Personalized virtual-real interactive cognitive game system based on ai adaptive brain cognitive enhancement

Abstract

The application discloses a personalized virtual-real interaction cognitive game system based on AI adaptive brain cognitive enhancement, comprising: a cognitive game design module for designing multiple cognitive tasks corresponding to attention, memory, executive function and perception function based on a dynamic difficulty adjustment algorithm according to user characteristics; the dynamic difficulty adjustment algorithm is used for dynamically adjusting the difficulty of a specific level task according to the behavior state and psychological state of the user; a cognitive task guiding module interacts with the user through a multi-modal guiding mechanism to guide the user to complete the cognitive task; a data tracking and analysis module is used for recording the performance and growth trajectory of the user in the completion process of each cognitive task; an instant feedback module provides feedback based on an instant feedback generation algorithm; a cognitive ability evaluation report module is used for generating a corresponding cognitive ability evaluation report according to the performance and data tracking and analysis algorithm; and a data storage module is used for storing the behavior data, emotion record and cognitive ability evaluation report of the user.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of AI adaptation, brain science and cognitive training, and particularly relates to an individualized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement. BACKGROUND

[0002] The adolescent period is an important stage of cognitive development for individuals, and cognitive ability training during this period has a profound impact on future development. Psychological research shows that adolescents experience rapid development of cognitive abilities during this period, especially in key cognitive skills such as attention, memory, executive function and perception. However, traditional cognitive training methods, such as classroom instruction and cognitive training software, while providing some help, are often monotonous and difficult to stimulate the interest and participation of adolescents, with the following shortcomings:

[0003] 1. Most cognitive training games lack systematic cognitive psychology design and cannot effectively guide adolescents in cognitive training;

[0004] 2. Lack of individualized feedback mechanism, unable to provide targeted guidance according to the behavior and psychological state of the player;

[0005] 3. Difficulty is fixed and cannot adapt to different players' cognitive level and psychological state. SUMMARY

[0006] The purpose of the present application is to provide an individualized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement, specifically a system for improving cognitive skills and cognitive abilities of adolescents based on scenario-based game design. The system, through structured task design, hierarchical difficulty system and standardized evaluation mechanism, combined with cognitive psychology principles, provides players with individualized cognitive training experience through game level design, task guidance, instant feedback, data tracking and evaluation reports, helping adolescents improve cognitive skills such as attention, memory, executive function and perception in a relaxed and enjoyable environment.

[0007] The purpose of the present application is to provide an individualized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement, including:

[0008] The cognitive game design module is used to design multiple specific level tasks based on user characteristics and a dynamic difficulty adjustment algorithm. These specific level tasks correspond to one or more cognitive tasks related to training in attention, memory, executive function, and perception. The cognitive tasks form a mapping with one or more of these functions. The user characteristics include the user's behavioral and psychological states. The dynamic difficulty adjustment algorithm dynamically adjusts the difficulty of the specific level tasks based on the user's behavioral and psychological states to ensure both challenge and adaptability.

[0009] The cognitive task guidance module interacts with the user through a multimodal guidance mechanism, thereby guiding the user to complete the cognitive task.

[0010] The data tracking and analysis module is used to record the user's performance and growth trajectory in the process of completing various cognitive tasks;

[0011] The real-time feedback module provides real-time feedback based on the user's performance in completing the cognitive task, using a real-time feedback generation algorithm to help the user understand their own cognitive abilities.

[0012] The cognitive ability assessment report module is used to generate a cognitive ability assessment report corresponding to the user's performance based on the user's performance and the data tracking and analysis algorithm. The cognitive ability assessment report also includes personalized cognitive training suggestions for the user. The data tracking and analysis algorithm is based on time series analysis and generates a detailed cognitive ability assessment report by analyzing the user's cognitive ability change trend. The cognitive ability assessment report includes task completion status, cognitive ability change trend and personalized training suggestions.

[0013] The data storage module is used to store the user's behavioral data, emotion records, and cognitive ability assessment report, so that the user's upstream and downstream stakeholders can refer to and analyze them.

[0014] Preferably, the algorithm for the dynamic difficulty adjustment is as follows: establishing a multi-objective optimization model based on reinforcement learning, and performing real-time parameter adjustment based on the multi-objective optimization model;

[0015] The input parameters of the multi-objective optimization model include:

[0016] (1) Behavioral state, which is characterized by: task response time, accuracy, operation trajectory accuracy and number of abandonments;

[0017] (2) Psychological state, which is characterized by: EEG focus index, eye movement anxiety frequency and heart rate variability (HRV).

[0018] Preferably, the task design workflow of the cognitive game design module includes:

[0019] S1, based on the data acquisition and fusion layer, implements data acquisition, data processing and data fusion. The data includes behavioral state and psychological state. The behavioral state is characterized by task response time, accuracy, operation trajectory precision and number of abandonments. The psychological state is characterized by EEG focus index, eye movement anxiety value, eye movement anxiety frequency and heart rate variability (HRV).

[0020] S2, based on the data fusion, uses a reinforcement learning model to evaluate the user's current state; the reinforcement learning model calculates the user's current state value S based on formula (1):

[0021] (1);

[0022] Where T is the task completion time, E is the error rate, F is the operation frequency, and α, β, and γ are the weighting coefficients of task completion time, error rate, and operation frequency, respectively (e.g., α=0.4, β=0.3, γ=0.3).

[0023] S3, Based on the dynamic difficulty adjustment algorithm, a generation difficulty coefficient K is obtained, which represents the difficulty of the specific level task dynamically adjusted according to the user's behavioral state and psychological state. At the same time, the difficulty of the specific level task is adjusted according to the user's current state value S to ensure the challenge and adaptability of the specific level task.

[0024] S4, using a task selector, cognitive function goals are selected based on the generated difficulty coefficient K and the rehabilitation goals; wherein, the selection of cognitive function goals includes adjusting the target tracking task parameters based on attention training goals, expanding the matrix memory complexity based on memory training goals, increasing the rule transformation frequency based on executive function training goals, and improving fuzzy recognition noise based on perceptual function training; then, based on the generated difficulty coefficient K, the number of task elements is generated based on the parameterized formula (2), which is:

[0025] (2);

[0026] The dynamic difficulty adjustment algorithm uses a parameterized formula. Generate task parameters, where N is the number of task elements, α' is the range scaling factor, β' is the cognitive load nonlinearity coefficient, and K is the generation difficulty coefficient.

[0027] S5, based on the cognitive function target selection, generate personalized level JSON and cognitive tasks, and load the cognitive tasks through the Unity engine;

[0028] S6. Update the model parameters of the reinforcement learning model once according to the rule of every fixed time interval to adapt to the user's long-term progress.

[0029] Preferably, the data processing includes:

[0030] A. Cross-modal adaptive processing, thereby supporting multi-device compatibility between brain-computer interfaces and VR controllers / eye controls;

[0031] B. Implement anti-fatigue adaptation measures, including: automatically switching to low-intensity tasks when the standard deviation of the heart rate variability (HRV) is >50ms;

[0032] C. Signal synchronization processing, including: the sampling period of the behavioral state is 100ms, the sampling frequency of the psychological state is 256Hz, and the timestamps of the behavioral state data and the psychological state data are aligned to form a synchronization signal;

[0033] D. Perform signal anomaly processing, including: when an eye movement anxiety value > 0.7 is detected, insert a 3-minute deep breathing guided task.

[0034] Preferably, the data fusion includes weighted fusion of the operation trajectory accuracy and the EEG focus index.

[0035] Preferably, the personalized level JSON and cognitive task in S4 are dynamically constructed using a programmatic content generation algorithm, including the following steps:

[0036] (1) Input seed parameters: user cognitive ability index V and training target weight W;

[0037] (2) Execute the generated rule set:

[0038] Spatial layout rule: Map = Voronoi_partition(V 1.2 );

[0039] Task complexity rules: ;

[0040] Output personalized level examples: Ensure that each generated cognitive challenge is unique.

[0041] Preferably, S4 further includes:

[0042] A long-term adaptive mechanism that automatically increases the proportion of executive function tasks based on weekly training data to incorporate the number of rehabilitation goals; and a dynamic downgrading mechanism that triggers emergency intervention if the accuracy of the selection of the cognitive function goals is less than 30% for two consecutive times, wherein the dynamic downgrading mechanism includes reducing the generation difficulty coefficient K to K×0.6 as the new generation difficulty coefficient.

[0043] Preferably, the cognitive task guidance module's operation includes:

[0044] S1. Interact with the user based on a multimodal guidance mechanism, wherein the multimodal guidance mechanism includes a combination of one or more of the following: cognitive task prompts, cognitive task selection questions, cognitive task guidance text input boxes, or cognitive task guidance image input boxes; wherein, the cognitive task prompts include: dynamically generating AR visual markers and overlaying them onto the VR scene; the cognitive task selection questions include: providing a cognitive task selector through voice broadcast and on-screen options, wherein the number of options in the voice broadcast and on-screen dynamically increases or decreases with the value of the generation difficulty coefficient K; the cognitive task guidance text input boxes or cognitive task guidance image input boxes include: performing functional training through handwritten formula recognition or performing perceptual training through noisy image annotation.

[0045] S2, start and execute the guidance logic engine, wherein the task type determines whether it is an attention task. If it is, the AR arrow guides the movement target. If not, it determines whether it is a memory task. If it is, the step-by-step voice prompt memory strategy is given. If not, the graphic instruction decomposition rule is determined. When the guidance logic engine is executed, the eye movement trajectory is monitored in real time. When the gaze deviation is greater than 3 seconds, the reinforcement prompt is triggered.

[0046] S3, dynamically adapt the difficulty in cognitive task guidance, including: enabling layered guidance when the error rate is >40%, the layered guidance includes text prompts to animation demonstrations and simplified task practice; switching to voice-stimulated guidance when EEG focus (θ / β) is <0.5;

[0047] S4, feeding back user interaction data to the bootstrap logic engine to form a data loop, thereby optimizing subsequent bootstrap strategies; wherein the user interaction data includes response latency and / or input accuracy.

[0048] Preferably, the working method of the data tracking and analysis module includes:

[0049] Collect data for tracking and analysis, including: collecting the user's historical behavioral data and cognitive ability assessment results;

[0050] The analysis of cognitive ability change trends includes: using a time series analysis model to extract cognitive ability change trends, wherein the time series analysis model calculates the trend value T based on the following formula (4):

[0051] (4);

[0052] Among them, S i For the score of the i-th evaluation, The average score is N, where N is the number of evaluations.

[0053] Preferably, the instant feedback generation algorithm is based on natural language processing technology, generating personalized feedback content according to the user's performance, and combining cognitive psychology theory to help users understand their cognitive abilities and provide improvement suggestions; wherein, the instant feedback generation algorithm includes:

[0054] Perform real-time data analysis: Analyze users' task completion status and extract key indicators, including completion time and error rate;

[0055] Generate instant feedback: Combining cognitive psychology theories, it generates positive incentives and improvement suggestions;

[0056] Adjust tone: Adjust the feedback tone according to the user's emotional state;

[0057] Generate natural language: Use a pre-trained language model to generate natural and fluent personalized feedback content.

[0058] The beneficial effects of the system of the present invention:

[0059] Through a series of scenario-based game tasks, combined with dynamic difficulty adjustment algorithms, real-time feedback generation algorithms, and data tracking and analysis algorithms, this helps teenage players improve their cognitive skills, such as attention, memory, executive function, and perception. Specific beneficial effects include:

[0060] (1) The dynamic difficulty adjustment algorithm is based on the reinforcement learning model. By analyzing the user's behavioral data (such as task completion time, error rate, operation frequency) and emotional state (such as focus, stress level), the game difficulty is dynamically adjusted to ensure that the player trains at a suitable level of challenge.

[0061] (2) The “psychological-behavioral double closed-loop feedback” forms a closed-loop feedback regulation effect in dynamic difficulty adjustment, which is closer to the real cognitive interaction situation.

[0062] (3) The interactive cognition is quantitatively described by the mathematical expression of the “cognitive task parameterization generation formula”, which makes the operation of the game system more reasonable.

[0063] (4) By forming a "real-time fusion of multimodal data" architecture, it is possible to achieve the technical effect of personalized adaptation.

[0064] (5) Deep integration of scenario-based game design and cognitive tasks: Through carefully designed game scenarios and task guidance, the principles of cognitive psychology are deeply integrated with scenario-based game design. Game scenarios such as "Rolling Ball Journey", "Lemon Match" and "Guardian of the Lighthouse" are not only visually appealing, but also help teenagers gradually improve their cognitive abilities through task guidance and instant feedback.

[0065] (6) Dynamic Difficulty Adjustment Algorithm: The system includes a dynamic difficulty adjustment algorithm that dynamically adjusts the game difficulty based on the player's behavior and psychological state. This design ensures the game's challenge and adaptability, allowing players of different skill levels to play at a difficulty level that suits them. For example, in the "Rolling Ball Journey" level, the average time players spend completing tasks is reduced by 23%, and the error rate is reduced by 18%.

[0066] (7) Real-time feedback generation algorithm: The system provides real-time feedback and evaluation based on the player's performance in the game and generates an evaluation report. The feedback content not only includes the task completion status, but also analyzes the player's cognitive ability in combination with the principles of cognitive psychology and provides targeted suggestions. For example, in the "Lemon Match" level, the player's accuracy rate increased by an average of 28%, and the task completion time was shortened by 15%.

[0067] (8) Data Tracking and Analysis Algorithm: The system records the player's performance and growth trajectory in each level through a data tracking and analysis algorithm, generating an evaluation report. This report can help players better understand their cognitive development and also provide a reference for parents and educators. For example, the adoption rate of training suggestions increased by 35%. Attached Figure Description

[0068] To more clearly illustrate the technical solutions in the specific embodiments or related technologies of the present invention, the drawings used in the description of the specific embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0069] Figure 1 This is a schematic diagram of the task design workflow of the cognitive game design module provided in an embodiment of the present invention;

[0070] Figure 2 This is a schematic diagram of the data flow architecture of a cognitive game design module provided according to an embodiment of the present invention;

[0071] Figure 3 This is a flowchart illustrating the attention and memory tasks of the guidance logic engine provided according to an embodiment of the present invention. Detailed Implementation

[0072] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0073] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0074] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0075] This embodiment provides a personalized virtual-real interactive cognitive game system based on AI adaptive brain cognition enhancement, including:

[0076] (1) A cognitive game design module, used to design multiple specific level tasks based on user characteristics and a dynamic difficulty adjustment algorithm. The multiple specific level tasks correspond to one or more cognitive tasks related to the training of attention, memory, executive function and perception function. The cognitive tasks form a mapping with one or more of attention, memory, executive function and perception function. The user characteristics include the user's behavioral state and psychological state. The dynamic difficulty adjustment algorithm is used to dynamically adjust the difficulty of the specific level tasks according to the user's behavioral state and psychological state to ensure the challenge and adaptability of the specific level tasks.

[0077] In this embodiment, the core of the dynamic difficulty adjustment algorithm is: to establish a multi-objective optimization model based on reinforcement learning, and to perform real-time parameter adjustment based on the multi-objective optimization model; the dynamic difficulty adjustment algorithm stabilizes the task success rate at 65%-80% (golden training range) to achieve accurate matching.

[0078] The input parameters of the multi-objective optimization model include:

[0079] (1) Behavioral state, which is characterized by: task response time, accuracy, operation trajectory accuracy and number of abandonments;

[0080] (2) Psychological state, which is characterized by: EEG focus index (EEG θ / β wave ratio), eye movement anxiety frequency (EM) and heart rate variability (HRV).

[0081] The algorithm flow of the dynamic difficulty adjustment algorithm:

[0082] def DDA_Algorithm(user_state, task_history):

[0083] # 1. Competency Assessment Model

[0084] skill_level = α * (0.6*accuracy + 0.2*speed + 0.2*EEG_focus) - β* EM_stress

[0085] # 2. Dynamic Difficulty Level Generation

[0086] if skill_level > threshold_high:

[0087] difficulty = base_difficulty * (1 + γ * tanh(skill_level)) # Increase the challenge

[0088] elif skill_level < threshold_low:

[0089] difficulty = base_difficulty * (1 - δ * (1 - sigmoid(skill_level))) # Reduce difficulty

[0090] else:

[0091] difficulty = base_difficulty

[0092] # 3. Real-time parameter tuning (based on reinforcement learning)

[0093] reward = calculate_reward(accuracy, EEG_focus)

[0094] difficulty += η * (reward - expected_reward) # η is the learning rate

[0095] return clamp(difficulty, min=0.1, max=0.9) # Limit the difficulty range

[0096] As a preferred embodiment, the mapping between the cognitive task formation and one or more of the functions of attention, memory, executive function and perception includes the mapping relationship between the cognitive task and one or more of the functions of attention, memory, executive function and perception shown in Table 1 below.

[0097] Table 1

[0098] Cognitive function Specific task name Task mechanism Difficulty adjustment parameter Attention Dynamic target tracking Locking color-changing target in moving background Target number / moving speed / background interference complexity Interference suppression challenge Ignoring flickering distractors and clicking specified icons Distractor density / effective target similarity Memory Spatial matrix memory Reproduce after memory flash grid sequence Grid size (3x3→7x7) / sequence length Multimodal associative memory Pairing sound-image-text combinations Pairing number / distractor similarity Executive function Rule switching task Switching classification rules according to color changes (e.g. red by shape / blue by number) Rule switching frequency / conflict item ratio Resource allocation decision Allocating resources to multiple target points within limited time Time pressure / target conflict degree Perception function Integration of space-time perception Estimating the arrival time of moving objects and intercepting them Object speed / trajectory nonlinearity Fuzzy pattern recognition Identifying hidden objects in noisy images Signal-to-noise ratio / feature ambiguity degree

[0099] As shown in Figure 1, the task design workflow of the cognitive game design module includes:

[0100] S1, based on the data acquisition and fusion layer, implements data acquisition, data processing and data fusion. The data includes behavioral state and psychological state. The behavioral state is characterized by task response time, accuracy, operation trajectory precision and number of abandonments. The psychological state is characterized by EEG focus index, eye movement anxiety value, eye movement anxiety frequency and heart rate variability (HRV).

[0101] In this embodiment, the data processing includes:

[0102] A. Cross-modal adaptive processing, thereby supporting multi-device compatibility between brain-computer interfaces (BCI) and VR controllers / eye controls;

[0103] B. Implement anti-fatigue adaptation measures, including: automatically switching to low-intensity tasks when the standard deviation of the heart rate variability (HRV) is >50ms;

[0104] C. Signal synchronization processing, including: the sampling period of the behavioral state is 100ms, the sampling frequency of the psychological state is 256Hz, and the timestamps of the behavioral state data and the psychological state data are aligned to form a synchronization signal;

[0105] D. Perform signal anomaly processing, including: when an eye movement anxiety value > 0.7 is detected, insert a 3-minute deep breathing guided task.

[0106] In this embodiment, the data fusion includes weighted fusion of the operation trajectory accuracy (measured value 0.92) and the EEG focus index (measured value 0.78). Of course, other data weighted fusions can also be performed, all of which are within the protection scope of this invention.

[0107] S2, based on the data fusion, uses a reinforcement learning model to evaluate the user's current state; the reinforcement learning model calculates the user's current state value S based on formula (1):

[0108] (1);

[0109] Where T is the task completion time, E is the error rate, F is the operation frequency, and α, β, and γ are the weighting coefficients of task completion time, error rate, and operation frequency, respectively (e.g., α=0.4, β=0.3, γ=0.3).

[0110] S3, Based on the dynamic difficulty adjustment algorithm, a generation difficulty coefficient K is obtained, which represents the difficulty of the specific level task dynamically adjusted according to the user's behavioral state and psychological state. At the same time, the difficulty of the specific level task is adjusted according to the user's current state value S to ensure the challenge and adaptability of the specific level task.

[0111] S4, using a task selector, cognitive function goals are selected based on the generated difficulty coefficient K and the rehabilitation goals; wherein, the selection of cognitive function goals includes adjusting the target tracking task parameters based on attention training goals, expanding the matrix memory complexity based on memory training goals, increasing the rule transformation frequency based on executive function training goals, and improving fuzzy recognition noise based on perceptual function training; then, based on the generated difficulty coefficient K, the number of task elements is generated based on the parameterized formula (2), which is:

[0112] (2);

[0113] The dynamic difficulty adjustment algorithm uses a parameterized formula. Generate task parameters, where N is the number of task elements, α' is the range scaling factor, β' is the cognitive load nonlinearity coefficient, and K is the generation difficulty coefficient.

[0114] This mathematical form is the core means to solve the technical problem of "personalized difficulty adaptation". As an example, the parameter relationship α'=5,β'=1.5 is used in the common application scenario of cognitive training, that is, equation (3):

[0115] Number of rehabilitation goals = ceil(5*K) 1.5 (3);

[0116] ceil(5*K 1.5 ) is a parameterized generation formula used to quantify the dynamic difficulty coefficient K into specific parameters of the game task (such as the number of objectives, sequence length, etc.). Its mathematical meaning and technical logic are as follows:

[0117] Mathematical decomposition:

[0118] 1. K 1.5 (Difficulty increases non-linearly)

[0119] K is the dynamic difficulty coefficient (usually ranging from 0.1 to 0.9).

[0120] The effect of the exponent 1.5: When K is small (e.g., 0.2), K... 1.5 =0.2 1.5 ≈0.089 → Suppresses the increase in difficulty;

[0121] When K is large (e.g., 0.8), K 1.5 =0.8 1.5 ≈0.716 → Accelerates the increase in difficulty;

[0122] Technical significance: A "progressive challenge curve" that matches human cognitive abilities (flat at low levels, steep at high levels).

[0123] 2.5*K 1.5 (Parameter range scaling)

[0124] Multiplier 5 normalizes K 1.5 Mapping to the actual task parameter range (e.g., number of targets 1-15):

[0125] K = 0.1 → 5 × 0.0316 ≈ 0.158;

[0126] K=0.9→5×0.853≈4.2655×0.853≈4.265;

[0127] 3. (Round up)

[0128] Ensure the output is an integer (e.g., the target quantity cannot be a decimal). For example:

[0129] ceil(0.158)=1;

[0130] ceil(4.265)=5ceil(4.265)=5.

[0131] Examples of the technology's applications are shown in Table 2 below (dynamic target tracking task).

[0132] User ability level (K) Calculation process Number of generated targets Cognitive load regulation logic Novice (K=0.2) ceil(5 x 0.2 1.5 = 1 1 target Avoiding attention overload Intermediate (K=0.5) ceil(5 x 0.5 1.5 = 3 3 targets Moderately challenging working memory capacity Advanced (K=0.8) ceil(5 x 0.8 1.5 = 5 5 targets Stimulating the multitasking potential of executive function

[0133] The basis for designing this quantitative formula:

[0134] 1. Cognitive psychology basis:

[0135] (1) Stevens Power Law: Human perception of stimulus intensity conforms to ψ(I)=kIn;

[0136] (2) The subjective intensity model of “visual attention load” with an exponent n=1.5 (experimental fit value).

[0137] 2. Algorithm Advantages

[0138] Anti-oscillation: Avoid jumps in difficulty (such as going directly from 2 to 5);

[0139] Boundary control: The parameters are constrained within the effective range (1~15) by ceil and multiplier 5.

[0140] Monotonicity: Ensures that the task difficulty increases strictly as K increases.

[0141] 3. Scalability (as an alternative solution)

[0142] Replacing the multiplier (e.g., changing 5 to 8) can adapt to different task types; and

[0143] Adjusting the index (e.g., K for memory tasks) 2.0 Matching functionality features.

[0144] The rehabilitation goals are used to dynamically allocate the weights of multiple training goals, thereby determining the training emphasis corresponding to different rehabilitation goals. For example, ADHD patients focus on attention training goals, so they are given a weight of 70%.

[0145] S5. Based on the cognitive function target selection, generate personalized level JSON and cognitive tasks, and load the cognitive tasks through the Unity engine.

[0146] In a preferred embodiment, the personalized level JSON and cognitive task in step S4 are dynamically constructed using a procedural content generation (PCG) algorithm, including the following steps:

[0147] (1) Input seed parameters: user cognitive ability index V and training target weight W;

[0148] (2) Execute the generated rule set:

[0149] Spatial layout rule: Map = Voronoi_partition(V 1.2 );

[0150] Task complexity rules: ;

[0151] Output personalized level examples: Ensure that each generated cognitive challenge is unique.

[0152] In this embodiment, the specific core concepts of the programmatic content generation algorithm are explained as follows:

[0153] (1) "Programmatic": refers to the automatic generation of content through predefined algorithm rules (not manually designed); for example: Perlin noise algorithm generates terrain, and L system syntax generates vegetation.

[0154] (2) "Content": encompasses digital assets such as game levels, missions, maps, textures, and story branches;

[0155] (3) "Generate": Dynamically create an infinite number of variants (e.g., 1 algorithm → 1000 level configurations).

[0156] Based on the aforementioned cognitive function target selection and procedural content generation (PCG) algorithm, a single task can generate more than 1,000 variations, achieving the technical goal of generating diversity.

[0157] S6. Update the model parameters of the reinforcement learning model once according to the rule of every fixed time interval to adapt to the user's long-term progress.

[0158] In this embodiment, the model parameters of the reinforcement learning model are updated every 10 seconds to adapt to the user's long-term progress.

[0159] like Figure 2 As shown, the data flow architecture of the cognitive game design module is as follows:

[0160] Starting from the user's end, real-time data flows into the status monitoring module, which outputs preprocessed signals to the AI ​​decision engine. At the same time, the user's historical database sends transfer learning data to the AI ​​decision engine. Subsequently, the AI ​​decision engine inputs difficulty instructions to the level generator and an assessment report to the therapist's dashboard. Then, the level generator inputs task parameters to the game rendering engine, which returns interactive feedback to the user's end.

[0161] In a preferred embodiment, S4 further includes:

[0162] A long-term adaptation mechanism that automatically increases the proportion of functional tasks based on weekly training data (e.g., +5% per week in this embodiment) to incorporate the number of rehabilitation goals; and a dynamic downgrading mechanism that triggers emergency intervention if the accuracy of the selection of the cognitive function goals is <30% for two consecutive times, wherein the dynamic downgrading mechanism includes reducing the generation difficulty coefficient K to K×0.6 as the new generation difficulty coefficient.

[0163] (2) The cognitive task guidance module interacts with the user through a multimodal guidance mechanism to guide the user to complete the cognitive task;

[0164] In a preferred embodiment, the operation of the cognitive task guidance module includes:

[0165] S1. Interact with the user based on a multimodal guidance mechanism, which includes a combination of one or more of the following: cognitive task prompts, cognitive task selection questions, cognitive task guidance text input boxes, or cognitive task guidance image input boxes. The cognitive task prompts include dynamically generating AR visual markers (such as highlighting target areas) and overlaying them onto the VR scene. The cognitive task selection questions include providing a cognitive task selector through voice prompts and on-screen options, wherein the number of voice prompts and on-screen options dynamically increases or decreases with the value of the generation difficulty coefficient K. The cognitive task guidance text input boxes or cognitive task guidance image input boxes include performing functional training through handwritten formula recognition or performing perceptual training through noisy image annotation.

[0166] S2, starts and executes the boot logic engine, such as Figure 3 The diagram shows the workflow of the guidance logic engine, taking attention and memory tasks as examples. It determines whether the task is an attention task based on the task type. If it is, the AR arrow guides the movement target. If it is not, it determines whether it is a memory task. If it is, the step-by-step voice prompt memory strategy is given. If it is not, the graphic instruction decomposition rule is determined. When the guidance logic engine is executed, the eye movement trajectory is monitored in real time. When the gaze deviation is greater than 3 seconds, the reinforcement prompt is triggered.

[0167] S3, Dynamic difficulty adaptation in cognitive task guidance, including: enabling layered guidance when the error rate is >40%, the layered guidance includes text prompts to animation demonstrations and simplified task practice; switching to voice-stimulated guidance when EEG focus (θ / β) is <0.5;

[0168] S4 feeds back user interaction data (response latency / input precision) to the bootstrapping logic engine to form a data loop, thereby optimizing subsequent bootstrapping strategies with an error convergence rate of >90%.

[0169] (3) Data tracking and analysis module, used to record the user's performance and growth trajectory in the process of completing various cognitive tasks;

[0170] In a preferred embodiment, the operation method of the data tracking and analysis module includes:

[0171] Collect data for tracking and analysis, including: collecting the user's historical behavioral data and cognitive ability assessment results;

[0172] The analysis of cognitive ability change trends includes: using a time series analysis model to extract cognitive ability change trends, wherein the time series analysis model calculates the trend value T based on the following formula (4):

[0173] (4);

[0174] Among them, S i For the score of the i-th evaluation, The average score is N, where N is the number of evaluations.

[0175] (4) Real-time feedback module: Based on the user’s performance in completing the cognitive task, it provides real-time feedback based on the real-time feedback generation algorithm, thereby helping the user understand their own cognitive abilities.

[0176] In a preferred embodiment, the instant feedback generation algorithm is based on natural language processing technology, generating personalized feedback content according to the user's performance, and combining cognitive psychology theory to help users understand their cognitive abilities and provide improvement suggestions; wherein, the instant feedback generation algorithm includes:

[0177] Perform real-time data analysis: Analyze users' task completion status and extract key metrics (such as completion time and error rate).

[0178] Generate instant feedback: Combining cognitive psychology theories, it generates positive incentives and improvement suggestions;

[0179] Adjust the tone of voice: Adjust the tone of feedback according to the user's emotional state (such as stress level);

[0180] Generate natural language: Use a pre-trained language model to generate natural and fluent personalized feedback content.

[0181] (5) Cognitive ability assessment report module, used to generate a cognitive ability assessment report corresponding to the user's performance based on the user's performance and data tracking analysis algorithm. The cognitive ability assessment report also includes personalized cognitive training suggestions for the user. The data tracking analysis algorithm is based on time series analysis and generates a detailed cognitive ability assessment report by analyzing the user's cognitive ability change trend. The cognitive ability assessment report includes task completion status, cognitive ability change trend and personalized training suggestions.

[0182] (6) Data storage module, used to store user behavior data, emotion records and the cognitive ability assessment report, so that the user's upstream and downstream stakeholders (e.g., parents and educators) can refer to and analyze them.

[0183] Application Examples

[0184] 1. Game level design module

[0185] The game level design module uses multiple level tasks to train attention, memory, executive function, and perception. Each level guides players through cognitive training through different task formats (such as the rolling ball challenge, lemon matching, and protecting the lighthouse). For example: Rolling Ball Challenge: By controlling the rolling ball to stay in the center of the ring, players train concentration and coordination; Lemon Match: By clicking on lemon slices, players train responsiveness; Protecting the Lighthouse: By controlling the cursor to stay in the target area, players train attention allocation, coordination, and spatial awareness.

[0186] 2. Dynamic Difficulty Adjustment Algorithm

[0187] The dynamic difficulty adjustment algorithm is based on a reinforcement learning model. It dynamically adjusts the game difficulty by analyzing player behavior data and emotional states. The specific implementation is as follows:

[0188] Inputs: Player behavioral data (such as task completion time, error rate, and operation frequency) and emotional state (such as focus and stress level);

[0189] Output: Adjusted game difficulty parameters;

[0190] Algorithm steps:

[0191] (1) Data collection: Collect players' behavioral data and emotional state in real time. For example, in the "Rolling Ball Journey" level, record the player's operation frequency (number of operations per second) and error rate (number of times the rolling ball deviates from the center of the ring);

[0192] (2) State Evaluation: The player's current state is evaluated using a reinforcement learning model. The model calculates the player's current state value S based on the following formula: Where T is the task completion time, E is the error rate, F is the operation frequency, and α, β, and γ are weighting coefficients (e.g., α=0.4, β=0.3, γ=0.3).

[0193] (3) Difficulty Adjustment: Adjust the game difficulty based on the state value S. For example, when S > 0.8, increase the game difficulty; when S < 0.2, decrease the game difficulty.

[0194] (4) Model update: The model parameters are updated every 10 seconds to adapt to the long-term progress of players.

[0195] Technical results: Through dynamic difficulty adjustment, players' average task completion time in the "Rolling Ball Journey" level was reduced by 23%, and the error rate was reduced by 18%.

[0196] 3. Real-time feedback generation algorithm

[0197] The real-time feedback generation algorithm is based on natural language processing technology and generates personalized feedback content based on the player's performance. The specific implementation is as follows:

[0198] Input: Player's task completion status and cognitive ability assessment results;

[0199] Output: Personalized feedback content;

[0200] Algorithm steps:

[0201] (1) Data analysis: Analyze the players' task completion status and extract key indicators (such as completion time and error rate). For example, in the "Lemon Match" level, record the players' clicking speed (number of clicks per second) and accuracy (percentage of correct clicks);

[0202] (2) Feedback Generation: Combining cognitive psychology theory, positive incentives and improvement suggestions are generated. For example, when a player's accuracy rate is below 70%, the system will prompt: "Your adaptability needs improvement. Try to identify targets faster!";

[0203] (3) Tone Adjustment: Adjust the feedback tone according to the player's emotional state (such as stress level). For example, when a player is detected to be under high stress, use a gentler tone: "You did a great job, keep up the good work!";

[0204] (4) Natural language generation: Use pre-trained language models to generate natural and fluent feedback content.

[0205] Technical Results: Through real-time feedback, players' accuracy in the "Lemon Match" level increased by an average of 28%, and the task completion time was reduced by 15%.

[0206] 4. Data tracking and analysis algorithms

[0207] The data tracking and analysis algorithm is based on time series analysis. By analyzing the changing trends of players' cognitive abilities, it generates detailed evaluation reports. The specific implementation is as follows:

[0208] Input: Player's historical behavioral data and cognitive ability assessment results;

[0209] Output: Report on trends in cognitive abilities;

[0210] Algorithm steps:

[0211] (1) Data collection: Collect players' historical behavioral data and cognitive ability assessment results. For example, record players' attention allocation ability score (110 points) in the "Guard the Lighthouse" level;

[0212] (2) Trend Analysis: A time series analysis model was used to extract the trend of cognitive ability changes. The model calculates the trend value T based on the following formula:

[0213] ;

[0214] Among them, S i For the score of the i-th evaluation, The average score is N, where N is the number of evaluations.

[0215] (3) Report generation: Generate detailed assessment reports, including task completion status, trends in cognitive ability changes, and personalized training suggestions;

[0216] (4) Parent Reference: Provide reference for parents and educators to help them understand the cognitive development of players.

[0217] Technical effects: Through data tracking and analysis, parents and educators are able to understand changes in players' cognitive abilities more accurately, and the adoption rate of training suggestions has increased by 35%.

[0218] Game Examples

[0219] Level 1: The Rolling Ball Journey (Pay attention to concentration and coordination)

[0220] (1) Game scene and task guidance:

[0221] After entering the Rolling Ball Journey scene, the system guides players through task prompts, helping them keep the ball in the center of the ring. Players record their actions via input boxes, and the system gradually unlocks more scenes from the Rolling Ball Journey based on the player's responses, providing real-time feedback.

[0222] (2) Dynamic difficulty adjustment:

[0223] The system analyzes the player's operation frequency and error rate through a dynamic difficulty adjustment algorithm, and adjusts the ring's movement speed and complexity in real time. For example, when the player's operation frequency is less than 2 times / second, the system will reduce the ring's movement speed; when the operation frequency is more than 5 times / second, the system will increase the ring's complexity.

[0224] (3) Immediate feedback and evaluation

[0225] Upon completing the task, players receive a "Concentration" badge, and the system provides immediate feedback emphasizing the importance of concentration. For example, the system might prompt the player: "Your concentration ability has improved, keep it up!" This feedback mechanism helps players better understand their cognitive abilities.

[0226] Level 2: Lemon Matching (Adaptability)

[0227] (1) Task design and goal setting:

[0228] Players train their reaction skills by completing lemon matching tasks. The system guides players to click on lemon slices through task prompts and provides real-time feedback based on the player's performance.

[0229] (2) Dynamic difficulty adjustment

[0230] The system analyzes players' clicking speed and accuracy using a dynamic difficulty adjustment algorithm, and adjusts the frequency and position of lemon slices in real time. For example, when a player's clicking speed is less than 3 times / second, the system reduces the frequency of lemon slice appearances; when the clicking speed is greater than 6 times / second, the system increases the complexity of the lemon slice appearance positions.

[0231] (3) Immediate feedback and evaluation

[0232] After completing a task, players receive an "Adaptability" badge, highlighting the importance of adaptability. For example, the system might prompt the player: "Your adaptability has improved; keep it up!" This feedback mechanism helps players build confidence and enhance their cognitive abilities.

[0233] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement, characterized in that, include: The cognitive game design module is used to design multiple specific level tasks based on user characteristics and a dynamic difficulty adjustment algorithm. These specific level tasks correspond to one or more cognitive tasks related to training in attention, memory, executive function, and perception. The cognitive tasks form a mapping with one or more of these functions. The user characteristics include the user's behavioral and psychological states. The dynamic difficulty adjustment algorithm dynamically adjusts the difficulty of the specific level tasks based on the user's behavioral and psychological states to ensure both challenge and adaptability. The cognitive task guidance module interacts with the user through a multimodal guidance mechanism, thereby guiding the user to complete the cognitive task. The data tracking and analysis module is used to record the user's performance and growth trajectory in the process of completing various cognitive tasks; The real-time feedback module provides real-time feedback based on the user's performance in completing the cognitive task, using a real-time feedback generation algorithm to help the user understand their own cognitive abilities. The cognitive ability assessment report module is used to generate a cognitive ability assessment report corresponding to the user's performance based on the user's performance and the data tracking and analysis algorithm. The cognitive ability assessment report also includes personalized cognitive training suggestions for the user. The data tracking and analysis algorithm is based on time series analysis and generates a detailed cognitive ability assessment report by analyzing the user's cognitive ability change trend. The cognitive ability assessment report includes task completion status, cognitive ability change trend and personalized training suggestions. The data storage module is used to store the user's behavioral data, emotion records, and cognitive ability assessment report, so that the user's upstream and downstream stakeholders can refer to and analyze them. The task design workflow of the cognitive game design module includes: S1, based on the data acquisition and fusion layer, implements data acquisition, data processing and data fusion. The data includes behavioral state and psychological state. The behavioral state is characterized by task response time, accuracy, operation trajectory precision and number of abandonments. The psychological state is characterized by EEG focus index, eye movement anxiety value, eye movement anxiety frequency and heart rate variability (HRV). S2, based on the data fusion, uses a reinforcement learning model to evaluate the user's current state; the reinforcement learning model calculates the user's current state value S based on formula (1): （1）； Where T is the task completion time, E is the error rate, F is the operation frequency, and α, β, and γ are the weighting coefficients of task completion time, error rate, and operation frequency, respectively. S3, Based on the dynamic difficulty adjustment algorithm, a generation difficulty coefficient K is obtained, which represents the difficulty of the specific level task dynamically adjusted according to the user's behavioral state and psychological state. At the same time, the difficulty of the specific level task is adjusted according to the user's current state value S to ensure the challenge and adaptability of the specific level task. S4, using a task selector, cognitive function goals are selected based on the generated difficulty coefficient K and the rehabilitation goals; wherein, the selection of cognitive function goals includes adjusting the target tracking task parameters based on attention training goals, expanding the matrix memory complexity based on memory training goals, increasing the rule transformation frequency based on executive function training goals, and improving fuzzy recognition noise based on perceptual function training; then, based on the generated difficulty coefficient K, the number of task elements is generated based on the parameterized formula (2), which is: （2）； The dynamic difficulty adjustment algorithm uses a parameterized formula. Generate task parameters, where N is the number of task elements, α' is the range scaling factor, β' is the cognitive load nonlinearity coefficient, and K is the generation difficulty coefficient. S5, based on the cognitive function target selection, generate personalized level JSON and cognitive tasks, and load the cognitive tasks through the Unity engine; S6. Update the model parameters of the reinforcement learning model once according to the rule of every fixed time interval to adapt to the user's long-term progress.

2. The personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 1, characterized in that, The algorithm for dynamic difficulty adjustment is as follows: establish a multi-objective optimization model based on reinforcement learning, and perform real-time parameter adjustment based on the multi-objective optimization model; The input parameters of the multi-objective optimization model include: (1) Behavioral state, which is characterized by: task response time, accuracy, operation trajectory accuracy and number of abandonments; (2) Psychological state, which is characterized by: EEG focus index, eye movement anxiety frequency and heart rate variability (HRV).

3. The personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 2, characterized in that, The data processing includes: A. Cross-modal adaptive processing, thereby supporting multi-device compatibility between brain-computer interfaces and VR controllers / eye controls; B. Implement anti-fatigue adaptation measures, including: automatically switching to low-intensity tasks when the standard deviation of the heart rate variability (HRV) is >50ms; C. Signal synchronization processing, including: the sampling period of the behavioral state is 100ms, the sampling frequency of the psychological state is 256Hz, and the timestamps of the behavioral state data and the psychological state data are aligned to form a synchronization signal; D. Perform signal anomaly processing, including: when an eye movement anxiety value > 0.7 is detected, insert a 3-minute deep breathing guided task.

4. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 3, characterized in that, The data fusion includes weighted fusion of the operation trajectory accuracy and the EEG focus index.

5. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 4, characterized in that, The personalized level JSON and cognitive task in S4 are dynamically constructed through a programmatic content generation algorithm, including the following steps: (1) Input seed parameters: user cognitive ability index V and training target weight W; (2) Execute the generated rule set: Spatial layout rule: Map = Voronoi_partition(V 1.2 ); Task complexity rules: ; Output personalized level examples: Ensure that each generated cognitive challenge is unique.

6. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 5, characterized in that, S4 further includes: A long-term adaptive mechanism that automatically increases the proportion of executive function tasks based on weekly training data to incorporate the number of rehabilitation goals; and a dynamic downgrading mechanism that triggers emergency intervention if the accuracy of the selection of the cognitive function goals is less than 30% for two consecutive times, wherein the dynamic downgrading mechanism includes reducing the generation difficulty coefficient K to K×0.6 as the new generation difficulty coefficient.

7. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 6, characterized in that, The working process of the cognitive task guidance module includes: S1, interacting with the user based on a multimodal guidance mechanism, wherein the multimodal guidance mechanism includes a combination of one or more of the following: cognitive task prompts, cognitive task multiple-choice questions, cognitive task guidance text input boxes, or cognitive task guidance image input boxes; wherein, the cognitive task prompts include: dynamically generating AR visual markers and overlaying them onto the VR scene; the cognitive task multiple-choice questions include: providing a cognitive task selector through voice broadcast and on-screen options, wherein the number of options in the voice broadcast and on-screen dynamically increases or decreases with the value of the generation difficulty coefficient K; the cognitive task guidance text input boxes or cognitive task guidance image input boxes include: performing functional training through handwritten formula recognition or performing perceptual training through noisy image annotation; S2, start and execute the guidance logic engine, wherein the task type determines whether it is an attention task. If it is, the AR arrow guides the movement target. If not, it determines whether it is a memory task. If it is, the step-by-step voice prompt memory strategy is given. If not, the graphic instruction decomposition rule is determined. When the guidance logic engine is executed, the eye movement trajectory is monitored in real time. When the gaze deviation is greater than 3 seconds, the reinforcement prompt is triggered. S3, dynamically adapt the difficulty in cognitive task guidance, including: enabling layered guidance when the error rate is >40%, the layered guidance includes text prompts to animation demonstrations and simplified task practice; switching to voice-stimulated guidance when EEG focus (θ / β) is <0.5; S4, feeding back user interaction data to the bootstrap logic engine to form a data loop, thereby optimizing subsequent bootstrap strategies; wherein the user interaction data includes response latency and / or input accuracy.

8. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 7, characterized in that, The working method of the data tracking and analysis module includes: Collect data for tracking and analysis, including: collecting the user's historical behavioral data and cognitive ability assessment results; The analysis of cognitive ability change trends includes: using a time series analysis model to extract cognitive ability change trends, wherein the time series analysis model calculates the trend value T based on the following formula (4): （4）； Among them, S i For the score of the i-th evaluation, denoted as the average score, and N represents the number of evaluations.

9. A personalized virtual-real interactive cognitive game system based on AI adaptive brain cognitive enhancement according to claim 8, characterized in that, The instant feedback generation algorithm is based on natural language processing technology, generating personalized feedback content according to the user's performance, and combining cognitive psychology theory to help users understand their cognitive abilities and provide improvement suggestions; wherein, the instant feedback generation algorithm includes: Perform real-time data analysis: Analyze users' task completion status and extract key indicators, including completion time and error rate; Generate instant feedback: Combining cognitive psychology theories, it generates positive incentives and improvement suggestions; Adjust tone: Adjust the feedback tone according to the user's emotional state; Generate natural language: Use a pre-trained language model to generate natural and fluent personalized feedback content.

Images