Data processing method and system based on monitoring bracelet

Abstract

The invention discloses a data processing method and system based on a monitoring bracelet, and the method comprises the steps: obtaining physiological monitoring data of a monitored object through a built-in chip module of the monitoring bracelet; performing abnormal interval identification and abnormal interval classification on the physiological monitoring data; according to the classification of all the abnormal intervals and the features of the abnormal intervals under each classification, constructing adaptive features of the monitoring object; performing parameter mapping through the self-adaptive features of the monitoring object to generate parameters of a data supplement model; and after data clearing is carried out on each abnormal interval, data supplementation is carried out by utilizing the data supplementation model after parameter assembly. Feature differences of individuals in different abnormal states can be reflected, and personalized data analysis is realized; more reasonable supplementary data is generated for different abnormal characteristics; the generated data is ensured to be consistent with normal data in numerical characteristics and dynamic trends, and the data continuity and physiological rationality are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring data processing, and specifically to a data processing method and system based on a monitoring bracelet. Background Art

[0002] With the development of wearable device technology, as a convenient physiological data acquisition tool, the monitoring bracelet is widely used in the fields of health management, sports tracking, and medical assistance. Existing monitoring bracelets usually obtain physiological signals such as the wearer's heart rate, blood oxygen saturation, and skin temperature in real time through built-in chip modules. However, in actual applications, due to poor contact between the bracelet and the skin, loose wearing, device detachment, or external environmental interference, there are phenomena of distortion, missing, or abnormal fluctuations in the collected physiological monitoring data. These abnormalities not only affect the accuracy of subsequent health assessments but may also mask potential health risks, reducing the data credibility and application effect.

[0003] Most of the existing technologies at present eliminate or roughly compensate for abnormal data by setting fixed thresholds or simple filtering algorithms, lacking in-depth analysis and classification of the abnormal generation mechanism, and it is difficult to adaptively optimize the data supplementation strategy according to different abnormal characteristics. In addition, traditional data supplementation methods are usually based on interpolation or regression speculation, and it is difficult to ensure the consistency of the supplemented data with the original waveform in terms of dynamic characteristics and physiological rationality.

[0004] In view of the above problems, there is an urgent need for a processing method that can accurately identify abnormal intervals in physiological data, classify abnormal types, dynamically generate supplementation parameters based on individual characteristics, and achieve high-quality data supplementation through an intelligent model. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A data processing method based on a monitoring bracelet, comprising:

[0007] Obtaining physiological monitoring data of a monitoring object through a chip module built in the monitoring bracelet;

[0008] Identifying and classifying abnormal intervals of the physiological monitoring data;

[0009] Constructing adaptive characteristics of the monitoring object according to all the classified abnormal intervals and the characteristics of the abnormal intervals under each classification;

[0010] Performing parameter mapping through the adaptive characteristics of the monitoring object to generate parameters of a data supplementation model;

[0011] After clearing the data in each abnormal interval, the data supplement model is used to supplement the data after parameter assembly.

[0012] As a preferred solution of the data processing method based on the monitoring bracelet of the present invention, wherein: the physiological monitoring data includes but is not limited to heart rate, blood oxygen saturation, and skin temperature;

[0013] Data processing is performed separately for different physiological monitoring data.

[0014] As a preferred embodiment of the data processing method based on the monitoring wristband of the present invention, the abnormal interval includes the abnormal part of the waveform after the physiological monitoring data is plotted into a waveform in a continuous time series;

[0015] Input the waveform data into the trained neural network, and output the start timestamp, end timestamp, interval classification results, and interval waveform features of each abnormal interval through identification of abnormal intervals;

[0016] The interval classification results include normal intervals and abnormal intervals; the abnormal intervals include but are not limited to poor wristband contact, wristband detachment and data distortion;

[0017] The interval waveform characteristics include the data average value, variance, maximum value, minimum value, peak rate and baseline drift rate within the interval.

[0018] As a preferred solution of the data processing method based on the monitoring bracelet of the present invention, wherein: within each category of abnormal interval, the seven interval waveform features are respectively used as seven dimensions;

[0019] Each abnormal interval of the waveform is located in the 7-dimensional data space, and the location set of all abnormal intervals is obtained: H = {h J , h L , h S}; where h J The coordinate set of the abnormal interval classified as poor contact of the bracelet, h L The coordinate set of the abnormal interval classified as the bracelet falling off is shown, h S A set of coordinates representing the classification of abnormal intervals as data distortion;

[0020] In the coordinate set of each abnormal interval classification, the coordinates of the midpoint of the set are integrated, and the mean of the discrete coordinates of the seven dimensions is calculated respectively to obtain the mean of the seven dimensions as the coordinates of the midpoint of the set; corresponding to the preset standard attention matrix;

[0021] The adaptive features include the coordinates of the midpoint of the set under each abnormal interval classification; expressed as: H = {h J0 , hL0 , h S0}; Among them, h J0 represents the coordinate of the midpoint of the set where the abnormal interval is classified as poor bracelet contact, and h L0 represents the coordinate of the midpoint of the set where the abnormal interval is classified as bracelet detachment, and h S0 represents the coordinate of the midpoint of the set where the abnormal interval is classified as data distortion.

[0022] As a preferred solution of the data processing method based on the monitoring bracelet described in the present invention, wherein: the parameter mapping includes, in the seven-dimensional space under each abnormal interval classification, presetting a standard coordinate point h0 to represent the preset standard value under each feature;

[0023] Let h x0 represent h J0 , h L0 , h S0 any one of the coordinates;

[0024] Compare the coordinates of h x0 and h0 in seven dimensions respectively. Based on the coordinates of h0 in each dimension, calculate the increase and decrease ratio of the coordinates of h x0 ;

[0025] In the standard attention matrix, adjust the parameters of the attention matrix according to the increase and decrease ratio and the mapping function to obtain the attention matrix corresponding to h x0 ;

[0026] As a preferred solution of the data processing method based on the monitoring bracelet described in the present invention, wherein: adjusting the parameters of the attention matrix includes, through the training set, fitting the increase and decrease ratio of the coordinate data of h0 in each dimension with the change of the attention matrix parameters to obtain the mapping function;

[0027] After adjusting the parameters of the attention matrix in each dimension respectively, normalize the adjusted parameters of the attention matrix to obtain the final attention matrix;

[0028] The data supplement model is an adversarial network.

[0029] As a preferred solution of the data processing method based on the monitoring bracelet described in the present invention, wherein: the data supplement includes using the trained adversarial network to supplement data for each abnormal interval respectively;

[0030] The input to the adversarial network is: the complete waveform, the start and end points of the abnormal interval, the adjusted attention matrix, and the classification of the abnormal interval;

[0031] Output layer: generate a supplementary waveform sequence with the same length as the original abnormal section;

[0032] The head of the generator of the adversarial network is a classifier, which is responsible for classifying the abnormal intervals into corresponding levels, and an independent data channel is connected after each level;

[0033] For the waveforms in the normal interval, they are segmented according to a fixed time window length, and 7-dimensional selection space coordinates are constructed for the data in each window;

[0034] For the abnormal interval p, let: the time window in the previous normal interval be q, and the time window in the next normal interval be h;

[0035] Connect p and h in the 7-dimensional selection space, and take the midpoint of the two as the coordinate position of p in the 7-dimensional selection space; calculate the relative distance by dividing the distance from p to each center by the spatial radius of the cluster represented by the cluster center; pg represents the relative distance from p to the cluster center g, and take the g corresponding to the minimum value of pg as the cluster center of p; among them, each cluster center represents a level, and in the adaptation module of each channel, adaptation is carried out in the corresponding cluster family;

[0036] The adaptation process of the adaptation module includes: presetting a standard coordinate h different from h0 0+ ;

[0037] In the corresponding cluster family of the adaptation module, randomly select A normal intervals; calculate the differences between the features of 7 dimensions in each interval and h 0+ respectively, so that each adaptation module adapts to the data of the corresponding level. After the adaptation is completed, physiological monitoring data is supplemented.

[0038] A data processing system based on a monitoring bracelet, where: a collection unit, through a chip module built in the monitoring bracelet, obtains physiological monitoring data of a monitoring object;

[0039] A classification unit, which identifies and classifies abnormal intervals in the physiological monitoring data;

[0040] An analysis unit, according to all abnormal interval classifications and the characteristics of abnormal intervals under each classification, constructs adaptive features for the monitoring object;

[0041] An adjustment unit, through the adaptive features of the monitoring object, performs parameter mapping to generate parameters of a data supplementation model;

[0042] A supplementation unit, after clearing the data for each abnormal interval, uses the data supplementation model assembled with parameters to perform data supplementation.

[0043] A computer device, comprising: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, the steps of any one of the methods of the present invention are implemented.

[0044] A computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0045] Advantages of the present invention: The data processing method based on a monitoring bracelet provided by the present invention can accurately identify abnormal intervals in physiological monitoring data on complete waveform data and perform fine-grained classification according to the type of abnormality, improving the accuracy and reliability of abnormality detection. By extracting multi-dimensional waveform features of each abnormal interval and constructing an adaptive feature space for the monitored object, the feature differences of an individual in different abnormal states can be reflected, realizing personalized data analysis. By comparing the adaptive features with the standard features, combining the increase and decrease ratios and the mapping function, the attention matrix is dynamically adjusted to guide the data supplementation model to generate more reasonable supplementary data for different abnormal characteristics. Using an adversarial network for data supplementation ensures that the generated data is consistent with normal data in terms of numerical characteristics and dynamic trends, effectively improving data continuity and physiological rationality. Description of the Drawings

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0047] Figure 1 It is the overall flowchart of a data processing method based on a monitoring bracelet provided by the first embodiment of the present invention. Detailed Embodiments

[0048] To make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0049] Embodiment 1, referring to Figure 1 , which is an embodiment of the present invention, provides a data processing method based on a monitoring bracelet, including:

[0050] S1: Obtain the physiological monitoring data of the monitored object by monitoring the chip module built in the bracelet.

[0051] The chip module integrates at least one physiological signal sensor, which may specifically include: a heart rate sensor (such as a photoplethysmography PPG sensor), a blood oxygen saturation sensor (such as a dual-wavelength near-infrared sensor), and a skin temperature sensor (such as a thermistor or a thermopile sensor).

[0052] The chip module outputs the above various physiological signals in the form of a digital data stream at a set sampling frequency (such as once per second or a higher frequency).

[0053] The different physiological monitoring data are respectively preprocessed according to their characteristics, including but not limited to: performing smoothing filtering on the heart rate signal to remove high-frequency noise; performing artifact removal on the blood oxygen saturation signal to enhance numerical stability; and performing baseline correction on the skin temperature signal to eliminate the influence of the external environment.

[0054] The preprocessed physiological data are standardized in a unified format to form a continuous time series, which serves as the basic data for subsequent anomaly recognition and supplementary processing.

[0055] S2: Identify and classify the abnormal intervals of the physiological monitoring data.

[0056] The abnormal intervals include the abnormal parts in the waveform after plotting the physiological monitoring data as a continuous time series. The waveform data is input into the trained neural network. Through the identification of the abnormal intervals, the start timestamp, end timestamp, interval classification result, and interval waveform characteristics of each abnormal interval are output.

[0057] First, by plotting the physiological monitoring data as a continuous time series, the dynamic change characteristics of the data can be comprehensively retained, intuitively reflecting the continuity, stability, and mutation of the physiological signal over time, and avoiding local misjudgment that may be caused by single-point sampling. Second, using the trained neural network to analyze the complete waveform can accurately identify the abnormal intervals in the signal based on the overall pattern of the time series, and has higher adaptability and recognition accuracy compared with traditional rule detection methods (such as threshold method, moving average method). Through the classification of the abnormal intervals, the abnormal types can be subdivided into different situations such as poor bracelet contact, bracelet detachment, and data distortion, so that subsequent data supplementation and correction can formulate differentiated processing strategies according to the abnormal reasons, thereby improving the rationality and effectiveness of data supplementation.

[0058] In addition, by extracting waveform features including the mean value, variance, maximum value, minimum value, peak rate, and baseline drift rate within each abnormal interval, a high-dimensional feature representation can be constructed for each abnormal section, which not only describes the overall statistical characteristics of the abnormal section but also provides a quantitative basis for subsequent adaptive feature modeling, parameter mapping, and supplementary model guidance.

[0059] In this embodiment, the neural network is a deep learning model for anomaly detection in time series data. Its main structure is as follows: Input layer: Receives the complete physiological monitoring data waveform sequence; Feature extraction layer: Extracts local waveform features through a one-dimensional convolutional neural network (1D-CNN) to capture the change trend, mutation points, and local stability in the time series; Time series modeling layer: Further learns the long-term dependence relationship of the sequence data using a recurrent neural network (such as LSTM, Long Short-Term Memory); Classification layer: Maps each time point or time period to normal or abnormal, and further subdivides it into specific abnormal types such as poor bracelet contact, bracelet detachment, and data distortion; Output layer: Outputs the start timestamp, end timestamp, interval classification result, and interval waveform features of each abnormal interval.

[0060] With the above structure, while ensuring the recognition accuracy, the operation efficiency of the model in resource-constrained environments such as bracelets can be taken into account. In other realizable embodiments, the neural network can also be replaced with the following structures, including but not limited to: Transformer structure, temporal convolutional network, and combination of graph convolutional network (GCN) with time series modeling, etc.

[0061] S3: Construct adaptive features for the monitored object according to all abnormal interval classifications and the features of abnormal intervals under each classification.

[0062] Within the abnormal intervals of each category, the seven interval waveform features are respectively used as seven dimensions.

[0063] Locate each abnormal interval of the waveform in a 7-dimensional data space to obtain a positioning set for all abnormal intervals: H = {h J , h L , h S}; where h J represents the coordinate set of the abnormal interval classified as poor bracelet contact, h L represents the coordinate set of the abnormal interval classified as bracelet detachment, and h S represents the coordinate set of the abnormal interval classified as data distortion.

[0064] In the set of coordinates classified for each abnormal interval, the coordinates of the points in the set are integrated. The mean values of the discrete coordinates in seven dimensions are calculated respectively to obtain the mean values of the seven dimensions (if a certain coordinate is 0 in all seven dimensions, this coordinate point is excluded), which are used as the coordinates of the midpoint of the set.

[0065] The adaptive features include the coordinates of the midpoint of the set under each abnormal interval classification, expressed as: H = {h J0 , h L0 , h S0}; where h J0 represents the coordinates of the midpoint of the set with the abnormal interval classification of poor bracelet contact, h L0 represents the coordinates of the midpoint of the set with the abnormal interval classification of bracelet detachment, and h S0 represents the coordinates of the midpoint of the set with the abnormal interval classification of data distortion.

[0066] Based on the waveform features of the abnormal intervals under different abnormal classifications, an adaptive feature that can truly reflect the individual differences and abnormal manifestation laws of the monitoring object is constructed, so as to provide reliable support for subsequent data supplementation and parameter mapping. By positioning each abnormal interval in the seven-dimensional feature space and integrating effective feature points under each classification, the feature centers of different abnormal types can be accurately extracted. In particular, during the mean value calculation process, all data points with feature values of zero in all feature dimensions are excluded to avoid interference of invalid data caused by bracelet detachment, signal loss, etc. on the feature center, thereby improving the authenticity and robustness of feature modeling. Through this method, personalized feature expression under abnormal classification can be achieved, enabling the parameter mapping and subsequent supplementary model to be dynamically adjusted according to actual characteristics, significantly enhancing the intelligence, self-adaptability and processing accuracy of the overall data processing system.

[0067] Since there are natural differences in the waveform feature manifestations of different individuals when facing the same type of abnormality (such as poor contact, detachment or data distortion), in this embodiment, through the mean value integration of the effective abnormal interval features, an adaptive feature center under abnormal classification is dynamically constructed, which can fully reflect the specific performance differences of different abnormal types in different populations. This design not only enhances the adaptability of the data processing method to individual differences, but also provides a more accurate and personalized feature basis for the subsequent parameter mapping and data supplementation process, thereby improving the robustness and effectiveness of the overall system in diverse application environments.

[0068] S4: Through the adaptive features of the monitoring object, parameter mapping is performed to generate the parameters of the data supplementation model.

[0069] In the seven-dimensional space under each abnormal interval classification, a preset standard coordinate point h0 is set to represent the preset standard value under each feature; corresponding to the preset standard attention matrix.

[0070] Let h x0 represent h J0 , h L0 , h S0 any one of the coordinates.

[0071] Compare the coordinates of h x0 and h0 in seven dimensions respectively. Based on the coordinates of h0 in each dimension, calculate the increase and decrease ratio of the coordinates of h x0 .

[0072] In the standard attention matrix, adjust the parameters of the attention matrix according to the increase and decrease ratio and the mapping function, and obtain the attention matrix corresponding to h x0 (used to focus on the missing part).

[0073] By setting preset standard coordinate points in the seven-dimensional feature space under each abnormal classification, and comparing the actually constructed adaptive features with the standard coordinates, the differences in each feature dimension are quantified, so as to dynamically adjust the standard attention matrix to achieve precise attention and compensation optimization for the missing data area. By calculating the increase and decrease ratio of the actual features relative to the standard features and combining the mapping function to adjust the attention parameters, the attention intensity of each feature dimension in the supplement process can be flexibly adjusted according to the abnormal type and individual differences. This design enables the supplement model to dynamically select the feature directions to focus on when facing different abnormal types and different individual performances, avoiding a one-size-fits-all unified supplement strategy, enhancing the pertinence, rationality and physiological consistency of data supplement, and thus significantly enhancing the self-adaptability, accuracy and data integrity of the overall system.

[0074] Furthermore, through the training set, fit the increase and decrease ratio of the coordinate data of h0 in each dimension and the change of the attention matrix parameters to obtain the mapping function. After adjusting the parameters of the attention matrix in each dimension respectively, normalize the parameters of the adjusted attention matrix to obtain the final attention matrix.

[0075] In this embodiment, first, based on the training set, for the seven-dimensional feature space under each abnormal interval classification, historical data of a large number of monitoring objects are collected, and the increase or decrease ratio between the adaptive feature and the standard feature of each sample on each feature dimension is calculated. At the same time, the effect change after adjusting the attention matrix parameters during the data supplementation process of each sample is recorded. Through statistical analysis and supervised learning, on each feature dimension, with the increase or decrease ratio as the input and the change of the attention parameter under the actual optimal compensation effect as the output, a fitting model is established to obtain the mapping function of each feature dimension. The mapping function can be a linear function, a piecewise function, or a non-linear function (such as a sigmoid function, a hyperbolic tangent function), which is determined according to the actual fitting accuracy. Subsequently, in actual applications, for each new abnormal interval, the increase or decrease ratio obtained by comparing the adaptive feature with the standard feature will be input into the corresponding mapping function respectively to output the adjusted attention matrix parameters. To ensure that the attention matrix meets the requirements of weighted sum consistency, after all dimension parameters are adjusted, the adjusted matrix is further normalized row by row so that the sum of weights in each row is 1, thereby obtaining the final attention matrix that can be used for data supplementation.

[0076] By fitting the mapping relationship between the increase or decrease ratio and the change of the attention matrix parameters in the training set, the problem of deviation in the supplementation effect caused by directly scaling linearly according to the ratio is avoided, enabling the attention adjustment to more accurately reflect the importance of each feature dimension for data supplementation in the abnormal state. Using the fitted mapping function, the increase or decrease amplitude of the attention can be flexibly controlled according to the change rules of different feature dimensions, improving the consistency of the supplemented data in terms of dynamic trend and physiological rationality. Through normalization processing, the numerical standardization of the attention matrix is ensured, avoiding the distortion of the attention weight caused by adjustment, and further enhancing the stability and robustness of the model in abnormal section compensation. Overall, this design enables the supplementation model to achieve more accurate and personalized missing data repair, improving the data processing ability of the system in complex abnormal situations.

[0077] S5: After clearing the data for each abnormal interval, use the data supplementation model after parameter assembly to perform data supplementation.

[0078] The data supplementation model is an adversarial network. The trained adversarial network is used to perform data supplementation for each abnormal interval respectively. The inputs to the adversarial network are: the complete waveform, the start and end points of the abnormal interval, the adjusted attention matrix, and the classification of the abnormal interval.

[0079] Output layer: Generate a supplementary waveform sequence with the same length as the original abnormal section.

[0080] For the detected abnormal intervals, through the trained adversarial network model, intelligent supplementation of missing or distorted physiological monitoring data is achieved, improving the integrity and continuity of the data. By taking the complete waveform data, the start and end positions of the abnormal intervals, the adjusted attention matrix, and the classification information of the abnormal intervals as inputs, the dynamic context features before and after the occurrence of the abnormality, the specific position range of the abnormal section, and the key features to be focused on corresponding to the abnormal type can be provided for the supplementation model, enabling the generator to generate supplementary waveforms that conform to physiological logic and seamlessly connect with the dynamic trend of the original data based on the real waveform trend and abnormal characteristics. The output layer generates a supplementary waveform sequence with the same length as the original abnormal section, ensuring the continuity of the data on the time axis and avoiding data jumps or logical breaks caused by mismatched supplementation lengths. Overall, by introducing the adversarial training mechanism, the present invention can effectively improve the performance of the supplementary data in terms of numerical rationality, dynamic trend continuity, and physiological feature consistency, significantly enhancing the credibility of the monitoring data and the robustness of the system.

[0081] The above-mentioned adversarial network is a multi-channel network, and the head of the network is used to determine the level suitable for each abnormal interval. Each level is configured with an adaptation module (there are multiple adaptation modules, and the more adaptation modules there are, the stronger the adaptation ability and the more accurate the data supplementation result), so that each abnormal interval is input into the corresponding channel.

[0082] Furthermore, through the design of multiple adaptation modules, it can be ensured that each waveform feature can be complemented specifically. Through the hierarchical random truncation and multi-channel adaptation mechanism proposed in the present invention, the characteristic laws of the normal intervals at the same level can be fully extracted before the data supplementation of the abnormal intervals, making the supplementary data not only conform to the overall state of the monitoring object in terms of numerical distribution but also highly consistent with the actual physiological characteristics in terms of local dynamic change trends. Overall, this design effectively improves the naturalness, coherence, and physiological rationality of the supplementary data, significantly enhancing the adaptability, stability, and supplementation accuracy of the system under diverse abnormal conditions.

[0083] Specifically: The head of the generator of the adversarial network is a classifier, which is responsible for classifying the abnormal intervals into the corresponding levels, and each level is connected to an independent data channel. The data within the abnormal intervals will be supplemented separately according to the input channels. The discriminator is a general discriminator, and its structure is a one-dimensional convolutional neural network.

[0084] For the waveforms in the normal intervals, they are segmented according to a fixed time window length, and a 7-dimensional data space coordinate (referred to as the "7-dimensional selection space" for distinction from the previous text) is constructed for the data within each window.

[0085] For the abnormal interval p, let: the time window in the previous normal interval be q, and the time window in the next normal interval be h.

[0086] Connect p and h in the 7-dimensional selection space, and take the midpoint of the two as the coordinate position of p in the 7-dimensional selection space; calculate the relative distance by dividing the distance from p to each center by the spatial radius of the cluster center-represented family; pg represents the relative distance from p to the cluster center g, and take the g corresponding to the minimum value of pg as the cluster center of p; where each cluster center represents a level, and in the adaptation module of each channel, adaptation is performed in the corresponding cluster family.

[0087] It should be noted that this solution is actually completed through two attention strategies. Before using the trained adversarial network model to supplement data for the abnormal interval, in each adaptation module, first extract the feature information of consecutive normal intervals from the collected physiological monitoring data. A normal interval refers to the waveform section determined to be normal after abnormal recognition and classification. For each normal section, extract seven waveform features, including the data average value, variance, maximum value, minimum value, peak rate, and baseline drift rate, and perform aggregated analysis on the features of multiple normal sections to form a description of the current global physiological state of the monitoring object. Subsequently, input the global features extracted from this normal section into the reserved adaptation module inside the trained adversarial network to dynamically adjust the attention mechanism inside the generator. The adversarial network configured with the attention parameters will perform subsequent analysis.

[0088] Before officially supplementing data for the abnormal section, make full use of the feature information of the monitoring object in the normal state to dynamically adapt the attention mechanism inside the adversarial network (pay attention to and adapt the attention for each adaptation module respectively), so as to achieve personalized guidance for the supplement generation process. Since there are natural differences in the physiological baseline states of different individuals and may fluctuate over time, directly supplementing data with fixed attention parameters is likely to cause the generated data to deviate from the actual state of the individual. By introducing an adaptive adjustment mechanism based on the features of the normal interval, the generator can more accurately capture the current individual's feature tendency during the supplement process, ensuring that the generated supplementary data is highly consistent with the individual's normal state in terms of numerical trend, fluctuation form, and physiological rationality, thereby improving the continuity, naturalness, and credibility of the supplementary data, and enhancing the stability and adaptability of the overall system in practical applications.

[0089] The adaptation process of the adaptation module includes: preset a standard coordinate h different from h0 0+ . In the cluster family corresponding to the adaptation module, randomly select A normal intervals; calculate the features of 7 dimensions in each interval and h respectively 0+The differences enable each adaptation module to adapt to data of the corresponding level. After the adaptation is completed, physiological monitoring data is supplemented.

[0090] Specifically, in each adaptation module, the adjustment method is similar to the method in the previous text. Multiple intervals are randomly intercepted within the normal range. Then a standard coordinate h different from h0 is preset. 0+ The differences between the features of the 7 dimensions in each interval and h are calculated respectively, and the average value is obtained. According to the average value of the differences in each dimension, an adjustment factor is calculated (this is different from the above scheme because in the above scheme, the personalization of individual abnormal manifestations is much greater than the regular pattern of monitoring data; while the scheme here is mainly based on regularity and adjusts the change range of features). r 0+ = 1 + α × tanh(β (i) × Δf i ) ; where α > 0 represents the adjustment amplitude coefficient, which controls the maximum increase and decrease ratio; β (i) > 0 represents the adjustment sensitivity coefficient under the i-th feature, which controls the offset response speed; tanh represents the hyperbolic tangent function, ensuring smooth and limited adjustment. Δf i represents the average value of the differences under the i-th feature; r (i) represents the adjustment factor under the i-th feature. After the adjustment factors in each dimension are input into the trained adversarial network model, the adversarial network model first needs to adapt to the wearer. After the adaptation is completed, subsequent monitoring and data supplementation are carried out. Among them, each cluster center represents a level; when "randomly intercepting multiple intervals" for each abnormal interval, it needs to be selected from the normal interval part in the same cluster family. (i) It should be noted that when adjusting the attention mechanism based on the features of the normal interval in this step, the calculation method of the adjustment factor used is significantly different from the adjustment scheme for the abnormal interval. The aforementioned attention adjustment for the abnormal interval mainly dynamically calculates the change range of the features of each abnormal instance according to the personalized offset between the abnormal features and the standard features, emphasizing the precise adaptation to the differences in individual abnormal manifestations to achieve personalized supplementation and correction of missing data or abnormal data.

[0091] It should be noted that when adjusting the attention mechanism based on the features of the normal interval in this step, the calculation method of the adjustment factor used is significantly different from the adjustment scheme for the abnormal interval. The aforementioned attention adjustment for the abnormal interval mainly dynamically calculates the change range of the features of each abnormal instance according to the personalized offset between the abnormal features and the standard features, emphasizing the precise adaptation to the differences in individual abnormal manifestations to achieve personalized supplementation and correction of missing data or abnormal data.

[0092] For the adjustment in the normal range in this step, the common rules of physiological monitoring data in the normal state are taken as the core. By randomly intercepting multiple local segments in the normal range, extracting features respectively and comparing them with the standard features, calculating the average difference in each feature dimension, and calculating a smooth adjustment factor based on this, the change range of the attention weight in the normal physiological state is controlled. This adjustment method aims to strengthen the generator's understanding of the normal physiological feature distribution of the current wearer, maintain attention to and compliance with the main physiological rules, rather than overemphasizing individual deviations, so as to improve the rationality and continuity of the overall supplementary data within the normal fluctuation range.

[0093] Therefore, the attention adjustment in this step belongs to an adaptive mechanism guided by conventional regularity, which is different from the mechanism based on personalized difference compensation in the abnormal range processing. The two are independent of each other and respectively adapt to the characteristic requirements of different stages in the data supplementation task, jointly ensuring the authenticity, continuity and physiological rationality of the supplementary data. That is to say, the first attention mechanism is to achieve user adaptation, so that the neural network focuses on the non-abnormal features, thus achieving rapid adaptation. The second attention mechanism is to pay attention to the "missing amount" or "excessive abnormal amount" of the abnormal part, so as to complete the data.

[0094] In order to further improve the accuracy and adaptive ability in the process of supplementing data in the abnormal range, when randomly intercepting the normal range for feature adaptation in this invention, a constraint mechanism based on level division is added to ensure that the randomly intercepted normal range segment and the target abnormal range are within the same level, maintaining the consistency and characteristic continuity of the feature space. By dividing the complete normal range with a fixed time window, constructing a 7-dimensional selection space, and using the density-based spatial clustering method (DBSCAN) for classification, the normal data with different stability features can be reasonably grouped.

[0095] On the other hand, this embodiment also provides a data processing system based on a monitoring bracelet, which includes:

[0096] A collection unit, through the chip module built in the monitoring bracelet, obtains the physiological monitoring data of the monitored object.

[0097] A classification unit, which identifies the abnormal range and classifies the abnormal range of the physiological monitoring data.

[0098] An analysis unit, according to all the abnormal range classifications and the features of the abnormal ranges under each classification, constructs the adaptive features of the monitored object.

[0099] An adjustment unit, through the adaptive features of the monitored object, performs parameter mapping to generate the parameters of the data supplementation model.

[0100] The supplementary unit clears the data for each abnormal interval and then uses the data supplementation model assembled with parameters to supplement the data.

[0101] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0102] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0103] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0104] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0105] Embodiment 2 is an embodiment of the present invention, which provides a data processing method based on a monitoring bracelet. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0106] Ten healthy adult volunteers were selected as the test subjects, and they wore the same model of monitoring bracelets respectively. They carried out regular daily activities within 24 consecutive hours, and some wearing looseness, short-term detachment and environmental interference situations were deliberately set during this period to simulate the occurrence of real abnormal states. The collected data included three physiological parameters: heart rate (bpm), blood oxygen saturation (%), and skin temperature (°C). The sampling frequency was set to 1 Hz to ensure high time resolution.

[0107] In the data processing stage, first, according to this method, the collected complete data waveform is input into the trained neural network to identify the abnormal intervals and classify them. Subsequently, seven characteristic parameters (average value, variance, maximum value, minimum value, peak rate, baseline drift rate, abnormal duration) are extracted under each abnormal classification to form a seven-dimensional feature space of the abnormal interval, and an adaptive feature midpoint is generated according to each classification. For adaptive adjustment, multiple data segments in the normal interval are randomly intercepted, features are extracted and compared with the preset standard features, and an adjustment factor is generated by using hyperbolic tangent mapping to dynamically adjust the attention mechanism of the generator inside the adversarial network to complete the rapid adaptation of individual physiological characteristics.

[0108] Subsequently, for the identified abnormal intervals, data supplementation is performed based on the adjusted adversarial network, and the length of the generated data sequence for supplementation is strictly consistent with the abnormal section. For comparison, the traditional interpolation method (linear interpolation) is used as the reference standard, and at the same time, the continuity index (such as root mean square error RMSE) of the supplemented data, the physiological feature consistency index (such as mean error), and the waveform smoothness score (scored by a third-party evaluator) after abnormal correction are recorded.

[0109] Finally, through the comparison of data indicators, the superiority of this method in terms of the quality of supplemented data and system adaptability is evaluated.

[0110] Specifically, in terms of the continuity index (RMSE), the average error of this method is between 2.1 and 2.5 bpm. Compared with the 4.5 - 5.1 bpm of the traditional interpolation method, the error is reduced by more than about 50%, showing an obvious advantage of the supplementary data in the time series continuity. In terms of the mean error and variance error of the supplementary section, the mean error of this method is generally controlled within 1.0 - 1.3 bpm, and the variance error is between 3.2 and 3.8 bpm 2 between, both of which are better than the corresponding error ranges of the traditional interpolation method (mean error 2.6 - 3.2 bpm, variance error 6.8 - 7.6 bpm 2 ), indicating that the supplementary data more accurately restores the dynamic change characteristics of physiological signals.

[0111] In terms of the waveform smoothness score, the supplementary waveforms are subjectively scored by third - party assessors. The waveforms supplemented by this method generally obtain a high score of 9, while the traditional interpolation supplementation is only 5 - 6 points. The physiological feature consistency score also shows an obvious advantage of this method, further corroborating the improvement effect of this method in the naturalness and rationality of supplementary data.

[0112] The above results fully demonstrate that by adapting the attention mechanism through normal interval feature guidance before data supplementation, the adversarial network can quickly adapt to the current physiological state of the wearer, so as to accurately capture key features during the abnormal section data supplementation process, improving the continuity, rationality and physiological feature consistency of the supplementary data.

[0113] [[ID=

Claims

1. A data processing method based on a monitoring wristband, characterized in that: include: Obtain the physiological monitoring data of the monitored object through the chip module built into the monitoring bracelet; Identifying and classifying abnormal intervals of the physiological monitoring data; According to all abnormal interval classifications and the characteristics of the abnormal intervals under each classification, adaptive features are constructed for the monitoring object; Perform parameter mapping through the adaptive characteristics of the monitored object to generate parameters of the data supplement model; After clearing the data in each abnormal interval, the data supplement model is used to supplement the data after parameter assembly.

2. The data processing method based on the monitoring bracelet according to claim 1, characterized in that: The physiological monitoring data include but are not limited to heart rate, blood oxygen saturation, and skin temperature; Data processing is performed separately for different physiological monitoring data.

3. The data processing method based on the monitoring wristband according to claim 2, characterized in that: The abnormal interval includes an abnormal portion of a waveform after the physiological monitoring data is plotted as a waveform in a continuous time series; Input the waveform data into the trained neural network, and output the start timestamp, end timestamp, interval classification results, and interval waveform features of each abnormal interval through identification of abnormal intervals; The interval classification results include normal intervals and abnormal intervals; the abnormal intervals include but are not limited to poor wristband contact, wristband detachment and data distortion; The interval waveform characteristics include the data average value, variance, maximum value, minimum value, peak rate and baseline drift rate within the interval.

4. The data processing method based on the monitoring bracelet according to claim 3, characterized in that: In each category of abnormal interval, the seven interval waveform features are respectively used as seven dimensions; Each abnormal interval of the waveform is located in the 7-dimensional data space, and the location set of all abnormal intervals is obtained: H = {h J , h L , h S }; where h J The coordinate set of the abnormal interval classified as poor contact of the bracelet, h L The coordinate set of the abnormal interval classified as the bracelet falling off is shown, h S A set of coordinates representing the classification of abnormal intervals as data distortion; In the coordinate set of each abnormal interval classification, the coordinates of the midpoint of the set are integrated, and the mean of the discrete coordinates of the seven dimensions is calculated respectively to obtain the mean of the seven dimensions as the coordinates of the midpoint of the set; The adaptive features include the coordinates of the midpoint of the set under each abnormal interval classification; expressed as: H = {h J0 , h L0 , h S0 }; where h J0 The coordinates of the midpoint of the set of abnormal intervals classified as poor contact of the bracelet, h L0 The coordinates of the midpoint of the set where the abnormal interval is classified as the bracelet falling off are shown, h S0 The coordinates of the points in the set representing the anomaly intervals classified as data distortions.

5. The data processing method based on the monitoring wristband according to claim 4, characterized in that: The parameter mapping includes, in the seven-dimensional space under each abnormal interval classification, a preset standard coordinate point h0, which is used to represent the preset standard value under each feature; corresponding to the preset standard attention matrix; Let h x0 Indicates h J0 、h L0 、h S0 Any coordinate in h x0 Compare the coordinates of seven dimensions with h0 respectively, and calculate h based on the coordinates of h0 in each dimension. x0 The ratio of increase or decrease of coordinates; In the standard attention matrix, the parameters of the attention matrix are adjusted according to the increase / decrease ratio and mapping function to obtain h x0 The corresponding attention matrix.

6. The data processing method based on the monitoring wristband according to claim 5, characterized in that: The adjusting of the parameters of the attention matrix includes fitting the increase and decrease ratio of the coordinate data of h0 in each dimension with the change of the attention matrix parameters through the training set to obtain the mapping function; After adjusting the parameters of the attention matrix in each dimension, the parameters of the adjusted attention matrix are normalized to obtain the final attention matrix; The data supplementation model is an adversarial network.

7. The data processing method based on the monitoring bracelet according to claim 6, characterized in that: The data supplementation includes using a trained adversarial network to supplement data for each abnormal interval; The input to the adversarial network is: the complete waveform, the start and end points of the abnormal interval, the adjusted attention matrix, and the classification of the abnormal interval; Output layer: Generates a supplementary waveform sequence with the same length as the original abnormal segment; The head of the generator of the adversarial network is a classifier, which is responsible for classifying abnormal intervals into corresponding levels, and each level is connected to an independent data channel; The waveform within the normal interval is segmented according to the fixed time window length, and the 7-dimensional selection space coordinates are constructed for the data in each window; For the abnormal interval p, let: the time window in the previous normal interval be q, and the time window in the next normal interval be h; Connect p and h in the 7-dimensional selection space, and take the midpoint between them as the coordinate position of p in the 7-dimensional selection space; calculate the distance from p to each center and divide it by the spatial radius of the cluster represented by the cluster center to obtain the relative distance; pg represents the distance from p to the cluster center g, and take the g corresponding to the minimum pg value as the cluster center of p; where each cluster center represents a level, and the adaptation module in each channel is adapted in the corresponding cluster family; The adaptation process of the adaptation module includes: presetting a standard coordinate h different from h0 0+ ; In the cluster family corresponding to the adaptation module, randomly select A normal intervals; calculate the features of the 7 dimensions in each interval and h 0+ The difference is determined by the adaptation module, so that each adaptation module adapts to the data of the corresponding level. After the adaptation is completed, the physiological monitoring data is supplemented.

8. A data processing system based on a monitoring wristband using the method according to any one of claims 1 to 7, characterized in that: The acquisition unit obtains the physiological monitoring data of the monitored object through the chip module built into the monitoring bracelet; a classification unit, for identifying and classifying abnormal intervals of the physiological monitoring data; An analysis unit constructs adaptive features for the monitoring object based on all abnormal interval classifications and the features of the abnormal intervals under each classification; An adjustment unit, which performs parameter mapping based on the adaptive characteristics of the monitored object to generate parameters of the data supplement model; The supplementation unit, after clearing the data in each abnormal interval, uses the data supplementation model after parameter assembly to supplement the data.

9. A computer device comprising: A memory and a processor; the memory stores a computer program, wherein the processor implements the steps of the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images