Method and system for improving the motor agility of adolescents in combination with audiovisual reaction training

Abstract

The present application relates to the technical field of sports training, and in particular to a method and system for improving the sports agility of teenagers in combination with audio-visual response training, which acquires basic response capability data to determine an initial interference proportion threshold value, presents a composite stimulation sequence mixed by target and interference stimulation, collects response action data and discriminates stimulation types to generate a response accuracy evaluation result, dynamically adjusts the interference proportion and shortens the presentation time length according to the evaluation result to form a progressive interference increment strategy, extracts attention bias mode data corresponding to the error response to adjust the interference feature distribution, so as to improve the individualization and adaptability of the sports agility training of teenagers.

Description

Technical Field

[0001] This invention relates to the field of sports training technology, and in particular to a method and system for improving athletic agility in adolescents by combining visual and auditory reaction training. Background Technology

[0002] In the field of agility training for adolescent athletes, traditional audiovisual response training methods typically rely on fixed patterns of stimulus presentation. For example, coaches or training systems pre-set a fixed ratio of target stimulus to distracting stimulus, and trainees repeatedly practice responses under a constant level of distraction. The core assumption of this approach is that by repeatedly exposing individuals to stimuli of the same difficulty, their reaction speed and accuracy can be gradually improved. Furthermore, existing practices often employ static stimulus presentation durations, meaning that the duration of all individual stimuli remains consistent, lacking dynamic management of the training load. While some methods introduce difficulty grading, the grading criteria are often based on macroscopic indicators such as age or physical fitness, rather than individual real-time response performance.

[0003] The conventional approach of using fixed-ratio stimulus sequences has significant drawbacks. First, it ignores the initial ability differences among trainees. High-level athletes, due to the low interference ratio, cannot obtain effective challenges, leading to low training efficiency; conversely, low-level athletes, due to the high interference ratio, make frequent mistakes, resulting in frustration and affecting their training motivation. Second, the static stimulus duration and interference ratio cannot respond to the adaptation phenomenon during training—as ability improves, the original level of difficulty will no longer constitute an effective stimulus, and the training effect tends to stagnate, i.e., a "plateau" occurs. This approach, lacking a dynamic adjustment mechanism, makes it difficult to achieve personalized, gradual load increases, limiting the physiological adaptation potential for agility improvement. Summary of the Invention

[0004] This invention provides a method and system for improving athletic agility in adolescents by combining audiovisual response training, which can solve the problems in the prior art.

[0005] A first aspect of the present invention provides a method for improving athletic agility in adolescents by combining audiovisual response training, comprising:

[0006] Acquire basic reaction ability data of the training subjects, and determine the initial interference ratio threshold based on the basic reaction ability data;

[0007] A composite stimulus sequence consisting of a target stimulus signal and an interference stimulus signal is presented within the training area. The target stimulus signal and the interference stimulus signal have distinguishable identification attributes in the dimension of perceptual features. The response action data of the training object to the composite stimulus sequence is collected, and the stimulus type corresponding to the response action data is determined based on the identification attributes. A response accuracy evaluation result including the correct response rate and the incorrect response rate is generated.

[0008] Based on the accuracy assessment results of the response and the preset advancement criteria, the proportion of the interference stimulus signal in the subsequent compound stimulus sequence is dynamically adjusted, and the presentation duration of a single stimulus in the compound stimulus sequence is shortened simultaneously, forming a progressive interference load increasing strategy.

[0009] The interference stimulus type corresponding to the error response rate is feature extracted to generate attention bias pattern data, and the feature distribution of interference stimulus signals in subsequent composite stimulus sequences is adjusted based on the attention bias pattern data.

[0010] Collect the response action data of the training subjects to the compound stimulus sequence, and determine the stimulus type corresponding to the response action data based on the identification attributes, generating a response accuracy evaluation result including correct response rate and incorrect response rate, including:

[0011] The input acquisition device records all the response actions of the training object during the presentation of the compound stimulus sequence. The response timestamp of each response action is matched with the presentation time window of each stimulus signal in the compound stimulus sequence to determine the trigger stimulus signal corresponding to the response action.

[0012] Extract the identification attributes of the triggering stimulus signal, determine whether the response operation type of the response action is consistent with the expected response type indicated by the identification attributes of the triggering stimulus signal, if consistent, classify the response action as a correct response, if inconsistent, classify it as an incorrect response;

[0013] The number of correct responses and the number of incorrect responses within a preset time period are statistically analyzed. The ratio of the number of correct responses to the total number of responses is calculated to obtain the correct response rate. The ratio of the number of incorrect responses to the total number of responses is calculated to obtain the incorrect response rate. The correct response rate and the incorrect response rate are then packaged into a response accuracy assessment result.

[0014] Based on the response accuracy assessment results and preset advancement criteria, the proportion of the interference stimulus signal in the subsequent complex stimulus sequence is dynamically adjusted, and the presentation duration of a single stimulus in the complex stimulus sequence is simultaneously shortened, forming a progressive interference load escalation strategy, including:

[0015] The correct response rate in the response accuracy assessment results is compared with the promotion threshold set in the promotion criteria. When the number of times the correct response rate continuously reaches or exceeds the promotion threshold meets the stability judgment condition, the load increment operation is triggered.

[0016] When the load increment operation is triggered, the increment of the interference ratio is calculated based on the current interference ratio and a preset nonlinear increment function, and the increment of the interference ratio is added to the current interference ratio to obtain the updated interference ratio.

[0017] Based on the coupling constraint relationship between the updated interference ratio and the current presentation duration, the presentation duration reduction amount is calculated, and the updated presentation duration is obtained by subtracting the presentation duration reduction amount from the current presentation duration.

[0018] When the correct response rate does not reach the advancement threshold, the current interference ratio and presentation duration are maintained unchanged, and a fine-tuning mechanism is activated to slightly extend the presentation duration. The updated interference ratio and the updated presentation duration are used as the generation control parameters for subsequent composite stimulus sequences. Through the dual-dimensional coordinated adjustment of increasing interference density and compressing the recognition time window, a progressive interference load increase strategy is formed.

[0019] When the load increment operation is triggered, the interference percentage increment is calculated based on the current interference percentage and a preset nonlinear increment function. The interference percentage increment is then added to the current interference percentage to obtain the updated interference percentage, including:

[0020] Obtain the number of interfering stimulus signals and the total number of stimulus signals in the current compound stimulus sequence, and calculate the ratio of the two as the current interference ratio;

[0021] The current interference ratio is substituted into a nonlinear increasing function as an input variable, and an inverse correlation mapping relationship between the current interference ratio and the interference ratio increment is established through the nonlinear increasing function.

[0022] The interference percentage increment is calculated through the reverse correlation mapping relationship. According to the reverse correlation mapping relationship, the interference percentage increment decreases as the current interference percentage increases. Thus, in the early stage of training, the interference density is rapidly increased with an increment greater than the preset increment threshold, and in the later stage of training, the difficulty gradient is finely adjusted with an increment less than the preset increment threshold.

[0023] The updated interference ratio is obtained by numerically superimposing the incremental interference ratio with the current interference ratio.

[0024] The error response rate is used to extract features from the types of interfering stimuli, generating attention-biased pattern data. Based on this attention-biased pattern data, the feature distribution of interfering stimulus signals in subsequent composite stimulus sequences is adjusted, including:

[0025] Extract all error response records from the response accuracy assessment results, obtain the trigger stimulus signal corresponding to each error response, identify the signal objects that belong to the interference stimulus signals in the trigger stimulus signals, and extract the feature vector of the signal objects.

[0026] Cluster analysis is performed on all extracted feature vectors to identify feature clusters whose frequency exceeds a preset frequency threshold, and these feature clusters are marked as confusing feature types.

[0027] The occurrence frequency of each obfuscation feature type in error responses is counted, its proportion in all error responses is calculated, a feature obfuscation distribution is generated, and the obfuscation feature type and its corresponding feature obfuscation distribution are encapsulated as attention-biased pattern data.

[0028] When generating subsequent compound stimulus sequences, the proportion of interference stimulus signals corresponding to the confusion feature types in the sequence is increased based on the feature confusion distribution in the attention bias pattern data.

[0029] The occurrence frequency of each obfuscation feature type in error responses is counted, and its proportion of all error responses is calculated to generate a feature obfuscation distribution. The obfuscation feature types and their corresponding feature obfuscation distributions are then encapsulated into attention-biased pattern data, including:

[0030] Traverse all error response records within a preset time period, extract the feature type identifier of the interference stimulus signal corresponding to each error response, establish a feature type statistics table, merge error responses with the same feature type identifier, and accumulate the number of times each confusion feature type appears in all error response records.

[0031] Calculate the total number of all error response records, divide the number of occurrences of each obfuscation feature type by the total number, and obtain the obfuscation level value corresponding to the obfuscation feature type.

[0032] Sort each confusing feature type and its corresponding confusion value from high to low according to the confusion value to generate a feature confusion distribution;

[0033] From the feature confusion distribution, filter out confusion feature types whose confusion values ​​exceed a preset significance threshold, and mark the filtered confusion feature types as confusion feature types;

[0034] Extract the multi-dimensional attribute parameters of the confusing feature type in the perceptual feature dimension, associate and bind the multi-dimensional attribute parameters with the corresponding confusion degree values ​​to form a feature-confusion degree mapping structure, and encapsulate the feature confusion degree distribution and the feature-confusion degree mapping structure into attention bias pattern data.

[0035] A second aspect of the present invention provides a system for improving athletic agility in adolescents by combining audiovisual response training, comprising:

[0036] An initial threshold unit is used to acquire basic reaction ability data of the training object and determine an initial interference ratio threshold based on the basic reaction ability data.

[0037] The response evaluation unit is used to present a composite stimulus sequence consisting of a target stimulus signal and an interference stimulus signal in the training area, wherein the target stimulus signal and the interference stimulus signal have distinguishable identification attributes in the dimension of perceptual features, collect the response action data of the training object to the composite stimulus sequence, and determine the stimulus type corresponding to the response action data based on the identification attributes, and generate a response accuracy evaluation result including the correct response rate and the incorrect response rate.

[0038] The load escalation unit is used to dynamically adjust the proportion of the interference stimulus signal in the subsequent composite stimulus sequence based on the response accuracy evaluation result and the preset escalation criteria, and simultaneously shorten the presentation duration of a single stimulus in the composite stimulus sequence, forming a progressive interference load escalation strategy.

[0039] The bias adjustment unit is used to extract features of the interference stimulus type corresponding to the error response rate, generate attention bias pattern data, and adjust the feature distribution of interference stimulus signals in subsequent composite stimulus sequences based on the attention bias pattern data.

[0040] A third aspect of the present invention provides an electronic device, comprising:

[0041] processor;

[0042] Memory used to store processor-executable instructions;

[0043] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0044] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0045] By setting an initial interference ratio threshold based on the basic reaction ability data of the trainees, the training starting point is highly aligned with individual differences, avoiding frustration caused by excessively high training intensity or loss of challenge due to excessively low intensity. This improves the adaptability and safety of training from the source, and is especially suitable for the sensitivity needs of adolescents in the developmental stage to stimulus responses.

[0046] By using composite stimulus sequences with distinguishable identifiable attributes, the system can accurately differentiate between target and interference responses, generating evaluation results that include both accuracy and error rates, quantifying the accuracy of trainees' responses in complex environments. Dynamically adjusting the proportion of interference stimuli and simultaneously shortening stimulus presentation time creates a continuously increasing cognitive load, effectively enhancing adolescents' screening and decision-making abilities in rapidly changing scenarios, and significantly improving their motor reaction agility and resistance to interference.

[0047] Feature extraction is performed on the interference types corresponding to erroneous responses to generate attention bias pattern data, identifying individuals' concentrated error tendencies in specific interference dimensions. Based on this, the feature distribution of subsequent interference stimuli is optimized to achieve targeted correction of attention biases, reduce erroneous responses caused by non-target stimuli, and concentrate training resources on individual weaknesses, thereby improving overall training efficiency and accelerating the release of adolescents' athletic agility potential. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating a method for improving athletic agility in adolescents that combines audiovisual response training, as described in an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0050] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0051] Figure 1 This is a flowchart illustrating a method for improving athletic agility in adolescents that combines audiovisual response training, as described in an embodiment of the present invention.

[0052] Methods for improving athletic agility in teenagers by combining audiovisual response training include:

[0053] Acquire basic reaction ability data of the training subjects, and determine the initial interference ratio threshold based on the basic reaction ability data;

[0054] A composite stimulus sequence consisting of a target stimulus signal and an interference stimulus signal is presented within the training area. The target stimulus signal and the interference stimulus signal have distinguishable identification attributes in the dimension of perceptual features. The response action data of the training object to the composite stimulus sequence is collected, and the stimulus type corresponding to the response action data is determined based on the identification attributes. A response accuracy evaluation result including the correct response rate and the incorrect response rate is generated.

[0055] Based on the accuracy assessment results of the response and the preset advancement criteria, the proportion of the interference stimulus signal in the subsequent compound stimulus sequence is dynamically adjusted, and the presentation duration of a single stimulus in the compound stimulus sequence is shortened simultaneously, forming a progressive interference load increasing strategy.

[0056] The interference stimulus type corresponding to the error response rate is feature extracted to generate attention bias pattern data, and the feature distribution of interference stimulus signals in subsequent composite stimulus sequences is adjusted based on the attention bias pattern data.

[0057] In one optional implementation, response action data of the training subjects to the compound stimulus sequence is collected, and the stimulus type corresponding to the response action data is determined based on the identification attribute. A response accuracy assessment result including correct response rate and incorrect response rate is generated, including:

[0058] The input acquisition device records all the response actions of the training object during the presentation of the compound stimulus sequence. The response timestamp of each response action is matched with the presentation time window of each stimulus signal in the compound stimulus sequence to determine the trigger stimulus signal corresponding to the response action.

[0059] Extract the identification attributes of the triggering stimulus signal, determine whether the response operation type of the response action is consistent with the expected response type indicated by the identification attributes of the triggering stimulus signal, if consistent, classify the response action as a correct response, if inconsistent, classify it as an incorrect response;

[0060] The number of correct responses and the number of incorrect responses within a preset time period are statistically analyzed. The ratio of the number of correct responses to the total number of responses is calculated to obtain the correct response rate. The ratio of the number of incorrect responses to the total number of responses is calculated to obtain the incorrect response rate. The correct response rate and the incorrect response rate are then packaged into a response accuracy assessment result.

[0061] The system continuously records all responses of the training subject during the presentation of the compound stimulus sequence using an input acquisition device. This device can be a pressure-sensitive pad, touch panel, motion-sensing camera, or a button with millisecond-level timing accuracy, and can be flexibly selected based on the training scenario. Each response is automatically accompanied by a response timestamp upon acquisition. This timestamp is generated by a synchronization clock shared with the stimulus presentation system, ensuring that the response timestamp and the stimulus presentation time window are on the same time base, thus avoiding timing matching errors caused by clock offsets.

[0062] During the timing matching phase, the response timestamp of each action is compared with the presentation time windows of each stimulus signal in the composite stimulus sequence. The presentation time window for each stimulus signal is defined by its start and end times. If the response timestamp of an action falls within the presentation time window of a stimulus signal, a correspondence is established between the action and the signal, confirming that the triggering stimulus is that stimulus signal. Considering the inherent delays in nerve conduction and muscle execution, the presentation time window can be extended by an allowable response delay during actual matching. This allowable delay is pre-set based on the age and basic reaction ability data of the training subjects, typically ranging from 150 to 400 milliseconds to cover the normal physiological response delay distribution range of adolescents. When the response timestamp of an action fails to fall within the valid matching window of any stimulus signal, the action is marked as invalid and not included in subsequent accuracy statistics, but it is recorded separately for analyzing the impulsive response tendency of the training subjects.

[0063] After timing matching is completed, the identifying attributes of the triggering stimulus signal are extracted, and the expected response type indicated by the identifying attribute is compared with the response operation type actually performed by the training subject. The identifying attributes are distinguishable in the perceptual feature dimension. For example, the target stimulus signal can be identified by a specific color, shape, tone, or direction, while the interfering stimulus signal uses a combination of features that are significantly different from the target stimulus signal in the above dimensions, enabling the training subject to complete perceptual differentiation within a short presentation time. The expected response type is predefined by the training rules. For example, when the triggering stimulus signal is the target stimulus signal, the expected response type is to perform a specific limb action (such as stepping on a designated area or pressing a corresponding button); when the triggering stimulus signal is the interfering stimulus signal, the expected response type is to inhibit the action, i.e., no response is produced. If the type of response action actually performed by the training object is consistent with the expected type of response, the response action is classified as a correct response; if it is inconsistent, it is classified as an incorrect response. Incorrect responses are further subdivided into two categories: one is that no response is generated to the target stimulus signal (missed report), and the other is that an undue response is generated to the interference stimulus signal (false report). The two types of incorrect responses are recorded separately for use in subsequent attention bias pattern analysis.

[0064] In the statistical calculation phase, the number of correct responses and the number of incorrect responses are summarized using a preset time period as the statistical unit. The preset time period can be the entire duration of a single training session or several sub-phases divided within a training session. Dividing the session into sub-phases helps to observe the changing trend of the trainees' attention span over time during continuous training. Let the number of correct responses within the preset time period be... The number of error responses is The total number of responses is ,in Correct response rate With error response rate Calculate as follows:

[0065] ;

[0066] ;

[0067] The two ratios mentioned above together constitute the core quantitative indicators for evaluating the accuracy of the response. It is important to note that... This represents the total number of actual response actions, excluding cases where no response was generated. Therefore, when handling false negative errors, the number of false negatives must be added separately to the total number of error responses. In the meantime, corresponding adjustments will be made. The calculation base is determined to ensure that the sum of the correct response rate and the incorrect response rate is always equal to 1, thus guaranteeing the mathematical consistency of the evaluation results.

[0068] Will and The detailed distribution information of each subclass of error response (including the number of missed calls, the number of false calls, and their corresponding triggering stimulus signal types) is encapsulated into a response accuracy assessment result. This assessment result is stored in structured data format and can be directly accessed in subsequent progressive load adjustment steps that dynamically adjust the proportion of interfering stimulus signals and shorten the stimulus presentation time. The assessment result also includes the actual response delay of each response action, i.e., the difference between the response timestamp and the start time of the triggering stimulus signal presentation. Although this delay data does not directly participate in the calculation of the correct response rate and the incorrect response rate, it can serve as an auxiliary dimension to reflect the change in the response speed of the training object, and together with the accuracy index, constitutes a multi-dimensional quantitative description of motor agility.

[0069] In real-world training scenarios, the presentation rate of compound stimulus sequences is relatively fast, requiring trainees to complete the entire process of perception, judgment, and action execution within a very short time. Therefore, the sampling frequency of the input acquisition device must be sufficiently high to ensure the accuracy of the response timestamp recording. When using a motion-sensing camera as the input acquisition device, keyframes need to be extracted from the acquired limb movement trajectories. These keyframe moments are used as response timestamps, and motion amplitude thresholds are used to filter out false responses caused by posture adjustments or unconscious micro-movements, ensuring that all response actions involved in timing matching are valid actions consciously performed by the trainee. Through this complete process, the final response accuracy assessment result can objectively and accurately reflect the trainee's cognitive inhibition and target recognition abilities under the current interference load conditions, providing reliable data for the dynamic adjustment of subsequent personalized training programs.

[0070] In one optional implementation, based on the response accuracy assessment results and preset escalation criteria, the proportion of the interference stimulus signal in the subsequent composite stimulus sequence is dynamically adjusted, and the presentation duration of a single stimulus in the composite stimulus sequence is simultaneously shortened, forming a progressive interference load escalation strategy, including:

[0071] The correct response rate in the response accuracy assessment results is compared with the promotion threshold set in the promotion criteria. When the number of times the correct response rate continuously reaches or exceeds the promotion threshold meets the stability judgment condition, the load increment operation is triggered.

[0072] When the load increment operation is triggered, the increment of the interference ratio is calculated based on the current interference ratio and a preset nonlinear increment function, and the increment of the interference ratio is added to the current interference ratio to obtain the updated interference ratio.

[0073] Based on the coupling constraint relationship between the updated interference ratio and the current presentation duration, the presentation duration reduction amount is calculated, and the updated presentation duration is obtained by subtracting the presentation duration reduction amount from the current presentation duration.

[0074] When the correct response rate does not reach the advancement threshold, the current interference ratio and presentation duration are maintained unchanged, and a fine-tuning mechanism is activated to slightly extend the presentation duration. The updated interference ratio and the updated presentation duration are used as the generation control parameters for subsequent composite stimulus sequences. Through the dual-dimensional coordinated adjustment of increasing interference density and compressing the recognition time window, a progressive interference load increase strategy is formed.

[0075] After obtaining the response accuracy assessment results, the correct response rate needs to be compared sequentially with the pre-set advancement threshold in the advancement criteria. The advancement threshold is a reference value determined during the training program design phase based on the average cognitive load tolerance of the adolescent group, typically set between 75% and 90%. The specific value can be personalized based on the age group and initial baseline response ability data of the trainees. After each round of training with the compound stimulus sequence, the correct response rate for that round is recorded and compared with the advancement threshold. The load escalation operation is only triggered when the correct response rate reaches or exceeds the advancement threshold for several consecutive rounds, and the number of consecutive successful rounds meets the stability judgment condition. The stability judgment condition is introduced to prevent trainees from prematurely entering a higher difficulty stage due to occasional high performance, thereby ensuring that the increase in training load is based on a stable ability foundation. The number of consecutive successful rounds in the stability judgment condition can be set to 3 to 5 rounds, with specific parameters uniformly configured by the training program.

[0076] After triggering the load increment operation, the adjustment of the interference percentage does not use a linear increment method, but instead introduces a non-linear increment function to calculate the increase in the interference percentage. The reason for using a non-linear increment function is that: in the early stages of training, the trainees' resistance to interference stimuli is relatively weak, so the increase in the interference percentage should be relatively small to avoid sudden high loads causing learned helplessness in the trainees; as training progresses, the trainees' resistance to interference gradually increases, and the increase in the interference percentage can be appropriately increased to maintain sufficient training challenge. Let the current interference percentage be... The incremental proportion of interference generated by the nonlinear increasing function is ,but This can be determined through the following relationship:

[0077] ;

[0078] in The base growth rate coefficient is used to control the overall growth rate. This is a non-linear adjustment index used to control the degree to which the increase decays as the current disturbance proportion changes. When When smaller, When it is close to 1, the increment is relatively large; when When it is large, As the value approaches zero, the increment converges, thus achieving a smooth transition during periods of high interference ratio. (Updated interference ratio) The result is obtained by adding the current interference percentage to the increment:

[0079] ;

[0080] To prevent the proportion of interference from exceeding a reasonable upper limit, it is necessary to... Setting a hard upper bound, typically not exceeding 70%, ensures that the trainees still have a sufficient number of target stimulus signals to respond to in each round of training, thus maintaining the effectiveness of the training.

[0081] After updating the interference ratio, the reduction in presentation duration needs to be calculated simultaneously. There is a coupled constraint relationship between presentation duration and interference ratio. The core logic of this constraint is that as the interference ratio increases, the training subject needs to quickly distinguish between the target stimulus and the interference stimulus within a shorter time window, thereby achieving a synergistic increase in cognitive load across both interference density and recognition time. Let the presentation duration of the current single stimulus be... The updated interference ratio Compared with the current interference ratio The difference between them is This will show the reduction in duration. Calculated using coupling constraint relationships:

[0082] ;

[0083] in This is the coupling coefficient, reflecting the strength of the impact of changes in the interference ratio on the degree of presentation time reduction. This coefficient needs to be configured in segments according to the age group of the training subjects and the current training stage. Updated presentation time for:

[0084] ;

[0085] To ensure the physiological safety of training, a lower limit needs to be set for the updated presentation duration to avoid over-compression that would prevent the training subjects from completing basic perceptual judgments. This lower limit is typically no less than 150 milliseconds. When the calculated... When the value is below the lower limit, the presentation duration is fixed at the lower limit, and only the interference ratio is adjusted.

[0086] When the trainee's correct response rate does not reach the advancement threshold, the load escalation operation is not triggered, and the current interference ratio and presentation duration remain unchanged. Simultaneously, a fine-tuning mechanism is activated to slightly extend the current presentation duration, providing the trainee with additional perceptual processing time to help further solidify their reaction ability at the current difficulty level. The extension amount of the fine-tuning mechanism... Set a fixed step size, typically 20 to 50 milliseconds, with the specific value determined based on the age and current performance level of the training subjects. Fine-tuning the presentation duration. for:

[0087] ;

[0088] The number of times the fine-tuning mechanism can be extended is also subject to an upper limit. If the correct response rate still does not improve after multiple rounds of fine-tuning, it will indicate that the training scheme needs to be manually reviewed to assess whether the current interference ratio setting exceeds the actual capability range of the training object.

[0089] Updated interference percentage With the updated presentation duration (or the slightly adjusted presentation duration) This parameter, used as a control parameter for generating subsequent compound stimulus sequences, is input into the stimulus sequence generation stage. During the generation of subsequent compound stimulus sequences, it follows... The prescribed proportions are used to allocate the number of interfering stimulus signals within the total number of stimuli, and according to... or The maximum presentation duration of each stimulus signal is controlled. Through a two-dimensional synergistic adjustment of increasing interference density and compressing the recognition time window, the trainee faces a combined challenge from both the quantity of interference and reaction speed at each stage, thereby gradually improving motor agility and attention selection efficiency under continuous adaptive pressure. The core advantage of this progressively increasing interference load strategy lies in controlling the load increase through a nonlinear function and linking the two adjustment dimensions with a coupled constraint relationship. This results in a smooth overall training difficulty curve with individual adaptability, avoiding the training stagnation or overload problems commonly found in fixed step size adjustment schemes.

[0090] In one optional implementation, when the load increment operation is triggered, the interference percentage increment is calculated based on the current interference percentage and a preset nonlinear increment function, and the interference percentage increment is added to the current interference percentage to obtain an updated interference percentage, including:

[0091] Obtain the number of interfering stimulus signals and the total number of stimulus signals in the current compound stimulus sequence, and calculate the ratio of the two as the current interference ratio;

[0092] The current interference ratio is substituted into a nonlinear increasing function as an input variable, and an inverse correlation mapping relationship between the current interference ratio and the interference ratio increment is established through the nonlinear increasing function.

[0093] The interference percentage increment is calculated through the reverse correlation mapping relationship. According to the reverse correlation mapping relationship, the interference percentage increment decreases as the current interference percentage increases. Thus, in the early stage of training, the interference density is rapidly increased with an increment greater than the preset increment threshold, and in the later stage of training, the difficulty gradient is finely adjusted with an increment less than the preset increment threshold.

[0094] The updated interference ratio is obtained by numerically superimposing the incremental interference ratio with the current interference ratio.

[0095] When triggering a load increment operation, it is necessary to first determine the actual proportion of interfering stimulus signals in the current composite stimulus sequence. Specifically, the number of interfering stimulus signals in the current composite stimulus sequence is calculated and divided by the total number of all stimulus signals in the sequence. The resulting ratio, which is between 0 and 1, represents the current proportion of interference. For example, if a complex stimulus sequence contains 20 stimulus signals, of which 8 are interference signals, then the current interference percentage is... This calculation process is directly based on the sequence structure, without the need for additional estimation or approximation, ensuring the accuracy and traceability of the interference ratio, and providing a reliable input basis for subsequent calculations of nonlinear increasing functions.

[0096] The current interference percentage obtained As input variables, a non-linear increasing function is substituted to establish an inverse correlation mapping relationship between the current interference percentage and the increment of the interference percentage. The so-called inverse correlation mapping relationship refers to the increment of the interference percentage... According to the current proportion of interference The relationship is a monotonically decreasing one where the value increases and decreases. The core design logic of this relationship lies in the fact that when training is in its initial stages, the trainee has not yet adapted to the interference stimulus; at this time... Smaller, corresponding A larger interference density can quickly increase the training difficulty to the effective stimulation range; however, as training progresses into the later stages, the trainee has developed a certain level of resistance to interference, at which point... Larger, corresponding The difficulty level is relatively small, thus enabling precise control of the difficulty gradient and avoiding stress-induced avoidance or a decline in movement quality caused by a sudden increase in difficulty.

[0097] The nonlinear increasing function uses a basic increase coefficient. With nonlinear adjustment index The interference accounts for an incremental proportion of the calculation. The calculation method is as follows .in, The base amplification coefficient controls the magnitude of the overall increase. Its value range is usually preset based on the age group and basic reaction ability data of the training subjects. For example, for teenagers with moderate reaction ability, It can be set between 0.05 and 0.15; The control increment is a nonlinear adjustment exponent. The changing decay rate, The larger the value, the larger the increment. The larger the gradient, the more dramatic the decay, and the more significant the effect of fine-tuning the difficulty gradient. The value ranges from 1.5 to 3.0. When Smaller (e.g.) )hour, Approaching 1, near The increment is at a relatively large level; when Larger (e.g.) )hour, Significantly less than 1, Significant reduction, enabling precise control. Through adjustments... and The combination of these can flexibly adapt to the ability levels and training cycle requirements of different training subjects.

[0098] After confirming that the calculation results of the reverse correlation mapping relationship meet the reasonableness constraints, the increment of the interference proportion will be... Compared with the current interference ratio Perform numerical superposition calculations to obtain the updated interference ratio. ,Right now After the superposition operation is completed, it is necessary to... Perform boundary constraint checks: If If the interference percentage exceeds the preset upper limit (e.g., 0.80), then... The truncation to this upper limit prevents the proportion of interfering stimulus signals in the composite stimulus sequence from becoming too high, which could lead to excessive compression of the perception opportunities of the target stimulus signal and affect the normal recognition and response of the training subject to the target stimulus. This upper limit can be dynamically configured according to the training stage and individual differences of the training subject, and has a certain degree of adaptive space.

[0099] In actual training scenarios, the aforementioned dynamic adjustment mechanism for interference ratio and the presentation duration reduction mechanism operate in tandem. (Updated interference ratio) This will be directly used to construct subsequent compound stimulus sequences, that is, when generating the next round of stimulus sequences, according to... The corresponding ratio allocates the number of interfering stimulus signals to the number of target stimulus signals. For example, if the total number of stimuli in the subsequent sequence is 20, and The sequence will contain 10 interfering stimulus signals and 10 target stimulus signals. The order of these signals in the sequence can be allocated according to a preset randomization strategy to prevent the training subjects from avoiding the influence of interfering stimuli through regular prediction.

[0100] To ensure the consistency and repeatability of the calculation results for the increase in the proportion of interference across different training batches, and The value remains fixed within the same training cycle and is not modified in real time based on fluctuations in the results of a single training session. Only after completing a full training cycle is the evaluation performed, based on the overall progress of the trainees, by the training administrator or the automated evaluation module. and Periodic updates are performed to ensure the adaptability of the training program during long-term execution.

[0101] It is worth noting that the inverse correlation mapping established by the nonlinear increasing function is not a simple linear inverse relationship, but rather a diminishing marginal effect characterized by an exponential decay curve. This design aligns with the physiological and cognitive adaptation patterns of agility training in adolescents: in the early stages of training, the trainees' ability to distinguish interfering stimuli improves rapidly, making it suitable to increase the interfering density with larger increments; while in the later stages, the trainees' ability to distinguish stimuli approaches their individual stage limit, at which point fine-tuning with small increments helps to continuously consolidate and strengthen the training effect without exceeding the cognitive load threshold, avoiding stagnation or even regression in training effects due to overstimulation. Through the combined effect of the nonlinear increasing function and the boundary constraint mechanism, the entire interfering load increasing strategy can maintain a reasonable pace of difficulty progression throughout the entire training cycle, providing a stable training load management framework for the systematic improvement of adolescents' agility.

[0102] In one optional implementation, features are extracted from the interference stimulus type corresponding to the error response rate to generate attention bias pattern data, and the feature distribution of interference stimulus signals in subsequent composite stimulus sequences is adjusted based on the attention bias pattern data, including:

[0103] Extract all error response records from the response accuracy assessment results, obtain the trigger stimulus signal corresponding to each error response, identify the signal objects that belong to the interference stimulus signals in the trigger stimulus signals, and extract the feature vector of the signal objects.

[0104] Cluster analysis is performed on all extracted feature vectors to identify feature clusters whose frequency exceeds a preset frequency threshold, and these feature clusters are marked as confusing feature types.

[0105] The occurrence frequency of each obfuscation feature type in error responses is counted, its proportion in all error responses is calculated, a feature obfuscation distribution is generated, and the obfuscation feature type and its corresponding feature obfuscation distribution are encapsulated as attention-biased pattern data.

[0106] When generating subsequent compound stimulus sequences, the proportion of interference stimulus signals corresponding to the confusion feature types in the sequence is increased based on the feature confusion distribution in the attention bias pattern data.

[0107] After one or more rounds of training, all erroneous response records are extracted from the response accuracy assessment results. Each erroneous response record includes the stimulus signal identifier that triggered the response, the response timestamp, and the response type label. For each erroneous response record, its corresponding triggering stimulus signal is traced back to determine whether the stimulus signal belongs to the interference stimulus signal category. The determination is based on the label attributes pre-set during the compound stimulus sequence generation stage, i.e., each stimulus signal carries a type label of "target" or "interference" during generation. If the type label of the stimulus signal triggering the erroneous response is "interference," it is listed as a signal object to be analyzed; if the type label is "target," it indicates that the erroneous response belongs to the missed detection type and is not included in the interference feature extraction process, but is recorded separately for subsequent missed detection rate analysis.

[0108] For all identified interfering signal objects, feature vectors are extracted along the perceptual feature dimension. The feature vectors are composed of visual channel attributes (such as color channel values, shape contour descriptors, motion direction angles, and flashing frequencies) and auditory channel attributes (such as pitch frequency, timbre category encoding, beat duration, and loudness envelope parameters). For combined audiovisual stimulus signals, the feature vector is formed by concatenating the visual and auditory sub-vectors, creating a high-dimensional joint feature representation. Each interfering signal object corresponds to a complete feature vector, and the feature vectors corresponding to all erroneous responses constitute the feature vector set to be clustered.

[0109] Cluster analysis is performed on the aforementioned feature vector set. The clustering algorithm preferably uses an adaptive density approach to accommodate the different error pattern distributions of different training objects. During clustering, feature vectors that are close in distance in the feature space are grouped into the same feature cluster. Each feature cluster represents a set of interfering stimulus signals with similar attributes in the perceptual dimension. After clustering, the number of feature vectors contained in each feature cluster is counted, and its proportion to the total number of feature vectors to be clustered is calculated. This proportion is used as the frequency of occurrence of the feature cluster. Feature clusters with a frequency exceeding a preset frequency threshold are marked as confusing feature types. The preset frequency threshold can be set according to the training stage and individual differences of the training objects. In the initial stage, a relatively loose value can be set, which is gradually tightened as training progresses to identify more granular confusion patterns. If the frequency of occurrence of a feature cluster is lower than the preset frequency threshold, the interference of the interfering stimulus signals in that cluster on the training objects is considered to be an occasional event without a systematic confusion pattern, and it is not included in the attention bias analysis at this time.

[0110] For each feature cluster labeled as an obfuscated feature type, count its occurrence count in all error responses, let the i-th... The frequency of occurrence of each confusion feature type is: The total number of all error responses is Then the feature confusion degree of this confusing feature type Calculate as follows:

[0111] ;

[0112] in, Index numbering for obfuscated feature types, For the first The number of times each obfuscation feature type appears in the error response. The total number of all error responses. For the first The feature confusion degree for each confusion feature type represents the relative weight of the erroneous response caused by that type of interfering stimulus. After calculating for all confusion feature types, each... The values ​​are arranged in descending order to form a feature confusion distribution. This distribution intuitively reflects the weak areas of the training object's recognition of different types of interfering stimuli at the perceptual level, i.e., the specific direction of attentional bias. All feature descriptors of all confusing feature types and their corresponding... Values ​​are encapsulated and stored to form attention-biased pattern data.

[0113] The generation of attention bias pattern data exhibits a time-accumulated property, meaning that the attention bias pattern data is updated after each training cycle. The update method employs a sliding weighted accumulation strategy, assigning higher weights to error responses from recent training cycles and gradually decreasing weights to older data. This ensures that the attention bias pattern data reflects the cognitive weaknesses of the trainee at the current stage in real time, rather than being dominated by early training data. This dynamic update mechanism allows for continuous and precise adjustments to subsequent compound stimulus sequences as training progresses.

[0114] When generating subsequent compound stimulus sequences, the feature distribution of the interfering stimulus signal is adjusted according to the feature confusion distribution in the attention-biased pattern data. Specifically, The higher the confusion feature type value, the higher the proportion of the interfering stimulus signal appearing in subsequent sequences. Let the proportion of the interfering stimulus signal of a certain confusion feature type before adjustment be... Adjusted occurrence ratio according to The relative sizes of the features are proportionally amplified so that feature types with higher confusion levels appear more frequently in the sequence, thereby providing targeted repetitive stimulus exposure to the weak perceptual dimensions of the training subjects. The sum of the proportions of each confused feature type after adjustment must satisfy a normalization constraint, that is, the sum of the proportions of all interfering stimulus signals equals 1, to ensure the overall structural rationality of the composite stimulus sequence.

[0115] For interfering stimulus signals not labeled as confusing features, their proportion in subsequent sequences is reduced accordingly, freeing up sequence space for confusing features. This dynamic adjustment of feature distribution is not a simple random replacement, but rather, while maintaining the overall perceptual load level of the composite stimulus sequence, it achieves targeted training and reinforcement by altering the internal structure of the interfering stimuli. As trainees are frequently exposed to interfering stimulus signals corresponding to their cognitive weaknesses, they are forced to activate more refined perceptual discrimination mechanisms, gradually building the ability to accurately identify this type of interfering stimulus, thereby driving the attention bias pattern towards a more balanced direction.

[0116] After several training cycles, a certain type of confusing feature... When the value drops below the confusion level corresponding to a preset frequency threshold, this type is removed from the active list of attention-biased pattern data, its occurrence ratio in subsequent sequences returns to normal, and the training focus shifts to areas that still maintain a high level. Other obfuscated feature types. This dynamic exit mechanism ensures that training resources are always focused on the perceptual dimensions that need the most strengthening at the current stage, avoiding inefficient repetitive training on already mastered obfuscation types, thereby maximizing the improvement in motion agility within a limited training time.

[0117] In one optional implementation, the frequency of occurrence of each obfuscation feature type in error responses is counted, its proportion of all error responses is calculated, and a feature obfuscation distribution is generated. The obfuscation feature types and their corresponding feature obfuscation distributions are then encapsulated into attention-biased pattern data, including:

[0118] Traverse all error response records within a preset time period, extract the feature type identifier of the interference stimulus signal corresponding to each error response, establish a feature type statistics table, merge error responses with the same feature type identifier, and accumulate the number of times each confusion feature type appears in all error response records.

[0119] Calculate the total number of all error response records, divide the number of occurrences of each obfuscation feature type by the total number, and obtain the obfuscation level value corresponding to the obfuscation feature type.

[0120] Sort each confusing feature type and its corresponding confusion value from high to low according to the confusion value to generate a feature confusion distribution;

[0121] From the feature confusion distribution, filter out confusion feature types whose confusion values ​​exceed a preset significance threshold, and mark the filtered confusion feature types as confusion feature types;

[0122] Extract the multi-dimensional attribute parameters of the confusing feature type in the perceptual feature dimension, associate and bind the multi-dimensional attribute parameters with the corresponding confusion degree values ​​to form a feature-confusion degree mapping structure, and encapsulate the feature confusion degree distribution and the feature-confusion degree mapping structure into attention bias pattern data.

[0123] When traversing all error response records within a preset time period, for each error response record, the feature type identifier of its corresponding interfering stimulus signal is extracted. The feature type identifier refers to a unique coded label pre-assigned to each type of interfering stimulus signal during the composite stimulus sequence design phase. This label covers the classification attributes of the interfering stimulus signal in perceptual feature dimensions such as color, shape, tone, direction, and speed. For example, if the interfering stimulus signals are divided into red, blue, and green categories in the color dimension, then each error response record carries a color category label to which the interfering stimulus belongs, which serves as the basis for subsequent merging and statistical analysis.

[0124] After extracting the feature type identifiers of all error response records, a feature type statistics table is created. This table uses the feature type identifier as the row index and the occurrence count as the statistical field, with all values ​​initialized to 0. During the traversal, for each error response record read, the count field of its corresponding feature type identifier in the statistics table is incremented by 1. If the feature type identifier corresponding to a certain error response record does not yet exist in the statistics table, a new row is created and the count is initialized to 1. After the traversal is complete, the statistics table records the cumulative occurrence count of each obfuscation feature type in all error response records within a preset time period, completing the merging operation for error responses with the same feature type identifier.

[0125] After completing the merge statistics, the total number of all erroneous response records is calculated. The confusion level value corresponding to each confusing feature type is obtained by dividing the frequency of occurrence of each type in the statistics table by this total number. The confusion level value reflects the relative intensity of the interference caused by this type of interfering stimulus signal to the training object. A higher value indicates that the training object is more likely to produce an erroneous response when faced with interfering stimuli of this type of feature; that is, the tendency for attention to be attracted to this type of feature and to make a misjudgment is more significant. The calculation of the confusion level value is based on the ratio of the frequency of occurrence of each feature type to the total number of erroneous responses, ensuring that the sum of the confusion level values ​​of each confusing feature type equals 1, forming a normalized probability distribution, which facilitates subsequent threshold filtering and sorting operations.

[0126] All obfuscated feature types are sorted from highest to lowest according to their obfuscation scores, generating a feature obfuscation distribution. After sorting, the obfuscated feature type with the highest obfuscation score is placed at the top of the distribution, representing the feature dimension most susceptible to interference in the current training phase. The feature obfuscation distribution is stored as an ordered list, containing the identifier code for each obfuscated feature type and its corresponding obfuscation score, forming a quantitative description of the attentional bias patterns of the training object. This distribution structure provides foundational data for subsequent saliency screening and also provides direct evidence for visualizing the distribution of attentional weaknesses in the training object.

[0127] The algorithm filters out confusing feature types whose confusion values ​​exceed a preset significance threshold from the feature confusion distribution, and marks these filtered confusing feature types as significant confusing feature types. The preset significance threshold is determined based on the complexity of the training scenario and the total number of interfering stimulus categories. For example, in a scenario with 8 interfering stimulus categories, if the categories are evenly distributed, the expected confusion value per category is approximately 0.125. The significance threshold can be set to a multiple higher than the expected value of the even distribution to identify feature types with significantly higher confusion levels. This filtering operation ensures that subsequent attention bias pattern data focuses on the interfering feature types that have the most prominent impact on the training object, avoiding the introduction of noise from low-confusion feature types that interfere with subsequent training adjustment strategies.

[0128] For the selected significant confusion feature types, multi-dimensional attribute parameters are extracted along the perceptual feature dimension. These multi-dimensional attribute parameters encompass the color channel values, brightness contrast, and shape contour descriptors of the interfering stimulus signal in the visual dimension, and quantitative descriptions such as frequency range, pitch category, and volume intensity in the auditory dimension. These attribute parameters characterize the physical features of the interfering stimulus signal at the perceptual level, providing precise operational targets for subsequent adjustments to the feature distribution of the interfering stimulus signal. For example, if the confusion value of a certain type of interfering stimulus signal is significantly high, and its multi-dimensional attribute parameters show that its color is highly similar to the target stimulus signal, then the training object's discrimination ability in that perceptual dimension can be targeted by increasing the proportion of this type of interfering stimulus signal in subsequent composite stimulus sequences, or by further reducing its color difference with the target stimulus signal.

[0129] Each significantly confusing feature type has its multi-dimensional attribute parameters associated with its corresponding confusion level value, forming a feature-confusion level mapping structure. This mapping structure uses the identifier encoding of the significantly confusing feature type as the key and a composite object containing the multi-dimensional attribute parameter vector and the confusion level value as the value, forming a dictionary-like data structure. The association between the multi-dimensional attribute parameter vector and the confusion level value ensures that when adjusting the distribution of the interference stimulus signal features in subsequent operations, the physical description information of this type of feature and its quantified degree of interference to the training object can be obtained simultaneously, achieving an integrated expression of feature description and interference intensity.

[0130] The feature confusion distribution and feature-confusion mapping structure are encapsulated into attention-biased pattern data. This encapsulated attention-biased pattern data comprises two core components: first, a feature confusion distribution arranged from highest to lowest confusion value, reflecting the relative interference intensity ranking of each confusing feature type; and second, a feature-confusion mapping structure, recording the binding relationship between the multi-dimensional perceptual attributes of each significant confusing feature type and its confusion value. These two parts complement each other: the feature confusion distribution provides a macroscopic overview of attention bias, while the feature-confusion mapping structure provides microscopic details of feature attributes, together forming a complete attention-biased pattern data package. This data package is used in subsequent training cycles to adjust the feature distribution of interfering stimulus signals in the compound stimulus sequence, enabling the training content to continuously focus on the training object's attention-deficient areas, achieving personalized, progressive adjustment of interference load.

[0131] In practical applications, the generation cycle of attention bias pattern data is synchronized with the training and evaluation cycle. An update process is triggered after each preset training period. The update process re-examines the error response records within that period, recalculates the confusion value of each confusing feature type, and compares it with historical attention bias pattern data to identify feature types with significant changes in confusion value. If the confusion value of a certain feature type continuously decreases and falls below the significance threshold over multiple training periods, it is removed from the list of significant confusing feature types, indicating that the training object's discrimination ability in that perceptual dimension has been effectively improved. If a new feature type with a confusion value exceeding the significance threshold appears, it is added to the list of significant confusing feature types, its multi-dimensional attribute parameters are extracted, and the feature-confusion mapping structure is updated to ensure that the attention bias pattern data always reflects the current attention bias state of the training object.

[0132] The method further includes:

[0133] When acquiring basic reaction ability data from trainees, the trainees stand at the center marked position of the training area, facing the main display screen or stimulus generator. The acquisition equipment includes a depth camera, an RGB camera, and a wearable inertial measurement unit (IMU). The cameras are mounted directly in front of the training area at a height of 1.5m to 2m above the ground, with a field of view covering the entire range of motion within the training area. The wearable IMU is fixed to the trainee's lower back or chest, with a sampling frequency set to 100Hz to 200Hz. The system presents a benchmark stimulus sequence to the trainee, containing 10 to 20 sets of single-target stimulus signals, with an interval of 2s to 5s between each set. The target stimulus signals are presented as specific color spots or graphic symbols via an LED array or display screen, lasting 0.5s to 1s. After the stimulus signal appears, the trainee must quickly move in the direction corresponding to the stimulus signal and touch the target area or return to the center position.

[0134] A depth camera captures real-time 3D skeletal keypoint coordinate sequences of the training subject, including keypoints on the head, shoulders, elbows, wrists, hips, knees, and ankles. An inertial measurement unit (IMU) acquires triaxial acceleration and triaxial angular velocity data. For each stimulus response, the system records the timestamps of the stimulus signal emission, the start of the subject's center of mass displacement, and the moment the subject touches the target area or completes the specified action. The determination of the start of center of mass displacement is based on the horizontal displacement of the hip center point exceeding a preset threshold (0.05m to 0.1m), or the acceleration amplitude detected by the IMU exceeding 3 to 5 times the standard deviation of the resting state. The time difference between the stimulus emission and the start of center of mass displacement is the single reaction delay time. The time difference between the start of center of mass displacement and the completion of the action is the single movement execution time.

[0135] The system statistically analyzes the reaction delay times of all valid responses in the benchmark stimulus sequence. After removing outliers exceeding three standard deviations above or below the average, the arithmetic mean of the remaining data is calculated as the average reaction time, and the standard deviation of the remaining data is calculated as the reaction stability index. The same processing is applied to the movement execution time of all valid responses to obtain the average movement time and movement stability index. The system uses the average reaction time, reaction stability index, average movement time, movement stability index, and the ratio of correct responses to the total number of stimuli to constitute the basic reaction ability data. In this data, a correct response is defined as the training subject completing an action matching the direction or type of the stimulus signal within a preset time window after the stimulus signal is emitted. The preset time window is set to 0.3s to 3s.

[0136] When determining the initial interference ratio threshold based on basic reaction ability data, the system compares the average reaction time with a preset reaction time grading standard. The reaction time grading standard is divided into fast reaction level, medium reaction level, and slow reaction level. The fast reaction level corresponds to an average reaction time of less than 0.4s, the medium reaction level corresponds to an average reaction time between 0.4s and 0.6s, and the slow reaction level corresponds to an average reaction time greater than 0.6s. If the training subject belongs to the fast reaction level and the correct response ratio is higher than 0.9, the initial interference ratio threshold is set to 0.3 to 0.4, indicating that the proportion of interference stimulus signals in the subsequent composite stimulus sequence is 30% to 40% of the total stimulus signals. If the training subject belongs to the medium reaction level and the correct response ratio is between 0.7 and 0.9, the initial interference ratio threshold is set to 0.2 to 0.3. If the training subject belongs to the slow reaction level or the correct response ratio is lower than 0.7, the initial interference ratio threshold is set to 0.1 to 0.2. The system also makes fine adjustments based on the reaction stability index. When the reaction stability index, i.e., the standard deviation, is greater than 30% of the average reaction time, the initial interference ratio threshold is reduced by 0.05 to 0.1 from the above range to adapt to the situation where the consistency of the training subjects' reactions is poor.

[0137] When a composite stimulus sequence is presented within the training area, the system generates a mixed sequence containing both the target stimulus signal and the interference stimulus signal based on an initial interference ratio threshold. The target stimulus signal is set as a green light spot or an upward arrow graphic in the perceptual feature dimension, and the interference stimulus signal is set as a red light spot or a downward arrow graphic. Alternatively, the target stimulus signal can be set as a high-frequency sound signal with a frequency of 1000Hz to 2000Hz, and the interference stimulus signal can be set as a low-frequency sound signal with a frequency of 200Hz to 500Hz. The presentation position of each stimulus signal in the composite stimulus sequence is randomly distributed within the training area, or randomly selected according to a preset spatial grid. The time interval between stimulus signals is randomly sampled within the range of 1s to 4s to avoid the training subject developing a fixed rhythm expectation. The duration of each stimulus signal is set to 0.5s to 1.5s, after which the stimulus signal disappears.

[0138] When collecting data on the trainees' response to the compound stimulus sequence, a depth camera continuously records the skeletal keypoint coordinate sequence at a frame rate of 30fps to 60fps. An inertial measurement unit (IMU) acquires triaxial acceleration, triaxial angular velocity, and triaxial magnetometer data at a frequency of 100Hz to 200Hz. The training area is covered with a pressure-sensitive mat or smart mat, each containing several pressure-sensing units, with each unit covering an area of ​​0.1m². 2 up to 0.2m 2The sampling frequency is from 50Hz to 100Hz. When the trainee's foot contacts a sensing unit on the mat, the unit outputs a pressure value exceeding a preset trigger threshold, and the system records the trigger time timestamp and trigger location coordinates. The system simultaneously records the timestamp of each stimulus signal emission, the stimulus signal type identifier, and the spatial location coordinates of the stimulus signal.

[0139] When determining the stimulus type corresponding to the response action data based on the identified attributes, the system first identifies the action response segments of the training subject after each stimulus signal is emitted. The start time of the action response segment is defined as the moment when the training subject's center of mass first undergoes a significant displacement after the stimulus signal is emitted. The displacement is determined by a change in the coordinates of the hip center point exceeding 0.05m or an inertial measurement unit acceleration amplitude exceeding the mean of the resting state plus three standard deviations. The end time of the action response segment is defined as the moment when the training subject completes the movement and enters a relatively static state. The static state is determined by a change in the coordinates of the hip center point within a continuous time window of 0.5s to 1s of less than 0.03m and an inertial measurement unit acceleration amplitude lower than the mean of the resting state plus two standard deviations.

[0140] The system analyzes the movement direction vector of the training subject in the action response segment. The movement direction vector is obtained by calculating the displacement vector of the hip center point coordinates between the start and end times of the action response segment. If the stimulus signal is visual, the system extracts the spatial position coordinates of the stimulus signal within the training area and calculates the direction vector between the training subject's starting position and the stimulus signal position as the target direction vector. If the stimulus signal is auditory, the system determines the target direction vector based on the semantic content or audio characteristics of the auditory stimulus; for example, high frequencies correspond to the forward direction, and low frequencies correspond to the backward direction. The system calculates the angle between the movement direction vector and the target direction vector. When the angle is less than a preset angle threshold, such as 45% to 60%, it determines that the stimulus type of the training subject's response matches the type identifier of the stimulus signal. If the type identifier of the stimulus signal is the target stimulus signal, the response action is recorded as a correct response. If the type identifier of the stimulus signal is an interference stimulus signal but the training subject still responds, or if the direction of the training subject's response is inconsistent with the direction of the target stimulus signal, the response action is recorded as an incorrect response.

[0141] The system statistically analyzes all response data within a preset time period, such as a training round or a training day. It calculates the correct response rate as the ratio of correct responses to the total number of stimuli, and the incorrect response rate as the ratio of incorrect responses to the total number of stimuli. The total number of stimuli includes the number of target stimulus signals and the number of interfering stimulus signals. Correct responses include correct responses to the target stimulus signal and correct ignoring of interfering stimulus signals, where correct ignoring is defined as the training subject not producing a significant movement response within a preset time window after the interfering stimulus signal is emitted. Incorrect responses include incorrect responses to interfering stimulus signals and incorrect response direction or failure to respond to the target stimulus signal within the specified time. The system uses the correct response rate, incorrect response rate, average reaction delay time, and standard deviation of reaction delay time to construct the response accuracy evaluation result.

[0142] When dynamically adjusting the proportion of interference stimuli based on the response accuracy assessment results and preset advancement criteria, the system compares the correct response rate of the current training round with the advancement threshold set in the advancement criteria. The advancement threshold is set to 0.8 to 0.9, indicating that the correct response rate needs to reach 80% to 90%. The system maintains a sliding window to record the correct response rate of the most recent training rounds, with the sliding window length set to 3 to 5 training rounds. When the correct response rate of all training rounds within the sliding window reaches or exceeds the advancement threshold, the system determines that the training object meets the advancement conditions and triggers the interference proportion adjustment operation. The interference proportion adjustment operation increases the proportion of the current interference stimuli in the compound stimulus sequence by an increment step, set to 0.05 to 0.1, that is, the interference proportion increases from the current value, for example, 0.3, to 0.35 or 0.4. If the correct response rate of any training round within the sliding window is lower than the advancement threshold, the system maintains the current interference proportion unchanged and continues training at the current difficulty level.

[0143] When synchronously shortening the presentation duration of a single stimulus in a compound stimulus sequence, the system calculates the reduction in presentation duration based on the interference ratio adjustment. A linear or non-linear correlation is established between the presentation duration reduction and the increase in the interference ratio; for example, for every 0.1 increase in the interference ratio, the presentation duration is reduced by 0.1s to 0.2s. The updated presentation duration is obtained by subtracting the reduction from the current presentation duration. The updated presentation duration has a lower limit constraint, set at 0.3s to 0.5s, to prevent the presentation duration from being too short and causing the training subjects to be unable to effectively perceive the stimulus signal. The system applies the updated interference ratio and the updated presentation duration to the control parameters for generating subsequent compound stimulus sequences. The number of interference stimulus signals in the subsequent compound stimulus sequence is calculated based on the product of the updated interference ratio and the total number of stimulus signals, and the presentation duration of each stimulus signal is set to the updated presentation duration.

[0144] When extracting features from the interference stimulus types corresponding to the error response rate, the system iterates through all error response records within a preset time period and extracts the feature type identifier of the interference stimulus signal corresponding to each error response. The feature type identifier includes the perceptual modality category, color attribute, shape attribute, audio frequency range, spatial location region, and temporal occurrence pattern of the interference stimulus signal. The system establishes a feature type statistics table, merges error responses with the same feature type identifier, and accumulates the number of occurrences of each confusing feature type in all error response records. The total number of all error response records is calculated, and the number of occurrences of each confusing feature type is divided by the total number to obtain the confusion value corresponding to that confusing feature type. All confusing feature types and their corresponding confusion values ​​are sorted from high to low according to the confusion value to generate a feature confusion distribution. Confusion feature types with confusion values ​​exceeding a preset significance threshold are selected from the feature confusion distribution. The significance threshold is set to 0.15 to 0.25, indicating that the proportion of this confusing feature type in the error response exceeds 15% to 25%. The selected confusing feature types are marked as high-confusion feature types.

[0145] The system extracts multi-dimensional attribute parameters of highly confusing feature types across the perceptual feature dimension. These parameters include hue, saturation, brightness, shape complexity, and spatial coordinate distribution range for visual stimuli, and frequency center value, frequency bandwidth, sound pressure level, and temporal envelope features for auditory stimuli. These multi-dimensional attribute parameters are then associated with their corresponding confusion values ​​to form a feature confusion mapping structure. The system encapsulates the feature confusion distribution and feature confusion mapping structure into attention-biased pattern data, which is stored in the training subjects' profiles.

[0146] When adjusting the feature distribution of interfering stimulus signals in subsequent compound stimulus sequences based on attention-biased pattern data, the system reads the highly confusing feature types and their confusion values ​​from the attention-biased pattern data. For the several highly confusing feature types with the highest confusion values, the system increases the frequency of these feature types in the interfering stimulus signals in subsequent compound stimulus sequences. The increase in frequency is directly proportional to the confusion value; the higher the confusion value, the greater the increase in the frequency of this feature type in the interfering stimulus signal. Specifically, the adjustment strategy is to increase the proportion of the number of interfering stimulus signals of this feature type to the total number of interfering stimulus signals by 0.1 to 0.2. Simultaneously, the system adjusts the feature similarity between the interfering stimulus signal and the target stimulus signal based on the multi-dimensional attribute parameters in the feature confusion mapping structure. For the multi-dimensional attribute parameters corresponding to the highly confusing feature types, the differences between the system-generated interfering stimulus signal and the target stimulus signal in these attribute dimensions gradually decrease. For example, the difference in color hue value gradually decreases from the initial 60% to 90% to 30% to 45%, and the difference in audio frequency gradually decreases from the initial 500Hz to 800Hz to 200Hz to 400Hz. By increasing the training intensity of highly confusing feature types and increasing the difficulty of feature similarity, the training subjects are encouraged to strengthen their ability to identify and suppress these easily confused interference stimuli.

[0147] A second aspect of the present invention provides a system for improving athletic agility in adolescents by combining audiovisual response training, comprising:

[0148] An initial threshold unit is used to acquire basic reaction ability data of the training object and determine an initial interference ratio threshold based on the basic reaction ability data.

[0149] The response evaluation unit is used to present a composite stimulus sequence consisting of a target stimulus signal and an interference stimulus signal in the training area, wherein the target stimulus signal and the interference stimulus signal have distinguishable identification attributes in the dimension of perceptual features, collect the response action data of the training object to the composite stimulus sequence, and determine the stimulus type corresponding to the response action data based on the identification attributes, and generate a response accuracy evaluation result including the correct response rate and the incorrect response rate.

[0150] The load escalation unit is used to dynamically adjust the proportion of the interference stimulus signal in the subsequent composite stimulus sequence based on the response accuracy evaluation result and the preset escalation criteria, and simultaneously shorten the presentation duration of a single stimulus in the composite stimulus sequence, forming a progressive interference load escalation strategy.

[0151] The bias adjustment unit is used to extract features of the interference stimulus type corresponding to the error response rate, generate attention bias pattern data, and adjust the feature distribution of interference stimulus signals in subsequent composite stimulus sequences based on the attention bias pattern data.

[0152] A third aspect of the present invention provides an electronic device, comprising:

[0153] processor;

[0154] Memory used to store processor-executable instructions;

[0155] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0156] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0157] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0158] 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 method for improving athletic agility in adolescents by combining audiovisual response training, characterized in that: include: Acquire basic reaction ability data of the training subjects, and determine the initial interference ratio threshold based on the basic reaction ability data; A composite stimulus sequence consisting of a target stimulus signal and an interference stimulus signal is presented within the training area. The target stimulus signal and the interference stimulus signal have distinguishable identification attributes in the dimension of perceptual features. The response action data of the training object to the composite stimulus sequence is collected, and the stimulus type corresponding to the response action data is determined based on the identification attributes. A response accuracy evaluation result including the correct response rate and the incorrect response rate is generated. Based on the accuracy assessment results of the response and the preset advancement criteria, the proportion of the interference stimulus signal in the subsequent compound stimulus sequence is dynamically adjusted, and the presentation duration of a single stimulus in the compound stimulus sequence is shortened simultaneously, forming a progressive interference load increasing strategy. The interference stimulus type corresponding to the error response rate is feature extracted to generate attention bias pattern data, and the feature distribution of interference stimulus signals in subsequent composite stimulus sequences is adjusted based on the attention bias pattern data.

2. The method according to claim 1, wherein, The process involves collecting the training subjects' response action data to the compound stimulus sequence, determining the stimulus type corresponding to the response action data based on the identified attributes, and generating a response accuracy assessment result including correct response rate and incorrect response rate, including: The input acquisition device records all the response actions of the training object during the presentation of the compound stimulus sequence. The response timestamp of each response action is matched with the presentation time window of each stimulus signal in the compound stimulus sequence to determine the trigger stimulus signal corresponding to the response action. Extract the identification attributes of the triggering stimulus signal, determine whether the response operation type of the response action is consistent with the expected response type indicated by the identification attributes of the triggering stimulus signal, if consistent, classify the response action as a correct response, if inconsistent, classify it as an incorrect response; The number of correct responses and the number of incorrect responses within a preset time period are statistically analyzed. The ratio of the number of correct responses to the total number of responses is calculated to obtain the correct response rate. The ratio of the number of incorrect responses to the total number of responses is calculated to obtain the incorrect response rate. The correct response rate and the incorrect response rate are then packaged into a response accuracy assessment result.

3. The method according to claim 1, wherein, The step involves dynamically adjusting the proportion of the interference stimulus signal in the subsequent complex stimulus sequence based on the response accuracy assessment results and preset advancement criteria, and simultaneously shortening the presentation duration of a single stimulus in the complex stimulus sequence, forming a progressive interference load escalation strategy, including: The correct response rate in the response accuracy assessment results is compared with the promotion threshold set in the promotion criteria. When the number of times the correct response rate continuously reaches or exceeds the promotion threshold meets the stability judgment condition, the load increment operation is triggered. When the load increment operation is triggered, the increment of the interference ratio is calculated based on the current interference ratio and a preset nonlinear increment function, and the increment of the interference ratio is added to the current interference ratio to obtain the updated interference ratio. Based on the coupling constraint relationship between the updated interference ratio and the current presentation duration, the presentation duration reduction amount is calculated, and the updated presentation duration is obtained by subtracting the presentation duration reduction amount from the current presentation duration. When the correct response rate does not reach the advancement threshold, the current interference ratio and presentation duration are maintained unchanged, and a fine-tuning mechanism is activated to slightly extend the presentation duration. The updated interference ratio and the updated presentation duration are used as the generation control parameters for subsequent composite stimulus sequences. Through the dual-dimensional coordinated adjustment of increasing interference density and compressing the recognition time window, a progressive interference load increase strategy is formed.

4. The method according to claim 3, wherein, When the load increment operation is triggered, the step of calculating the interference percentage increment based on the current interference percentage and a preset nonlinear increment function, and then adding the interference percentage increment to the current interference percentage to obtain the updated interference percentage, includes: Obtain the number of interfering stimulus signals and the total number of stimulus signals in the current compound stimulus sequence, and calculate the ratio of the two as the current interference ratio; The current interference ratio is substituted into a nonlinear increasing function as an input variable, and an inverse correlation mapping relationship between the current interference ratio and the interference ratio increment is established through the nonlinear increasing function. The interference percentage increment is calculated through the reverse correlation mapping relationship. According to the reverse correlation mapping relationship, the interference percentage increment decreases as the current interference percentage increases. Thus, in the early stage of training, the interference density is rapidly increased with an increment greater than the preset increment threshold, and in the later stage of training, the difficulty gradient is finely adjusted with an increment less than the preset increment threshold. The updated interference ratio is obtained by numerically superimposing the incremental interference ratio with the current interference ratio.

5. The method according to claim 1, wherein, The step of extracting features from the interference stimulus type corresponding to the error response rate to generate attention bias pattern data, and adjusting the feature distribution of interference stimulus signals in subsequent composite stimulus sequences based on the attention bias pattern data, includes: Extract all error response records from the response accuracy assessment results, obtain the trigger stimulus signal corresponding to each error response, identify the signal objects that belong to the interference stimulus signals in the trigger stimulus signals, and extract the feature vector of the signal objects. Cluster analysis is performed on all extracted feature vectors to identify feature clusters whose frequency exceeds a preset frequency threshold, and these feature clusters are marked as confusing feature types. The occurrence frequency of each obfuscation feature type in error responses is counted, its proportion in all error responses is calculated, a feature obfuscation distribution is generated, and the obfuscation feature type and its corresponding feature obfuscation distribution are encapsulated as attention-biased pattern data. When generating subsequent compound stimulus sequences, the proportion of interference stimulus signals corresponding to the confusion feature types in the sequence is increased based on the feature confusion distribution in the attention bias pattern data.

6. The method according to claim 5, wherein, The process of counting the occurrence frequency of each obfuscation feature type in error responses, calculating its proportion of all error responses, generating a feature obfuscation distribution, and encapsulating the obfuscation feature types and their corresponding feature obfuscation distributions into attention-biased pattern data includes: Traverse all error response records within a preset time period, extract the feature type identifier of the interference stimulus signal corresponding to each error response, establish a feature type statistics table, merge error responses with the same feature type identifier, and accumulate the number of times each confusion feature type appears in all error response records. Calculate the total number of all error response records, divide the number of occurrences of each obfuscation feature type by the total number, and obtain the obfuscation level value corresponding to the obfuscation feature type. Sort each confusing feature type and its corresponding confusion value from high to low according to the confusion value to generate a feature confusion distribution; From the feature confusion distribution, filter out confusion feature types whose confusion values ​​exceed a preset significance threshold, and mark the filtered confusion feature types as confusion feature types; Extract the multi-dimensional attribute parameters of the confusing feature type in the perceptual feature dimension, associate and bind the multi-dimensional attribute parameters with the corresponding confusion degree values ​​to form a feature-confusion degree mapping structure, and encapsulate the feature confusion degree distribution and the feature-confusion degree mapping structure into attention bias pattern data.

7. A youth agility enhancement system combining audiovisual response training, used to implement the method as described in any one of claims 1-6, characterized in that, include: An initial threshold unit is used to acquire basic reaction ability data of the training object and determine an initial interference ratio threshold based on the basic reaction ability data. The response evaluation unit is used to present a composite stimulus sequence consisting of a target stimulus signal and an interference stimulus signal in the training area, wherein the target stimulus signal and the interference stimulus signal have distinguishable identification attributes in the dimension of perceptual features, collect the response action data of the training object to the composite stimulus sequence, and determine the stimulus type corresponding to the response action data based on the identification attributes, and generate a response accuracy evaluation result including the correct response rate and the incorrect response rate. The load escalation unit is used to dynamically adjust the proportion of the interference stimulus signal in the subsequent composite stimulus sequence based on the response accuracy evaluation result and the preset escalation criteria, and simultaneously shorten the presentation duration of a single stimulus in the composite stimulus sequence, forming a progressive interference load escalation strategy. The bias adjustment unit is used to extract features of the interference stimulus type corresponding to the error response rate, generate attention bias pattern data, and adjust the feature distribution of interference stimulus signals in subsequent composite stimulus sequences based on the attention bias pattern data.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images