A concentration analysis and training method and device and storage medium

Abstract

This invention relates to a method, device, and storage medium for attention analysis and training, comprising the following steps: Step 1, data acquisition step: real-time acquisition of the user's electroencephalogram (EEG) signals and behavioral interaction data related to the training task; Step 2, index calculation step: calculation and generation of a comprehensive attention index based on the EEG signals and behavioral interaction data; Step 3, content matching step: matching and calling corresponding attention training interactive programs from a preset content library based on the real-time value and temporal variation characteristics of the comprehensive attention index and a state judgment model; Step 4, difficulty adjustment step: adaptively adjusting the difficulty level of attention training interactive programs in subsequent training based on the user's attention performance data formed in historical training cycles; Step 5, report generation step: recording training process data and generating a report on the user's attention development and training effect. This invention enables better and more accurate attention analysis.

Description

Technical Field

[0001] This invention relates to the field of data analysis, and more specifically to a method, apparatus, and storage medium for attention analysis and training. Background Technology

[0002] With the development of network and AI technologies, people often face the problem of poor concentration in their studies and work. Some groups (such as children) also suffer from insufficient concentration. The reasons for poor concentration can be summarized into four aspects: physiological factors (fatigue, lack of sleep, hunger), psychological factors (low interest, poor mood), environmental factors (many distractions), and personal habits. Existing concentration training devices (such as traditional brain rings) have obvious shortcomings: they only have simple sound and voice interaction, with limited interactive elements and a lack of rich visual and contextual interactive forms; they cannot intuitively present various concentration-related data and changes through display devices, lacking visual and data visualization feedback; and they lack precise monitoring, real-time feedback, and personalized adaptation mechanisms for concentration states, making it difficult to achieve a scientific concentration cycle of "monitoring-recovery-training," and failing to meet users' needs for targeted training to improve concentration and develop good concentration habits. Therefore, a concentration analysis and training method, device, and storage medium are proposed. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a method for attention analysis and training, comprising the following steps: Step 1, Data Acquisition Steps: Real-time acquisition of the user's electroencephalogram (EEG) signals and behavioral interaction data related to the training task; Step 2, Index Calculation Step: Based on the aforementioned electroencephalographic signals and behavioral interaction data, a comprehensive attention index is calculated and generated; Step 3, Content Matching Step: Based on the real-time value of the comprehensive focus index and its temporal change characteristics, and based on the state judgment model, match and call the corresponding focus training interactive program from the preset content library; Step 4, Difficulty Adjustment Step: Based on the user's attention performance data formed in the historical training cycle, adaptively adjust the difficulty level of the interactive program in subsequent training. Step 5, Report Generation Steps: Record training process data and generate a report on user focus development and training effectiveness.

[0004] Furthermore, in step one, the acquisition of electroencephalographic signals specifically involves: acquiring multi-channel electroencephalographic signals from the user's frontal and parietal lobes using a signal acquisition device worn on the head; the behavioral interaction data includes the user's response time to visual or auditory stimuli and the accuracy of operation selection.

[0005] Furthermore, in step two, calculating and generating the comprehensive focus index FI specifically includes: The EEG signal was preprocessed and subjected to spectral analysis to extract the signal energy values ​​E in the θ, α, and β bands, respectively. θ E α E β， , i.e., physiological indicators; The behavioral interaction data is processed to obtain normalized reaction speed index R and accuracy index A, i.e., behavioral indicators. The aforementioned physiological and behavioral indicators are combined using a weighted fusion formula to form a single comprehensive attention index FI, the calculation formula of which is as follows: ; in, This is used to prevent small positive numbers with a denominator of zero; , and The preset weighting coefficients are used, and they satisfy the following conditions: ; C is a constant term used to adjust the scoring criteria.

[0006] Furthermore, in step three, the interaction matching program based on the state determination model specifically includes: Feature extraction: Calculate the rate of change of the comprehensive attention index FI within the latest time window. The current comprehensive focus index (FI) and its rate of change Combined into attention state feature vector ; State determination: The attention state feature vector The input is fed into the state determination model; the state determination model is a classification model pre-trained based on historical training data, which internally stores feature center vectors for multiple attention state categories. , where j represents the state category index, and the state categories include at least dispersed state, stable state, and highly concentrated state; The calculation process for state determination is as follows: First, calculate the current attention state feature vector. With the feature center vector of each attention state category The Euclidean distance between them; Then, the distance is converted into similarity weights based on the Gaussian kernel function; finally, by normalizing the similarity weights of all categories, the probability Pj that the user currently belongs to the j-th state is obtained, and the probability calculation formula is as follows: ; in, Representing vectors and The Euclidean distance between them Representing vectors and Direct European distance, N is the total number of preset state categories, which is a model parameter used to control the smoothness of the probability distribution. Program matching: Select the state with the highest probability value Pj as the dominant state, and select and start the corresponding attention training interactive program from the preset content library according to the preset "state-program" mapping relationship.

[0007] Furthermore, in step four, the adaptive adjustment of the difficulty level specifically includes: Based on the user's comprehensive attention index sequence over multiple recent training cycles, a quantitative value Level representing their long-term attention level is calculated. An adaptive control algorithm is employed, based on the quantified value of long-term attention level (Level) and the current interactive program difficulty (Diff). now The appropriate theoretical level value The difference between the two is used to calculate and set the recommended difficulty level (Diff) for the next training cycle. next The specific calculation process is as follows: ; Where α is the adjustment step size coefficient, used to control the magnitude of difficulty adjustment; The difficulty of implementing adaptive control algorithms increases with the improvement of user capabilities, resulting in adaptive adjustment.

[0008] Furthermore, in step five, generating a report on the development and training effectiveness of user focus specifically includes: A personal training database is established for each user to continuously store relevant data for each training session. The relevant data includes at least a timestamp, the comprehensive focus index, the identifier of the training interaction program, and the difficulty level. Based on the personal training database, a visual report is generated through data analysis. The report includes: a curve showing the change of the comprehensive focus index over time, performance analysis at different difficulty levels, and personalized improvement suggestions based on training history.

[0009] Furthermore, the method also includes device connection and interaction steps: The signal acquisition device establishes a connection with one or more display processing terminals via wireless communication. The attention training interactive program runs on the display processing terminal and provides the user with an interactive interface that includes real-time feedback on the comprehensive attention index.

[0010] A focus analysis and training device, the device comprising: The data acquisition unit is used to collect users' electroencephalographic signals and behavioral interaction data in real time. The index calculation unit is used to calculate and generate a comprehensive focus index based on the collected data. The content matching unit is used to match and invoke the attention training interactive program based on the index and the state judgment model. Difficulty adjustment unit is used to adaptively adjust the program difficulty based on historical performance data; The report generation unit is used to record data and generate training effect reports.

[0011] A storage medium storing a computer program, which, when executed by a processor, implements the steps of the method.

[0012] The beneficial effects of this invention are: By precisely acquiring multi-channel EEG signals from the frontal and parietal lobes of the user through a head-worn signal acquisition device, and combining this with behavioral interaction data such as the user's response time to visual or auditory stimuli and the accuracy of operation selection, a weighted fusion formula is used to fuse the frequency band energy values ​​of the EEG signals with normalized behavioral indicators to generate a comprehensive attention index that accurately reflects the level of concentration, solving the problem of the single-faceted nature of traditional assessments. Based on a state judgment model, through Euclidean distance calculation and Gaussian kernel function processing, it accurately determines the user's attention state, such as scattered state, stable state, and highly concentrated state, and matches it with rich interactive games such as suspended balls, replacing the single-sound voice interaction of traditional brain rings, improving the fun and targeting of training. It supports wireless connection to various display terminals such as mobile phones, tablets, and TVs via Bluetooth or network, and presents the comprehensive attention index in real time and visually. The change curve allows users to intuitively understand their own status, overcoming the shortcomings of traditional devices where data is not visible. Based on the user's historical training cycle attention performance data, an adaptive control algorithm calculates a quantitative value of long-term attention level, dynamically adjusting the difficulty of the training program so that the difficulty increases with ability, adapting to the training needs of different stages. A personalized training database is established for each user, continuously storing data such as training timestamps, attention index, program identifiers, and difficulty, generating a visual report that includes index change curves, performance analysis at different difficulty levels, and personalized improvement suggestions. Combined with dynamic monitoring of attention improvement trajectory through the system account, it helps users scientifically improve their attention and develop good attention habits through a "monitor-recovery-training" attention cycle, comprehensively covering attention problems caused by physiological, psychological, environmental, and personal habits. Attached Figure Description

[0013] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] like Figure 1 As shown, a method for analyzing and training focus includes the following steps: Step 1, Data Acquisition Steps: Real-time acquisition of the user's electroencephalogram (EEG) signals and behavioral interaction data related to the training task; Step 2, Index Calculation Step: Based on the aforementioned electroencephalographic signals and behavioral interaction data, a comprehensive attention index is calculated and generated; Step 3, Content Matching Step: Based on the real-time value of the comprehensive focus index and its temporal change characteristics, and based on the state judgment model, match and call the corresponding focus training interactive program from the preset content library; Step 4, Difficulty Adjustment Step: Based on the user's attention performance data formed in the historical training cycle, adaptively adjust the difficulty level of the interactive program in subsequent training. Step 5, Report Generation Steps: Record training process data and generate a report on user focus development and training effectiveness.

[0016] In step one, collecting electroencephalographic signals specifically involves acquiring multi-channel electroencephalographic signals from the user's frontal and parietal lobes using a signal acquisition device worn on the head; the behavioral interaction data includes the user's response time to visual or auditory stimuli and the accuracy of operation selection. By explicitly collecting multi-channel EEG signals from the user's frontal and parietal lobes (core brain regions related to attention), and simultaneously acquiring behavioral interaction data that directly reflects training performance, such as response time and operational accuracy under visual / auditory stimuli, this approach achieves dual-dimensional data support for attention assessment, encompassing both physiological and behavioral aspects. This ensures the targeted and comprehensive nature of data collection while avoiding the biased assessments caused by traditional single-dimensional data collection. It provides high-quality, multi-dimensional raw data for the accurate calculation of the subsequent comprehensive attention index, making attention analysis more aligned with the user's actual training scenarios and enhancing the scientific rigor of subsequent training matching and difficulty adjustment.

[0017] The user wears a head-mounted signal acquisition device to perform a visual stimulation training task (numbers flash randomly on the screen, and the user needs to quickly click on odd numbers). The device acquires 4-channel EEG signals from the frontal and parietal lobes in real time. After preprocessing such as filtering and noise reduction, and spectral analysis, the theta band energy value is extracted. α band energy value β band energy value ; The training task included 12 visual stimuli, and the users' response times were 0.36s, 0.42s, 0.39s, 0.45s, 0.37s, 0.41s, 0.38s, 0.43s, 0.35s, 0.40s, 0.34s, and 0.44s, respectively. The average response time was calculated as: (0.36+0.42+0.39+0.45+0.37+0.41+0.38+0.43+0.35+0.40+0.34+0.44)÷12=4.84÷12≈0.403s; The normalized acceptable range for response time is set to [0.3s, 0.6s]. The normalization formula is R = (0.6 - actual response time) / (0.6 - 0.3). Substituting into the formula, we get R = (0.6 - 0.403) / 0.3 = 0.197 / 0.3 ≈ 0.657. If 10 out of 12 operations are correct, the accuracy rate A = 10 / 12 ≈ 0.833. Since the normalization range of accuracy rate is [0,1], we directly take the normalized A = 0.833. Combining the comprehensive focus index formula Set preset weights (satisfy To prevent the denominator from being a tiny positive number δ=0.001 with zero, the constant term C=50 of the scoring criterion is adjusted. First, calculate the relevant terms of the electroencephalogram (EEG) signal: ; Substituting these values ​​into the formula, we can calculate the overall focus index: FI = 0.4 × 0.784 + 0.3 × 0.657 + 0.3 × 0.833 + 50 = 0.3136 + 0.1971 + 0.2499 + 50 = 50.7606; If only EEG signals are collected, the calculated FI = 0.4 × 0.784 + 50 = 50.3136, which cannot reflect the response speed and accuracy of the user's operation; If only behavioral data is collected, the calculated FI = 0.3 × 0.657 + 0.3 × 0.833 + 50 = 50.447 lacks objective support from the physiological level and has obvious biases. However, the dual-dimensional data collection in this case allows the FI value to more accurately reflect the user's true level of focus.

[0018] Step two, specifically, involves calculating and generating the comprehensive focus index FI, including: The EEG signal was preprocessed and subjected to spectral analysis to extract the signal energy values ​​E in the θ, α, and β bands, respectively. θ E α E β， , i.e., physiological indicators; The behavioral interaction data is processed to obtain normalized reaction speed index R and accuracy index A, i.e., behavioral indicators. The aforementioned physiological and behavioral indicators are combined using a weighted fusion formula to form a single comprehensive attention index FI, the calculation formula of which is as follows: ; in, This is used to prevent small positive numbers with a denominator of zero; and The preset weighting coefficients are used, and they satisfy the following conditions: ; C is a constant term used to adjust the scoring criteria; By extracting the energy values ​​of the θ, α, and β frequency bands related to attention from EEG signals and combining them with normalized behavioral interaction indicators (reaction speed R and accuracy A), a single comprehensive attention index FI is synthesized using a weighted fusion formula. This incorporates objective EEG data from the physiological level while also taking into account actual behavioral performance in training tasks. The weight coefficients satisfy the constraint that the sum is 1, and the δ parameter avoids the denominator being zero, while C adjusts the scoring benchmark. This solves the problems of the traditional single-indicator assessment of attention being one-sided and unstable, making attention measurement more accurate, comprehensive, and reliable. It provides a standardized and directly usable core analytical basis for subsequent state judgment models.

[0019] For example, in a visual stimulus training scenario (numbers flash randomly on the screen, and the user needs to quickly click on odd numbers), the relevant data is as follows: EEG signal processing results: After preprocessing and spectral analysis, the energy value of the theta band was extracted. α band energy value β band energy value ; Behavioral interaction data processing results: The average response time is 0.403s. After normalization, R≈0.657 is calculated using the formula R=(0.6-actual response time) / (0.6-0.3); 10 out of 12 operations were correct, and the normalized accuracy A≈0.833. Preset parameters: weighting coefficients (satisfy To prevent the denominator from being a tiny positive number δ=0.001 with zero, the constant term C=50 of the scoring criterion is adjusted. Calculation of Overall Focus Index (FI): The first step is to calculate the proportion of EEG signals: ; The second step is to substitute the formula into the weighted fusion formula.

[0020] ; The third step is to calculate each value step by step: 0.4 × 0.784 = 0.3136, 0.3 × 0.657 = 0.1971, 0.3 × 0.833 = 0.2499; Fourth step, sum to get the final FI value: FI = 0.3136 + 0.1971 + 0.2499 + 50 = 50.7606; Effect verification: If calculated solely based on EEG signals (ignoring behavioral indicators), then FI = 0.4 × 0.784 + 50 = 50.3136, which cannot reflect the user's response speed and accuracy during operation; if calculated solely based on behavioral data (ignoring EEG signals), then FI = 0.3 × 0.657 + 0.3 × 0.833 + 50 = 50.447, which lacks objective support from the physiological level. Both of these values ​​deviate significantly from the FI = 50.7606 calculated using the dual-dimensional fusion method, fully demonstrating that the calculation method in this case can more comprehensively and accurately quantify the user's level of concentration.

[0021] Step three, which involves matching the interaction program based on the state determination model, specifically includes: Feature extraction: Calculate the rate of change of the comprehensive attention index FI within the latest time window. The current comprehensive focus index (FI) and its rate of change Combined into attention state feature vector ; State determination: The attention state feature vector The input is fed into the state determination model; the state determination model is a classification model pre-trained based on historical training data, which internally stores feature center vectors for multiple attention state categories. , where j represents the state category index, and the state categories include at least dispersed state, stable state, and highly concentrated state; The calculation process for state determination is as follows: First, calculate the current attention state feature vector. With the feature center vector of each attention state category The Euclidean distance between them; Then, the distance is converted into similarity weights based on the Gaussian kernel function; finally, by normalizing the similarity weights of all categories, the probability Pj that the user currently belongs to the j-th state is obtained, and the probability calculation formula is as follows: ; in, Representing vectors and The Euclidean distance between them N is the total number of preset state categories, which is a model parameter used to control the smoothness of the probability distribution. Program matching: Select the state with the highest probability value Pj as the dominant state, and select and start the corresponding attention training interactive program from the preset content library according to the preset "state-program" mapping relationship. By integrating the real-time value of the comprehensive focus index with its temporal rate of change to form an attention state feature vector, and using a state judgment model pre-trained based on historical training data, the probability of a user belonging to a scattered state, stable state, or highly focused state is accurately quantified through Euclidean distance calculation, Gaussian kernel function similarity conversion, and probability normalization. This solves the problems of traditional training programs lacking specificity and relying on a single indicator, leading to adaptation bias. Based on the "state-program" mapping relationship, the corresponding interactive program is called, allowing the training content to be accurately matched with the user's real-time focus state. This not only improves the fun and adaptability of the training but also ensures its scientific nature and effectiveness, providing core technical support for personalized focus training.

[0022] The visual stimulation training scenario (numbers flash randomly on the screen, and the user needs to quickly click on odd numbers) has the following data and parameter settings: The baseline data remains the same: the current overall focus index FI is calculated to be 50.7606; the latest time window is set to 10 seconds, and the overall focus index 10 seconds ago is... Then the rate of change This forms the attention state feature vector. ; State determination model parameters: The preset total number of state categories N=3 (representing dispersed state j=1, stable state j=2, and highly concentrated state j=3 respectively), and the feature center vectors of each state stored internally in the model are... (Dispersed state: FI is low, and the rate of change is negative) (Steady state: FI is moderate, and the rate of change is gradual) (Highly concentrated state: FI is high, and the rate of change is positive), the parameter σ=2 controls the smoothness of the probability distribution; Euclidean distance calculation: Calculate the eigenvectors separately. With each state center vector The Euclidean distance is given by the formula: : With dispersed state Distance: ; With steady state Distance: ; With highly concentrated state Distance: ; Probability calculation: Substitute into the Gaussian kernel function probability formula, where : Molecular calculations: (P1 molecule); (P2 molecule); (P3 molecule); Denominator calculation: ; Probability results: P1=0.0157 / 1.0525≈0.015 (1.5), P2=0.9305 / 1.0525≈0.884 (88.4%), P3=0.1063 / 1.0525≈0.101 (10.1%). Program matching and effect verification: The stable state with the highest probability (P2=88.4%) was selected as the dominant state. Based on the preset "state-program" mapping relationship, the corresponding "suspended ball balance training program" was called (the suspended ball maintains its height when the user's concentration is stable, and does not fall when there are slight fluctuations). If this judgment method is not used and the "rapid number combo program" corresponding to the highly concentrated state is called blindly, the user will experience more operational errors and increased training frustration because the user's current concentration is not at a high level. The matching method in this case accurately matches the user's real-time state, effectively improving training participation and concentration enhancement effect.

[0023] In step four, the adaptive adjustment of the difficulty level specifically includes: Based on the user's comprehensive attention index sequence over multiple recent training cycles, a quantitative value Level representing their long-term attention level is calculated. An adaptive control algorithm is employed, based on the quantified value of long-term attention level (Level) and the current interactive program difficulty (Diff). now The appropriate theoretical level value The difference between the two is used to calculate and set the recommended difficulty level (Diff) for the next training cycle. next The specific calculation process is as follows: ; Where α is the adjustment step size coefficient, used to control the magnitude of difficulty adjustment; The difficulty of implementing adaptive control algorithms increases with the improvement of user capabilities; By calculating a quantitative value, Level, representing the long-term attention level based on the comprehensive attention index sequence of the user's recent continuous training cycles, and combining it with an adaptive control algorithm, the difficulty of subsequent training is dynamically adjusted according to the gap between Level and the theoretical level value that matches the current interactive program difficulty. This solves the problem of poor adaptability caused by fixed or manually adjusted training difficulty in traditional methods. It achieves precise increase in difficulty as the user's ability improves, avoiding situations where the difficulty is too low and lacks training effect, or too high and causes frustration. It also ensures the targeting and continuity of training, allowing attention levels at different stages to receive matched reinforcement training and improving overall training efficiency.

[0024] The visual stimulation training scenario (numbers flash randomly on the screen, and the user needs to quickly click on odd numbers) has the following data and parameter settings: Basic data continuation and supplement: The user's comprehensive focus index for the recent 5 consecutive training cycles (10 minutes each) is as follows: ( (The current period's FI value is calculated previously). The long-term focus level quantification value, Level, is taken as the average value of this sequence (representing a stable long-term level). The result is: ; Difficulty-related parameter settings: Difficulty level of the current training interactive program (Diff) now =3 (Preset difficulty level range is 1-10, the higher the value, the greater the difficulty), the theoretical level value corresponding to this difficulty. (i.e., the ideal concentration index level that is suitable for difficulty level 3), adjust the step size coefficient α=0.5 (to control the adjustment range of difficulty and avoid excessive fluctuations); Next round of difficulty calculation: Substituting into the adaptive adjustment formula, we get: ; Because the difficulty level is an integer, the final Diff setting is... next =4; Effect verification: If this adaptive adjustment method is not used, continue using Diff. now =3, since the user's long-term Level=50.11212 is already higher than the theoretical level of 49.5 for this difficulty, the user can maintain a high accuracy rate without conscious concentration during training, lacking the effect of strengthening focus; blindly increasing the difficulty to Level 5 (Level 3) would be ineffective. target (5)=51), the user's current Level is not suitable, the actual response time is extended to 0.52s, the accuracy is reduced, and the frustration is obvious; while the Diff adjusted in this case next =4, corresponding to the theoretical level Level target(4) = 50.2, which is highly matched with user Level = 50.11212. During training, the user needs to maintain stable focus in order to maintain a high accuracy rate. It is challenging but not frustrating, and effectively promotes the continuous improvement of focus.

[0025] Step five, generating a report on user attention development and training effectiveness, specifically includes: A personal training database is established for each user to continuously store relevant data for each training session. The relevant data includes at least a timestamp, the comprehensive focus index, the identifier of the training interaction program, and the difficulty level. Based on the personal training database, a visual report is generated through data analysis. The report includes: the curve of the change of the comprehensive focus index over time, performance analysis at different difficulty levels, and personalized improvement suggestions based on training history. By establishing a personalized training database for each user, the system comprehensively retains core data such as training timestamps, overall focus index, training program identifiers, and difficulty levels. Based on data analysis, it generates visual reports that include index change curves, performance analysis at different difficulty levels, and personalized improvement suggestions. This solves the problems of traditional training, such as lack of systematic data retention, inability to quantify and track effects, and unclear improvement directions. It allows users and relevant personnel (such as parents and instructors) to intuitively grasp the development trajectory of focus and the suitability at different difficulty levels. Personalized suggestions are precisely tailored to the training history, providing a scientific basis for adjusting subsequent training plans. This helps users continuously optimize their training programs and improve the relevance and long-term effectiveness of focus training.

[0026] The visual stimulus training scenario (numbers flash randomly on the screen, and the user needs to quickly click on odd numbers), and the related data and report generation process are as follows: Personal training database construction: The system creates a dedicated account for each user, continuously storing training data for four consecutive weeks (three training sessions per week, one training cycle per session). Key data is as follows: Week 1, first training session: Overall focus index FI = 49.2, training program is number click training, difficulty level Diff = 3, average response time 0.42s, accuracy 0.80; Week 1, second training session: FI = 49.5, training program and difficulty level remain the same, average response time 0. 0.41s, accuracy 0.82; Week 1, 3rd training session, FI=49.8, program and difficulty remain the same (number click training, level 3), average response time 0.40s, accuracy 0.83; Week 2, 1st training session, FI=50.1, program and difficulty remain unchanged, average response time 0.39s, accuracy 0.85; Week 2, 2nd training session, FI=50.3, program and difficulty remain unchanged, average response time 0.38s, accuracy 0.86; Week 2, 3rd training session... Training session 1: FI=50.5, training program unchanged, difficulty level increased to 4, average response time 0.41s, accuracy 0.82; Week 3, first training session: FI=50.6, program is number click training, difficulty level 4, average response time 0.40s, accuracy 0.84; Week 3, second training session: FI=50.7, program and difficulty unchanged, average response time 0.39s, accuracy 0.85; Week 3, third training session: FI=50.9, program... With the same order and difficulty, the average response time was 0.38s and the accuracy was 0.87. In the first training session of week 4, FI=51.0, with the same order and difficulty, the average response time was 0.37s and the accuracy was 0.88. In the second training session of week 4, FI=51.2, with the same order and difficulty, the average response time was 0.36s and the accuracy was 0.89. In the third training session of week 4, FI=51.5, with the same order and difficulty, the average response time was 0.35s and the accuracy was 0.90. Visual report generation: The curve of the overall focus index change: The curve is generated with the timestamp as the horizontal axis and the FI value as the vertical axis. It can be seen that the FI steadily increased from 49.2 once in the first week to 51.5 three times in the fourth week, showing a continuous upward trend. In particular, after switching to difficulty level 4 three times in the second week, the FI stabilized briefly and then continued to climb, indicating that the difficulty was well adapted. Performance analysis at different difficulty levels: For difficulty level 3, a total of 5 training sessions were conducted, with an average FI of [missing value]. The average response time was 0.40s, and the average accuracy was 0.83. The training for difficulty level 4 consisted of 7 runs, with an average FI of [missing value]. The average response time is 0.38s and the average accuracy is 0.86. It can be seen that after the difficulty is increased, the user's FI value, response speed and accuracy continue to optimize, indicating that the current difficulty is within the appropriate range. Personalized improvement suggestions: Based on the curve trend and difficulty performance analysis, the report suggests that "continue to maintain difficulty level 4 training for the next 1-2 weeks to consolidate the stability of focus; if the FI value is stable above 51.5 in the 5th week of training, you can try to increase the difficulty to level 5 to further enhance the upper limit of focus; it is recommended to maintain a calm state for 10 minutes before training. According to historical data, the FI value of training after calming down is on average 0.3 higher, and the effect is better." Without this report generation mechanism, users would only perceive immediate feedback from a single training session, unable to know that their FI value improved by 2.3 over four weeks, nor would they understand the adaptation differences between difficulty levels 3 and 4. This could lead to blindly increasing or maintaining the original difficulty, resulting in low training efficiency. However, with the report generated in this case, users clearly understand their progress trajectory and clarify their subsequent training direction. After consolidating training at difficulty level 4 as recommended in week 5, their FI value stabilized at 51.6-51.8. When subsequently increasing to difficulty level 5, they quickly adapted and maintained high accuracy, fully demonstrating the scientific guiding role of the report in training optimization.

[0027] The method also includes device connection and interaction steps: The signal acquisition device establishes a connection with one or more display processing terminals via wireless communication. The attention training interactive program runs on the display processing terminal and provides the user with an interactive interface that includes real-time feedback of the comprehensive attention index; By supporting wireless communication connections between signal acquisition devices and multiple display processing terminals such as mobile phones, tablets, and TVs, the interactive focus training program can run flexibly on different terminals. It also provides an interactive interface with real-time feedback of comprehensive focus index, solving the problems of single connection, delayed data feedback, and lack of visibility in traditional devices. This not only improves the flexibility and convenience of the usage scenarios, but also allows users to instantly grasp changes in their focus state and adjust the training pace in a timely manner. Combined with the previous steps of state matching and difficulty adjustment, it further enhances the targeting and immediacy of training, improving the overall training experience and effectiveness.

[0028] The visual stimulus training scenario (numbers flash randomly on the screen, and the user needs to quickly click on odd numbers) involves the following data and device interaction process: Device connection: The EEG signal acquisition device (brain ring) worn by the user establishes a connection with both mobile phone (personal terminal) and TV (home terminal) via Bluetooth wireless communication. The signal quality is good, ensuring real-time transmission of EEG signals and behavioral interaction data without delay. Program execution and real-time feedback: When training at home, users can choose to run the number click training program on their TV (current difficulty). The mobile app simultaneously displays the interactive feedback interface, showcasing key data in real time. The current comprehensive focus index FI=50.7606, the index change rate ΔFI=0.09406, the current attention state (steady state, probability P2=88.4%), and the remaining training time (8 minutes) are presented intuitively in the form of numbers and dynamic curves. Scene switching and status adjustment: After 5 minutes of training, if the user needs to go out, they can directly close the TV program and open the same training program on their mobile phone. Due to the wireless connection, the data is seamlessly connected, and the mobile phone will instantly synchronize the current FI value, difficulty level Diff=4, training time and historical interaction data without re-initialization. While out and about, the user noticed on their phone that the FI value had dropped to 50.2 (ΔFI=-0.005), indicating a slight loss of focus. They then adjusted their state by taking deep breaths, and one minute later, the interface updated in real time that the FI had risen back to 50.8 (ΔFI=0.012), quickly restoring a stable state and preventing a continued decline in focus from affecting the training effect. Multi-terminal adaptation verification: If the user trains in the office, a wireless connection can be established between the brainband and the tablet. The tablet can also load the user's training data, historical FI change curves, and current adaptation difficulty in real time, achieving "training in one place, synchronized across multiple devices". Without this solution, the user can only train on a single device, and cannot continue the training process when away from home. Furthermore, the user cannot know the FI changes in real time, and cannot adjust in time when concentration is distracted, causing the average FI of the training cycle to drop to 50.1, thus reducing the training effect. With the solution in this case, the user can flexibly switch terminals according to the scenario and respond to changes in status in real time. The average FI of the cycle is maintained above 50.6, and the continuity and effectiveness of training are significantly improved.

[0029] A focus analysis and training device, the device comprising: The data acquisition unit is used to collect users' electroencephalographic signals and behavioral interaction data in real time. The index calculation unit is used to calculate and generate a comprehensive focus index based on the collected data. The content matching unit is used to match and invoke the attention training interactive program based on the index and the state judgment model. Difficulty adjustment unit is used to adaptively adjust the program difficulty based on historical performance data; The report generation unit is used to record data and generate training effect reports.

[0030] A storage medium storing a computer program, which, when executed by a processor, implements the steps of the method.

[0031] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for analyzing and training focus, characterized in that, Includes the following steps: Step 1, Data Acquisition Steps: Real-time acquisition of the user's electroencephalogram (EEG) signals and behavioral interaction data related to the training task; Step 2, Index Calculation Steps: Based on EEG physiological signals and behavioral interaction data, a comprehensive attention index is calculated and generated; Step 3, Content Matching Step: Based on the real-time value of the comprehensive focus index and its temporal change characteristics, and based on the state judgment model, match and call the corresponding focus training interactive program from the preset content library; Step 4, Difficulty Adjustment Step: Based on the user's attention performance data formed in the historical training cycle, adaptively adjust the difficulty level of the attention training interactive program in subsequent training. Step 5, Report Generation Steps: Record training process data and generate a report on user focus development and training effectiveness.

2. The concentration analysis and training method according to claim 1, characterized in that: In step one, the collection of electroencephalographic signals specifically involves: acquiring multi-channel electroencephalographic signals from the user's frontal and parietal lobes using a signal acquisition device worn on the head; behavioral interaction data includes the user's response time to visual or auditory stimuli and the accuracy of operation selection.

3. A method for attention analysis and training according to claim 3, characterized in that: Step two, specifically calculating and generating the comprehensive focus index FI, includes: Preprocessing and spectral analysis were performed on the EEG signals to extract the signal energy values ​​E in the theta, alpha, and beta bands, respectively. θ E α E β , i.e., physiological indicators; The behavioral interaction data is processed to obtain the normalized reaction speed index R and accuracy index A, which are the behavioral indicators. The aforementioned physiological and behavioral indicators are combined into a single comprehensive attention index FI using a weighted fusion formula.

4. The concentration analysis and training method according to claim 3, characterized in that: Step three, the interaction matching program based on the state determination model, specifically includes: Feature extraction: Calculate the rate of change of the comprehensive attention index FI within the latest time window. The current comprehensive focus index (FI) and its rate of change Combined into attention state feature vector ; State determination: The attention state feature vector The input is fed into the state determination model; the state determination model is a classification model pre-trained based on historical training data, which internally stores feature center vectors for multiple attention state categories. , where j represents the state category index, and the state categories include at least dispersed state, stable state, and highly concentrated state; The calculation process for state determination is as follows: First, calculate the current attention state feature vector. With the feature center vector of each attention state category The Euclidean distance between them; Then, the distance is converted into similarity weights based on the Gaussian kernel function; finally, by normalizing the similarity weights of all categories, the probability Pj that the user currently belongs to the j-th state is obtained. Program matching: Select the state with the highest probability value Pj as the dominant state, and select and start the corresponding attention training interactive program from the preset content library according to the preset state-program mapping relationship.

5. The concentration analysis and training method according to claim 4, characterized in that: Step four, the adaptive adjustment of the difficulty level specifically includes: Based on the user's comprehensive attention index sequence over multiple recent training cycles, a quantitative value Level representing their long-term attention level is calculated. An adaptive control algorithm is used, based on the quantified value of long-term attention level (Level) and the difficulty of the current interactive program (Diff). now The appropriate theoretical level value The difference between the two is used to calculate and set the recommended difficulty level (Diff) for the next training cycle. next ; The difficulty of implementing adaptive control algorithms increases with the improvement of user capabilities, resulting in adaptive adjustment.

6. The concentration analysis and training method according to claim 5, characterized in that: Step five, generating a report on the user's attention development and training effectiveness, specifically includes: Establish a personal training database for each user to continuously store relevant data for each training session. The relevant data includes at least timestamps, comprehensive focus index, identification of training interaction programs, and difficulty level. Based on a personal training database, a visual report is generated through data analysis. The report includes: a curve showing the change of the comprehensive focus index over time, performance analysis at different difficulty levels, and personalized improvement suggestions based on training history.

7. The concentration analysis and training method according to claim 6, characterized in that: The method also includes device connection and interaction steps: The signal acquisition device establishes a connection with one or more display processing terminals via wireless communication; The attention training interactive program runs on a display processing terminal and provides users with an interactive interface that includes real-time feedback on a comprehensive attention index.

8. A concentration analysis and training device, wherein the device is used in the training method of any one of claims 1-7, characterized in that: The device includes: The data acquisition unit is used to collect users' electroencephalographic signals and behavioral interaction data in real time. The index calculation unit is used to calculate and generate a comprehensive focus index based on the collected data. The content matching unit is used to match and invoke the attention training interactive program based on the index and the state judgment model. Difficulty adjustment unit is used to adaptively adjust the program difficulty based on historical performance data; The report generation unit is used to record data and generate training effect reports.

9. A storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images