Fitness data sharing system based on digital twinning

Abstract

The invention discloses a fitness data sharing system based on digital twinning, and relates to the technical field of data sharing. A data analysis module analyzes a data type needing to be supplemented by an interpolation point according to user fitness data, and performs space-time analysis on multiple times of fitness data of the same data type; the interpolation supplement module generates a corresponding weight index for each piece of fitness data according to the time correlation and the fitness platform correlation, and performs weighted calculation on the fitness data of the same data type to obtain estimation data of the interpolation point; and after all interpolation points are supplemented, integrating and sharing the fitness data from different fitness platforms to a fitness coach. The sharing system is combined with space-time analysis, the correlation between interpolation points and other data is calculated through multiple times of fitness data, more accurate data completion is achieved, reasonable weights are given to data of different sources, errors between equipment and a fitness platform are reduced, and data fusion scientificity is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data sharing, and particularly to a fitness data sharing system based on digital twin. Background Art

[0002] Modern people are increasingly concerned about health and fitness. Smart wearable devices (such as smart bracelets, smart watches, etc.) can collect data such as users' heart rate, steps, movement trajectories, calorie consumption, etc. in real time. These data can not only help users intuitively understand their own exercise conditions, but also be used to optimize training plans and improve fitness effects. The sharing system mainly involves multiple aspects such as health management, smart fitness equipment, social interaction, and data security. With the development of smart devices and Internet of Things technology, the demand for this system has gradually increased.

[0003] The prior art has the following defects:

[0004] Users' fitness data is distributed in different devices (such as smart bracelets, treadmills, smart gym equipment) and different applications. Moreover, fitness equipment is affected by factors such as sensor accuracy, sampling interval, network latency, etc., and there may be data acquisition breakpoints. For example, a smart bracelet may cause the loss of exercise data in certain time periods due to low battery or Bluetooth disconnection, thus affecting users' subsequent scientific exercise.

[0005] Based on this, the present invention proposes a fitness data sharing system based on digital twin. By combining spatio-temporal analysis, calculating the correlation between interpolation points and other data using multiple fitness data, more accurate data completion is achieved, reasonable weights are assigned to data from different sources, the error between devices and fitness platforms is reduced, and the scientific nature of data fusion is improved. Summary of the Invention

[0006] The purpose of the present invention is to provide a fitness data sharing system based on digital twin to solve the deficiencies in the background art.

[0007] To achieve the above purpose, the present invention provides the following technical solution: A fitness data sharing system based on digital twin, including a data processing module, a data analysis module, and an interpolation and supplementation module;

[0008] Data processing module: Collect users' fitness data within a period of time through multi-source data fusion, remove outliers from the fitness data using statistical methods, mark both data missing points and outlier removal points as interpolation points, and perform standardization processing on fitness data from different fitness platforms;

[0009] Data analysis module: Based on users' fitness data, analyze the data types that need to be supplemented at the interpolation points, perform spatio-temporal analysis on multiple fitness data of the same data type to obtain the time correlation between each fitness data and the interpolation points as well as the fitness platform relevance;

[0010] Interpolation and supplementation module: Generate a corresponding weight index for each fitness data according to time correlation and fitness platform relevance. Estimated data of interpolation points are obtained by weighted calculation of fitness data of the same data type. After all interpolation points are supplemented, fitness data from different fitness platforms are integrated and shared.

[0011] In a preferred embodiment, the interpolation and supplementation module takes the fitness platform's adaptability to the interpolation point as the adaptability of historical fitness data to the interpolation point;

[0012] Calculate the correlation coefficient between historical fitness data and the interpolation point based on the time correlation index between historical fitness data and the interpolation point and the adaptability of historical fitness data to the interpolation point. The expression is:

[0013] GSR = γ1 * TE + γ2 * HE, where GSR is the correlation coefficient, HE is the adaptability of historical fitness data to the interpolation point, TE is the time correlation index between historical fitness data and the interpolation point, γ1 and γ2 are proportionality coefficients, and both γ1 and γ2 are greater than 0;

[0014] After obtaining the correlation coefficients of all historical fitness data with the same data type as the interpolation point data, calculate the weight index of each historical fitness data with respect to the interpolation point. The expression is: In the formula, M is the number of historical fitness data with the same data type as the interpolation point data, GSR f is the correlation coefficient of the current historical fitness data, GSR i is the correlation coefficient of the i-th historical fitness data, ω f is the weight index of the current historical fitness data.

[0015] In a preferred embodiment, after the interpolation and supplementation module obtains the weight indices of all historical fitness data with the same data type as the interpolation point data, weighted calculation is performed on all historical fitness data with the same data type as the interpolation point data from different fitness platforms to obtain the fitness data value of the interpolation point. The expression is: In the formula, jsz c is the fitness data value of the interpolation point, M is the number of historical fitness data with the same data type as the interpolation point data, ω i is the weight index of the i-th historical fitness data, jsz i is the i-th historical fitness data value.

[0016] In a preferred embodiment, the data analysis module classifies the types of user fitness data and calculates the time correlation index between historical fitness data and the interpolation point. The expression is: In the formula, TE is the time correlation index between historical fitness data and the interpolation point, T interpolation is the interpolation point time, Thistory where \(t\) is the historical fitness data time and \(\lambda\) is the time decay factor;

[0017] Calculate the fitness platform - interpolation point fitness, and the expression is: In the formula, \(PE\) is the fitness between the fitness platform and the interpolation point, \(N\) is the number of fitness data records of the user on the current fitness platform, \(X_i\) platform,i represents the fitness data with the same data type as the interpolation point in the \(i\) - th exercise record of the user on the current fitness platform, and \(X\) total represents the sum of fitness data with the same data type as the interpolation point of the user on all fitness platforms.

[0018] In a preferred embodiment, the calculation logic of the time decay factor \(\lambda\) is as follows: Obtain the time distribution difference and exercise frequency of the user's fitness data within a period of time, perform normalization processing on the time distribution difference and exercise frequency of the fitness data, map the value ranges of the time distribution difference and exercise frequency of the fitness data to the range of \([0,1]\), obtain the normalized value of the time distribution difference of the fitness data and the normalized value of the exercise frequency, and subtract the normalized value of the exercise frequency from the normalized value of the time distribution difference of the fitness data to obtain the time decay factor \(\lambda\).

[0019] In a preferred embodiment, the data processing module collects data from multiple fitness platforms, parses the data formats of different fitness platforms, extracts fitness indicators, unifies the timestamp format, aligns the data from different time sources, performs user identity association, and attributes the fitness data of different devices to the same user;

[0020] Use the box - plot statistical method to detect outliers, mark the outliers as interpolation points, detect data missing situations, including time - interval breakpoints and indicator missing, analyze the missing data that needs to be filled according to the user's historical fitness data, and mark the missing points as interpolation points;

[0021] Perform standardization processing on the fitness data from different devices and fitness platforms, including unifying units and performing normalization processing on the values using the min - max normalization method, and perform smoothing processing on the fitness data using the moving average method.

[0022] In a preferred embodiment, the data processing module uses the box - plot statistical method to detect outliers, which includes the following steps:

[0023] Sort the collected fitness data according to the numerical size. After sorting, calculate the first quartile \(Q1\) and the third quartile \(Q3\);

[0024] Calculate the inter - quartile range \(IQR\) based on the first quartile \(Q1\) and the third quartile \(Q3\). The inter - quartile range \(IQR\) is used to represent the variation range of the fitness data, and the expression is: \(IQR = Q3−Q1\);

[0025] Generate the upper and lower bounds of fitness data based on the interquartile range, and the expression is:

[0026] After obtaining the upper and lower bounds of the fitness data, analyze each fitness data of the same data type. If the fitness data value is less than the lower bound value, mark this data point as an outlier. If the fitness data value is greater than the upper bound value, mark this data point as an outlier.

[0027] In a preferred embodiment, the data processing module smooths the fitness data by a moving average method, including the following steps:

[0028] After obtaining the data values of the fitness data at multiple data points, calculate the standard deviation of the data values, and the expression is: In the formula, σ is the standard deviation of the data values, m is the number of data points, S i is the fitness data value of the i-th data point, S avg is the average value of the fitness data;

[0029] Compare the obtained standard deviation of the data values with a preset standard deviation threshold. If the standard deviation of the data values of the fitness data is less than or equal to the standard deviation threshold, analyze that the data fluctuation of this fitness data is small. If the standard deviation of the data values of the fitness data is greater than the standard deviation threshold, analyze that the data fluctuation of this fitness data is large;

[0030] Select the fitness data with large fluctuations for smoothing processing. After selecting the time window, perform a moving average calculation on the fitness data within the time window to obtain the moving average value at the current time t, and the expression is: In the formula, SMA t is the moving average value at the current time t, X t is the fitness data value at the current time t, n is the window size. After obtaining the moving average value at the current time t, replace the original data with the moving average value.

[0031] In the above technical solution, the technical effects and advantages provided by the present invention:

[0032] The present invention analyzes the data types that need to be supplemented at the interpolation points based on the user's fitness data through a data analysis module, performs spatio-temporal analysis on multiple fitness data of the same data type to obtain the time correlation between each fitness data and the interpolation points and the fitness platform relevance. The interpolation and supplementation module generates corresponding weight indices for each fitness data based on the time correlation and the fitness platform relevance, and the fitness data of the same data type are weighted and calculated to obtain the estimated data at the interpolation points. After all interpolation points are supplemented, the fitness data from different fitness platforms are integrated and shared with the fitness coach. The sharing system combines spatio-temporal analysis, calculates the correlation between the interpolation points and other data using multiple fitness data, realizes more accurate data completion, assigns reasonable weights to data from different sources, reduces the error between devices and fitness platforms, and improves the scientific nature of data fusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0034] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0036] Embodiment 1: Please refer to Figure 1 As shown, the fitness data sharing system based on digital twin in this embodiment includes a data processing module, a data analysis module, and an interpolation and supplementation module;

[0037] Data processing module: Collects the user's fitness data over a period of time through multi-source data fusion, uses statistical methods to remove outliers from the fitness data, marks both the data missing points and the outlier removal points as interpolation points, performs standardization processing on the fitness data from different fitness platforms, and sends the standardized fitness data and interpolation point information to the data analysis module and the interpolation and supplementation module;

[0038] Data analysis module: Based on the types of data that need to be supplemented at the interpolation points analyzed from the user's fitness data, perform spatio-temporal analysis on multiple fitness data of the same data type to obtain the temporal correlation between each fitness data and the interpolation points as well as the fitness platform relevance, and send the temporal correlation and fitness platform relevance to the interpolation and supplementation module;

[0039] Interpolation and supplementation module: Generate corresponding weight indices for each fitness data based on the temporal correlation and fitness platform relevance, calculate the estimated data of the interpolation points by weighted calculation of fitness data of the same data type, and after all interpolation points are supplemented, integrate and share the fitness data from different fitness platforms to the fitness coach.

[0040] The working process of the sharing system is as follows:

[0041] The sharing system collects the user's fitness data over a period of time through multi-source data fusion, uses statistical methods to remove outliers from the fitness data, marks both the data missing points and the outlier removal points as interpolation points, performs standardization processing on the fitness data from different fitness platforms, based on the types of data that need to be supplemented at the interpolation points analyzed from the user's fitness data, perform spatio-temporal analysis on multiple fitness data of the same data type to obtain the temporal correlation between each fitness data and the interpolation points as well as the fitness platform relevance, generate corresponding weight indices for each fitness data based on the temporal correlation and fitness platform relevance, calculate the estimated data of the interpolation points by weighted calculation of fitness data of the same data type, and after all interpolation points are supplemented, integrate and share the fitness data from different fitness platforms to the fitness coach.

[0042] In this application, the data analysis module performs spatio-temporal analysis on multiple fitness data of the same data type based on the types of data that need to be supplemented at the interpolation points analyzed from the user's fitness data to obtain the temporal correlation between each fitness data and the interpolation points as well as the fitness platform relevance. The interpolation and supplementation module generates corresponding weight indices for each fitness data based on the temporal correlation and fitness platform relevance, calculates the estimated data of the interpolation points by weighted calculation of fitness data of the same data type, and after all interpolation points are supplemented, integrates and shares the fitness data from different fitness platforms to the fitness coach. The sharing system combines spatio-temporal analysis, calculates the correlation between the interpolation points and other data using multiple fitness data, realizes more accurate data completion, assigns reasonable weights to data from different sources, reduces the error between devices and fitness platforms, and improves the scientificity of data fusion.

[0043] Examples of the application scenarios of this application include the following:

[0044] 1. Integration and optimization of gym member training data

[0045] Scenario description: Members of a gym use smart fitness equipment of different brands (such as treadmills, smart spinning bikes, strength training equipment, etc.), and also use different wearable devices (such as smart watches, heart rate monitors) to record exercise data. However, the data format and data sampling frequency of each platform are different, and some training data may be missing.

[0046] Application method: Through multi-source data fusion, collect members' sports data (heart rate, calorie consumption, cadence, etc.) on different devices. Use outlier detection methods to eliminate erroneous data, such as abnormal heart rate values ​​caused by equipment failure. Perform data standardization to make data from different devices compatible with each other and supplement missing values ​​(such as cadence data not recorded by a treadmill can be supplemented by a wristband). Generate a spatiotemporal weight index, calculate the spatiotemporal correlation of data, and improve data integrity. The integrated data is provided to fitness coaches to help them analyze the training status of members and optimize training plans.

[0047] Application effect: Fitness coaches can develop more accurate fitness plans based on more complete user data. Members can obtain cross-device, integrated fitness data reports to avoid data fragmentation.

[0048] 2. Fusion analysis of home fitness and outdoor sports data

[0049] Scenario description: A fitness enthusiast uses a smart fitness mirror and yoga mat to train at home, and also runs, rides, and other sports outdoors. Due to different devices, data is stored in multiple independent fitness platforms, making it impossible to fully view his long-term exercise trends.

[0050] Application method: The sharing system collects users' indoor and outdoor sports data (yoga training duration, running distance, cycling power, etc.). Unify the data format of different devices and supplement missing values ​​(for example, if the fitness mirror does not provide step data, it can be combined with the data of the smart watch for interpolation calculation). Calculate the weights of different sports data based on time and platform relevance and perform data fusion. Generate a complete sports report to show the user's overall fitness status and provide training optimization suggestions.

[0051] Application effect: Users can view all fitness data on one platform to form a unified training file. The integrity of exercise data is improved, helping users better evaluate their fitness progress.

[0052] 3. Fitness competition and intelligent scoring system

[0053] Scenario description: An online fitness platform held a virtual fitness challenge. The participants were from different countries and used smart fitness devices of different brands (such as Peloton, Garmin, Apple Watch). Due to different data formats, it was difficult for the competition organizers to fairly evaluate the performance of all participants.

[0054] Application method: The sharing system collects data from different fitness platforms and standardizes the formats to have a unified evaluation standard. Identify and remove abnormal data. For example, some devices may have incorrectly recorded extremely high calorie consumption. Adopt time correlation analysis to fill in the missing data points to ensure the completeness of all participants' data. Calculate the comprehensive exercise scores of the participants and provide personalized training suggestions based on data analysis.

[0055] Application effect: The fitness data of the participants can be fairly evaluated, improving the fairness of the competition. The platform can provide a consistent competition experience for users of different devices, enhancing user participation.

[0056] 4. Sports rehabilitation and medical health monitoring

[0057] Scenario description: A patient after knee surgery is undergoing rehabilitation training. Their exercise data comes from the motion monitoring devices in the hospital rehabilitation center, smart fitness devices at home, and the remote exercise guidance platform provided by the doctor. These data are scattered in different systems, making it difficult for the doctor to comprehensively track the patient's rehabilitation situation.

[0058] Application method: The sharing system collects the patient's exercise data in different scenarios (such as walking distance, knee flexion and extension angles, heart rate changes, etc.). Unify the data format and interpolate to fill in the missing data points (for example, the daily walking data not recorded by the hospital device can be supplemented by the smart watch). Calculate the spatio-temporal correlation of the data and screen the most representative rehabilitation data. Generate a rehabilitation progress report to help the doctor more accurately evaluate the patient's recovery and adjust the rehabilitation plan.

[0059] Application effect: The doctor can obtain more complete rehabilitation data, improving the accuracy of the treatment plan. The patient can understand their rehabilitation progress through data feedback, enhancing their confidence in rehabilitation.

[0060] 5. Sports training and data analysis of professional athletes

[0061] Scenario description: The coach of a professional sports team hopes to integrate the players' exercise data during training, in the gym, and during competitions to optimize the training plan. However, these data come from a wide range of sources, including smart training bracelets, competition sensors, strength training equipment, etc. The data is inconsistent and some data is missing.

[0062] Application method: The shared system integrates the training data of team members (heart rate, cadence, VO2 max, muscle load, etc.). Data standardization is carried out, and interpolation methods are used to fill in the missing data to ensure the integrity of the training data. According to time and platform weight index, data analysis is optimized to extract key training indicators. Personalized training suggestions are generated to help coaches formulate more scientific training plans.

[0063] Application effect: The sports team can adjust training based on complete data to improve sports performance. Athletes can view long-term training trends and plan their training rhythms more scientifically.

[0064] Example 2: The data processing module collects the fitness data of users over a period of time through multi-source data fusion, uses statistical methods to remove outliers from the fitness data, and marks both the data missing points and the outlier removal points as interpolation points, and performs standardization processing on the fitness data from different fitness platforms;

[0065] The data processing module collects data from multiple fitness platforms (such as smart bracelets, fitness apps, gym equipment, smart watches, etc.), analyzes the data formats of different fitness platforms, extracts key fitness indicators (such as steps, heart rate, calorie consumption, exercise duration, etc.), unifies the timestamp format, aligns the data from different time sources to ensure time synchronization, and performs user identity association to attribute the fitness data of different devices to the same user;

[0066] Use box plot statistical methods to detect outliers and mark the outliers as interpolation points instead of directly deleting them to prevent data loss.

[0067] Detect data missing situations, including time interval breakpoints (such as no data on a certain day) and indicator missing (such as some devices not providing heart rate data), and based on the user's historical fitness data, analyze which data missing needs to be filled in and mark the missing points as interpolation points.

[0068] Since there are differences in units, dimensions, and calculation methods for the fitness data from different devices and fitness platforms, standardization is required, including:

[0069] Unit conversion:

[0070] Unify the calorie consumption calculation method (such as converting kilocalories to joules).

[0071] Unify the measurement methods of steps and distance (such as converting kilometers to miles).

[0072] Unify the time format (such as adjusting the time in different time zones to UTC time).

[0073] The minimum-maximum normalization method is used to normalize the values so that the value ranges of fitness data from different fitness platforms are the same. The moving average method is used to reduce measurement fluctuations and improve data smoothness.

[0074] The data processing module uses the box plot statistical method to detect outliers, including the following steps:

[0075] Sort the collected fitness data (such as steps, heart rate, calorie consumption, etc.) according to the numerical size;

[0076] After sorting, calculate the first quartile Q1 (25% quantile): that is, the value at the first 25% position in the data, calculate the third quartile Q3 (75% quantile): that is, the value at the first 75% position in the data, and calculate the median Q2 (50% quantile): that is, the middle value of the data (used to understand the central tendency of the data);

[0077] Calculate the interquartile range based on the first quartile Q1 and the third quartile Q3. The interquartile range is used to represent the range of data variation, and the expression is: IQR = Q3 - Q1. The interquartile range reflects the degree of concentrated distribution of the data and can be used to judge whether the data is abnormal;

[0078] Generate the upper and lower bounds of the fitness data based on the interquartile range. The expression is:

[0079] After obtaining the upper and lower bounds of the fitness data, analyze each fitness data of the same data type. If the fitness data value is less than the lower bound value, it indicates that it may be an abnormally low outlier (such as an incorrect heart rate measurement resulting in a too low heart rate), and mark this data point as an outlier. If the fitness data value is greater than the upper bound value, it indicates that it may be an abnormally high outlier (such as a misjudged sudden increase in steps), and mark this data point as an outlier. Mark the outlier as an interpolation point;

[0080] The data processing module detects data missing situations, including time interval breakpoints and indicator missing. Based on the user's historical fitness data, analyze the missing data that needs to be filled in, and mark the missing points as interpolation points, including the following steps:

[0081] Define a time interval threshold (such as 1 hour, 1 day), detect breakpoints in the time series, traverse the time series data, and check the time difference between adjacent data points: if the time interval between adjacent data points exceeds the set threshold, it is determined that there is a time breakpoint in this time period, and record the start time and end time of the time breakpoint;

[0082] Statistically determine whether each time point contains complete fitness data. If some metrics are empty or abnormal, it is determined that the metrics at this data point are missing. Record the names and time positions of the missing metrics, obtain the user's historical fitness data (exercise records for the past week and month), analyze the user's exercise pattern, and determine whether the missing data conforms to their exercise habits. For example, if a user usually runs from 6 am to 7 am and the data for this time period is missing, then data such as steps and calories need to be supplemented. Mark the detected time breakpoints and metric missing points as interpolation points, and record the types of interpolation points (time breakpoints or metric missing).

[0083] The data processing module smooths the fitness data through the moving average method, including the following steps:

[0084] The moving average method is used to smooth the short-term fluctuations of fitness data, improve the stability and continuity of the data, and reduce the impact of mutation values on data analysis.

[0085] Select fitness data with large fluctuations for smoothing, such as: heart rate data (the heart rate sensor may be affected by jitter). Step data (the pedometer error may cause sudden increases or decreases in steps). Calorie consumption (different calculation methods may cause numerical fluctuations). Identify abnormal fluctuation points to avoid error propagation. After obtaining the data values of fitness data at multiple data points, calculate the standard deviation of the data values. The expression is: In the formula, σ is the standard deviation of the data values, m is the number of data points, S i is the fitness data value of the i-th data point, S avg is the mean of the fitness data. The larger the standard deviation of the data values of the fitness data, the greater the fluctuation of the fitness data. Compare the obtained standard deviation of the data values with the preset standard deviation threshold. If the standard deviation of the data values of the fitness data is less than or equal to the standard deviation threshold, analyze that the data fluctuation of the fitness data is small. If the standard deviation of the data values of the fitness data is greater than the standard deviation threshold, analyze that the data fluctuation of the fitness data is large.

[0086] After selecting the time window, perform a moving average calculation on the fitness data within the time window to obtain the moving average value at the current time t. The expression is:

[0087] In the formula, SMA t is the moving average value at the current time t, X t is the fitness data value at the current time t, and n is the window size;

[0088] After obtaining the moving average value at the current time t, replace the original data with the moving average value to ensure that the data is stable and the trend is clear.

[0089] The data analysis module conducts spatio-temporal analysis on multiple fitness data of the same data type according to the data types that need to be supplemented at the interpolation points based on the user's fitness data, so as to obtain the time correlation between each fitness data and the interpolation point and the relevance of the fitness platform;

[0090] The data analysis module classifies the types of user fitness data. Common data types include: steps (pedometer data), heart rate (data measured by wearable devices), calorie consumption (calculated based on steps, exercise time, and weight), and exercise duration (start and end times of exercise).

[0091] Analyze the missing situation of the interpolation points to judge the data types that need to be supplemented. If only the steps are missing, the interpolation points need to supplement step data. If multiple indicators are missing (such as steps and heart rate), joint supplementation is required.

[0092] Calculate the time correlation index between the historical fitness data and the interpolation point. The expression is:

[0093] In the formula, TE is the time correlation index between the historical fitness data and the interpolation point, T interpolation is the time of the interpolation point, T history is the time of the historical fitness data, and λ is the time decay factor. The smaller the time correlation index, the stronger the time correlation between the historical fitness data and the interpolation point, and the greater the weight;

[0094] The time decay factor λ reflects the time stability of the user's fitness data and is used to control the decay speed of the time correlation weight. The larger λ is, the more stable the user's fitness data changes over time, that is, the data within a longer period still has a higher reference value; the smaller λ is, the faster the data changes, and the earlier data contributes less to the current interpolation.

[0095] Obtain the time distribution difference and exercise frequency of the user's fitness data within a period of time, normalize the time distribution difference and exercise frequency of the fitness data, map the value range of the time distribution difference and exercise frequency of the fitness data to between [0,1], obtain the normalized value of the time distribution difference of the fitness data and the normalized value of the exercise frequency, subtract the normalized value of the exercise frequency from the normalized value of the time distribution difference of the fitness data to obtain the time decay factor λ. The time decay factor determines the influence weight of the historical fitness data on the interpolation point data, playing a role in balancing the long-term trend and short-term changes. It can improve the accuracy of the interpolated data, enabling the fitness data sharing system to more accurately predict and complete the missing data;

[0096] The calculation expression of the time distribution difference of the fitness data is: In the formula, is the time distribution difference of the fitness data, A is the total number of user exercise records, T i is the exercise duration of the user's i-th exercise, It is the average exercise duration of users. If the time distribution difference of fitness data is large, it indicates that the users' exercise time is not fixed. Then the time decay factor takes a large value, making the influence of historical data more persistent. Even if the time interval is large, it can still play a role.

[0097] The exercise frequency is obtained by dividing the number of exercise records of the user within the statistical time range by the total duration of the statistical time range. If the exercise frequency is high (the user's exercise habit is regular), then λ takes a small value, making the weight of recent data higher.

[0098] Calculate the fitness degree of fit between the fitness platform and the interpolation point. The expression is: In the formula, PE is the fitness degree of fit between the fitness platform and the interpolation point, N is the number of fitness data records of the user on the current fitness platform, and X platform,i represents the fitness data (such as steps, heart rate, etc.) of the user's i-th exercise record on the current fitness platform that has the same data type as the interpolation point data, and X total represents the total sum of fitness data of the user on all fitness platforms that has the same data type as the interpolation point data.

[0099] The greater the fitness degree of fit between the fitness platform and the interpolation point, the higher the matching degree of the fitness data on this fitness platform with the interpolation point, that is, the greater the weight.

[0100] The interpolation and supplementation module generates a corresponding weight index for each fitness data based on time correlation and fitness platform relevance, calculates the estimated data of the interpolation point by weighted calculation of fitness data of the same data type, and after all interpolation points are supplemented, integrates and shares the fitness data from different fitness platforms with the fitness coach;

[0101] The interpolation and supplementation module takes the fitness degree of fit between the fitness platform and the interpolation point as the fitness degree of fit between the historical fitness data and the interpolation point. That is, if the historical fitness data comes from a certain fitness platform, then the fitness degree of fit between this fitness platform and the interpolation point is the fitness degree of fit between the historical fitness data and the interpolation point;

[0102] Calculate the correlation coefficient between the historical fitness data and the interpolation point based on the time correlation index between the historical fitness data and the interpolation point and the fitness degree of fit between the historical fitness data and the interpolation point. The expression is:

[0103] GSR = γ2 * HE - γ1 * TE. In the formula, GSR is the correlation coefficient, HE is the fitness degree of fit between the historical fitness data and the interpolation point, TE is the time correlation index between the historical fitness data and the interpolation point, γ1 and γ2 are proportionality coefficients, and both γ1 and γ2 are greater than 0;

[0104] After obtaining the correlation coefficients of all historical fitness data with the same data type as the interpolation point, calculate the weight index of each historical fitness data with the interpolation point. The expression is: where M is the number of historical fitness data of the same data type as the interpolation point data, and GSR f is the correlation coefficient of the current historical fitness data, and GSR i is the correlation coefficient of the i-th historical fitness data, and ω f is the weight index of the current historical fitness data.

[0105] After the interpolation supplement module obtains the weight indexes of all historical fitness data of the same data type as the interpolation point data, it calculates the fitness data value of the interpolation point by weighted calculation of all historical fitness data of the same data type from different fitness platforms. The expression is: where jsz c is the fitness data value of the interpolation point, M is the number of historical fitness data of the same data type as the interpolation point data, and ω i is the weight index of the i-th historical fitness data, and jsz i is the i-th historical fitness data value.

[0106] The above formulas are all dimensionless and take their numerical calculations. The formula is a formula obtained by collecting a large amount of data for software simulation to get the closest real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0107] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0108] The above-disclosed preferred embodiments of the present invention are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to only the specific implementation manners. Obviously, according to the content of this specification, many modifications and changes can be made. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art in the technical field can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. Fitness data sharing system based on digital twins, characterized by: It includes data processing module, data analysis module and interpolation supplement module; Data processing module: collects users' fitness data over a period of time through multi-source data fusion, uses statistical methods to remove outliers from the fitness data, and marks data missing points and outlier removal points as interpolation points, and standardizes fitness data from different fitness platforms; Data analysis module: Analyze the data types that need to be supplemented at the interpolation points based on the user's fitness data, and perform spatiotemporal analysis on multiple fitness data of the same data type to obtain the time correlation between each fitness data and the interpolation points and the relevance of the fitness platform; Interpolation supplement module: Generate corresponding weight index for each fitness data according to time correlation and fitness platform relevance. Fitness data of the same data type are weighted to obtain estimated data of interpolation points. After all interpolation points are supplemented, fitness data from different fitness platforms are integrated and shared.

2. The digital twin-based fitness data sharing system according to claim 1, characterized in that: The interpolation supplement module uses the fitness degree between the fitness platform and the interpolation point as the fitness degree between the historical fitness data and the interpolation point; The correlation coefficient between the historical fitness data and the interpolation point is obtained by calculating the time correlation index between the historical fitness data and the interpolation point and the fitness between the historical fitness data and the interpolation point. The expression is: GSR = γ2*HE-γ1*TE, where GSR is the correlation coefficient, HE is the degree of fit between the historical fitness data and the interpolation point, TE is the temporal correlation index between the historical fitness data and the interpolation point, γ1 and γ2 are proportional coefficients, and both γ1 and γ2 are greater than 0; After obtaining the correlation coefficients of all historical fitness data with the same data type as the interpolation point, the weight index of each historical fitness data and the interpolation point is calculated. The expression is: Where M is the number of historical fitness data with the same data type as the interpolation point, GSR f is the correlation coefficient of the current historical fitness data, GSR i is the correlation coefficient of the i-th historical fitness data, ω f is the weight index of the current historical fitness data.

3. The digital twin-based fitness data sharing system according to claim 2, characterized in that: After the interpolation supplement module obtains the weight index of all historical fitness data with the same data type as the interpolation point, all historical fitness data with the same data type as the interpolation point from different fitness platforms are weighted and calculated to obtain the fitness data value of the interpolation point. The expression is: In the formula, jsz c is the fitness data value of the interpolation point, M is the number of historical fitness data with the same data type as the interpolation point, ω i is the weight index of the i-th historical fitness data, jsz i is the i-th historical fitness data value.

4. The digital twin-based fitness data sharing system according to claim 3 is characterized in that: The data analysis module classifies the user's fitness data type and calculates the time correlation index between the historical fitness data and the interpolation point. The expression is: Where TE is the time correlation index between historical fitness data and interpolation points, T interpolation is the interpolation point time, T history is the historical fitness data time, λ is the time decay factor; Calculate the fitness of the fitness platform and the interpolation point. The expression is: Where PE is the degree of fit between the fitness platform and the interpolation point, N is the number of fitness data records of the user on the current fitness platform, and X platform,i represents the user's fitness data of the i-th exercise record on the current fitness platform, which has the same data type as the interpolation point. total Represents the sum of the user's fitness data across all fitness platforms with the same interpolation point data type.

5. The digital twin-based fitness data sharing system according to claim 4, characterized in that: The calculation logic of the time decay factor λ is as follows: obtaining the time distribution difference and exercise frequency of the user's fitness data within a period of time, normalizing the time distribution difference and exercise frequency of the fitness data so that the value ranges of the time distribution difference and exercise frequency of the fitness data are mapped to between [0,1], obtaining the normalized value of the time distribution difference of the fitness data and the normalized value of the exercise frequency, and subtracting the normalized value of the exercise frequency from the normalized value of the time distribution difference of the fitness data to obtain the time decay factor λ.

6. The digital twin-based fitness data sharing system according to claim 5, characterized in that: The data processing module collects data from multiple fitness platforms, parses the data formats of different fitness platforms, extracts fitness indicators, unifies the timestamp format, aligns data from different time sources, associates user identities, and attributes fitness data from different devices to the same user; Use box plot statistics to detect outliers, mark outliers as interpolation points, detect missing data, including time interval breakpoints and missing indicators, analyze missing data that need to be completed based on the user's historical fitness data, and mark missing points as interpolation points; Standardize fitness data from different devices and fitness platforms, including unifying units and normalizing values ​​using the min-max normalization method, and smoothing fitness data using the sliding average method.

7. The digital twin-based fitness data sharing system according to claim 6, characterized in that: The data processing module uses the box plot statistical method to detect outliers, including the following steps: Sort the collected fitness data by numerical value. After sorting, calculate the first quartile Q1 and the third quartile Q3; The interquartile range IQR is calculated based on the first quartile Q1 and the third quartile Q3. The interquartile range IQR is used to represent the range of variation of fitness data. The expression is: IQR = Q3 - Q1; The upper and lower bounds of fitness data are generated based on the interquartile range, and the expression is: After obtaining the upper and lower bounds of the fitness data, each fitness data of the same data type is analyzed. If the fitness data value is less than the lower bound value, the data point is marked as an outlier. If the fitness data value is greater than the upper bound value, the data point is marked as an outlier.

8. The digital twin-based fitness data sharing system according to claim 7, characterized in that: The data processing module performs smoothing processing on the fitness data by a sliding average method, including the following steps: After obtaining the data values ​​of the fitness data at multiple data points, the standard deviation of the data values ​​is calculated. The expression is: In the formula, σ is the standard deviation of the data value, m is the number of data points, S i is the fitness data value of the ith data point, S avg is the mean of fitness data; Compare the acquired data value standard deviation with a preset standard deviation threshold value, if the data value standard deviation of the fitness data is less than or equal to the standard deviation threshold value, analyze that the data fluctuation of the fitness data is small, if the data value standard deviation of the fitness data is greater than the standard deviation threshold value, analyze that the data fluctuation of the fitness data is large; Select the fitness data with large fluctuations for smoothing. After selecting the time window, perform sliding average calculation on the fitness data in the time window to obtain the sliding average at the current time t. The expression is: In the formula, SMA t is the sliding average at the current time t, X t is the fitness data value at the current time t, n is the window size, after obtaining the sliding average at the current time t, the sliding average is used to replace the original data.

Images