Child body fat rate adaptive prediction method and system based on trajectory feature cooperation

Abstract

The present application relates to the technical field of children's health assessment, in particular to a children's body fat rate adaptive prediction method and system based on trajectory feature cooperation, comprising the following steps: obtaining children's physical examination data set, analyzing height and weight change to calculate difference ratio to screen samples, extracting waist circumference change slope to evaluate stability to screen samples, calculating body sign proportion difference to adjust training weight, identifying age stage residual difference gap to adjust model parameters, combining body fat trend and body weight trend direction to adjust prediction slope to generate results. In the present application, samples are screened through height and weight trajectory, combined with the fluctuation continuity of waist circumference change slope to eliminate abnormal data, the feature weighting coefficient is set according to the proportion deviation of height and weight and muscle mass, the response coefficient is adjusted according to the distance relationship of the grouping mean and the central interval of the age stage residual, the change path is judged by the consistency of the body fat and body weight prediction trend, the non-trend prediction deviation is avoided, and the stability and sample adaptability of the model are improved.

Description

Technical Field

[0001] This invention relates to the field of children's health assessment technology, and in particular to a method and system for adaptive prediction of children's body fat percentage based on trajectory feature collaboration. Background Technology

[0002] The field of pediatric health assessment technology encompasses various methods and tools for the quantitative analysis and comprehensive evaluation of the physical health status of children. This includes monitoring children's growth and development, assessing physiological indicators, determining nutritional status, and identifying potential health risks. By collecting basic physical data such as height, weight, waist circumference, and BMI, and combining this data with medical standards and health models, personalized health reference information is provided for children. Furthermore, information technologies such as electronic health records and health data modeling are used to continuously track and dynamically assess children's health status. Among these methods, pediatric body fat prediction refers to constructing a system based on children's physical data... The body fat percentage prediction model uses a supervised learning mechanism to establish a mapping relationship between variables to estimate body fat levels. It covers the collection of children's physical characteristics data, construction of training data, selection of feature variables, setting of model structure, and prediction output. Specifically, it includes constructing a data sample set based on basic variables such as children's height, weight, age, gender, and BMI, using a regression model as the core prediction architecture, and combining data standardization, multivariate linear fitting, residual analysis optimization, and other methods to train the model. The model accuracy is evaluated by cross-validation, and the estimated value of children's body fat percentage is obtained through the model output, realizing indirect prediction of body fat percentage based on static physical characteristics data.

[0003] Traditional children's body fat prediction technology uses static physical characteristics variables to construct a single-period sample set and fits the body fat output with a global model structure. It does not dynamically segment the physical characteristics trajectory of the samples during periodic growth changes, and cannot remove training data with structural inconsistencies during developmental fluctuations. This results in a mixed sample input structure and unstable convergence during training. In groups with unbalanced feature ratios, it causes systematic errors in the prediction results. For example, the BMI input of groups with abnormal muscle proportions amplifies the bias in fat determination. During continuous monitoring, the lack of a trend adjustment mechanism causes directional distortion of body fat prediction values, affecting the reliability of the model and the range of data credibility in health monitoring scenarios. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a method and system for adaptive prediction of children's body fat percentage based on trajectory feature collaboration. The technical solution is as follows:

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an adaptive prediction method for children's body fat percentage based on trajectory feature collaboration, comprising the following steps:

[0006] S1: Obtain children's physical examination dataset, analyze the height and weight change trajectory of each sample, calculate the difference ratio between height and weight growth of each sample in a continuous period, and filter the samples according to the distribution characteristics of the difference ratio in multiple intervals to generate a rhythm-screened sample set.

[0007] S2: Call the rhythm screening sample set, analyze the waist circumference change data of each sample in multiple consecutive detection cycles after screening, calculate the difference in waist circumference change slope between adjacent cycles, evaluate the stability of waist circumference change of the sample, further screen the sample, and generate waist circumference rhythm screening set.

[0008] S3: Using the waist circumference rhythm screening set, calculate the ratio difference between height, weight and muscle mass for each sample after further screening. Based on the distribution characteristics of the ratio difference in multiple intervals, calculate the degree of deviation and adjust the feature weights in the training phase to generate training weight adjustment data.

[0009] S4: Extract the training weight adjustment data, train the body fat prediction model, calculate the mean of the prediction residuals of the samples in each age group, identify the distance from the overall sample residual concentration interval, screen the age groups with abnormal residuals, adjust the residual response adjustment coefficient, and generate the model parameter adjustment results.

[0010] As a further embodiment of the present invention, the rhythm screening sample set includes height growth trajectory, weight growth trajectory, and difference ratio distribution interval; the waist circumference rhythm screening set includes waist circumference change slope, slope fluctuation continuity, and waist circumference change trend structure; the training weight adjustment data includes the main interval of the ratio difference, the degree of sample deviation, and the feature weighting coefficient; and the model parameter adjustment results include prediction residual grouping, age group error level, and residual response adjustment coefficient.

[0011] As a further aspect of the present invention, the step of obtaining the rhythm screening sample set specifically includes:

[0012] S101: Obtain the children's physical examination dataset, extract the height and weight data of each sample in multiple consecutive testing periods, call the height and weight changes of each sample in adjacent periods, calculate the difference ratio between height growth and weight growth, and generate height and weight difference ratio data.

[0013] S102: Based on the height-weight difference ratio data, by identifying the number of samples in each difference ratio interval, analyze the density difference of the number of samples in each interval, and generate difference ratio distribution density data;

[0014] S103: Call the difference ratio distribution density data, identify the interval with the highest distribution density according to the concentration of sample distribution density in each interval, perform preliminary screening of samples, and generate a rhythm screening sample set.

[0015] As a further aspect of the present invention, the step of obtaining the waist circumference rhythm filter set specifically includes:

[0016] S201: Call the rhythm screening sample set, extract the waist circumference data of each sample in multiple consecutive detection cycles, calculate the waist circumference change rate of each sample in multiple cycles according to the recording order between adjacent detection cycles, and generate waist circumference change slope sequence data.

[0017] S202: Based on the waist circumference change slope sequence data, select the slope difference between adjacent periods of each sample, and evaluate the stability of waist circumference change by analyzing the slope change trend of each sample, thereby generating waist circumference fluctuation structure data.

[0018] S203: Call the waist circumference fluctuation structure data, and further filter the samples based on the stability of the waist circumference data changes to generate a waist circumference rhythm filter set.

[0019] As a further aspect of the present invention, the specific formula for evaluating the stability of waist circumference changes in the sample is as follows:

[0020] ;

[0021] Calculate the stability structural value of the slope difference of waist circumference change;

[0022] in, For the first The sample at the th Normalized value of the slope of waist circumference change over each period For the first The sample at the th Normalized value of the slope of waist circumference change over each period For the first The sample at the th The horizontal detection frequency weighting factor for each cycle, For the first The sample at the th Normalized value of waist circumference change rate over a period of time For the first The age smoothing adjustment factor for each sample For the first The number of consecutive testing cycles contained in a sample. For periodic index variables, For sample index variables, For the first The stability structure value of the slope difference of waist circumference change corresponding to each sample.

[0023] As a further aspect of the present invention, the step of obtaining the training weight adjustment data specifically includes:

[0024] S301: Using the waist circumference rhythm screening set, extract the height, weight and muscle mass data of each sample after further screening, calculate the ratio of weight to height of each sample, and obtain the ratio of weight to muscle mass. By analyzing the distribution of the ratio values ​​in multiple continuous numerical intervals, generate body proportion distribution data.

[0025] S302: Call the body proportion distribution data, identify the interval with the highest proportion distribution density and use it as the body proportion benchmark, calculate the difference between the proportion value of each sample and the benchmark, and generate body structure deviation data;

[0026] S303: Based on the body structure deviation data, assess the degree of deviation of each sample according to the difference between the sample ratio and the benchmark, adjust the feature weighting coefficients in the training phase, construct a set of weight parameters for the training samples, and generate training weight adjustment data.

[0027] As a further aspect of the present invention, the step of obtaining the model parameter adjustment result specifically includes:

[0028] S401: Extract the training weight adjustment data, use the samples to train the body fat prediction model, calculate the residual value between the sample prediction result and the actual result, and combine the age group label of each sample to group the residual data to generate age group residual group data.

[0029] S402: Call the age group residual grouping data, calculate the residual mean within each age group, identify the concentration interval of the overall training sample residuals, analyze the distance between the residual mean of each age group and the overall residual interval, evaluate the deviation range of the residual mean, and generate age group residual deviation data.

[0030] S403: Based on the residual deviation data of the age group, according to the distance between the average residual level of each age group and the overall sample residual concentration interval, screen the age groups with abnormal residuals, and adjust the residual response adjustment coefficient of the target age group sample to generate the model parameter adjustment results.

[0031] As a further aspect of the present invention, the method further includes:

[0032] S5: Based on the model parameter adjustment results, collect continuous vital sign data records of the target user, use the body fat prediction model to predict the user's body fat percentage, identify the changing trend of body fat prediction results over multiple periods, compare it with the user's weight data, analyze the consistency of the changing trend, use the trend direction relationship to set the slope adjustment range of the body fat prediction change path, and generate body fat trend prediction results.

[0033] The body fat trend prediction results specifically include body fat prediction data, body fat change trend, and trend slope adjustment range.

[0034] As a further aspect of the present invention, the steps for obtaining the body fat trend prediction result are specifically as follows:

[0035] S501: Based on the model parameter adjustment results, collect the target user's height, weight, waist circumference and other physical characteristics data for multiple consecutive periods, call the body fat prediction model, output the user's body fat prediction results for each period, and generate body fat percentage sequence data.

[0036] S502: Based on the body fat percentage sequence data, extract the weight change trend of the target user in each period, analyze the direction of change of the body fat percentage sequence and the weight data, analyze the degree of consistency of the change trend, obtain the trend consistency index, and generate trend relationship data;

[0037] The specific formula for the degree of consistency in the analysis of changing trends is as follows:

[0038] ;

[0039] Calculate the trend consistency index;

[0040] in, This represents an indicator of trend consistency. Representing the Normalized values ​​of body fat percentage changes over a period of time Representing the Normalized values ​​of weight change over a period of time This indicates the total number of consecutive detection cycles. Indicates the current cycle number. This represents a non-zero minimum constant set to prevent the denominator from being zero;

[0041] S503: Call the trend relationship data, adjust the slope of the predicted body fat change path according to the relationship between the body fat percentage change trend and the weight change trend, and obtain the body fat trend prediction result.

[0042] On the other hand, an adaptive prediction system for children's body fat percentage based on trajectory feature collaboration is provided. This system is applied to the adaptive prediction method for children's body fat percentage based on trajectory feature collaboration. The system includes:

[0043] The vital signs screening module acquires children's physical examination datasets, analyzes the height and weight change trajectories of each sample, calculates the difference ratio between height and weight growth for each sample in a continuous period, and filters the samples based on the distribution characteristics of the difference ratio in multiple intervals to generate a rhythm screening sample set.

[0044] The waist circumference stability module calls the rhythm screening sample set, analyzes the waist circumference change data of each sample in multiple consecutive detection cycles after screening, calculates the difference in waist circumference change slope between adjacent cycles, evaluates the stability of waist circumference change of the sample, further screens the sample, and generates a waist circumference rhythm screening set.

[0045] The proportional weight adjustment module uses the waist circumference rhythm screening set to calculate the proportional difference between height, weight and muscle mass of each sample after further screening. Based on the distribution characteristics of the proportional difference in multiple intervals, it calculates the degree of deviation and adjusts the feature weights in the training phase to generate training weight adjustment data.

[0046] The age error correction module extracts the training weight adjustment data, trains the body fat prediction model, calculates the mean of the prediction residuals of the samples in each age group, identifies the distance from the overall sample residual concentration interval, filters out age groups with abnormal residuals, adjusts the residual response adjustment coefficient, and generates model parameter adjustment results.

[0047] The trend output module collects continuous vital sign data records of the target user based on the model parameter adjustment results, uses the body fat prediction model to predict the user's body fat percentage, identifies the changing trend of body fat prediction results over multiple periods, compares it with the user's weight data, analyzes the consistency of the changing trend, uses the trend direction relationship to set the slope adjustment range of the body fat prediction change path, and generates body fat trend prediction results.

[0048] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0049] Based on the difference ratio of height and weight trajectories in children's physical examination data, a sample selection process is constructed by constructing a distribution interval of change rhythm. Abnormal data is eliminated by combining the fluctuation continuity of waist circumference change slope. Feature weighting coefficients are set according to the ratio deviation of height, weight and muscle mass. A training parameter control path oriented towards sample structural differences is established. The response coefficient is adjusted by the distance relationship between the group mean of age group residuals and the concentration interval. The change path is judged based on the consistency of body fat and weight prediction trends. The slope adjustment range of body fat percentage results is constructed to dynamically enhance the time response performance of body fat change direction, avoid non-trend prediction bias, improve the stability and sample adaptability of the model in continuous prediction over time periods, enhance the input tolerance range of abnormal body shape samples, and strengthen the rationality of prediction results output under physical sign perturbation. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0052] Figure 2 This is a detailed flowchart of S1 of the present invention;

[0053] Figure 3 This is a detailed flowchart of the S2 process of the present invention;

[0054] Figure 4 This is a detailed flowchart of the S3 process of the present invention;

[0055] Figure 5 This is a detailed flowchart of the S4 process of the present invention;

[0056] Figure 6 This is a detailed flowchart of S5 of the present invention;

[0057] Figure 7 This is a system flowchart of the present invention. Detailed Implementation

[0058] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0059] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0060] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0061] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0062] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0063] Please see Figure 1 This invention provides a technical solution for an adaptive prediction method of children's body fat percentage based on trajectory feature collaboration, comprising the following steps:

[0064] S1: Obtain children's physical examination dataset, analyze the height and weight change trajectory of each sample, calculate the difference ratio between height and weight growth of each sample in a continuous period, and filter the samples according to the distribution characteristics of the difference ratio in multiple intervals to generate a rhythm-screened sample set.

[0065] S2: Call the rhythm-filtered sample set, analyze the waist circumference change data of each sample in multiple consecutive detection cycles after filtering, calculate the difference in waist circumference change slope between adjacent cycles, evaluate the stability of waist circumference change of the sample, further filter the sample, and generate a waist circumference rhythm-filtered set.

[0066] S3: Using the waist circumference rhythm screening set, calculate the ratio difference between height, weight and muscle mass for each sample after further screening. Based on the distribution characteristics of the ratio difference in multiple intervals, calculate the degree of deviation and adjust the feature weights in the training phase to generate training weight adjustment data.

[0067] S4: Extract training weight adjustment data, train the body fat prediction model, calculate the mean of the prediction residuals of the sample in each age group, identify the distance from the overall sample residual concentration interval, screen the age groups with abnormal residuals, adjust the residual response adjustment coefficient, and generate the model parameter adjustment results.

[0068] S5: Based on the model parameter adjustment results, collect continuous vital sign data records of the target user, use the body fat prediction model to predict the user's body fat percentage, identify the changing trend of body fat prediction results over multiple periods, compare it with the user's weight data, analyze the consistency of the changing trend, use the trend direction relationship to set the slope adjustment range of the body fat prediction change path, and generate body fat trend prediction results.

[0069] The rhythm screening sample set includes height growth trajectory, weight growth trajectory, and difference ratio distribution interval. The waist circumference rhythm screening set includes waist circumference change slope, slope fluctuation continuity, and waist circumference change trend structure. The training weight adjustment data includes the main interval of the ratio difference, the degree of sample deviation, and the feature weighting coefficient. The model parameter adjustment results include prediction residual grouping, age group error level, and residual response adjustment coefficient. The body fat trend prediction results specifically include body fat prediction data, body fat change trend, and trend slope adjustment magnitude.

[0070] Please see Figure 2 The specific steps for obtaining the rhythm screening sample set are as follows:

[0071] S101: Obtain the children's physical examination dataset, extract the height and weight data of each sample in multiple consecutive testing periods, call the height and weight changes of each sample in adjacent periods, calculate the difference ratio between height growth and weight growth, and generate height and weight difference ratio data.

[0072] To obtain the height and weight data of each sample in the children's physical examination dataset, we will use three consecutive physical examinations of a certain grade in a school as an example. We will extract some student data to illustrate the specific calculation process for changes in height and weight. The example is shown in Table 1.

[0073] Table 1 Student Continuous Periodic Physical Examination Data Table

[0074] ;

[0075] As shown in Table 1, taking student A as an example, the height increase between adjacent periods is calculated as follows: subtracting the second and first test results gives 124cm - 120cm = 4cm, and subtracting the third and second test results gives 128cm - 124cm = 4cm. Thus, the height increase for consecutive periods is 4cm and 4cm respectively. The weight increase for consecutive periods is calculated in the same way. The weight difference between the second and first test results is 25kg - 23kg = 2kg, and the weight difference between the third and second test results is 27kg - 25kg = 2kg. Thus, the weight increase for consecutive periods is 2kg. Next, the difference ratio between height increase and weight increase is calculated. Taking student A as an example, the difference ratio is calculated by dividing the height increase by the weight increase. The difference ratio between the first and second test periods is 4cm ÷ 2kg = 2cm / kg, and the difference ratio between the second and third test periods is also 2cm / kg. The same calculation method is used to process other students in the table to obtain the difference ratio for consecutive periods for each sample, thus obtaining the height and weight difference ratio data.

[0076] S102: Based on the height-weight difference ratio data, by identifying the number of samples in each difference ratio interval, analyze the density difference of the number of samples in each interval, and generate difference ratio distribution density data;

[0077] Based on the height-weight difference ratio data, taking the continuous periodic difference ratio data of three students in Table 1 as an example: Student A's difference ratio for both periods was 2cm / kg; Student B's difference ratio for the first and second periods was 3cm ÷ 1kg = 3cm / kg, and for the second and third periods it was 2cm ÷ 1.5kg ≈ 1.33cm / kg; Student C's difference ratio for the first and second periods was 4cm ÷ 2kg = 2cm / kg, and for the second and third periods it was 5cm ÷ 3kg ≈ 1.67cm / kg. The difference ratios of each sample are categorized into the corresponding distribution intervals, as shown in Table 2.

[0078] Table 2. Statistical table of distribution density of difference ratios

[0079] ;

[0080] As shown in Table 2, the number of samples appearing in each interval is counted, and the percentage density of the total number of occurrences is calculated. Taking the difference ratio interval of 2 to 2.5 cm / kg as an example, it appears 3 times in the six difference ratio data of three samples. The calculated density is 3 times ÷ 6 times × 100% = 50%. In this way, statistical analysis is performed on each interval one by one to complete the calculation of the sample density in each difference ratio interval and generate difference ratio distribution density data.

[0081] S103: Call the difference ratio distribution density data, identify the interval with the highest distribution density based on the concentration of sample distribution density in each interval, perform preliminary screening of samples, and generate a rhythm screening sample set;

[0082] Using the difference ratio distribution density data, and continuing with the statistical data in Table 2 as an example, the density comparison method was used to identify the interval with the highest density proportion in each interval. From the density data shown in Table 2, it was found that the density of the difference ratio interval 2 to 2.5 cm / kg was 50%, which far exceeded the density of other intervals (1 to 1.5 cm / kg was 16.7%, 1.5 to 2 cm / kg was 16.7%, and 3 to 3.5 cm / kg was 16.7%). The interval of 2 to 2.5 cm / kg was determined to be the interval with the highest sample density. Subsequently, student samples in this interval were marked and extracted. Taking student A as an example, the difference ratio of all cycles was 2 cm / kg, which belongs to the interval with the highest density. The difference ratio of student C's first and second cycles was also 2 cm / kg, which belongs to this interval. Student B did not enter this interval and was therefore excluded. By using this method to screen all samples, the sample group with the difference ratio located in the interval with the highest density was extracted, and a rhythm screening sample set was generated.

[0083] Please see Figure 3 The specific steps for obtaining the waist circumference rhythm filter set are as follows:

[0084] S201: Call the rhythm-filtered sample set, extract the waist circumference data of each sample in multiple consecutive detection cycles, calculate the waist circumference change rate of each sample in multiple cycles according to the recording order between adjacent detection cycles, and generate waist circumference change slope sequence data.

[0085] The rhythm screening sample set was called up, and waist circumference measurement data of each sample in three consecutive physical examination cycles were extracted. Taking the waist circumference data of students A and C in the same grade rhythm screening sample set as an example, the waist circumference of student A was 55cm in the first measurement, 56.5cm in the second measurement, and 58cm in the third measurement. The waist circumference of student C was 56cm in the first measurement, 58cm in the second measurement, and 61cm in the third measurement, as shown in Table 3.

[0086] Table 3. Student Waist Circumference Data over Continuous Periods

[0087] ;

[0088] As shown in Table 3, the waist circumference change rate for each sample was calculated. The rate was defined as the difference in waist circumference between adjacent periods divided by the time interval of the corresponding detection period. Taking student A as an example, the difference between the first and second waist circumference data was calculated as 56.5cm − 55cm = 1.5cm, and the difference between the second and third waist circumference data was calculated as 58cm − 56.5cm = 1.5cm. Assuming the detection period interval is 6 months, the waist circumference change rate for student A in two adjacent periods is 1.5cm / 6 months = 0.25cm / month. The waist circumference change rate for student C was calculated in the same way. The rate between the first and second detection periods was (58cm − 56cm) / 6 months ≈ 0.33cm / month, and the rate between the second and third detection periods was (61cm − 58cm) / 6 months = 0.5cm / month. The waist circumference change rate of all samples in adjacent periods was calculated and recorded through the above process to generate waist circumference change slope sequence data.

[0089] S202: Based on the waist circumference change slope sequence data, select the slope difference between adjacent periods of each sample, and evaluate the stability of waist circumference change by analyzing the slope change trend of each sample, thereby generating waist circumference fluctuation structure data.

[0090] The specific formula for assessing the stability of waist circumference changes in the sample is as follows:

[0091] ;

[0092] Calculate the stability structural value of the slope difference of waist circumference change;

[0093] in, For the first The sample at the th Normalized value of the slope of waist circumference change over each period For the first The sample at the th Normalized value of the slope of waist circumference change over each period For the first The sample at the th The horizontal detection frequency weighting factor for each cycle, For the first The sample at the th Normalized value of waist circumference change rate over a period of time For the first The age smoothing adjustment factor for each sample For the first The number of consecutive testing cycles contained in a sample. For periodic index variables, For sample index variables, For the first The stability structure value of the slope difference of waist circumference change corresponding to each sample.

[0094] formula:

[0095] ;

[0096] Detailed explanation of the formula and its calculation derivation:

[0097] This formula is used to calculate the first... The dynamic index of waist circumference changes in a child sample over a continuous testing period reflects the stability of the waist circumference change trend during the periodic variation. The results are used to assess whether the sample has structural rhythmic characteristics, thereby further screening samples with good stability to generate waist circumference fluctuation structure data.

[0098] Parameter meanings and settings:

[0099] Indicates the first The sample at the th The normalized value of the rate of change in waist circumference between cycles is based on the maximum rate of change within the cycle. The unit is dimensionless. In actual testing, the waist circumference was set at 56.8 cm in cycle 3 and 55.2 cm in cycle 2, with a cycle interval of 30 days. The rate of change is... The maximum recorded rate is 0.0667, so the normalized value is... ;

[0100] Indicates the first The sample at the th The cross-sectional detection frequency weighting factor for the period is set based on the number of detections per unit time. Given a maximum detection frequency of 10 times, and setting the sample to 8 times in the 3rd period, the corresponding weighting factor is... ;

[0101] Indicates the first The sample at the th The stability coefficient of the cyclical waist circumference change trend is set as the normalized form of the slope standard deviation, with the standard deviation set at 0.012. After normalization, we get... ;

[0102] Indicates the first The age smoothing modifier for a sample of 10-year-olds, with a reference sample mean of 12-years, is defined as follows: ;

[0103] This indicates that the sample was collected in a total of 5 periods, therefore ;

[0104] Substitute the parameters into the formula to calculate:

[0105] No. The sample at the th to The normalized values ​​of the periodic slope are as follows:

[0106] , , , ;

[0107] The differences are as follows:

[0108] , , ;

[0109] Substitute the weighting factors in sequence , , ;

[0110] ;

[0111] ;

[0112] ;

[0113] ;

[0114] The result of 0.0682 indicates that the waist circumference change trend of this sample is at a low level within a continuous period. The smaller the value, the more stable the waist circumference change. This value is directly used as a quantitative reference indicator in the subsequent screening process and input into the rhythm stability judgment process to determine whether it belongs to an acceptable sample and enter the next stage of calculation process.

[0115] S203: Call the waist circumference fluctuation structure data, and further filter the samples based on the stability of the waist circumference data changes to generate a waist circumference rhythm filter set;

[0116] Using waist circumference fluctuation data, and taking students A and C as examples, further filtering is performed. Specifically, the fluctuation amplitude in the fluctuation data is assessed. Samples with weak fluctuation amplitudes are directly added to the filter set, samples with moderate fluctuation amplitudes require further analysis, and samples with strong fluctuation amplitudes are directly excluded. For example, student A is retained due to their weak fluctuation amplitude. Student C's fluctuation data shows moderate fluctuation amplitude, so additional waist circumference measurement data or growth data is used for further analysis to determine if it meets the criteria. Assuming student C provides additional waist circumference data of 53.5cm from the previous year, the average annual waist circumference change rate is calculated to be (61cm−53.5cm) / 12 months≈0.625cm / month. Comparing this with single-cycle data, it is found that the annual average fluctuation amplitude exceeds the stable range, indicating a long-term fluctuation trend in waist circumference growth; therefore, this student is excluded. After completing the above process, all samples with weak fluctuation amplitudes or those that pass further analysis are retained, constructing the final waist circumference rhythm filter set.

[0117] Please see Figure 4 The specific steps for obtaining training weight adjustment data are as follows:

[0118] S301: Using the waist circumference rhythm filter set, extract the height, weight and muscle mass data of each sample after further filtering, calculate the ratio of weight to height for each sample, and obtain the ratio to muscle mass. By analyzing the distribution of the ratio values ​​in multiple continuous numerical intervals, generate body proportion distribution data.

[0119] Using a waist circumference rhythm screening set, height, weight, and muscle mass data of each selected sample were analyzed. Taking student A as an example, student A's height is 128cm, weight is 27kg, and muscle mass is 8.5kg. The ratio of weight to height was calculated using the formula: weight / height, which is approximately 27kg ÷ 128cm ≈ 0.211kg / cm. Then, the ratio of muscle mass to weight / height was calculated using the formula: muscle mass / (weight / height), which is approximately 8.5kg ÷ 0.21. 1 kg / cm ≈ 40.28 cm. The same method was used to calculate other samples and summarize them. The proportion values ​​were classified into different continuous numerical ranges, such as 0 to 20 cm, 20 to 40 cm, and 40 to 60 cm. The proportion of the sample in each range was counted. The sample distribution in each range was clarified by the proportion of the sample. For example, it was calculated that the 0 to 20 cm range accounted for 5% of the total sample, the 20 to 40 cm range accounted for 25%, and the 40 to 60 cm range accounted for 70%. Then, the distribution proportion statistics of all ranges were completed, and the body size proportion distribution data was generated.

[0120] S302: Call the body proportion distribution data, identify the interval with the highest proportion distribution density and use it as the body proportion benchmark, calculate the difference between the proportion value of each sample and the benchmark, and generate body structure deviation data.

[0121] The system retrieves body proportion distribution data and uses the sample density within each proportion interval as a basis. Specifically, it analyzes the sample density of each interval, compares and identifies the interval with the highest density, and determines this interval as the body proportion benchmark. Taking the data example from paragraph S301, the density of the 0-20cm interval is 5%, the density of the 20-40cm interval is 25%, and the density of the 40-60cm interval is 70%. By directly comparing the density percentages, the 40-60cm interval is identified as the highest density interval and determined as the proportion benchmark interval. Then, the difference between the proportion value of each sample and the median of the benchmark interval is calculated. The specific difference calculation method is to subtract the median of the interval from the sample proportion value. The median of the interval is (40cm+60cm) / 2=50cm. Taking student A as an example, its proportion value is 40.28cm, and the calculated difference is 40.28cm−50cm=−9.72cm, indicating that the sample proportion value is smaller than the median of the benchmark interval. Similarly, the difference values ​​of other samples are calculated and recorded. After summarizing the difference calculation results of all samples, body structure deviation data is generated.

[0122] S303: Based on body structure deviation data, assess the degree of deviation of each sample according to the difference between the sample ratio and the benchmark, adjust the feature weighting coefficients in the training phase, construct a set of weight parameters for training samples, and generate training weight adjustment data.

[0123] Based on body shape deviation data, the magnitude of the difference between each sample's proportion and the benchmark median is analyzed. The absolute value of the difference represents the degree of deviation of the sample. The degree of deviation is defined as follows: 0 to 5 cm is slight deviation, 5 to 10 cm is moderate deviation, and greater than 10 cm is significant deviation. The degree of deviation is evaluated, and the feature weighting coefficients for the training stage are determined based on the degree of deviation of the sample. Taking student A as an example, the absolute value of the difference is 9.72 cm, which falls into the moderate deviation category. Therefore, the feature weighting coefficient for BMI and waist circumference is set to 0.8, indicating that the feature influence is weakened compared to samples with slight deviation (coefficient of 1) during the training process. Similarly, if the absolute value of the difference for other samples is 3 cm, which is a slight deviation, the feature weighting coefficient remains at 1. If the difference is 12 cm, which is a significant deviation, the feature weighting coefficient is further reduced to 0.6. The above evaluation is performed on all samples and the corresponding coefficients are configured. Finally, the feature weighting coefficients of each sample are integrated to construct a set of weight parameters for the training samples and generate training weight adjustment data.

[0124] Please see Figure 5 The specific steps for obtaining the model parameter adjustment results are as follows:

[0125] S401: Extract training weight adjustment data, use samples to train the body fat prediction model, calculate the residual value between the sample prediction result and the actual result, combine the age group label of each sample, group the residual data, and generate age group residual group data.

[0126] The training weight adjustment data is extracted. Based on the samples with labeled feature weight coefficients, body fat percentage is predicted using the height, weight, and waist circumference feature data of each sample. The actual body fat measurement value is used as the real data. The difference between the predicted body fat value and the actual body fat measurement value of each sample is calculated. This difference is defined as the residual. For example, if the feature weight coefficient of a student sample is set to 0.8, its predicted body fat percentage is 22%, while the actual measurement value is 20%. Then its residual is 22% − 20% = 2%. The residual value is associated with the age label to which the sample belongs and grouped. For example, the residual of an 8-year-old sample is 1.5%, and the residual of a 9-year-old sample is 2.3%. The residuals are recorded into the corresponding age groups according to the age label, forming a residual set by age. Finally, the residual data in all age groups are counted to complete the age segmentation processing of the residual values, thereby generating age segment residual grouping data.

[0127] S402: Call the age group residual grouping data, calculate the residual mean within each age group, identify the concentration interval of the overall training sample residuals, analyze the distance between the residual mean of each age group and the overall residual interval, evaluate the deviation range of the residual mean, and generate age group residual deviation data.

[0128] The calculation process involves calling age-group residual grouping data and calculating the mean residual within each age group. A specific example demonstrates the calculation process: In the 8-year-old group, there are 3 samples with residuals of 1.5%, 2.0%, and 1.0%, and the mean is calculated as (1.5% + 2.0% + 1.0%) / 3 ≈ 1.5%. Similarly, in the 9-year-old group, there are 4 samples with residuals of 2.3%, 1.8%, 2.0%, and 2.5%, and the mean is calculated as (2.3% + 1.8% + 2.0% + 2.5%) / 4 = 2.15%. This calculation method is repeated for all age groups to obtain the mean residual within each age group, and then... The first step involves performing an overall analysis of the residual data for all age groups to identify the central distribution interval of the overall sample residuals. Based on the residuals of all samples, the median and standard deviation of the residual distribution are calculated. For example, if the median of the overall residual distribution is 2% and the standard deviation is 0.3%, then the central distribution interval is defined as median ± standard deviation, i.e., 1.7% to 2.3%. After that, the distance between the mean of the residuals for each age group and the central distribution interval of the overall residuals is calculated. For example, if the mean of the residuals for the 8-year-old age group is 1.5%, the distance between it and the minimum value of the central distribution interval of 1.7% is 0.2%. The deviation of each age group is calculated and evaluated accordingly, and the deviation distance analysis of all groups is completed to generate age group residual deviation data.

[0129] S403: Based on the residual deviation data of each age group, according to the distance between the average residual level of each age group and the residual concentration interval of the whole sample, the age groups with abnormal residuals are screened, and the residual response adjustment coefficient of the target age group sample is adjusted to generate the model parameter adjustment results.

[0130] Based on the residual deviation data for each age group, the deviation distance between the mean residual of each age group and the overall residual concentration interval is determined. The specific comparison and judgment process is as follows: a deviation distance between 0 and 0.1% is defined as slight deviation, a distance between 0.1% and 0.3% is defined as moderate deviation, and a distance greater than 0.3% is defined as severe deviation. Taking the 8-year-old age group as an example, its distance is 0.2%, which falls into the moderate deviation category. The distance for the 9-year-old age group is 0.15%, which also falls into the moderate deviation category. Further, taking a certain age group with a distance of 0.35% as an example, it falls into the severe deviation category. Subsequently, the severely deviated age group is identified as the abnormal group and marked. The residual response adjustment coefficient is then adjusted accordingly. For example, for the severely deviated group, the response adjustment coefficient is reduced from the standard coefficient of 1.0 to 0.6, for the moderate deviation group it is adjusted to 0.8, and for the slightly deviated group it remains unchanged at 1.0. The adjustment and configuration of the coefficients for all age groups are completed. The residual response adjustment coefficients for each age group are summarized to form the adjusted training parameter set, generating the model parameter adjustment results.

[0131] Please see Figure 6 The specific steps for obtaining the body fat trend prediction results are as follows:

[0132] S501: Based on the model parameter adjustment results, collect the target user's height, weight, waist circumference and other physical characteristics data for multiple consecutive periods, call the body fat prediction model, output the user's body fat prediction results for each period, and generate body fat percentage sequence data.

[0133] Based on the model parameter adjustment results, continuous periodic vital sign data collection was conducted on the target users. In each continuous detection cycle, the user's height, weight, and waist circumference were recorded. The collection cycle was set to once every 30 days. Assuming the user's height in cycles 1 to 4 were 125.2cm, 125.8cm, 126.1cm, and 126.5cm respectively; weights were 27.0kg, 27.8kg, 28.5kg, and 29.0kg respectively; and waist circumferences were 57.4cm, 57.9cm, 58.5cm, and 59.1cm respectively, based on each... The complete vital signs parameters collected periodically are input into the trained body fat prediction model. The model uses the feature weighting mechanism and age group adjustment coefficient from the previous training phase to automatically output the predicted body fat values ​​for the corresponding period. Assuming the predicted results are 19.6%, 20.2%, 21.0%, and 21.7%, the predicted results for the above consecutive periods are arranged in a time series to form an ordered sequence of body fat percentage values. At the same time, the original vital signs data corresponding to each period are retained to ensure the integrity of the data chain structure, which facilitates subsequent trend analysis and synchronous processing, and finally generates body fat percentage sequence data.

[0134] S502: Based on body fat percentage sequence data, extract the weight change trend of the target user in each period, analyze the direction of change of body fat percentage sequence and weight data, analyze the degree of consistency of change trend, obtain trend consistency index, and generate trend relationship data;

[0135] The specific formula for analyzing the consistency of trends is as follows:

[0136] ;

[0137] Calculate the trend consistency index;

[0138] in, This represents an indicator of trend consistency. Representing the Normalized values ​​of body fat percentage changes over a period of time Representing the Normalized values ​​of weight change over a period of time This indicates the total number of consecutive detection cycles. Indicates the current cycle number. This represents a non-zero minimum constant set to prevent the denominator from being zero.

[0139] formula:

[0140] ;

[0141] Detailed explanation of the formula and its calculation derivation:

[0142] The formula is used to calculate the directional consistency between the trend of body fat percentage change and the trend of weight change. The results are used to evaluate the stability of the body fat trend relationship data and to adjust the prediction.

[0143] Parameter meanings and settings:

[0144] The total number of consecutive testing cycles is set to 4, representing data from 4 consecutive child physical examinations.

[0145] The normalized value of body fat percentage change in the l-th cycle is obtained by subtracting the sample mean from the change in body fat percentage in each cycle and then dividing by the standard deviation. The actual measured body fat percentage values ​​are set to 17.0%, 17.8%, 18.2%, and 18.6%, with changes of 0.8%, 0.4%, and 0.4%, respectively. After normalization, the values ​​are 1.27, -0.22, and -0.22, respectively.

[0146] The normalized value of weight change in the l-th period is obtained by subtracting the sample mean from the weight change in each period and then dividing by the standard deviation. The actual measured weight values ​​are set to 31.4 kg, 32.2 kg, 33.0 kg, and 33.8 kg, with changes of 0.8 kg, 0.8 kg, and 0.8 kg, respectively. After normalization, the value is 1.73.

[0147] The minimum constant set to prevent the denominator from being zero is 0.01.

[0148] Substitute the parameters into the formula to calculate:

[0149] Period 1:

[0150] , ;

[0151] ;

[0152] Cycle 2:

[0153] , ;

[0154] ;

[0155] Period 3:

[0156] , ;

[0157] ;

[0158] 4th cycle:

[0159] , ;

[0160] ;

[0161] Average value calculation:

[0162] ;

[0163] Substitute into the formula:

[0164] ;

[0165] The results show that the trend consistency index is 0.5811, indicating that the changes in body fat percentage and weight show moderate consistency over the four periods. The closer the value is to 0, the more consistent the trend direction is; the closer it is to 1, the greater the difference in the trend direction is. The results can be used to determine the relationship between body fat trends and to provide a basis for subsequent adjustments.

[0166] S503: Call trend relationship data, adjust the slope of the predicted body fat change path based on the relationship between the trend of body fat percentage change and the trend of weight change, and obtain the prediction result of body fat trend.

[0167] By calling the trend relationship data, the slope of the body fat percentage change path is adjusted according to the trend consistency index. If the trend relationship is "high", the original slope trend is maintained. For example, the original slope of the T1-T4 cycle is set to (21.7%-19.6%) / 3≈0.7% / cycle. If the trend consistency is "medium", the slope is reduced to 0.75 times the original value. If it is "low", it is reduced to 0.5 times. In the current case, the trend is "high", so the slope remains unchanged. The predicted endpoint is extended by one cycle. The estimated body fat percentage of the T5 cycle is 21.7%+0.7%=22.4%. The extended predicted value is added to the original body fat percentage sequence to complete the time trend curve and establish the change path in order to continuously track the dynamic changes of body fat and finally obtain the body fat trend prediction result.

[0168] Please see Figure 7 An adaptive prediction system for children's body fat percentage based on trajectory feature collaboration is used to execute the aforementioned adaptive prediction method for children's body fat percentage based on trajectory feature collaboration. The system includes:

[0169] The vital signs screening module acquires children's physical examination datasets, analyzes the height and weight change trajectories of each sample, calculates the difference ratio between height and weight growth for each sample in a continuous period, and filters the samples based on the distribution characteristics of the difference ratio in multiple intervals to generate a rhythm screening sample set.

[0170] The waist circumference stability module calls the rhythm screening sample set, analyzes the waist circumference change data of each sample in multiple consecutive detection cycles after screening, calculates the difference in waist circumference change slope between adjacent cycles, evaluates the stability of waist circumference change of the sample, further screens the sample, and generates the waist circumference rhythm screening set.

[0171] The proportional weight adjustment module uses the waist circumference rhythm screening set to calculate the proportional difference between height, weight and muscle mass for each sample after further screening. Based on the distribution characteristics of the proportional difference in multiple intervals, it calculates the degree of deviation and adjusts the feature weights in the training phase to generate training weight adjustment data.

[0172] The age error correction module extracts training weight adjustment data, trains the body fat prediction model, calculates the mean of the prediction residuals of the samples in each age group, identifies the distance from the overall sample residual concentration interval, filters out age groups with abnormal residuals, adjusts the residual response adjustment coefficient, and generates model parameter adjustment results.

[0173] The trend output module collects continuous vital sign data records of the target user based on the model parameter adjustment results, uses the body fat prediction model to predict the user's body fat percentage, identifies the changing trend of body fat prediction results over multiple periods, compares it with the user's weight data, analyzes the consistency of the changing trend, uses the trend direction relationship to set the slope adjustment range of the body fat prediction change path, and generates body fat trend prediction results.

[0174] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0175] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0176] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0177] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0178] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0179] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0180] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0181] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0182] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0183] If the aforementioned functions are implemented as 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 this invention, or the part that contributes to the prior art, or a part of the 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0184] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An adaptive prediction method for children's body fat percentage based on trajectory feature collaboration, characterized in that, The method includes: S1: Obtain children's physical examination dataset, analyze the height and weight change trajectory of each sample, and generate a rhythm screening sample set; The specific steps for generating the rhythm screening sample set are as follows: S101: Obtain the children's physical examination dataset, extract the height and weight data of each sample in multiple consecutive testing periods, call the height and weight changes of each sample in adjacent periods, calculate the difference ratio between height growth and weight growth, and generate height and weight difference ratio data. S102: Based on the height-weight difference ratio data, by identifying the number of samples in each difference ratio interval, analyze the density difference of the number of samples in each interval, and generate difference ratio distribution density data; S103: Call the difference ratio distribution density data, identify the interval with the highest distribution density according to the concentration of sample distribution density in each interval, perform preliminary screening of samples, and generate a rhythm screening sample set; S2: Call the rhythm screening sample set to generate the waist circumference rhythm screening set; The specific steps for generating the waist circumference rhythm filter set are as follows: S201: Call the rhythm screening sample set, extract the waist circumference data of each sample in multiple consecutive detection cycles, calculate the waist circumference change rate of each sample in multiple cycles according to the recording order between adjacent detection cycles, and generate waist circumference change slope sequence data. S202: Based on the waist circumference change slope sequence data, select the slope difference between adjacent periods of each sample, and evaluate the stability of waist circumference change by analyzing the slope change trend of each sample, thereby generating waist circumference fluctuation structure data. S203: Call the waist circumference fluctuation structure data, and further filter the samples based on the stability of the waist circumference data changes to generate a waist circumference rhythm filter set; S3: Using the waist circumference rhythm filter set, generate training weight adjustment data; The specific steps for generating training weight adjustment data are as follows: S301: Using the waist circumference rhythm screening set, extract the height, weight and muscle mass data of each sample after further screening, calculate the ratio of weight to height of each sample, and obtain the ratio of weight to muscle mass. By analyzing the distribution of the ratio values ​​in multiple continuous numerical intervals, generate body proportion distribution data. S302: Call the body proportion distribution data, identify the interval with the highest proportion distribution density and use it as the body proportion benchmark, calculate the difference between the proportion value of each sample and the benchmark, and generate body structure deviation data; S303: Based on the body structure deviation data, assess the degree of deviation of each sample according to the difference between the sample ratio and the benchmark, adjust the feature weighting coefficients in the training phase, construct a set of weight parameters for the training samples, and generate training weight adjustment data. S4: Extract the training weight adjustment data, train the body fat prediction model, and generate the model parameter adjustment results; The specific steps for generating the model parameter adjustment results are as follows: S401: Extract the training weight adjustment data, use the samples to train the body fat prediction model, calculate the residual value between the sample prediction result and the actual result, and combine the age group label of each sample to group the residual data to generate age group residual group data. S402: Call the age group residual grouping data, calculate the residual mean within each age group, identify the concentration interval of the overall training sample residuals, analyze the distance between the residual mean of each age group and the overall residual interval, evaluate the deviation range of the residual mean, and generate age group residual deviation data. S403: Based on the residual deviation data of the age group, according to the distance between the average residual level of each age group and the overall sample residual concentration interval, screen the age groups with abnormal residuals, and adjust the residual response adjustment coefficient of the target age group sample to generate the model parameter adjustment results. S5: Based on the model parameter adjustment results, collect continuous vital sign data records of the target user, use the body fat prediction model to predict the user's body fat percentage, identify the changing trend of body fat prediction results over multiple periods, compare it with the user's weight data, analyze the consistency of the changing trend, use the trend direction relationship to set the slope adjustment range of the body fat prediction change path, and generate body fat trend prediction results.

2. The adaptive prediction method for children's body fat percentage based on trajectory feature collaboration according to claim 1, characterized in that, The specific formula for evaluating the stability of waist circumference changes in the sample in step S202 is as follows: ； Calculate the stability structural value of the slope difference of waist circumference change; in, For the first The sample at the th Normalized value of the slope of waist circumference change over each period For the first The sample at the th Normalized value of the slope of waist circumference change over each period For the first The sample at the th The lateral detection frequency weighting factor for each cycle, For the first The sample at the th Normalized value of waist circumference change rate over a period of time For the first The age smoothing adjustment factor for each sample For the first The number of consecutive testing cycles contained in a sample. For periodic index variables, For sample index variables, For the first The stability structure value of the slope difference of waist circumference change corresponding to each sample.

3. The adaptive prediction method for children's body fat percentage based on trajectory feature collaboration according to claim 1, characterized in that, The specific steps for obtaining the body fat trend prediction results are as follows: S501: Based on the model parameter adjustment results, collect vital sign data of the target user for multiple consecutive periods, call the body fat prediction model, output the body fat prediction results of the user for each period, and generate body fat percentage sequence data. The vital signs data include height, weight, and waist circumference; S502: Based on the body fat percentage sequence data, extract the weight change trend of the target user in each period, analyze the direction of change of the body fat percentage sequence and the weight data, analyze the degree of consistency of the change trend, obtain the trend consistency index, and generate trend relationship data; The specific formula for the degree of consistency in the analysis of changing trends is as follows: ； Calculate the trend consistency index; in, This represents an indicator of trend consistency. Representing the Normalized values ​​of body fat percentage changes over a period of time Representing the Normalized values ​​of weight change over a period of time This indicates the total number of consecutive detection cycles. Indicates the current cycle number. This represents a non-zero minimum constant set to prevent the denominator from being zero; S503: Call the trend relationship data, adjust the slope of the predicted body fat change path according to the relationship between the body fat percentage change trend and the weight change trend, and obtain the body fat trend prediction result.

4. A child body fat percentage adaptive prediction system based on trajectory feature collaboration, characterized in that, The system is used to implement the adaptive prediction method for children's body fat percentage based on trajectory feature collaboration as described in any one of claims 1-3, and the system comprises: Vital signs screening module: acquire children's physical examination dataset, analyze the height and weight change trajectory of each sample, calculate the difference ratio between height and weight growth of each sample in a continuous period, and screen the samples based on the distribution characteristics of the difference ratio in multiple intervals to generate a rhythm screening sample set. Waist circumference stability module: calls the rhythm screening sample set, analyzes the waist circumference change data of each sample in multiple consecutive detection cycles after screening, calculates the difference in waist circumference change slope between adjacent cycles, evaluates the stability of waist circumference change of the sample, further screens the sample, and generates waist circumference rhythm screening set. Proportional weight adjustment module: Using the waist circumference rhythm screening set, calculate the proportional difference between height, weight and muscle mass of each sample after further screening. Based on the distribution characteristics of the proportional difference in multiple intervals, calculate the degree of deviation and adjust the feature weights in the training stage to generate training weight adjustment data. Age error correction module: Extracts the training weight adjustment data, trains the body fat prediction model, calculates the mean of the prediction residuals of the samples in each age group, identifies the distance from the overall sample residual concentration interval, filters out age groups with abnormal residuals, adjusts the residual response adjustment coefficient, and generates model parameter adjustment results. Trend Output Module: Based on the model parameter adjustment results, collect continuous vital sign data records of the target user, use the body fat prediction model to predict the user's body fat percentage, identify the changing trend of body fat prediction results over multiple periods, compare it with the user's weight data, analyze the consistency of the changing trend, use the trend direction relationship to set the slope adjustment range of the body fat prediction change path, and generate body fat trend prediction results.

Images