen signals of key brain regions in real time, and the spatial resolution reaches a millimeter level; the dynamic brain region weight analysis module is used for dynamically recognizing emotion and concentration degree related brain regions based on a deep learning model and automatically allocating weight parameters; and the data fusion module is used for constructing a two-dimensional neural network model, synchronously extracting spatial and temporal characteristics of emotion titer, awakening degree and concentration degree, and integrating eye movement tracking and heart rate variability data. Through high-precision sensing, dynamic brain region analysis and multi-modal closed-loop feedback, the accuracy and real-time performance of monitoring are improved, and the accuracy and the real-time performance of eye movement tracking are improved. And accurate brain region identification and signal processing can be better carried out for different individuals.

Description

Technical Field

[0001] The present invention relates to the technical field of emotion and concentration monitoring, and particularly to a student emotion and concentration monitoring and feedback system based on fNIRS. Background Art

[0002] With the development of educational technology, how to more effectively monitor students' emotions and concentration has become an important challenge. Traditional monitoring methods often rely on subjective evaluations based on questionnaires or observations, which are easily affected by the attitudes of teachers and students and environmental factors, and the objectivity and real-time nature of the data are poor. In recent years, functional near-infrared spectroscopy imaging, i.e., fNIRS technology, has gradually become an effective tool for monitoring students' brain activities due to its non-invasive nature, high spatial resolution, and real-time monitoring ability of brain activities.

[0003] As the application scope of fNIRS technology continues to expand, how to accurately identify key brain regions to capture changes in emotions and concentration has become a crucial issue. Although ordinary fNIRS systems can monitor blood oxygen signals, they are insufficient in dealing with physiological differences between individuals, which limits the accuracy of their monitoring in practical applications. In addition, existing multimodal data fusion methods have deficiencies in feature extraction and registration, resulting in poor effects in comprehensively evaluating students' cognitive states. Therefore, a student emotion and concentration monitoring and feedback system based on fNIRS is proposed. Summary of the Invention

[0004] (1) Technical Problems to be Solved

[0005] Aiming at the deficiencies of the existing technology, the present invention provides the problems that although existing fNIRS systems can monitor blood oxygen signals, they are insufficient in dealing with physiological differences between individuals, and existing multimodal data fusion methods have deficiencies in feature extraction and registration, resulting in poor effects in comprehensively evaluating students' cognitive states.

[0006] (2) Technical Solutions

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A student emotion and concentration monitoring and feedback system based on fNIRS, comprising: an fNIRS sensing module for continuously and real-time monitoring the blood oxygen signals of key brain regions with a spatial resolution reaching the millimeter level;

[0009] A dynamic brain region weight analysis module for dynamically identifying emotion- and concentration-related brain regions based on a deep learning model and automatically assigning weight parameters;

[0010] The data fusion module constructs a two-dimensional neural network model to synchronously extract the spatio-temporal features of emotional valence, arousal, and attention, integrates eye movement tracking and heart rate variability data, and enhances the comprehensive assessment ability of cognitive states through feature-level fusion;

[0011] The feedback execution module adopts an adaptive threshold triggering mechanism to provide multimodal feedback when a decrease in attention or negative emotions are detected;

[0012] The optimization processing module includes teaching strategy optimization and data processing optimization. The local edge device completes signal preprocessing and feature extraction, and the cloud performs model training and optimization to reduce data transmission while ensuring the privacy and security of the original biological data.

[0013] Furthermore, the fNIRS sensing module sets the light source-detector layout, uses a dual-wavelength laser diode array, i.e., 16 light sources + 24 detectors, integrates an inertial measurement unit to detect head movement, and combines accelerometer data to dynamically compensate for motion artifacts, improving the signal-to-noise ratio; it sets up wireless transmission and realizes multi-channel data synchronous transmission based on the Bluetooth 5.2 protocol, supporting the simultaneous connection of multiple devices.

[0014] On the basis of the foregoing solution, the dynamic brain region weight analysis module constructs a deep learning model, including a spatial attention layer for locating key brain regions and a temporal convolutional layer for extracting temporal features, dynamically allocates brain region weights in an end-to-end manner, introduces transfer learning and cross-subject calibration techniques. The transfer learning strategy includes a pre-training stage and a fine-tuning stage. In the pre-training stage, the basic model is trained on a public dataset to learn the common brain function patterns across subjects; in the fine-tuning stage, the model is adapted using 5 minutes of calibration data from new users to reduce the influence of individual differences; for the cross-subject calibration technique, adversarial domain adaptation is introduced, and the feature distributions of different users are aligned through a generative adversarial network.

[0015] As a further solution of the present invention, the data fusion module adopts multimodal synchronous acquisition. When performing eye movement tracking, an infrared camera is integrated to capture the fixation point and pupil diameter, and the visual attention distribution is judged through ROI analysis; for heart rate variability, a PPG sensor is used to extract the heart rate signal, and the ratio of the low-frequency component to the high-frequency component is calculated to evaluate the activity of the autonomic nervous system.

[0016] Furthermore, in the data fusion module, the feature fusion algorithm uses a multi-scale spatio-temporal convolutional network to synchronously process fNIRS, eye movement, and HRV data; during spatial alignment, the multimodal data is mapped to a unified brain region coordinate system; during temporal alignment, the temporal deviation of multiple sensors is eliminated based on dynamic time warping; during feature-level fusion, the multimodal features are weighted and fused through a cross-attention mechanism.

[0017] On the basis of the foregoing solution, a personalized policy library is set in the feedback execution module, which contains a variety of teaching intervention solutions and matches the best feedback mode according to the historical data of students.

[0018] As a further solution of the present invention, the feedback execution module adopts a multi-channel feedback form, integrating educational robots, intelligent desks or AR and VR devices to provide multi-modal feedback such as voice, vision and touch (i.e., vibration). The adaptive threshold trigger mechanism adopts a dual-threshold strategy: the primary threshold, when the concentration continuously drops below the baseline value by 60% for more than 30 seconds, triggers a mild reminder; the secondary threshold, when high negative emotions are detected, immediately starts a strong intervention.

[0019] Furthermore, in the optimization processing module, the teaching strategy is optimized to generate class and individual learning status reports based on the monitoring data, provide suggestions for teachers to adjust the course rhythm, and provide visual brain function maps and emotion-concentration change trend charts; when optimizing data processing, edge computing technology is adopted to complete data preprocessing and feature extraction locally, and only the desensitized analysis results are uploaded to the cloud.

[0020] (III) Beneficial effects

[0021] Compared with the prior art, the present invention provides a student emotion and concentration monitoring and feedback system based on fNIRS, which has the following beneficial effects:

[0022] 1. Through high-precision sensing, dynamic brain region analysis and multi-modal closed-loop feedback, the present invention not only improves the accuracy and real-time performance of monitoring, but also can better perform accurate brain region recognition and signal processing for different individuals. By integrating multi-modal data, the system can provide a more comprehensive and detailed cognitive state assessment, improve the monitoring accuracy. At the same time, positive emotions and supportive feedback can effectively improve the learning efficiency and initiative of students.

[0023] 2. In the present invention, the dynamic brain region weight analysis module realizes the accurate analysis and personalized adaptation of individual brain functions by optimizing the brain region weight distribution, combining transfer learning and adversarial domain adaptation, and improves the accuracy of brain function analysis.

[0024] 3. The data fusion module of the present invention can achieve efficient data fusion, align the data spaces from different modalities to a unified brain region coordinate system, ensure consistency in the spatial dimension, and in the time dimension, use dynamic time warping technology to eliminate the timing deviation of each sensor and ensure synchronization in the time dimension. During feature-level fusion, through the cross-attention mechanism, key feature information is extracted and weighted and fused from multi-modal data, thereby obtaining a weighted comprehensive feature representation, which can more comprehensively describe the concentration and emotional changes of students. A multi-modal synchronous acquisition module is also set up to capture various physiological signals of students during the learning process, facilitating the understanding of the visual attention distribution and overall emotional state of students for different contents during the learning process.

[0025] 4. The feedback execution module of the present invention can, through the collaborative work of the personalized policy library, multi-channel feedback forms, and adaptive threshold triggering mechanism, monitor the changes in students' emotions and concentration in real time, and adjust intervention measures immediately according to feedback requirements, thereby effectively improving the effects of learning and classroom management. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic flow structure diagram of a student emotion and concentration monitoring and feedback system based on fNIRS proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] Embodiment 1

[0029] Referring to Figure 1 , a student emotion and concentration monitoring and feedback system based on fNIRS includes:

[0030] The fNIRS sensing module uses a low-power wireless head-mounted device to continuously monitor the blood oxygen signals of key brain regions in real time, with a spatial resolution reaching the millimeter level. A light source-detector layout is set up, using a dual-wavelength laser diode array, namely 16 light sources + 24 detectors, and a 3D-printed flexible headband is used to achieve close fitting with the scalp to ensure coverage of the prefrontal cortex and the key brain region of the dorsolateral prefrontal cortex. An inertial measurement unit is integrated to detect head movement, and motion artifacts are dynamically compensated in combination with accelerometer data to improve the signal-to-noise ratio. Wireless transmission is set up, and multi-channel data synchronous transmission is realized based on the Bluetooth 5.2 protocol, supporting simultaneous connection of multiple devices to meet the deployment requirements at the class level.

[0031] The dynamic brain region weight analysis module dynamically identifies emotion- and concentration-related brain regions based on a deep learning model, automatically assigns weight parameters, and constructs a deep learning model. It includes a spatial attention layer for localizing key brain regions and a temporal convolutional layer for extracting temporal features, dynamically assigning brain region weights in an end-to-end manner. It introduces transfer learning and cross-subject calibration techniques. The transfer learning strategy includes a pre-training stage and a fine-tuning stage. In the pre-training stage, a basic model is trained on a public dataset to learn the common brain function patterns across subjects. In the fine-tuning stage, the model is adapted using 5-minute calibration data of new users to reduce the influence of individual differences. The cross-subject calibration technique introduces adversarial domain adaptation to align the feature distributions of different users through a generative adversarial network. In the present invention, the dynamic brain region weight analysis module realizes the accurate analysis and personalized adaptation of individual brain functions by optimizing brain region weight allocation, combining transfer learning and adversarial domain adaptation, improving the accuracy of brain function analysis;

[0032] The data fusion module constructs a two-dimensional neural network model to synchronously extract the spatio-temporal features of emotion valence, arousal, and concentration, integrates eye movement tracking and heart rate variability data, enhances the comprehensive evaluation ability of cognitive states through feature-level fusion, and adopts multi-modal synchronous acquisition. When performing eye movement tracking, an infrared camera is integrated to capture the fixation points and pupil diameters, and the visual attention distribution is judged through ROI analysis; for heart rate variability, a PPG sensor is used to extract the heart rate signal, and the ratio of the low-frequency component to the high-frequency component is calculated to evaluate the activity of the autonomic nervous system. The feature fusion algorithm uses a multi-scale spatio-temporal convolutional network to synchronously process fNIRS, eye movement, and HRV data; during spatial alignment, multi-modal data is mapped to a unified brain region coordinate system; during temporal alignment, the dynamic time warping is used to eliminate the temporal bias of multi-sensors; during feature-level fusion, cross-attention mechanism is used to weight and fuse multi-modal features. In the present invention, the data fusion module can achieve efficient data fusion, align the data spaces from different modalities to a unified brain region coordinate system to ensure consistency in the spatial dimension. In the temporal dimension, the dynamic time warping technique is used to eliminate the temporal bias of each sensor to ensure synchronization in the temporal dimension. During feature-level fusion, through the cross-attention mechanism, key feature information is extracted and weighted and fused from multi-modal data, thereby obtaining a weighted comprehensive feature representation, which can more comprehensively describe the concentration and emotion changes of students. It also sets a multi-modal synchronous acquisition module that can capture various physiological signals of students during the learning process, facilitating the understanding of the visual attention distribution and overall emotion state of students for different contents during the learning process;

[0033] The feedback execution module adopts an adaptive threshold triggering mechanism. When a decrease in attention or negative emotions are detected, it provides multimodal feedback, such as dynamically adjusting the difficulty of the virtual scene and tactile vibration reminders. It sets up a personalized strategy library with various teaching intervention plans built-in, and matches the best feedback mode according to the student's historical data, such as splitting knowledge points for explanation and providing motivational voice prompts. It uses a multi-channel feedback form, integrating educational robots, smart desks, or AR and VR devices to provide multimodal feedback including voice, vision, and touch (i.e., vibration), enhancing the intervention effect. The adaptive threshold triggering mechanism adopts a dual-threshold strategy: a primary threshold where the focus continuously drops below 60% of the baseline value for more than 30 seconds, triggering a mild reminder, such as a slight vibration of the desk; a secondary threshold where when high negative emotions are detected, a strong intervention is immediately initiated, such as switching the AR scene to a relaxation mode. Through the collaborative work of the personalized strategy library, multi-channel feedback form, and adaptive threshold triggering mechanism, the feedback execution module of the present invention can continuously monitor the changes in students' emotions and focus, and instantaneously adjust the intervention measures according to the feedback requirements, thus effectively improving the effects of learning and classroom management;

[0034] The optimization processing module includes teaching strategy optimization and data processing optimization. The local edge device completes signal preprocessing and feature extraction, and the cloud performs model training and optimization, reducing data transmission while ensuring the privacy and security of the original biological data. The teaching strategy optimization is to generate class and individual learning status reports based on the monitored data, providing suggestions for teachers to adjust the course rhythm, such as repeating knowledge point explanations and optimizing the frequency of classroom interactions, and providing visual brain function maps and emotion-focus change trend charts to enhance the transparency of the technology. When performing data processing optimization, edge computing technology is used. The data preprocessing and feature extraction are completed locally, and only the desensitized analysis results are uploaded to the cloud to avoid the leakage of the original biological information. Through high-precision sensing, dynamic brain region analysis, and multimodal closed-loop feedback, the present invention not only improves the accuracy and real-time performance of monitoring, but also can better perform accurate brain region identification and signal processing for different individuals. By integrating multimodal data, the system can provide a more comprehensive and detailed cognitive state assessment. At the same time, positive emotions and supportive feedback can effectively improve students' learning efficiency and initiative.

[0035] Embodiment 2

[0036] Refer to Figure 1 , a student emotion and focus monitoring and feedback system based on fNIRS, including;

[0037] The fNIRS sensing module uses a low-power wireless head-mounted device to continuously monitor the blood oxygen signals of key brain regions in real time, with a spatial resolution reaching the millimeter level;

[0038] Dynamic brain region weight analysis module, which dynamically identifies emotion- and concentration-related brain regions based on a deep learning model, automatically assigns weight parameters, reduces individual physiological differences, and constructs a deep learning model. It includes a spatial attention layer for localizing key brain regions and a temporal convolutional layer for extracting temporal features, dynamically assigning brain region weights in an end-to-end manner, introducing transfer learning and cross-subject calibration techniques. The transfer learning strategy includes a pre-training stage and a fine-tuning stage. In the pre-training stage, a basic model is trained on a public dataset to learn the common brain function patterns across subjects; in the fine-tuning stage, the model is adapted using 5-minute calibration data of new users to reduce the influence of individual differences. The cross-subject calibration technique introduces adversarial domain adaptation, aligns the feature distributions of different users through a generative adversarial network, and through optimizing the brain region weight allocation, combines transfer learning and adversarial domain adaptation to achieve accurate analysis and personalized adaptation of individual brain functions, improving the accuracy of brain function analysis;

[0039] Data fusion module, which constructs a two-dimensional neural network model to synchronously extract the spatio-temporal features of emotion valence, arousal, and concentration, integrates eye movement tracking and heart rate variability data, enhances the comprehensive assessment ability of cognitive states through feature-level fusion, and uses multi-modal synchronous acquisition. When performing eye movement tracking, an infrared camera is integrated to capture the fixation point and pupil diameter, and the visual attention distribution is judged through ROI analysis; for heart rate variability, a PPG sensor is used to extract the heart rate signal, and the ratio of the low-frequency component to the high-frequency component is calculated to evaluate the activity of the autonomic nervous system;

[0040] Feedback execution module, which adopts an adaptive threshold triggering mechanism. When a decrease in attention or negative emotion is detected, it provides multi-modal feedback, such as dynamically adjusting the difficulty of the virtual scene and tactile vibration reminders;

[0041] Optimization processing module, including teaching strategy optimization and data processing optimization. The local edge device completes signal preprocessing and feature extraction, and the cloud performs model training and optimization, reducing data transmission while ensuring the privacy and security of the original biological data.

[0042] In particular, the fNIRS sensing module sets the light source-detector layout, uses a dual-wavelength laser diode array, i.e., 16 light sources + 24 detectors, and realizes close fitting with the scalp through a 3D-printed flexible headband to ensure coverage of the prefrontal cortex and the key brain region of the dorsolateral prefrontal cortex. An inertial measurement unit is integrated to detect head movement, and motion artifacts are dynamically compensated by combining accelerometer data, improving the signal-to-noise ratio; wireless transmission is set, and multi-channel data synchronous transmission is realized based on the Bluetooth 5.2 protocol, supporting simultaneous connection of multiple devices to meet the class-level deployment requirements.

[0043] It should be noted that in the data fusion module, the feature fusion algorithm uses a multi-scale spatio-temporal convolutional network to synchronously process fNIRS, eye movement and HRV data; when performing spatial alignment, the multi-modal data is mapped to a unified brain region coordinate system; when performing temporal alignment, the dynamic time warping is used to eliminate the temporal deviation of multiple sensors; during feature-level fusion, the cross-attention mechanism is used to weight and fuse multi-modal features. In the feedback execution module, a personalized policy library is set up, which contains a variety of teaching intervention programs, and the best feedback mode is matched according to the student's historical data, such as splitting and explaining knowledge points and providing incentive voice prompts. The data fusion module can achieve efficient data fusion, align the data from different modalities to a unified brain region coordinate system in the spatial dimension to ensure consistency in the spatial dimension. In the temporal dimension, the dynamic time warping technology is used to eliminate the temporal deviation of each sensor to ensure synchronization in the temporal dimension. During feature-level fusion, through the cross-attention mechanism, key feature information is extracted and weighted and fused from multi-modal data, so as to obtain a weighted comprehensive feature representation, which can more comprehensively describe the student's concentration and emotional changes. A multi-modal synchronous acquisition module is also set up to capture various physiological signals of students during the learning process, which is convenient for understanding the visual attention distribution and overall emotional state of students for different contents during the learning process. The feedback execution module adopts a multi-channel feedback form, integrating educational robots, intelligent desks or AR and VR devices to provide multi-modal feedback such as voice, vision and touch (i.e., vibration) to enhance the intervention effect. The adaptive threshold triggering mechanism adopts a dual-threshold strategy: the primary threshold, when the concentration continuously drops below 60% of the baseline value for more than 30 seconds, triggers a mild reminder, such as a slight vibration of the desk; the secondary threshold, when a high negative emotion is detected, immediately starts a strong intervention, such as switching the AR scene to a relaxation mode. Through the collaborative work of the personalized policy library, multi-channel feedback form and adaptive threshold triggering mechanism, the feedback execution module can monitor the changes in students' emotions and concentration in real time and adjust the intervention measures immediately according to the feedback needs, so as to effectively improve the learning and classroom management effects. In the optimization processing module, the teaching strategy is optimized to generate class and individual learning status reports based on the monitoring data, and provide suggestions for teachers to adjust the course rhythm, such as repeating the explanation of knowledge points and optimizing the frequency of classroom interaction, and provide a visual brain function map and an emotion-concentration change trend map to enhance the transparency of the technology; when optimizing data processing, edge computing technology is used to complete data preprocessing and feature extraction locally, and only the desensitized analysis results are uploaded to the cloud to avoid the leakage of original biological information. Through high-precision sensing, dynamic brain region analysis and multi-modal closed-loop feedback, the present invention not only improves the accuracy and real-time performance of monitoring, but also can better perform accurate brain region identification and signal processing for different individuals. By integrating multi-modal data, the system can provide a more comprehensive and detailed cognitive state assessment, and at the same time, positive emotions and supportive feedback can effectively improve the learning efficiency and initiative of students.

[0044] In the description in this article, it should be noted that relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

Claims

1. A student emotion and concentration monitoring and feedback system based on fNIRS, characterized in that: include; fNIRS sensor module, which continuously monitors blood oxygen signals in key brain areas in real time, with a spatial resolution of millimeters; Dynamic brain area weight analysis module, which dynamically identifies emotion and concentration related brain areas based on deep learning models and automatically assigns weight parameters; The data fusion module builds a two-dimensional neural network model to simultaneously extract the spatiotemporal characteristics of emotional valence, arousal, and concentration, integrates eye tracking and heart rate variability data, and enhances the comprehensive assessment of cognitive status through feature-level fusion; The feedback execution module uses an adaptive threshold trigger mechanism to provide multimodal feedback when a decrease in attention or negative emotions is detected; The optimization processing module includes teaching strategy optimization and data processing optimization. The local edge device completes signal preprocessing and feature extraction, and the cloud performs model training and optimization to reduce data transmission while ensuring the privacy and security of original biological data.

2. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 1, characterized in that: The fNIRS sensor module is configured with a light source-detector layout, adopts a dual-wavelength laser diode array, i.e., 16 light sources + 24 detectors, integrates an inertial measurement unit to detect head movement, and dynamically compensates for motion artifacts in combination with accelerometer data to improve the signal-to-noise ratio; wireless transmission is configured to achieve multi-channel data synchronous transmission based on the Bluetooth 5.2 protocol, and supports simultaneous connection of multiple devices.

3. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 1, characterized in that: The dynamic brain region weight analysis module constructs a deep learning model, including a spatial attention layer for locating key brain regions and a temporal convolution layer for extracting temporal features, dynamically allocating brain region weights in an end-to-end manner, and introducing transfer learning and cross-subject calibration technology. The transfer learning strategy includes a pre-training stage and a fine-tuning stage. In the pre-training stage, a basic model is trained on a public data set to learn common brain function patterns across subjects. During the fine-tuning phase, the 5-minute calibration data of new users is used to adapt the model to reduce the impact of individual differences; Cross-subject calibration technology introduces adversarial domain adaptation and aligns the feature distributions of different users through generative adversarial networks.

4. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 1, characterized in that: The data fusion module adopts multi-modal synchronous acquisition, integrates infrared camera to capture gaze point and pupil diameter during eye tracking, and determines visual attention distribution through ROI analysis; Heart rate variability uses a PPG sensor to extract heart rate signals and calculates the ratio of low-frequency components to high-frequency components to evaluate the activity of the autonomic nervous system.

5. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 4, characterized in that: The feature fusion algorithm in the data fusion module uses a multi-scale spatiotemporal convolutional network to synchronously process fNIRS, eye movement and HRV data; during spatial alignment, the multimodal data is mapped to a unified brain region coordinate system; during temporal alignment, multi-sensor timing deviations are eliminated based on dynamic time warping; during feature-level fusion, multimodal features are weightedly fused through a cross-attention mechanism.

6. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 1, characterized in that: The feedback execution module is provided with a personalized strategy library, with a variety of teaching intervention programs built in, and the best feedback mode is matched according to the student's historical data.

7. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 1, characterized in that: The feedback execution module adopts a multi-channel feedback form, integrates educational robots, smart desks or AR, VR devices, and provides multi-modal feedback such as voice, vision and touch, i.e. vibration. The adaptive threshold trigger mechanism adopts a dual threshold strategy: primary threshold, when the concentration level is continuously lower than the baseline value by 60% for more than 30 seconds, a mild reminder is triggered; The secondary threshold, when high negative emotions are detected, immediately initiates strong intervention.

8. The fNIRS-based student emotion and concentration monitoring and feedback system according to claim 1, characterized in that: The teaching strategy optimization in the optimization processing module is to generate class and individual learning status reports based on monitoring data, provide teachers with course rhythm adjustment suggestions, and provide visual brain function maps and emotion-concentration change trend maps; Edge computing technology is used for data processing optimization to complete data preprocessing and feature extraction locally, and only upload the desensitized analysis results to the cloud.

Images