fNIRS adaptive feedback emotion cognitive collaborative training device and method

Abstract

The application discloses a kind of fNIRS adaptive feedback emotion cognitive collaborative training device and method, it is related to emotional disorder cognitive training technical field.The device includes: data acquisition module, for obtaining original double-wavelength light intensity signal from brain;Data processing module, for the original double-wavelength light intensity signal is carried out multistage pretreatment;Neural feedback module, for calculating multidimensional evaluation index according to hemoglobin concentration data, and using dynamic threshold control algorithm adjusts the difficulty level and feedback threshold of training task;Training module, for training participant based on emotional stimulation task and cognitive training task;Evaluation module, for training evaluation according to comprehensive performance score and feedback threshold comparison result.The present application closely combines brain function equipment with cognitive training and emotional stimulation task, adds dynamic evaluation and real-time feedback to traditional task, and realizes the precise evaluation of user training effect by brain network analysis technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mood disorder cognitive training, and particularly relates to an fNIRS adaptive feedback mood cognitive collaborative training device and method. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] Emotion regulation dysfunction often leads to impaired social adaptation of individuals, especially affecting their cognitive function and quality of life. Developing effective emotion regulation training methods to improve the emotional management ability of individuals is crucial for mental health promotion and clinical intervention.

[0004] Traditional emotion regulation training methods mainly include psychological treatment (such as cognitive behavioral therapy, mindfulness training, etc.), biofeedback therapy and drug treatment, etc. However, these methods have obvious limitations: first, traditional psychological treatment relies mainly on patient subjective reports and physician experience, lacking objective and quantifiable physiological indicators as evaluation support, making it difficult to achieve dynamic and precise regulation of the training process; second, due to the limitations of time, place and professional personnel configuration, the accessibility and scalability of traditional training mode are insufficient; in addition, existing biofeedback methods are mostly based on single physiological signals (such as skin electricity, heart rate variability, etc.), which cannot fully capture the complex activity patterns of the brain in emotional cognitive processing, especially lacking systematic evaluation of brain region collaboration and whole brain network efficiency.

[0005] With the progress of neuroimaging technology and computational analysis methods, neural feedback training has gradually become a new research direction in the field of emotion regulation. As a non-invasive brain imaging technology, functional near-infrared spectroscopy (fNIRS) has been widely used in the study of emotion and cognitive processes due to its good portability, anti-motion interference ability and high correlation with neural oxygen metabolism. Combined with virtual reality, interactive interface and other technologies, the fNIRS neural feedback system can monitor the changes in hemoglobin concentration in different brain regions during user task execution in real time, and convert it into visual feedback information, thereby helping users to enhance their self-awareness and regulation ability of emotional state, and improve the participation and persistence of training.

[0006] However, the existing fNIRS neurofeedback system still has obvious deficiencies in the mechanism construction and adaptive regulation of emotion-cognition collaborative training. On the one hand, most systems use fixed feedback thresholds and static task paradigms, which cannot be dynamically adjusted according to the real-time brain activity state of the user, resulting in a mismatch between the training difficulty and the actual emotion-cognition load of the user, affecting the training effect and user compliance. On the other hand, the existing scheme often uses brain signals as passive recording indicators for emotion or cognitive tasks, without deeply integrating emotion stimulation tasks and cognitive training tasks, and lacks a comprehensive evaluation system based on multi-dimensional neural indicators, especially in the following three aspects: first, most existing systems focus on the activation intensity of a single brain region or a few brain regions, and fail to effectively integrate indicators reflecting the overall activation level of brain regions, making it difficult to fully represent the overall mobilization degree and resource investment of the brain during emotion-cognition processing, and unable to provide a core basis for quantitative evaluation of training intensity; second, the existing scheme lacks sufficient description of the dynamic collaboration relationship between brain regions, and lacks systematic monitoring and analysis of the functional connection strength or pattern between brain regions, so it cannot reveal the dynamic interaction mechanism between different brain networks in emotion and cognitive tasks and its influence on emotion regulation efficiency; third, the existing evaluation is limited to local brain regions or simple connection analysis, lacking comprehensive consideration of indicators such as global efficiency and local efficiency to measure the information transmission and integration capacity of brain networks, making it difficult to evaluate the efficiency and flexibility of emotion-cognition collaborative processing from the network level, and also unable to achieve precise optimization of the training scheme. The above limitations result in the inability of the existing system to systematically quantify the neural collaboration state during emotion-cognition interaction and achieve closed-loop regulation. SUMMARY

[0007] In view of the deficiencies of the prior art, the purpose of the present application is to provide a fNIRS adaptive feedback emotion-cognition collaborative training device and method, which breaks through the limitations of traditional single brain region or isolated indicators by deeply integrating brain function imaging devices with cognitive training and emotion stimulation tasks, and real-time captures the overall activation characteristics and cross-brain region dynamic collaboration patterns of the brain in emotion-cognition collaborative tasks, and based on multi-dimensional brain network collaboration analysis, introduces a dynamic evaluation and real-time feedback mechanism for traditional tasks to achieve precise evaluation and individualized regulation of user training effect.

[0008] To achieve the above purpose, the present application is realized by the following technical solutions:

[0009] The first aspect of the present application provides a fNIRS adaptive feedback emotion-cognition collaborative training device, comprising:

[0010] A data acquisition module for acquiring raw dual-wavelength light intensity signals from the brain;

[0011] The data processing module is used to perform multi-level preprocessing on the original dual-wavelength light intensity signal to obtain hemoglobin concentration data that can reflect the trend of oxyhemoglobin concentration changes.

[0012] The neurofeedback module is used to calculate multidimensional evaluation indicators based on hemoglobin concentration data, design a comprehensive performance score calculation formula based on the interrelationships between the multidimensional evaluation indicators, and use a dynamic threshold control algorithm to adjust the difficulty level and feedback threshold of the training task. The multidimensional evaluation indicators include energy value, brain functional connectivity indicators, and brain network efficiency indicators.

[0013] The training module is used to train participants based on emotional stimulus tasks and cognitive training tasks, and to calculate a comprehensive performance score based on the training results and multidimensional evaluation indicators.

[0014] The evaluation module is used to evaluate the training based on the comparison results of the overall performance score and the feedback threshold.

[0015] Furthermore, it also includes a user information module for storing demographic data and training history.

[0016] Furthermore, it also includes a device connection module, which is used to coordinate data interaction between external devices and the main control unit.

[0017] Furthermore, the data acquisition module is a head-mounted brain scanning device that emits near-infrared light into the brain and receives the returned raw dual-wavelength light intensity signals to monitor the working status of different areas of the brain in real time when the user performs a task.

[0018] Furthermore, the data processing module is also used for:

[0019] The original dual-wavelength light intensity signal was smoothed using a moving average filter.

[0020] The modified Beer-Lambert law was used to convert the preprocessed signal into the relative concentration change of oxyhemoglobin and deoxyhemoglobin, thus obtaining the hemoglobin concentration signal.

[0021] Frequency domain filtering is performed on the hemoglobin concentration signal.

[0022] Furthermore, in the data processing module, the process of converting the preprocessed signal into the relative concentration changes of oxyhemoglobin and deoxyhemoglobin introduces path length factor correction and scattering coefficient compensation.

[0023] Furthermore, in the data processing module, all processing results are timestamped and stored in a buffer, supporting real-time waveform display and offline data backtracking analysis.

[0024] Further, the neural feedback module is further used for calculating the energy value of the target signal by integration based on the time-frequency spectrogram, using the phase lock value to quantify the cooperation degree between brain regions as a brain function connection index, and introducing a graph analysis algorithm to calculate the small world property of the brain network as a network efficiency index.

[0025] Further, the evaluation module is further used for generating a multi-dimensional training effect evaluation report according to the evaluation indexes, and the evaluation indexes include the neural function index, the behavior performance index and the physiological regulation efficiency index.

[0026] The second aspect of the application provides an fNIRS adaptive feedback emotion-cognition collaborative training method, comprising the following steps:

[0027] Obtaining original dual-wavelength light intensity signals from the brain;

[0028] Multi-stage preprocessing is performed on the original dual-wavelength light intensity signals to obtain hemoglobin concentration data capable of reflecting the change trend of oxyhemoglobin concentration;

[0029] According to the hemoglobin concentration data, multi-dimensional evaluation indexes are calculated, a comprehensive performance score calculation formula is designed according to the mutual relationship between the multi-dimensional evaluation indexes, and a dynamic threshold control algorithm is used to adjust the difficulty level and feedback threshold of the training task, and the multi-dimensional evaluation indexes include energy value, brain function connection index and brain network efficiency index;

[0030] Based on the emotion stimulation task and the cognitive training task, the participants are trained, and the comprehensive performance score is calculated according to the training results and the multi-dimensional evaluation indexes;

[0031] According to the comprehensive performance score and the feedback threshold comparison result, the training evaluation is performed.

[0032] The above one or more technical solutions have the following beneficial effects:

[0033] The application discloses an fNIRS adaptive feedback emotion-cognition collaborative training device and method, and the core is that through an adaptive neural feedback mechanism, the feedback threshold and the training task difficulty are dynamically adjusted in real time, so that the training is more suitable for the current neural state and cognitive load of an individual, thereby improving the accuracy of evaluation and the effectiveness of training.

[0034] The application generates a multi-dimensional training effect evaluation report by fusing multi-modal data, especially comprehensively analyzing the overall activation characteristics and cross-brain region dynamic cooperation mode exhibited by the brain in performing collaborative tasks. The report can more comprehensively and intuitively reflect the training progress of the individual emotion-cognition collaboration ability, and provides a strong objective basis for accurately evaluating and intervening emotion-related problems.

[0035] Advantages of the additional aspects of the application will be partially given in the following description, partially will become apparent from the following description, or will be understood by the practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, without creative labor, other drawings can also be obtained from these drawings.

[0037] Figure 1 The structural framework diagram of the fNIRS adaptive feedback emotion-cognition collaboration training device in embodiment one of the application is shown in the figure.

[0038] Figure 2 The working principle flow chart of the fNIRS adaptive feedback emotion-cognition collaboration training device in embodiment one of the application is shown in the figure. DETAILED DESCRIPTION

[0039] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0040] It should be noted that the terms used herein are only for the purpose of describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form, and in addition, it should be understood that when the terms "comprise" and / or "include" are used in the specification, there is a feature, step, operation, device, component and / or combination thereof;

[0041] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0042] Embodiment one:

[0043] The embodiment one of the present application provides a fNIRS adaptive feedback emotion cognitive collaborative training device, as shown in Figure 1 and Figure 2 , comprising a data acquisition module, a data processing module, a neural feedback module, a training module, an evaluation module, a user information module and a device connection module.

[0044] The data acquisition module is used for acquiring original dual-wavelength light intensity signals from the brain.

[0045] Specifically, the data acquisition module is a safe head-mounted brain scanning device based on fNIRS, which monitors the working state of different regions of the brain in real time when the user performs a task by emitting near-infrared light to the brain and receiving the returned original dual-wavelength light intensity signals. The state information and signal intensity are sent to the device connection module, and the collected data are sent to the data processing module.

[0046] The data processing module is used for multi-stage preprocessing of the original dual-wavelength light intensity signals to obtain hemoglobin concentration data capable of reflecting the change trend of oxyhemoglobin concentration.

[0047] Specifically, the data processing module adopts a real-time streaming processing architecture and is responsible for multi-stage preprocessing of the original dual-wavelength light intensity signals collected by the fNIRS device. Specifically, the following steps are included:

[0048] Step 1.1: smoothing the original dual-wavelength light intensity signals by using a sliding average filter.

[0049] The module first applies a sliding average filter to smooth the original signals, and the formula is:

[0050] .

[0051] Where I( ) represents the signal intensity after the sliding average filter processing, represents the time index of the current moment, L is the window length of the sliding average, and i represents the window index. The window length can be configured to 5-10 sampling points, which can effectively suppress high-frequency electronic noise and random interference.

[0052] Step 1.2: converting the preprocessed signals into the relative concentration change amount of oxyhemoglobin and deoxyhemoglobin by using a modified Beer-Lambert law to obtain the hemoglobin concentration signal.

[0053] The classic Beer-Lambert law describes the attenuation of light in a homogeneous medium, which is proportional to the concentration of the substance and the path length. However, the brain has strong scattering properties, and the propagation path of photons in it is much larger than the straight-line distance between the light source and the detector, which makes the direct application of the classic law produce a huge error. The present application introduces the process of converting the preprocessed signal into the relative concentration change of oxyhemoglobin and deoxyhemoglobin into path length factor correction and scattering coefficient compensation. The formula is:

[0054] .

[0055] Where OD is the optical density, which is the logarithm of the ratio of incident light intensity to outgoing light intensity, reflecting the total attenuation of light; is the absorption coefficient, which is a known constant of the absorption ability of hemoglobin to different wavelengths of light; ΔC is the change in hemoglobin concentration that needs to be solved; DPF is the differential path factor, which is the key parameter for path length correction; G is the loss factor for compensating for scattering effects. Using the average light intensity in the resting state as the baseline, the optical density corresponding to the light intensity at each time point during the task is calculated. Then, for the light signals of two specific wavelengths (750nm and 830nm) emitted by the fNIRS device, the above correction equation is established respectively, forming a binary linear equation set. By solving this equation set, the relative concentration change of oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) can be uniquely separated and calculated, and the systematic errors caused by individual head structure differences and photon scattering are effectively compensated by introducing DPF and G, significantly improving the quantitative accuracy of hemoglobin concentration change.

[0056] Step 1.3: Frequency domain filtering of hemoglobin concentration signal.

[0057] Finally, the hemoglobin concentration signal is filtered in the frequency domain through a fourth-order Butterworth band-pass filter with a passband frequency of 0.01-0.1Hz, retaining the low-frequency oscillation components related to neural activity while suppressing high-frequency physiological noise such as heart rate and respiration. The processed hemoglobin concentration data is transmitted in real time to the neurofeedback module and evaluation module through a low-latency data bus at a refresh rate of 20Hz.

[0058] It needs to be particularly pointed out that the above-mentioned multi-stage preprocessing steps constitute a step-by-step refining, targeted and coordinated processing chain. The smoothing processing of step 1.1 provides low-noise input for the high-precision concentration inversion of step 1.2; the accurate concentration of step 1.2 provides reliable physical quantities for the effective band separation of step 1.3; and the filtering of step 1.3 finally ensures the high correlation between signal characteristics and emotional cognitive task neural mechanisms. The design of the entire preprocessing process closely combines the core technical problems solved by this module, namely, how to provide high-quality and high-timeliness brain activity data for subsequent emotional cognitive collaborative training and adaptive neural feedback. The quality of the preprocessed data directly affects the accuracy of the brain activation state judgment of the neural feedback module, and then affects the adaptive adjustment of the task difficulty in the training module and the precise evaluation of the training effect in the evaluation module, which is the key data cornerstone for realizing the emotional-cognitive collaborative training closed-loop system. In the data processing module, the processing results are all time-stamped and stored in the buffer area, supporting real-time waveform display and offline data backtracking analysis.

[0059] The neural feedback module is used for calculating multi-dimensional evaluation indexes according to the hemoglobin concentration data, designing a comprehensive performance score calculation formula according to the mutual relationship between the multi-dimensional evaluation indexes, and adjusting the difficulty level and feedback threshold of the training task using a dynamic threshold control algorithm. The multi-dimensional evaluation indexes include energy value, brain function connection index and brain network efficiency index.

[0060] Specifically, the neural feedback module is the core control unit of the device and has an adaptive adjustment function. This module analyzes the oxygenated hemoglobin concentration change trend transmitted from the data processing module in real time, specifically including the following steps:

[0061] Step 2.1: Calculate the energy value, brain function connection index, and brain network efficiency index.

[0062] The energy value of the target signal is calculated by integration based on the time-frequency spectrum, aiming to convert the original brain oxygen signal into a quantifiable overall brain activation level indicator. The energy value can effectively reflect the resource investment and mobilization intensity of a specific brain area or the whole brain during emotional cognitive task execution, thereby overcoming the limitations of traditional methods that only focus on a single time point or narrow-band power while ignoring the overall energy distribution. This provides an objective basis for evaluating the brain's resource consumption and activation state during the training process. The phase lock value is used to quantify the degree of cooperation between brain regions as a brain functional connectivity indicator, aiming to reveal the dynamic synchronization and collaboration patterns of different brain regions during emotional cognitive processing. This indicator can capture the timing correlation characteristics of information transmission between brain regions and identify the brain network modules that play a key role in emotional cognitive collaboration and their interaction strength, thereby addressing the problem of existing solutions that lack a dynamic collaboration relationship between brain regions and are difficult to reveal network interaction mechanisms. The small-world property of brain networks is calculated using graph theory analysis algorithms as a network efficiency indicator, aiming to evaluate the efficiency and integration capacity of brain information processing from a macro network level. The small-world property and other network efficiency indicators can reflect the optimal balance of brain networks in information transmission, thereby quantifying the flexibility and robustness of brain networks during emotional cognitive collaboration processing. This makes up for the limitations of existing evaluation methods, which are mainly focused on local brain regions or simple connection analysis and cannot evaluate the efficiency and flexibility of collaborative processing from a network level.

[0063] Step 2.1.1: Calculate the energy value.

[0064] Step 2.1.1.1: Perform continuous wavelet transform on the oxyhemoglobin signal using Morlet wavelet to obtain the time-frequency spectrum.

[0065] Step 2.1.1.2: The neural feedback module calculates the signal energy value within the required target frequency band that completely overlaps in time with the task execution period based on the time-frequency spectrum obtained from the aforementioned wavelet transform. The formula is:

[0066] .

[0067] where, represents the total energy in a specific frequency band; t represents the time variable; f represents the frequency variable; represents a double summation operation on time t and frequency f within the specific frequency band band; CWT(t, f) is the continuous wavelet transform, which describes the local characteristics of the signal at time t and frequency f.

[0068] Step 2.1.2: Calculate the brain functional connectivity indicator.

[0069] Meanwhile, the module calculates the brain functional connectivity index based on the phase lock value (PLV). The oxygenated hemoglobin signals of the emotion-related brain region (prefrontal cortex PFC) and the cognition-related brain region (dorsolateral prefrontal cortex DLPFC) are selected, and the PLV of the signals of the two brain regions in the theta and alpha bands is calculated to quantify the degree of cooperation between the brain regions. The PLV calculation formula is:

[0070] .

[0071] wherein, Δ (tk) is the instantaneous phase difference of the signals of the two brain regions at time tk, and N is the number of sampling points. The closer the PLV value is to 1, the stronger the functional connectivity is.

[0072] Step 2.1.3: Calculate the brain network efficiency index.

[0073] In order to more systematically evaluate the processing and analysis ability of each brain region, the present embodiment introduces a graph theory analysis algorithm to calculate the small-world property (C and L) of the brain network as the network efficiency index. Each brain region is regarded as a node, and the PLV value between the brain regions is regarded as the weight of the edge to construct a brain function network. The small-world property C and L are calculated according to the following formula:

[0074] .

[0075] wherein, C and L are the clustering coefficient and characteristic path length of the actual brain network, and are the corresponding values of the random network. >1 indicates that the network has the efficient small-world property.

[0076] The above three types of indexes are not isolated, but constitute a progressive and three-dimensional evaluation framework from local activation energy to inter-regional cooperative connection, and then to global optimization network. The multi-band energy reflects the independent working strength of each brain region; the functional connectivity describes the information exchange quality between the key brain regions; and the small-world property reveals the overall efficiency of the whole brain information processing from the system level. The combination of the three can comprehensively and reliably quantify the brain activation strength and functional state related to the task, and provide more accurate and multi-dimensional basis for neurofeedback than a single index.

[0077] Step 2.2: Design the comprehensive performance score calculation formula according to the mutual relationship between the multi-dimensional evaluation indexes.

[0078] ​​After obtaining the multi-dimensional evaluation index of brain activation, the ratio of the number of correct responses to the total number of tasks in each task period is taken as the accuracy according to the behavior data such as accuracy and reaction time transmitted by the training module, the time interval from stimulus presentation to user response is recorded as the reaction time through a high-precision timer, and the comprehensive performance score is calculated, the formula is:

[0079] .

[0080] Wherein, P represents the comprehensive performance score, which is a single index for quantifying user training performance; Acc represents the accuracy of the user in the training task; RT represents the reaction time of the user, and the reciprocal of the reaction time is used to represent the reaction speed, and the faster the reaction, the larger the value; is the target frequency band brain activation energy value obtained by time-frequency analysis;

[0081] is the functional connectivity index between the prefrontal cortex and the dorsolateral prefrontal cortex, and the weight γ2 is used to strengthen the contribution of emotion-cognition brain area cooperation. σ is the small world property of brain network, and the weight γ3 is used to evaluate the information integration efficiency of the brain. α, β, , , are the weight coefficients of accuracy, reaction speed, brain activation energy, brain function connection index and brain network efficiency index, respectively, which are used to adjust the contribution proportion of each index in the total score. The comprehensive score model not only simply adds each index, but also considers the internal synergy and antagonism relationship. For example, when PLV increases (good brain area cooperation), it is often accompanied by the optimization of and the improvement of Acc, which indicates that the user is in a good "emotion-cognition" cooperative state, and the system will give a higher score. On the contrary, if is high but PLV is low, it may mean that the user tries to concentrate attention but the brain area cooperation is not good, and the system may determine that the cognitive load is too heavy or the emotional interference is too large, so the score will not be significantly improved, or even trigger the difficulty adjustment mechanism. This multi-index cross-validation mechanism greatly improves the accuracy and robustness of the evaluation.

[0082] Step 2.3: Adjust the difficulty level and feedback threshold of the training task using a dynamic threshold control algorithm.

[0083] The threshold calculation formula is:

[0084] .

[0085] Wherein, Th represents the adaptive feedback threshold; represents the average value of the user's recent comprehensive performance score P; represents the standard deviation of the user's recent comprehensive performance score; a sensitivity coefficient for controlling the range of influence of the standard deviation; a basic offset for setting a minimum reference for the threshold.

[0086] By monitoring the deviation amplitude and duration of the comprehensive performance score P value from the adaptive feedback threshold Th in consecutive task trials, and introducing high performance offset and low performance offset as the judgment boundary, a multi-parameter coupled decision model is established. In the model, the grade promotion condition formula is:

[0087] .

[0088] wherein, is the high performance offset.

[0089] When the number of consecutive trials that meet the grade promotion condition formula reaches the preset value, the difficulty level is increased, and the new difficulty level calculation formula is:

[0090] .

[0091] wherein is a configurable difficulty adjustment coefficient, and represent the new difficulty level and the current difficulty level respectively, is the ceiling function, which ensures that the difficulty level is positively correlated with the degree of user performance exceeding the threshold. Conversely, the grade reduction condition formula is:

[0092] .

[0093] wherein is the low performance offset, and when the detected consecutive trials reach the threshold, the new difficulty reduction level calculation formula is:

[0094] .

[0095] wherein and represent the new difficulty level and the current difficulty level respectively, and the max function is the maximum value function, which ensures that the difficulty level is not lower than the basic level. This feedback control logic based on quantitative parameters and continuous monitoring effectively avoids false adjustment caused by accidental fluctuations, and realizes the adaptive and precise regulation of training difficulty.

[0096] The training module is used for training participants based on emotional stimulation tasks and cognitive training tasks, and calculating the comprehensive performance score according to the training results and multi-dimensional evaluation indexes.

[0097] The training module is developed based on Python and PyTorch framework, and a complete localization training system is constructed for task training. The task design follows the principle of integrating emotion regulation and cognitive function, mainly including task A and task B. Task A is an "emotion stimulus task". This task presents standardized visual stimulus materials with emotional valence (positive, negative, neutral) to activate the user's emotion-related brain neural network, which can enhance emotion recognition accuracy, improve emotion state awareness, and promote the integration of emotion response and cognitive evaluation system. At the beginning of the task, the system interface presents emotion stimulus pictures retrieved from the database in the center. The pictures are fixed in time and divided into positive, negative and neutral categories. Below the interface, there are positive and negative emotion selection buttons. The user needs to judge the emotional attributes of the pictures. Task B is a "cognitive training task". This task uses a dynamic working memory paradigm to impose cognitive load through sequence number memory and reproduction operations, which can strengthen working memory capacity, improve attention control ability, and enhance cognitive control function in emotionally challenging situations. At the beginning of the task, the system interface presents the numbers 0-9 in sequence, and 6 numbers form a complete sequence. Participants need to reproduce the number sequence according to the task stage requirements. When the participant taps the keyboard to input the number, the interface synchronously displays the input content and triggers the corresponding piano scale sound.

[0098] The training module constructs a dual-task cooperative training system with a deep coupling mechanism. The core is to establish a bidirectional real-time data linkage and parameter mutual adjustment mechanism between task A and task B. The specific cooperation logic is as follows:

[0099] The emotion stimulus task and the cognitive training task are deeply coupled through the dynamic threshold control algorithm provided by the neural feedback module. Based on the deviation relationship between the feedback threshold Th updated by the neural feedback module in real time and the user's comprehensive performance score P, the training module dynamically adjusts the difficulty parameters and presentation logic of the dual tasks. Specifically, at the beginning of each training round, the system first calls the historical training data in the user information module, combines the latest feedback threshold generated by the neural feedback module, and sets the initial difficulty level of this round through the dynamic threshold control algorithm.

[0100] The cooperative training alternately executes the emotion stimulus task (task A) and the cognitive training task (task B) in rounds. During the execution of task A, the system monitors the reaction time and recognition accuracy of the user to the emotional pictures in real time, and obtains the oxygenated hemoglobin concentration data of the prefrontal cortex (PFC) through the data processing module. The neural feedback module dynamically updates the feedback threshold based on the real-time acquisition of brain function connection indicators and brain network efficiency indicators. The update formula is based on the original dynamic threshold control algorithm, and introduces the inter-task cooperation factor ξ:

[0101] .

[0102] where ξ is the synergistic sensitivity coefficient, is the baseline functional connectivity reference value of the user, is the current adaptive feedback threshold. This formula ensures that the threshold adjustment can reflect the real-time changes in the synergistic state of the emotional cognitive brain region.

[0103] When the user's reaction to high arousal negative pictures is detected, the RT is significantly lower than the historical average, and the PFC brain region energy value exceeds the current threshold , the system triggers the difficulty increasing mechanism of task B. The new difficulty level of task B Introduce the dual-task coupling coefficient λ, dynamically adjust by the following formula:

[0104] 。

[0105] where is the adjusted classification complexity of task B, is the current classification complexity of task B, the default value is 0.6, is the baseline brain activation energy of the user. This formula integrates the real-time brain activation energy and the degree of performance score deviation threshold, so that the cognitive task difficulty more accurately matches the current emotional arousal level of the user.

[0106] If the user's working memory accuracy during the execution of task B continues to be higher than 85%, but the PLV value is lower than the synergistic requirement corresponding to the dynamic threshold, it indicates that the cognitive performance is good but the emotional cognitive synergy is insufficient. At this time, the system starts the adaptive challenge mode in the next round of task A, and calculates the emotional picture classification complexity level by the following formula:

[0107] .

[0108] where is the adjusted classification complexity of task A, is the current classification complexity of task A, is the complexity adjustment coefficient, the default value is 0.6. This design takes advantage of the gain effect of brain network efficiency σ on information processing capacity to specifically strengthen emotional information processing load when synergy is insufficient.

[0109] The training module also establishes a bidirectional parameter linkage mechanism triggered by the threshold. When the neural feedback module detects that the brain network small-world property σ exceeds the threshold for three consecutive rounds, and the P value is continuously higher than the new threshold, it is determined that the user enters a high-efficiency synergistic state. At this time, the difficulty of the dual task is increased simultaneously: the length of the number sequence of task B is increased, and the picture switching interval of task A is compressed. This parameter adjustment based on dynamic threshold ensures that the emotional cognitive load always keeps pace with the evolution of the user's neural state.

[0110] an evaluation module configured to evaluate the training according to the comprehensive performance score and the comparison result of the feedback threshold.

[0111] Specifically, the evaluation module is further configured to receive and fuse multi-source data from the data processing module and the training module, and generate a multi-dimensional training effect evaluation report according to evaluation indexes, the evaluation indexes including a neurological function index, a behavior performance index, and a physiological regulation efficiency index. The neurological function index is, for example, the activation intensity of a target brain region and the brain network functional connectivity. The behavior performance index is, for example, the completion accuracy, the reaction time, and the difficulty level of each task. The physiological regulation efficiency index is, for example, the speed of recovery of the physiological signal of the user to the baseline level when facing a challenging task. The module integrates the three evaluation indexes and generates a structured evaluation report through a weighted fusion algorithm. The report includes a quantitative score (such as a neurological regulation gain score and a behavior adaptation index), a visual trend chart (a performance progress curve and a brain function activation pattern comparison), and individualized insights based on a machine learning model (such as identifying the user's dominant training mode and pointing out the short board of coordinated regulation to be optimized), thereby providing a scientific basis for the trainer to adjust the intervention scheme.

[0112] a user information module configured to store demographic data and training history records.

[0113] Specifically, the user information module is responsible for the whole-cycle management of the user's personal archives, covering basic information including demographic data such as age and gender, complete training history records including metadata such as the start and end time, duration, and training task type of each training, individualized system parameter configurations including adjustable parameters such as the initial difficulty level, feedback sensitivity threshold, and visual prompt intensity, and baseline brain function data collected before each training, including physiological indicators such as the changes in cerebral oxygen saturation and hemoglobin concentration in the resting state. All data are stored and managed using a time-series database to ensure data consistency and traceability. The user information module and the training module interact with each other, and before each training, the training module calls the historical performance data and individualized parameters in the user information module to set the initial difficulty and feedback threshold of the current training. During the training process, the performance data generated in real time is continuously written into the user information module for updating the user's training performance.

[0114] a device connection module configured to coordinate the data interaction between external devices and the host end.

[0115] Specifically, the device connection module, as the core communication hub of the system, adopts a multi-thread asynchronous communication architecture and is responsible for coordinating the data interaction between all external devices and the host end. The module first automatically identifies the connected fNIRS device model and firmware version through the device enumeration protocol, and initializes the high-speed data channel based on the TCP / IP protocol. During the data transmission stage, the module monitors key indicators such as signal strength, data packet loss rate, and transmission delay in real time. When the signal quality is detected to be below the preset threshold, such as a data packet loss rate > 5% or a delay > 100 ms, an alarm is automatically triggered, and a prompt is given through the software interface. The module has a built-in data buffer pool and retransmission mechanism to effectively deal with transient data loss caused by network fluctuations. At the same time, the module adopts a modular design and reserves standardized multi-modal interfaces, including Bluetooth 5.0 and USB-C, supporting plug-and-play extension of other physiological signal acquisition devices. It also supports synchronous acquisition and timestamp alignment of multi-source heterogeneous data through JSON format nested binary data packets, and sends the data to the data processing module and the neural feedback module for processing after connection is completed.

[0116] Embodiment two

[0117] The embodiment two of the present application provides a fNIRS adaptive feedback emotion cognitive collaborative training method, comprising the following steps:

[0118] Obtain the original dual-wavelength light intensity signal from the brain;

[0119] Multi-stage preprocessing is performed on the original dual-wavelength light intensity signal to obtain hemoglobin concentration data reflecting the change trend of oxygenated hemoglobin concentration;

[0120] According to the hemoglobin concentration data, a multi-dimensional evaluation index is calculated, a comprehensive performance score calculation formula is designed according to the mutual relationship between the multi-dimensional evaluation indexes, and a dynamic threshold control algorithm is used to adjust the difficulty level and feedback threshold of the training task. The multi-dimensional evaluation indexes include energy value, brain function connection index and brain network efficiency index;

[0121] Based on the emotion stimulation task and the cognitive training task, the participants are trained, and the comprehensive performance score is calculated according to the training results and the multi-dimensional evaluation indexes;

[0122] According to the comparison result of the comprehensive performance score and the feedback threshold, the training evaluation is performed.

[0123] The steps involved in the above embodiment two correspond to the embodiment one, and the specific implementation can refer to the related description part of the embodiment one.

[0124] Those skilled in the art can be aware that units and algorithm steps of each example described in combination with the embodiments disclosed in the application can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0125] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the embodiments of the application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in or transmitted by a computer readable storage medium. The computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired or wireless manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data processing device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media, optical media or semiconductor media, etc.

[0126] The above description is only a specific implementation of the application, but the protection scope of the application is not limited thereto. Any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the application, which should be covered within the protection scope of the application. Therefore, the protection scope of the application should be subject to the protection scope of the claims.

Claims

1. An fNIRS adaptive feedback emotion cognitive collaborative training device, characterized in that, Comprise: a data acquisition module for acquiring original dual-wavelength light intensity signals from the brain; a data processing module for multi-stage preprocessing of the original dual-wavelength light intensity signals to obtain hemoglobin concentration data reflecting the trend of changes in the concentration of oxygenated hemoglobin; a neural feedback module for calculating multi-dimensional evaluation indexes including energy value, brain functional connectivity index and brain network efficiency index based on the hemoglobin concentration data; specifically, the energy value of the target signal is calculated by integration based on the time-frequency spectrogram, the degree of cooperation between brain regions is quantified as the brain functional connectivity index using the phase-locked value, and the small-world property of the brain network is calculated as the network efficiency index using the graph theory analysis algorithm; a comprehensive performance score calculation formula is designed according to the mutual relationship between the multi-dimensional evaluation indexes, and a dynamic threshold control algorithm is used to adjust the difficulty level and feedback threshold of the training task, with the threshold calculation formula being: ， wherein Th represents an adaptive feedback threshold value; an average value representing a recent comprehensive performance score of the user; a standard deviation representing a recent comprehensive performance score of the user; a sensitivity coefficient for controlling a range of influence of the standard deviation; a base offset value; the grade promotion condition formula being: ， wherein, is a high performance offset, is a high performance offset, when the number of consecutive attempts that meet the grade promotion condition formula reaches a preset value, a difficulty level promotion algorithm is started, with the new difficulty level promotion formula being: ， wherein is a configurable difficulty adjustment coefficient, and represent the new difficulty level and the current difficulty level, respectively, is a rounding up function, and the level decrease condition formula is: ， wherein is a low performance bias, and when the detected consecutive trials reach a threshold, then the new difficulty level is calculated by the formula: ， where the max function is the maximum value function; a training module for training participants based on emotional stimulation tasks and cognitive training tasks, and calculating a comprehensive performance score based on the training results and multi-dimensional evaluation indexes; an evaluation module for training evaluation based on the comparison results of the comprehensive performance score and the feedback threshold. 2.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 1, wherein, It also includes a user information module for storing demographic data and training history records. 3.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 1, wherein, It also includes a device connection module for coordinating data interaction between external devices and the host. 4.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 1, wherein, The data acquisition module is a head-mounted brain scanning device that monitors the working state of different regions of the brain in real time when the user performs tasks by emitting near-infrared light to the brain and receiving the original dual-wavelength light intensity signals returned. 5.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 1, wherein, The data processing module is also used for: using a moving average filter to smooth the original dual-wavelength light intensity signals; using a modified Beer-Lambert law to convert the preprocessed signals into relative concentration changes of oxygenated hemoglobin and deoxygenated hemoglobin to obtain hemoglobin concentration signals; performing frequency domain filtering on the hemoglobin concentration signals. 6.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 5, wherein, In the data processing module, the process of converting the preprocessed signals into relative concentration changes of oxygenated hemoglobin and deoxygenated hemoglobin introduces path length factor correction and scattering coefficient compensation. 7.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 1, wherein, In the data processing module, the processing results are all time-stamped and stored in a buffer, supporting real-time waveform display and offline data backtracking analysis. 8.The fNIRS adaptive feedback emotion cognitive collaborative training device of claim 1, wherein, The evaluation module is also used to generate multi-dimensional training effect evaluation reports based on evaluation indexes including neural function indexes, behavioral performance indexes and physiological regulation efficiency indexes.

Images