An intelligent management method for interactive data of reading pens based on AI analysis

Abstract

The present invention is applicable to the field of human-computer interaction technology, and provides an intelligent management method for interactive data of a reading pen based on AI analysis, the method comprising: obtaining the target individual's reading interaction data, physiological signal data, and initial content level parameters corresponding to the current content to be pushed; screening out a number of historical samples from the reading interaction data, requiring that the content level parameters corresponding to each historical sample be consistent with the initial content level parameters. The present invention proposes a method for dynamically correcting content level parameters by combining a main correction coefficient and an auxiliary correction coefficient by constructing an analysis mechanism based on the correlation law between an individual's physiological state and the reading effect score. The system first determines whether there is a correlation law that the worse the physiological state, the lower the score and the difference amplification trend is significant. On this basis, a main correction coefficient is generated to reflect the degree of deviation of the current state, and the auxiliary correction coefficient quantifies its evolution direction and speed based on the trend of physiological state changes.

Description

Technical Field

[0001] The present invention belongs to the field of human-computer interaction technology, and in particular relates to an intelligent management method for interactive data of a reading pen based on AI analysis. Background Art

[0002] Currently, reading pens, as a typical interactive learning tool for children, are widely used in scenarios such as early childhood enlightenment and reading assistance. Traditional reading pen systems often use fixed graphic content or preset voice feedback as push content, usually performing coarse-grained content grading based on age groups or textbook progress, and lack the ability to deeply identify individual user differences and dynamically adapt. Although some more advanced systems have introduced behavior recording functions, such as counting clicks, identifying frequently read areas, or generating reading preferences, the content push strategy still relies primarily on preset rules and cannot be intelligently judged based on the user's real-time status, making it difficult to effectively match the content with the user's current receptive ability or attention state.

[0003] In existing point-to-point content push logic, the system often makes push decisions based solely on static behavioral tags or single indicators (such as click frequency or error rate), failing to dynamically adjust based on factors such as the user's current physiological state, emotional level, or interaction rhythm. Even if some products integrate wearable devices, the physiological signals collected are only used for single-point abnormality alerts or post-analysis, and a real-time closed-loop mechanism between physiological state and content push has not yet been established. This limitation results in the system's lack of flexible adjustment capabilities in the face of individual state fluctuations, which can easily lead to problems such as content intensity mismatch, excessive learning pressure, or insufficient motivation, affecting learning outcomes and user experience. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent management method for interactive data of a reading pen based on AI analysis, aiming to solve the problems raised in the background technology.

[0005] The present invention is implemented as follows: a method for intelligent management of interactive data of a reading pen based on AI analysis, the method comprising:

[0006] Obtain the target individual's click-and-read interaction data, physiological signal data, and initial content level parameters corresponding to the content to be pushed;

[0007] Filter out several historical samples from the click-and-read interaction data, require that the content level parameters corresponding to each historical sample are consistent with the initial content level parameters, and determine the click-and-read effect score corresponding to each historical sample;

[0008] Analyze several historical samples to determine whether there is a trend of change in the same direction between the target individual's physiological state and the reading effect score, and whether the reading effect score shows a correlation pattern of enhanced response to physiological state deviation;

[0009] On the premise of confirming the existence of a correlation pattern, the historical sample with the highest reading effect score is selected as the reference sample, the reference physiological state value associated with the reference sample is extracted, and the real-time physiological state value of the target individual at the current moment is collected. The main correction coefficient is calculated based on the deviation between the two;

[0010] Collect continuous physiological state values within a preset time interval before the current moment, and calculate and generate auxiliary correction coefficients based on their change trends;

[0011] The initial content level parameters are dynamically corrected by combining the main correction coefficient and the auxiliary correction coefficient.

[0012] As a further limitation of the technical solution of the embodiment of the present invention, the steps of screening a plurality of historical samples from the click-and-read interaction data, requiring that the content level parameters corresponding to each historical sample are consistent with the initial content level parameters, and determining the click-and-read effect score corresponding to each historical sample include:

[0013] Parse the click-and-read interaction data, extract the local click-and-read interaction data corresponding to all historical push events, and obtain the content level parameters of the push content associated with each local click-and-read interaction data;

[0014] Filter out local click-and-read interaction data with content-level parameters that are consistent with initial content-level parameters and use them as historical samples;

[0015] Analyze each historical sample and obtain the behavioral response data of the target individual within the preset time window after the push content is sent. The behavioral response data specifically includes at least one or more indicators of the target individual's point reading accuracy, point reading response duration and repeated click frequency within the preset time window, and calculate the point reading effect score of the historical sample based on these indicators through weighted calculation.

[0016] As a further limitation of the technical solution of the embodiment of the present invention, the steps of analyzing a plurality of historical samples to determine whether there is a trend of change in the same direction between the physiological state of the target individual and the reading effect score, and whether the reading effect score shows a correlation regularity of enhanced response to the physiological state deviation include:

[0017] Extract the click-and-read effect score corresponding to each historical sample, and obtain the physiological state value of the historical sample at the time the pushed content was sent based on the physiological signal data;

[0018] Sort the physiological state values of several historical samples from large to small to generate a physiological state sequence, and extract the corresponding reading effect score of each historical sample in the sorted order to generate a score sequence. Determine whether the physiological state sequence and the score sequence both show a decreasing trend. If so, it is determined that there is a trend of change in the same direction between the physiological state of the target individual and the reading effect score;

[0019] On the basis of confirming the existence of a trend of change in the same direction, we analyze whether there is an enhanced response effect in the physiological state sequence and the score sequence, in which the greater the difference in the physiological state value, the more significant the increase in the difference in the point-reading effect score. If so, we confirm that a correlation law is established.

[0020] As a further limitation of the technical solution of the embodiment of the present invention, the physiological state value refers to: the physiological parameter data of the target individual collected by the wearable device, the numerical indicator used to characterize the physiological state of the target individual obtained after feature extraction and standardization processing, the physiological parameter data is a weighted combination of one or more of heart rate, skin conductivity, body surface temperature, body movement amplitude and respiratory rate; wherein, the physiological state value is a monotonic positive indicator, and the higher the value, the better the physiological state of the target individual.

[0021] As a further limitation of the technical solution of the embodiment of the present invention, under the premise of confirming the existence of a correlation law, the steps of selecting the historical sample with the highest reading effect score as a reference sample, extracting the reference physiological state value associated with the reference sample, and collecting the real-time physiological state value of the target individual at the current moment, and calculating and generating the main correction coefficient based on the deviation between the two include:

[0022] After confirming the existence of correlation patterns, the historical samples with the highest reading effect scores are selected as reference samples, and their corresponding physiological state values are extracted as reference physiological state values;

[0023] Collect the real-time physiological status value of the target individual at the current moment;

[0024] The deviation between the real-time physiological state value and the reference physiological state value is calculated, and a main correction coefficient is generated based on the deviation.

[0025] As a further limitation of the technical solution of the embodiment of the present invention, the step of collecting continuous physiological state values within a preset time interval before the current moment and calculating and generating the auxiliary correction coefficient based on the change trend thereof includes:

[0026] Collecting several sets of physiological status values of the target individual within a preset time interval before the current moment, wherein the collection time interval of the physiological status values is preset and consistent;

[0027] Constructing a physiological state value time series from a number of groups of physiological state values in chronological order, and calculating the change slope of the physiological state value time series;

[0028] A corresponding auxiliary correction coefficient is generated based on the value of the change slope.

[0029] As a further limitation of the technical solution of the embodiment of the present invention, when the initial content level parameter is dynamically corrected, a preset parameter correction model is called based on the main correction coefficient and the auxiliary correction coefficient to achieve joint correction;

[0030] The parameter correction model is: ,in Refers to the revised content level parameters, Refers to the initial content level parameters, Refers to the main correction coefficient, that is, the deviation between the real-time physiological state value and the reference physiological state value. Refers to the adjustment weight corresponding to the main correction coefficient, Refers to the auxiliary correction coefficient, that is, the slope of the change of the physiological state value time series, Refers to the adjustment weight corresponding to the auxiliary correction coefficient, Refers to the trigger threshold of the main correction factor.

[0031] As a further limitation of the technical solution of the embodiment of the present invention, after the revised content level parameters are generated, push content of the corresponding level is matched in a preset content corpus based on the content level parameters and pushed to the target individual.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] This paper proposes a dynamic correction method for content rating parameters by building an analysis mechanism based on the correlation between individual physiological status and reading effect scores, combining primary and secondary correction coefficients. The system first determines whether there is a correlation pattern whereby worse physiological status leads to lower scores, with a significant trend of differential amplification. Based on this correlation, it generates a primary correction coefficient to reflect the degree of deviation from the current status. The secondary correction coefficient quantifies the direction and speed of physiological status evolution based on the changing trend.

[0034] This joint correction mechanism effectively improves the rationality and foresight of content adjustments, enabling the system to respond to the current status and perceive risk trends when pushing content, thereby achieving a more accurate, robust and personalized content level adjustment effect, which is significantly different from the single indicator-driven solution in existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A flowchart of a method provided by an embodiment of the present invention;

[0036] Figure 2 A flowchart of determining historical samples in the method provided in an embodiment of the present invention;

[0037] Figure 3 A flow chart of establishing correlation rules in the method provided in an embodiment of the present invention;

[0038] Figure 4 A flow chart of generating a main correction coefficient in a method provided in an embodiment of the present invention;

[0039] Figure 5 This is a flow chart of generating auxiliary correction coefficients in the method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0041] Figure 1 A flow chart of a method provided by an embodiment of the present invention is shown.

[0042] Specifically, a method for intelligent management of interactive data of a reading pen based on AI analysis includes the following steps:

[0043] Step S100: obtaining the target individual's point-reading interaction data, physiological signal data, and initial content level parameters corresponding to the content to be pushed.

[0044] In this embodiment of the present invention, the target individual refers to a specific user currently using the reading pen for learning or reading, typically a child user whose device account is bound or whose identity is recognized by the system. When performing intelligent management, the system uses this target individual as the core unit, collecting and analyzing their interactive behavior and physiological status to dynamically adjust the personalized content push strategy.

[0045] The current content to be pushed refers to the specific content fragments that the system is about to present or feedback to the target individual in the current point-reading interaction process. It may include text paragraphs, knowledge points, prompt statements or encouraging voices, etc. Its content structure has been preset in the system content resource library and is equipped with corresponding level annotations. The corresponding initial content level parameter refers to the default level value pre-annotated in the content resource library for the content to be pushed, which is used to describe the difficulty or adaptation level of the content in terms of language complexity, information density, expression or psychological guidance. The level parameter can be derived from the existing content annotation mechanism in existing education or human-computer interaction systems, and is often automatically generated through manual expert evaluation, semantic complexity calculation, text readability analysis or corpus label learning model.

[0046] The point-reading interaction data is collected in real time by the reading pen and its supporting recognition module during the point-reading operation of the target individual. Specifically, the target object can be identified through the linkage between the optical recognition component at the bottom of the reading pen and the coded paper surface. At the same time, the internal timing processing unit and the interactive control logic are combined to record the time, location, frequency and response behavior of the click event, and synchronize it to the interactive data management module through local cache or wireless transmission.

[0047] Physiological signal data comes from wearable physiological monitoring devices worn by the target individual, such as smart bracelets, smart watches, or somatosensory modules. These devices utilize built-in multimodal sensors (such as photoplethysmography sensors, skin resistance sensors, temperature sensors, and triaxial accelerometers) to synchronously acquire dynamic parameters related to the target individual's physiological state while the target individual is performing a reading. These devices are typically linked to the reading pen system via Bluetooth or a proprietary protocol. Data collection automatically begins when the pen is activated or a task begins, and is then transmitted to the system backend for combined behavioral and physiological analysis.

[0048] It should be noted that the point-reading interaction data and physiological signal data are data collected by the system during the period when the target individual last started to use the reading pen continuously and continued to perform effective point-reading operations. This data interval is real-time and continuous, and can reflect the interactive behavior characteristics and physiological reaction status exhibited by the target individual during the actual point-reading process, avoiding analysis distortion caused by discontinuous operations or breakpoints.

[0049] Click-to-read interaction data includes at least the target individual's click location, click timestamp, clicked object identifier, click frequency, click response duration, any misclicks, and any repeated clicks during each click-to-read event. This data can be used to assess the target individual's understanding of the current content, their level of focus, and the fluency of their interaction. Physiological signal data includes at least one or more parameters, such as heart rate, skin conductivity, body temperature, body movement amplitude, and respiratory rate, acquired through wearable devices during the click-to-read period, to reflect the target individual's real-time physiological state and emotional level.

[0050] Furthermore, the method for intelligent management of interactive data of the reading pen based on AI analysis further includes the following steps:

[0051] Step S200 , screen out a number of historical samples from the click-and-read interaction data, require that the content level parameters corresponding to each historical sample are consistent with the initial content level parameters, and determine the click-and-read effect score corresponding to each historical sample.

[0052] Specifically, Figure 2 A flow chart for determining historical samples is shown.

[0053] The process of selecting a number of historical samples from the click-and-read interaction data, requiring the content level parameters corresponding to each historical sample to be consistent with the initial content level parameters, and determining the click-and-read effect score corresponding to each historical sample specifically includes the following steps:

[0054] Step S201: parsing the click-and-read interaction data, extracting the local click-and-read interaction data corresponding to all historical push events, and obtaining the content level parameters of the push content associated with each local click-and-read interaction data;

[0055] Step S202, filtering out local click-and-read interaction data whose content level parameters are consistent with the initial content level parameters, and using them as historical samples;

[0056] Step S203: Analyze each historical sample to obtain the behavioral response data of the target individual within the preset time window after the push content is sent. The behavioral response data specifically includes at least one or more indicators of the target individual's point reading accuracy, point reading response duration, and repeated click frequency within the preset time window, and calculate the point reading effect score of the historical sample based on these indicators through weighted calculation.

[0057] In this embodiment of the present invention, the selection of historical push events involves extracting all data segments from the click-and-read interaction data that possess content-rating parameter identifiers and complete click-and-read behavior, and constructing a candidate historical sample set from these segments. To ensure the credibility of trend determination and score calculation, the system only performs historical sample screening and subsequent analysis when the candidate data volume meets a set threshold; otherwise, the content-rating parameter correction process will not be triggered. This setting is intended to prevent statistical distortion of scores or invalid trend determinations due to insufficient sample size.

[0058] The content rating parameter associated with each local click-through interaction is a synchronized data field generated and written by the system each time content is pushed. This field is recorded in the system log as part of that interaction data and can therefore be directly extracted when parsing the click-through interaction data. The system parses this field and compares it with the initial content rating parameters of the content to be pushed. It then selects data segments with consistent content rating parameters, which serve as historical samples for subsequent scoring modeling.

[0059] The purpose of selecting historical samples is to construct a subset of samples corresponding to the initial content level from the current click-and-read behavior process, thereby ensuring a consistent basis for comparison between click-and-read behavior and content complexity. Because click-and-read effectiveness scores are significantly affected by content difficulty, selecting only samples with the same level as the currently pushed content eliminates content-related interference and ensures that score differences more accurately reflect the target individual's physiological state or attention level.

[0060] The system uses a preset time window to extract the target individual's behavioral response segment in each historical sample. Its length is set to cover the click-and-read feedback cycle, for example, within 5 seconds of the push content being sent, encompassing both the initial click-and-read response and potential repeat clicks. This time window, serving as the boundary for extracting behavioral response data, directly impacts the stability and sensitivity of the scoring calculation.

[0061] In specific implementation, the system first identifies the push time point of each historical sample and extracts all interactive behavior data within the time window after that time point. The behavioral response data includes at least the following three dimensions: (1) click-through accuracy, which is the ratio of the number of valid clicks to the total number of clicks within the time period; (2) click-through response duration, which is the time interval between the push and the first valid click; and (3) repeated click frequency, which is the cumulative number of clicks on the same target content within the window. After the system standardizes the above indicators, it uses a weighted calculation model to generate a click-through effect score for the historical sample.

[0062] Furthermore, the method for intelligent management of interactive data of the reading pen based on AI analysis further includes the following steps:

[0063] Step S300 , analyzing several historical samples to determine whether there is a trend of change in the same direction between the target individual's physiological state and the reading effect score, and whether the reading effect score shows a correlation law of enhanced response to the physiological state deviation.

[0064] Specifically, Figure 3 A flow chart for establishing correlation rules is shown.

[0065] The analysis of several historical samples to determine whether there is a trend of change in the same direction between the physiological state of the target individual and the reading effect score, and whether the reading effect score shows a correlation law of enhanced response to the physiological state deviation specifically includes the following steps:

[0066] Step S301: extracting the click-reading effect score corresponding to each historical sample, and obtaining the physiological state value of the historical sample at the time when the pushed content was sent based on the physiological signal data;

[0067] Step S302: Sort the physiological state values of the historical samples from largest to smallest to generate a physiological state sequence, extract the corresponding reading effect score of each historical sample in the sorted order, generate a score sequence, and determine whether the physiological state sequence and the score sequence both show a decreasing trend. If so, it is determined that there is a trend of change in the same direction between the physiological state of the target individual and the reading effect score;

[0068] Step S303, based on the determination of the existence of a trend of change in the same direction, analyze whether there is an enhanced response effect in the physiological state sequence and the score sequence, in which the larger the difference in the physiological state value, the more significant the increase in the difference in the reading effect score. If so, confirm that a correlation law is established.

[0069] The physiological state value refers to: the physiological parameter data of the target individual collected by the wearable device, which is a numerical indicator used to characterize the physiological state of the target individual after feature extraction and standardization. The physiological parameter data is a weighted combination of one or more of heart rate, skin conductivity, body surface temperature, body movement amplitude and respiratory rate; among them, the physiological state value is a monotonically positive indicator, and the higher its value, the better the physiological state of the target individual.

[0070] In this embodiment of the present invention, the click-through effect score corresponding to historical samples can be calculated directly from the click-through interaction data. After constructing the historical sample set, the system extracts the click-through behavior response information of the target individual within a preset time window after the push content was sent for each sample. Based on behavioral parameters such as accuracy, response time, and repeat click frequency, the click-through effect score for each sample is calculated using a preset weighted scoring formula. Therefore, as a derivative indicator of interaction data, the click-through effect score is reproducible and objective, and can be directly derived from the interaction data without the need for additional external evaluation.

[0071] The physiological state value is a comprehensive numerical indicator used to quantify the current physiological state of a target individual. Its underlying data source is multi-source physiological parameter data collected by wearable devices worn by the target individual. The system performs feature extraction on this raw physiological signal data, including but not limited to extracting statistical features such as maximum, minimum, mean, standard deviation, and instantaneous rate of change. Through normalization, physiological parameters from different sources and dimensions are mapped to a unified numerical space. The resulting physiological state value can be a single-value score derived from a weighted fusion of multidimensional feature vectors, reflecting the target individual's overall physiological activity, stability, or level of stress at a specific moment. One or more of the following input features, heart rate, skin conductivity, surface temperature, body movement amplitude, and respiratory rate, are comprehensively considered. The weights can be set by expert experience or optimized through data-driven learning. Because this value is a monotonic positive indicator, higher values indicate a more stable physiological state and greater focus, while lower values indicate fatigue, anxiety, or distraction.

[0072] The specific implementation process of step S302 includes: the system extracts the physiological state value corresponding to each record in the historical sample and sorts them from largest to smallest to obtain a physiological state sequence; then, according to the order of the sorted results, it extracts the reading effect score of each sample in turn to form a score sequence. The system uses trend consistency judgment algorithms, such as decreasing trend identification, monotonicity testing, or correlation coefficient analysis, to determine whether the two sequences change in the same direction in the overall ranking. If higher physiological state values correspond to higher scores, and the overall trend is consistently decreasing, it is considered that there is a directional relationship between physiological state and reading effect score, indicating that at the current content level, the target individual's physiological state affects their reading performance and is exploitable.

[0073] Based on the trend consistency confirmed in the previous step, step S303 further analyzes whether there is an enhanced response effect between the physiological state value and the reading effect score. In specific implementation, the system calculates the difference between any two samples in the physiological state sequence and simultaneously calculates the difference between their corresponding scores. Statistical analysis is performed on multiple sample pairs to observe whether the following pattern exists: the larger the difference in physiological state value, the more significant the difference in reading effect score, showing an amplification effect. If this effect is significant, it indicates that the score change has a nonlinear characteristic in response to physiological state, and the scoring system is sensitive to state fluctuations, providing the foundation for building a dynamic regulation mechanism.

[0074] The purpose of establishing correlation patterns is to determine whether the target individual's behavior at the current content level is significantly influenced by their physiological state. If this correlation holds, it indicates that the current individual's reading performance is significantly affected by their physiological state, and the system needs to introduce a correction mechanism to adjust subsequent content push strategies. This pattern reflects the interactive relationship between reading behavior and individual state, and is a key prerequisite for the intelligent adjustment and content level adaptation achieved by the present invention.

[0075] Furthermore, the method for intelligent management of interactive data of the reading pen based on AI analysis further includes the following steps:

[0076] Step S400, on the premise of confirming the existence of a correlation law, select the historical sample with the highest reading effect score as the reference sample, extract the reference physiological state value associated with the reference sample, and collect the real-time physiological state value of the target individual at the current moment, and calculate and generate the main correction coefficient based on the deviation between the two.

[0077] Specifically, Figure 4 A flow chart for generating the main correction coefficients is shown.

[0078] Among them, under the premise of confirming the existence of a correlation law, the historical sample with the highest reading effect score is selected as the reference sample, the reference physiological state value associated with the reference sample is extracted, and the real-time physiological state value of the target individual at the current moment is collected. The main correction coefficient is calculated based on the deviation between the two, which specifically includes the following steps:

[0079] Step S401: After confirming the existence of a correlation pattern, select the historical sample with the highest reading effect score as a reference sample, and extract its corresponding physiological state value as a reference physiological state value;

[0080] Step S402, collecting the real-time physiological status value of the target individual at the current moment;

[0081] Step S403: Calculate the deviation between the real-time physiological state value and the reference physiological state value, and generate a main correction coefficient based on the deviation.

[0082] In this embodiment of the present invention, the significance of selecting the historical sample with the highest click-and-read effect score as the reference sample is that, from all historical samples that meet the correlation rule, the click-and-read interaction with the target individual's optimal performance and most stable state is prioritized as a reference benchmark to characterize the target individual's optimal response state under the current content level parameters. This reference sample not only possesses complete behavioral response characteristics and a high performance score, but also corresponds to a clear physiological state value. Based on this, the system can establish a "high performance-high state" correspondence, providing a reliable comparison standard for subsequent corrections to the current physiological state.

[0083] Based on the deviation between the physiological state value of the reference sample and the currently collected real-time physiological state value, the system can calculate and generate the main correction coefficient. This correction coefficient is obtained through relative deviation calculation, that is, the ratio of the change of the real-time physiological state value compared to the reference physiological state value. The reason for using the deviation as the basis for calculation is that it has the characteristics of unified dimension, good standardization, and strong adaptability, and can effectively measure the degree of deviation between the current state and the ideal state. The relative deviation method can also eliminate the influence of the absolute magnitude difference of the physiological state value, so that the system can maintain a stable judgment standard under different individuals or different physiological baselines.

[0084] The benefit of generating a master correction coefficient based on the deviation is that the system can dynamically adjust content level parameters based on the actual degree of deviation from the current state. This allows for flexible adaptation based on the real-time state changes of the target individual, even if the current content is initially fixed. This mechanism reflects the state-driven intelligent adjustment concept proposed in this invention: the further the individual's current physiological state deviates from the ideal state, the more the system tends to reduce the content complexity or information intensity to avoid excessive cognitive load. When the state approaches or exceeds the reference state, the system maintains or increases the current content level to enhance learning challenge and sense of achievement, thereby achieving more precise and flexible interactive strategy control.

[0085] Furthermore, the method for intelligent management of interactive data of the reading pen based on AI analysis further includes the following steps:

[0086] Step S500 , collecting continuous physiological state values within a preset time interval before the current moment, and calculating and generating an auxiliary correction coefficient based on the change trend thereof.

[0087] Specifically, Figure 5 A flow chart for generating auxiliary correction coefficients is shown.

[0088] The process of collecting continuous physiological state values within a preset time interval before the current moment and calculating and generating the auxiliary correction coefficient based on the change trend thereof specifically includes the following steps:

[0089] Step S501, collecting a plurality of sets of physiological status values of a target individual within a preset time interval before the current moment, wherein the collection time interval of the physiological status values is preset and consistent;

[0090] Step S502, constructing a physiological state value time series from a plurality of sets of physiological state values in chronological order, and calculating a change slope of the physiological state value time series;

[0091] Step S503: Generate a corresponding auxiliary correction coefficient based on the value of the change slope.

[0092] In an embodiment of the present invention, the specific implementation of step S502 is as follows: the system collects and organizes several groups of continuous physiological state values at a consistent sampling frequency within a preset time interval before the current moment, and these values are arranged in chronological order to form a complete physiological state value time series. In this sequence, each data point corresponds to a timestamp and its corresponding physiological state value. The system performs linear trend modeling on the time series, that is, with time as the horizontal axis and the physiological state value as the vertical axis, calculates the linear regression slope of the sequence, and obtains the direction and rate of change of the physiological state of the current target individual within the time interval. This slope is the so-called change slope. A positive value indicates an improvement in the state, and a negative value indicates a deterioration in the state. The larger the absolute value of the slope, the faster the change. The closer the slope is to zero, the more stable the state remains.

[0093] In step S503, the system generates a corresponding auxiliary correction coefficient based on the calculated change slope value. Specifically, the system can set a segmented mapping relationship or a weighting function according to the sign and amplitude of the change slope, and convert the trend direction (improvement, deterioration) and trend intensity (speed of change) into the adjustment amplitude of the content level parameters. For example, when the change slope is negative and the absolute value is large, the auxiliary correction coefficient takes a negative enhancement value, prompting the system to more strongly lower the content level; if the change slope is positive, the auxiliary correction coefficient can be positive or close to zero, indicating that the status is improved or maintained, and the content level can be maintained or slightly improved.

[0094] The principle behind generating the auxiliary correction coefficient from the changing slope is that it reflects the current state trend of the target individual rather than the state value at a single static moment, thus providing dynamic perception over time. Unlike the primary correction coefficient, the auxiliary correction coefficient does not rely on a reference sample but instead directly reflects the state evolution trend of the target individual in the period preceding the current reading behavior. Therefore, it fills the primary correction mechanism's blind spot in responding to short-term fluctuations and enhances the system's ability to perceive the direction and speed of physiological state changes.

[0095] The benefit of introducing this mechanism is that the system can not only make corrections based on the extent to which the current state deviates from the ideal state, but can also identify whether the state is improving or deteriorating, and dynamically adjust intervention strategies accordingly. For example, if the physiological state is still acceptable but declining, the system can preemptively lower the content level to prevent it from entering a critical state. Conversely, when the state is improving, even if the current value is slightly below the reference value, unnecessary intervention can be avoided, achieving smoother and more intelligent content adjustment. This move improves the sensitivity of personalized adaptation and the foresight of adjustment strategies, enhancing the system's dynamic adaptability to individual state fluctuations.

[0096] Furthermore, in the embodiments of the present invention, the setting of the auxiliary correction coefficient is highly consistent with the amplification effect in the aforementioned correlation law. It can even be said that the parameter adjustment mechanism was designed based on this phenomenon. According to the established correlation law, the target individual exhibits a distinct enhanced response characteristic during the point reading process. That is, the greater the gap between physiological state values, the more pronounced the gap in their point reading effect scores. This nonlinear amplification phenomenon indicates that individuals are highly sensitive to state changes. Therefore, generating the main correction coefficient solely based on the deviation between the current static physiological state and the reference state is insufficient to fully reflect the dynamic risk of an individual's current state.

[0097] The auxiliary correction coefficient is a supplementary parameter constructed to respond to this nonlinear trend. By tracking the state evolution trend and capturing the dynamic direction of the "improvement" or "deterioration" process, it can proactively intervene in state changes that may cause significant score fluctuations. The auxiliary correction mechanism not only effectively reinforces the main correction mechanism but also constitutes an independent feedback channel for the changing trends of the target individual's physiological state, making the final correction of the content-level parameters more forward-looking, continuous, and stable. By integrating the two, the system can identify state deviations while taking into account the changing trends, improving the adaptation accuracy and practical value of the overall adjustment strategy in the context of enhanced response.

[0098] Furthermore, the method for intelligent management of interactive data of the reading pen based on AI analysis further includes the following steps:

[0099] Step S600: Dynamically modify the initial content level parameter by combining the main modification coefficient and the auxiliary modification coefficient. After generating the modified content level parameter, match the push content of the corresponding level in the preset content corpus based on the content level parameter and push it to the target individual.

[0100] When dynamically correcting the initial content level parameters, based on the main correction coefficient and the auxiliary correction coefficient, the preset parameter correction model is called to achieve joint correction;

[0101] The parameter correction model is: ,in Refers to the revised content level parameters, Refers to the initial content level parameters, Refers to the main correction coefficient, that is, the deviation between the real-time physiological state value and the reference physiological state value. Refers to the adjustment weight corresponding to the main correction coefficient, Refers to the auxiliary correction coefficient, that is, the slope of the change of the physiological state value time series, Refers to the adjustment weight corresponding to the auxiliary correction coefficient, Refers to the trigger threshold of the main correction factor.

[0102] In an embodiment of the present invention, by jointly applying the main correction coefficient and the auxiliary correction coefficient to the adjustment process of the initial content level parameters, the system can simultaneously take into account the static deviation of the current physiological state and its changing trend in the time dimension. The significance of this joint correction method is that the main correction coefficient can accurately reflect the actual degree of deviation of the current physiological state of the target individual compared to the ideal state, while the auxiliary correction coefficient supplements the ability to identify the direction and rate of state evolution. Both make a comprehensive judgment on the target state from the two dimensions of cross-sectional and longitudinal changes, making the correction process both sensitive and robust.

[0103] Compared with the adjustment method driven by a single indicator, this joint mechanism can effectively avoid the problem of misjudgment or misadjustment due to short-term outliers, and also improve the system's ability to predict the potential risks behind state fluctuations. Specifically, when the main correction coefficient identifies that the state deviation is not significant but the auxiliary correction coefficient reflects that the state is rapidly deteriorating, the system can still actively initiate a level correction to achieve defensive content adjustment; conversely, if the current state deviates significantly but the trend shows a recovery direction, the system can also choose to delay intervention or ease the correction amplitude to avoid excessive intervention. It is this combination of static and dynamic judgment mechanisms that makes the system's intelligent adjustment behavior more fault-tolerant and adaptive.

[0104] The correction model employed in this invention is a weighted correction structure based on threshold triggering. Specifically, when the primary correction coefficient exceeds a set threshold, the system multiplies both the primary and secondary correction coefficients by preset weights, adds them together, and uses this sum to adjust the initial content rating parameters. This model boasts an intuitive structure, high computational efficiency, and ease of embedded implementation in actual devices. It also offers excellent interpretability and controllability, making it a preferred solution for engineering practice.

[0105] However, it should be made clear that this correction model is not the only way to implement the present invention. Although its structure is simple, it does not exclude other forms of substitution or expansion. For example, a nonlinear function modeling method of state difference and trend slope, such as exponential amplification, logarithmic compression, etc., can be used to adapt to more complex state response relationships; content level prediction models based on historical score sample training, such as decision trees, logistic regression or neural network models, can also be introduced to dynamically generate level correction results by learning the mapping relationship between historical states and scores. In addition, physiological state data and point reading behavior data can be fused and input into a multivariate adjustment model to achieve higher-level score prediction and content adjustment control. Therefore, the core of the present invention is not to limit a specific calculation formula, but to achieve response adjustment and adaptive management of individual states in content push by establishing a physiological state-driven level correction logic.

[0106] It should be understood that, although the various steps in the flow chart of each embodiment of the present invention are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0107] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0108] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0109] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for intelligent management of interactive data of a reading pen based on AI analysis, characterized in that: The method comprises: Obtain the target individual's click-and-read interaction data, physiological signal data, and initial content level parameters corresponding to the content to be pushed; Filter out several historical samples from the click-and-read interaction data, require that the content level parameters corresponding to each historical sample are consistent with the initial content level parameters, and determine the click-and-read effect score corresponding to each historical sample; Analyze several historical samples to determine whether there is a trend of change in the same direction between the target individual's physiological state and the reading effect score, and whether the reading effect score shows a correlation pattern of enhanced response to physiological state deviation; The steps of analyzing a number of historical samples to determine whether there is a trend of change in the same direction between the physiological state of the target individual and the reading effect score, and whether the reading effect score shows a correlation pattern of an enhanced response to the physiological state deviation include: Extract the click-and-read effect score corresponding to each historical sample, and obtain the physiological state value of the historical sample at the time the pushed content was sent based on the physiological signal data; Sort the physiological state values of several historical samples from large to small to generate a physiological state sequence, and extract the corresponding reading effect score of each historical sample in the sorted order to generate a score sequence. Determine whether the physiological state sequence and the score sequence both show a decreasing trend. If so, it is determined that there is a trend of change in the same direction between the physiological state of the target individual and the reading effect score; On the basis of confirming the existence of the same-direction change trend, we analyze whether there is an enhanced response effect in the physiological state sequence and the score sequence, where the larger the difference in physiological state values, the more significant the increase in the difference in the reading effect score. If so, we confirm that the correlation law is established. The physiological state value refers to: physiological parameter data of the target individual collected by the wearable device, which is obtained after feature extraction and normalization, and is used to obtain a numerical indicator for characterizing the physiological state of the target individual. The physiological parameter data is a weighted combination of one or more of heart rate, skin conductivity, body surface temperature, body movement amplitude, and respiratory rate. The physiological state value is a monotonically positive indicator, and a higher value indicates a better physiological state of the target individual. On the premise of confirming the existence of a correlation pattern, the historical sample with the highest reading effect score is selected as the reference sample, the reference physiological state value associated with the reference sample is extracted, and the real-time physiological state value of the target individual at the current moment is collected. The main correction coefficient is calculated based on the deviation between the two; Collect continuous physiological state values within a preset time interval before the current moment, and calculate and generate auxiliary correction coefficients based on their change trends; The initial content level parameters are dynamically corrected by combining the main correction coefficient and the auxiliary correction coefficient.

2. The method for intelligent management of interactive data of a reading pen based on AI analysis according to claim 1 is characterized in that: The steps of selecting a number of historical samples from the click-and-read interaction data, requiring the content level parameters corresponding to each historical sample to be consistent with the initial content level parameters, and determining the click-and-read effect score corresponding to each historical sample include: Parse the click-and-read interaction data, extract the local click-and-read interaction data corresponding to all historical push events, and obtain the content level parameters of the push content associated with each local click-and-read interaction data; Filter out local click-and-read interaction data with content-level parameters that are consistent with initial content-level parameters and use them as historical samples; Analyze each historical sample and obtain the behavioral response data of the target individual within the preset time window after the push content is sent. The behavioral response data specifically includes at least one or more indicators of the target individual's point reading accuracy, point reading response duration and repeated click frequency within the preset time window, and calculate the point reading effect score of the historical sample based on these indicators through weighted calculation.

3. The method for intelligent management of interactive data of a reading pen based on AI analysis according to claim 1, characterized in that: On the premise of confirming the existence of a correlation pattern, the steps of selecting the historical sample with the highest reading effect score as the reference sample, extracting the reference physiological state value associated with the reference sample, and collecting the real-time physiological state value of the target individual at the current moment, and calculating and generating the main correction coefficient based on the deviation between the two include: After confirming the existence of correlation patterns, the historical samples with the highest reading effect scores are selected as reference samples, and their corresponding physiological state values are extracted as reference physiological state values; Collect the real-time physiological status value of the target individual at the current moment; The deviation between the real-time physiological state value and the reference physiological state value is calculated, and a main correction coefficient is generated based on the deviation.

4. The method for intelligent management of interactive data of a reading pen based on AI analysis according to claim 3 is characterized in that: The steps of collecting continuous physiological state values within a preset time interval before the current moment and calculating and generating an auxiliary correction coefficient based on its change trend include: Collecting several sets of physiological status values of the target individual within a preset time interval before the current moment, wherein the collection time interval of the physiological status values is preset and consistent; Constructing a physiological state value time series from a number of groups of physiological state values in chronological order, and calculating the change slope of the physiological state value time series; A corresponding auxiliary correction coefficient is generated based on the value of the change slope.

5. The method for intelligent management of interactive data of a reading pen based on AI analysis according to claim 4 is characterized in that: When dynamically correcting the initial content level parameters, based on the main correction coefficient and the auxiliary correction coefficient, the preset parameter correction model is called to achieve joint correction; The parameter correction model is: ,in Refers to the revised content level parameters, Refers to the initial content level parameters, Refers to the main correction coefficient, that is, the deviation between the real-time physiological state value and the reference physiological state value. Refers to the adjustment weight corresponding to the main correction coefficient, Refers to the auxiliary correction coefficient, that is, the slope of the change of the physiological state value time series, Refers to the adjustment weight corresponding to the auxiliary correction coefficient, Refers to the trigger threshold of the main correction factor.

6. The method for intelligent management of interactive data of a reading pen based on AI analysis according to claim 5 is characterized in that: After the revised content level parameters are generated, push content of the corresponding level is matched in a preset content corpus based on the content level parameters, and pushed to the target individual.

Images