Method for constructing a moderate activity intensity assessment model for preschool children

Abstract

The present application provides a method for constructing a moderate activity intensity assessment model for preschool children, comprising: obtaining personalized indicators, three-axis acceleration signals, and one-dimensional vector sum signals of the three-axis acceleration signals for preschool children. Statistical feature indicators of the three-axis acceleration signals and the one-dimensional vector sum signals within a preset time window are extracted, and the multiple preset time windows are non-overlapping time windows. A feature set is constructed based on the extracted statistical feature indicators and personalized indicators. According to the relationship between the feature set and the exercise energy consumption, the exercise energy consumption data is calculated. According to the relationship between the exercise energy consumption data and the first moderate intensity threshold and the second moderate intensity threshold, the average daily moderate-to-strong activity index is determined. By using the exercise energy consumption monitoring model to monitor the exercise energy consumption of preschool children, the average daily moderate-to-strong activity index for daily assessment activities of preschool children can be obtained to improve the accuracy of exercise energy consumption monitoring.

Description

Technical Field

[0001] The present application relates to the medical field, specifically to the field of measuring physiological parameters of children, and more specifically to a method for constructing a moderate activity intensity assessment model for preschool children and related devices. Background Art

[0002] Physical activity (PA) refers to any bodily movement resulting from skeletal muscle contraction that increases energy expenditure. It is a key dimension supporting early childhood growth and development. It can be categorized as light physical activity (LPA), moderate physical activity (MPA), and vigorous physical activity (VPA). Growing evidence shows that lower levels of sedentary behavior (SB) and appropriate physical activity intensity can effectively improve children's physical health. The preschool period (typically encompassing children aged 3 to 6 years) is a critical period for developing good physical activity habits, and these habits influence physical activity behaviors in later developmental stages. Objectively monitoring physical activity is crucial for evaluating the effectiveness of physical activity interventions and quantifying the relationship between physical activity and health outcomes. Energy expenditure, as a fundamental determinant of physical activity levels, is a crucial dimension in promoting early childhood growth and development. Therefore, it is necessary to objectively monitor preschoolers' exercise energy expenditure and quantify their physical activity intensity.

[0003] Typically, the intensity of physical activity can be determined by monitoring the energy expenditure of preschoolers. Energy expenditure can be expressed in terms of absolute energy (EE) or relative energy expenditure (METs). Therefore, the intensity level of physical activity can be determined based on the EE or MET values. Currently, based on the MET values, the energy expenditure range for moderate-intensity activity is generally considered to be 3-6 METs. However, data collection and testing within this range is only applicable to adolescents and adults. Due to growth and developmental factors, the relationship between acceleration signals recorded simultaneously during exercise in preschoolers and METs differs significantly from that in children, adolescents, or adults. Furthermore, no specific energy expenditure thresholds for moderate-intensity activity are available for preschoolers aged 3, 4, 5, or 6 years.

[0004] Currently, the main method for monitoring exercise energy expenditure is to monitor individuals using accelerometer sensors, providing an estimated energy expenditure. These methods are based on accelerometer activity counts (AC). AC is the data obtained by processing the acceleration signal through a specific filter. This results in a significant loss of motion-related information, thus affecting the accuracy of the exercise energy expenditure monitoring method.

[0005] Accelerometer sensor devices can also record the raw acceleration signal data of the test person. Existing evidence shows that compared with the exercise energy consumption estimation model built based on AC, the estimation accuracy of the exercise energy consumption estimation model built based on the raw acceleration signal characteristics is significantly improved. For example, the method of determining the type of exercise based on the acceleration waveform to measure the energy consumption of exercise; or the method of inputting the acceleration signal into the convolutional neural network to extract the acceleration characteristics, and then fusing them with other features to predict energy consumption; or the method of inputting the filtered acceleration signal into the exercise intensity estimation formula to calculate the exercise intensity and then calculate the calories consumed; or the method of constructing an exercise energy consumption estimation model for foreign preschool children based on the raw acceleration signals of foreign preschool children. However, these exercise energy consumption estimation models for preschool children based on raw acceleration signal characteristics are all constructed with foreign children as the test subjects, and their reproducibility and accuracy are still not ideal.

[0006] Secondly, in the research on methods for estimating preschool children's exercise energy expenditure based on raw acceleration signal features, personalized indicators are not integrated or are only integrated, which can lead to inaccurate estimates of preschool children's exercise energy expenditure values ​​in the constructed models. In addition, during the model construction process, the contribution of each feature to the degree of influence of the trained model's predictive performance is generally not analyzed. As a result, although relevant models are constructed, because the key features with the main influence are not identified, the constructed relevant models may not be constructed based on all key features, resulting in room for improvement in the model's estimation accuracy.

[0007] Finally, previous methods typically output energy expenditure values, but have not proposed models for estimating moderate activity intensity in preschoolers. This model, which outputs indicators quantifying physical activity intensity—specifically, an average daily moderate-to-vigorous activity index—is more suitable for assessing whether preschoolers meet the required time for moderate activity intensity. Summary of the Invention

[0008] In view of this, the present application is committed to providing a method and device for constructing a moderate activity intensity assessment model for preschool children, so as to use the model to monitor the exercise energy consumption of preschool children and determine whether the daily physical activity intensity and corresponding activity duration of preschool children meet the standards.

[0009] In a first aspect, the present application provides a method for constructing a moderate activity intensity assessment model for preschool children, the method comprising:

[0010] Acquiring personalized indicators of preschool children, three-axis acceleration signals, and a one-dimensional vector sum signal of the three-axis acceleration signals;

[0011] Extracting statistical characteristic indicators of the three-axis acceleration signal and the one-dimensional vector sum signal within a preset time window, wherein the multiple preset time windows are non-overlapping time windows;

[0012] Constructing a feature set based on the extracted statistical feature indicators and the personalized indicators;

[0013] Calculating and obtaining exercise energy consumption data based on the relationship between the feature set and exercise energy consumption;

[0014] The average daily moderate-intensity activity index is determined based on the relationship between exercise energy consumption data and the first moderate-intensity threshold and the second moderate-intensity threshold.

[0015] The average daily moderate-intensity activity index for preschool children represents statistical information on average daily moderate-intensity activity. This statistical information may include the total duration of average daily moderate-intensity activity, and may also include the integral of the curve of metabolic equivalents (METs) values ​​changing over time in the corresponding time period under certain set conditions (i.e., the sum of the areas under the METs value curve in a certain time period).

[0016] In a possible implementation, the daily average moderate-strong activity index , t is the length of the preset time window, N is the total number of preset time windows that meet the requirements; the initial value of N is 0. When the exercise energy consumption data corresponding to the preset time window is greater than or equal to the first moderate intensity threshold and less than or equal to the second moderate intensity threshold, N is updated by adding 1.

[0017] In one possible implementation, the three-axis acceleration signal includes an x-axis acceleration signal, a y-axis acceleration signal, and a z-axis acceleration signal. The statistical characteristic indicators of the x-axis acceleration signal include at least the maximum value of multiple x-axis acceleration signals within the preset time window; the statistical characteristic indicators of the y-axis acceleration signal include at least the peak value of multiple y-axis acceleration signals within the preset time window; the statistical characteristic indicators of the z-axis acceleration signal include at least the minimum value of multiple z-axis acceleration signals; and the statistical characteristic indicators of the one-dimensional vector sum signal include at least the average value and variance of multiple one-dimensional vector sum signals within the preset time window.

[0018] In one possible implementation, the relationship between the feature set and exercise energy consumption is established through a trained machine learning model.

[0019] In one possible implementation, the trained machine learning model includes an extreme random tree regression model.

[0020] In a possible implementation, the exercise energy consumption data includes at least one of an energy consumption mean and a metabolic equivalent mean.

[0021] In a possible implementation, the personalized indicator includes at least one of gender, age, height, weight, and body mass index.

[0022] In one possible implementation, the first medium intensity threshold is 3.2 Mets, and the second medium intensity threshold is 5.3 Mets.

[0023] In one possible implementation, the first medium intensity threshold corresponding to a 3-year-old preschool child is 3.2 Mets, and the second medium intensity threshold corresponding to a 3-year-old preschool child is 4.4 Mets.

[0024] In one possible implementation, the first medium intensity threshold corresponding to a 4-year-old preschool child is 3.4 Mets, and the second medium intensity threshold corresponding to a 4-year-old preschool child is 4.6 Mets.

[0025] In one possible implementation, the first medium intensity threshold corresponding to a 5-year-old preschool child is 3.7 Mets, and the second medium intensity threshold corresponding to a 5-year-old preschool child is 4.8 Mets.

[0026] In one possible implementation, the first medium intensity threshold corresponding to a 6-year-old preschool child is 3.8 Mets, and the second medium intensity threshold corresponding to a 6-year-old preschool child is 5.1 Mets.

[0027] In one possible implementation, the age of a preschool child is calculated based on the age corresponding to the actual date of birth.

[0028] In a second aspect, the present application provides a device for constructing a moderate activity intensity assessment model for preschool children, the device comprising:

[0029] A data acquisition unit, configured to acquire personalized indicators of preschool children, a three-axis acceleration signal, and a one-dimensional vector sum signal of the three-axis acceleration signal;

[0030] a feature extraction unit, configured to extract statistical feature indicators of the three-axis acceleration signal and the one-dimensional vector sum signal within a preset time window, wherein the multiple preset time windows are non-overlapping time windows;

[0031] A feature construction unit, configured to construct a feature set based on the extracted statistical feature indicators and the personalized indicators;

[0032] A prediction unit, configured to calculate and obtain exercise energy consumption data based on a relationship between the feature set and exercise energy consumption;

[0033] The judgment unit is used to determine the daily average moderate-intensity activity index based on the relationship between the exercise energy consumption data and the first moderate-intensity threshold and the second moderate-intensity threshold.

[0034] In a third aspect, the present application provides an electronic device, the device comprising: a memory and a processor;

[0035] The memory is used to store relevant program codes;

[0036] The processor is used to call the program code to execute the method for constructing a moderate activity intensity assessment model for preschool children as described in any one of the implementations of the first aspect above.

[0037] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program for executing the method for constructing a moderate activity intensity assessment model for preschool children as described in any one of the implementations of the first aspect above.

[0038] In a fifth aspect, the present application provides a computer program product, which includes a computer program / instructions, and when the computer program / instructions are executed by a processor, it implements the method for constructing a moderate activity intensity assessment model for preschool children as described in any one of the implementation methods of the first aspect above.

[0039] The invention of the present invention is:

[0040] 1. Provides a moderate activity intensity threshold range suitable for Chinese preschool children; further provides moderate activity intensity threshold ranges suitable for 3, 4, 5, and 6 years old;

[0041] 2. Integrates personalized indicators that are specific to the development of Chinese preschool children;

[0042] 3. For the constructed model, sort the important features and find the key features that have the main influence.

[0043] In the above-mentioned implementation of the present application, in order to construct a moderate activity intensity assessment model for preschool children, the personalized indicators, three-axis acceleration signals, one-dimensional vector sum signals of the three-axis acceleration signals, and corresponding motion energy consumption values ​​of preschool children are first obtained. Statistical feature indicators of the three-axis acceleration signals and the one-dimensional vector sum signals within a preset time window are extracted, wherein the multiple preset time windows are non-overlapping time windows. A feature set is constructed based on the extracted statistical feature indicators and personalized indicators. According to the relationship between the feature set and the motion energy consumption, the motion energy consumption data is calculated. According to the relationship between the motion energy consumption data and the first moderate intensity threshold and the second moderate intensity threshold, the average daily moderate-to-strong activity index is determined.

[0044] The personalized indicators for preschoolers used in the model construction reflect the developmental characteristics of certain Chinese preschoolers. These seemingly routine indicators actually have a certain impact on the accuracy of the preschooler exercise assessment model. Among the personalized indicators, height has the greatest impact on the model's prediction accuracy. Furthermore, during the model construction process, the contribution of each feature to the predicted performance of the trained model is analyzed to identify the key features with the greatest impact. This ensures that the relevant model is constructed using all key features. These two factors also help improve the accuracy of exercise energy consumption monitoring.

[0045] Furthermore, the current globally accepted range is based on statistical analysis of international adolescents, while our threshold range is derived from data collected from Chinese preschoolers. Therefore, the threshold range we derived is consistent with the developmental characteristics of Chinese preschoolers. Furthermore, our proposed moderate activity intensity threshold range is more precise, helping to improve the accuracy of assessments of moderate-intensity activity in preschoolers. Furthermore, this model construction method also provides assessment tools and data support for the precise management of preschoolers' exercise, which has strong practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments provided in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0047] Figure 1 A flowchart of a method for constructing a moderate activity intensity assessment model for preschool children provided in an embodiment of the present application.

[0048] Figure 2 This is a flowchart of a method for monitoring exercise energy consumption of preschool children provided in an embodiment of the present application.

[0049] Figure 3A schematic diagram of a device for constructing a moderate activity intensity assessment model for preschool children provided in an embodiment of the present application.

[0050] Figure 4 This is a schematic diagram of an exercise energy consumption monitoring device for preschool children provided in an embodiment of the present application.

[0051] Figure 5 A schematic diagram of an electronic device provided in an embodiment of the present application.

[0052] Figure 6 The box plots are the statistical ranges of the average Mets values ​​obtained by averaging the measured Mets values ​​recorded synchronously for samples aged 4 to 6 years old while performing moderate intensity activities within a time window of a certain length, where (a) is the statistical range of the average Mets values ​​recorded synchronously for samples aged 4 years old while performing moderate intensity activities, (b) is the statistical range of the average Mets values ​​recorded synchronously for samples aged 5 years old while performing moderate intensity activities, and (c) is the statistical range of the average Mets values ​​recorded synchronously for samples aged 6 years old while performing moderate intensity activities. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. The described embodiments are only exemplary implementation methods of the present application and are not all implementation methods. Those skilled in the art can combine the embodiments of the present application to obtain other embodiments without creative work, and these embodiments are also within the scope of protection of the present application.

[0054] Because the preschool period (ages 3 to 6) is a critical period for developing good physical activity habits, and physical activity habits during this period may influence physical activity behaviors in later developmental stages, it is necessary to monitor the intensity of preschool children's physical activity to promote their early growth and development.

[0055] Typically, the intensity of physical activity can be determined by monitoring the energy expenditure of preschoolers. Currently, the main method for monitoring individuals using accelerometers is to estimate their energy expenditure. However, this method is primarily applicable to adolescents and adults. The physical activity habits of preschoolers differ significantly from those of adolescents (typically aged 11 to 17) and adults. Therefore, the accuracy of energy consumption monitoring for preschoolers using existing energy consumption monitoring models is low.

[0056] In addition, current methods for monitoring energy consumption during exercise are all based on accelerometer activity counts (AC). AC is the data obtained by processing the acceleration signal through a specific filter, which will lose a lot of exercise-related information, thus affecting the accuracy of the exercise energy consumption monitoring method.

[0057] There are currently some studies that have constructed models for estimating exercise energy expenditure in preschool children based on raw acceleration signal characteristics. However, these models were all constructed using children from abroad, and their reproducibility and accuracy are still not ideal.

[0058] Based on this, an embodiment of the present application provides a method for constructing a moderate activity intensity assessment model for preschool children to improve the accuracy of exercise energy consumption monitoring. In specific implementation, personalized indicators, three-axis acceleration signals, and one-dimensional vector sum signals of the three-axis acceleration signals are obtained for preschool children. Statistical feature indicators of the three-axis acceleration signals and the one-dimensional vector sum signals within a preset time window are extracted, wherein the multiple preset time windows are non-overlapping time windows. A feature set is constructed based on the extracted statistical feature indicators and personalized indicators. According to the relationship between the feature set and exercise energy consumption, exercise energy consumption data is calculated. According to the relationship between the exercise energy consumption data and the first moderate intensity threshold and the second moderate intensity threshold, the average daily moderate-strong activity index is determined. Subsequently, the exercise energy consumption monitoring model can be used to monitor the exercise energy consumption of preschool children. Since the exercise energy consumption monitoring model is constructed based on the exercise energy consumption data of preschool children, the exercise energy consumption monitoring model is used to monitor the exercise energy consumption of preschool children, and the average daily moderate-strong activity index for daily assessment activities of preschool children can be obtained, thereby improving the accuracy of exercise energy consumption monitoring.

[0059] In order to facilitate understanding of the technical methods provided by the embodiments of the present application, a detailed introduction will be given below in conjunction with the drawings in the embodiments.

[0060] See also Figure 1 As shown, Figure 1 A flowchart of a method for constructing a moderate activity intensity assessment model for preschool children provided in an embodiment of the present application.

[0061] Optionally, the method may be executed by a processing device, which may be an electronic device or other device, and is not limited in this embodiment of the present application.

[0062] The method may include the following steps:

[0063] S101: Obtain personalized indicators of preschool children, three-axis acceleration signals, and one-dimensional vector sum signals of the three-axis acceleration signals.

[0064] The triaxial acceleration signal can be obtained by a preschool child wearing an accelerometer for at least one day. The triaxial acceleration signal can include an x-axis acceleration signal, a y-axis acceleration signal, and a z-axis acceleration signal. The personalized indicator can include at least one of gender, age, height, weight, and body mass index (BMI).

[0065] Alternatively, the one-dimensional vector sum signal of the three-axis acceleration signal can be obtained by calculating the one-dimensional vector sum of the x-axis acceleration signal, the y-axis acceleration signal, and the z-axis acceleration signal. For example, the square sum of the x-axis acceleration signal, the y-axis acceleration signal, and the z-axis acceleration signal can be calculated, and then the arithmetic square root of the square sum can be calculated as the one-dimensional vector sum signal.

[0066] S102: Extracting statistical characteristic indicators of the three-axis acceleration signal and the one-dimensional vector sum signal within a preset time window.

[0067] That is, embodiments of the present application can use preset time windows to divide the three-axis acceleration signal and the one-dimensional vector sum signal, thereby obtaining three-axis acceleration signals and one-dimensional vector sum signals corresponding to multiple preset time windows. The multiple preset time windows are non-overlapping time windows. That is, the times corresponding to the multiple preset time windows do not overlap.

[0068] In one possible implementation, for each preset time window, there may be multiple three-axis acceleration signals and multiple one-dimensional vector sum signals. Based on this, the statistical characteristic indicators of the extracted x-axis acceleration signal include at least the maximum value of the multiple x-axis acceleration signals within the preset time window. The statistical characteristic indicators of the y-axis acceleration signal include at least the peak-to-peak value of the multiple y-axis acceleration signals within the preset time window. The statistical characteristic indicators of the z-axis acceleration signal include at least the minimum value of the multiple z-axis acceleration signals. The statistical characteristic indicators of the one-dimensional vector sum signal include at least the mean and variance of the multiple one-dimensional vector sum signals within the preset time window.

[0069] S103: Construct a feature set based on the extracted statistical feature indicators and personalized indicators.

[0070] After determining the statistical feature indicators, the feature set can be constructed using the statistical feature indicators and personalized indicators for preschool children.

[0071] S104: Calculate and obtain exercise energy consumption data based on the relationship between the feature set and exercise energy consumption.

[0072] In one possible implementation, the relationship between the feature set and exercise energy consumption is established using a trained machine learning model. Optionally, the trained machine learning model may include an extreme randomized tree regression model. The training process of the machine learning model can be found in subsequent embodiments and will not be described in detail here.

[0073] In one possible implementation, exercise energy consumption data may include energy expenditure (EE) or metabolic equivalents (METs). Different exercise energy consumption data may also include different indicators in the feature set corresponding to the exercise energy consumption data. This process will be described in subsequent embodiments.

[0074] S105: Determine the daily average moderate-intensity activity index based on the relationship between the exercise energy consumption data and the first moderate-intensity threshold and the second moderate-intensity threshold.

[0075] The first medium intensity threshold is smaller than the second medium intensity threshold.

[0076] In one possible implementation, the level of physical activity intensity can be determined based on metabolic equivalents (METs). Moderate activity intensity is more conducive to growth and development in preschoolers, so determining a preschooler's average daily moderate-to-vigorous activity index can be used to monitor their exercise energy expenditure.

[0077] The METs range corresponding to moderate activity intensity can be determined by a first moderate intensity threshold and a second moderate intensity threshold. Alternatively, the first and second moderate intensity thresholds can be determined in advance through statistical experiments. Therefore, after calculating the exercise energy consumption data for preschool children, the average daily moderate-to-vigorous activity index can be determined based on the relationship between the exercise energy consumption data and the first and second moderate intensity thresholds.

[0078] In one possible implementation, the first moderate intensity threshold may be 3.2 Mets, and the second moderate intensity threshold may be 5.3 Mets. That is, the range corresponding to the moderate activity intensity of preschool children is [3.2, 5.3]. Among them, the threshold range corresponding to the moderate activity intensity of preschool children aged 4, 5 and 6 is obtained by performing box plot statistics on the corresponding data. Among them, the box plot of the value range statistics of the corresponding measured average mets value when the samples aged 4 to 6 are performing moderate activity intensity activities is as follows Figure 6 Therefore, based on the numerical pattern corresponding to the moderate activity intensity threshold for preschool children aged 4-6, it is estimated that the moderate activity intensity threshold for 3-year-old preschool children is [3.2, 4.4] Mets;

[0079] Among them, the range of moderate activity intensity corresponding to preschool children of different ages is also different, as shown in the following description:

[0080] The first moderate intensity threshold for 3-year-old preschool children is 3.2 Mets, and the second moderate intensity threshold for 3-year-old preschool children is 4.4 Mets. That is, the range of moderate activity intensity for 3-year-old preschool children is [3.2, 4.4]. The overall standard deviation is plus or minus 0.2, and the standard deviation range is [0, 0.3]. In specific implementation, the value of the standard deviation depends on factors such as the height range of the measurement sample.

[0081] The first moderate intensity threshold for 4-year-old preschool children is 3.4 Mets, and the second moderate intensity threshold for 4-year-old preschool children is 4.6 Mets. That is, the range of moderate activity intensity for 4-year-old preschool children is [3.4, 4.6]. The overall standard deviation is plus or minus 0.2, and the standard deviation range is [0, 0.3]. In specific implementation, the value of the standard deviation depends on factors such as the height range of the measurement sample.

[0082] The first moderate intensity threshold corresponding to 5-year-old preschool children is 3.7 Mets, and the second moderate intensity threshold corresponding to 5-year-old preschool children is 4.8 Mets. That is, the range of moderate activity intensity corresponding to 5-year-old preschool children is [3.7, 4.8]. The overall standard deviation is plus or minus 0.2, and the standard deviation range is [0, 0.4]. In specific implementation, the value of the standard deviation depends on factors such as the height range of the measurement sample.

[0083] The first moderate-intensity threshold for a 6-year-old preschooler is 3.8 Mets, and the second moderate-intensity threshold for a 6-year-old preschooler is 5.1 Mets. This means the range of moderate activity intensity for a 6-year-old preschooler is [3.8, 5.1]. The overall standard deviation is plus or minus 0.2, and the standard deviation ranges from [0, 0.6]. The specific implementation of this standard deviation depends on factors such as the height range of the measured sample. It should be noted that the age of preschoolers is calculated based on their actual birth date.

[0084] In specific implementation, the daily average moderate-strong activity index can be determined by the following methods . Wherein, t is the length of the preset time window, and N is the total number of preset time windows that meet the requirements. Set the initial value of N to 0. For the exercise energy consumption data determined for each preset time window, when the exercise energy consumption data corresponding to the preset time window is greater than or equal to the first moderate intensity threshold and less than or equal to the second moderate intensity threshold, it indicates that the current preset time window meets the requirements, and N is updated by adding 1. Perform the above operation on multiple preset time windows, and the total number N of all preset time windows that meet the requirements can be finally determined. Multiply the length t of each preset time window by the total number N to obtain the average daily moderate-intensity activity index of preschool children.

[0085] Through the above steps, the execution process of the moderate activity intensity assessment model is determined, that is, the moderate activity intensity assessment model for preschool children is constructed.

[0086] In order to more clearly understand the technical solutions in the above embodiments, the relationship between the sample features and exercise energy consumption in step S104, that is, the trained machine learning model, is introduced in detail below.

[0087] To obtain a trained machine learning model, it is first necessary to obtain sample data for training. The sample data is then used to train the machine learning algorithm to obtain a trained machine learning model. Since the moderate activity intensity assessment model is designed to accurately monitor the exercise energy consumption of preschool children and calculate exercise energy consumption data, it is necessary to collect relevant data on preschool children. This method mainly includes the following steps A1-A4:

[0088] A1: Obtain personalized indicators, triaxial acceleration signals, and motion energy consumption label values ​​corresponding to the triaxial acceleration signals for preschool children.

[0089] Preschool children primarily include children between the ages of 3 and 6. The specific age range for data collection can be determined based on actual needs. The collected samples can be made to include as many age groups as possible, thereby increasing sample diversity and making the trained machine learning model more accurate.

[0090] Personalized indicators for preschool children include: gender, age, height, weight, and body mass index. The three-axis acceleration signal can be used to characterize the movement state of preschool children. For example, preschool children can wear an accelerometer to measure the three-axis acceleration signal of the preschool children in different movement states. In addition, the preschool children wear a gas metabolism analyzer. While using the accelerometer to measure the three-axis acceleration signal, the gas metabolism analyzer is also used to measure the exercise energy consumption in different movement states as the exercise energy consumption label value, that is, the sample true value, so that each three-axis acceleration signal corresponds to an exercise energy consumption label value. Among them, the gas metabolism analyzer is measured by exhalation analysis, and the exercise energy consumption label value is calculated by measuring oxygen intake, carbon dioxide output, etc.

[0091] Optionally, the exercise energy consumption label value may include EE or METs. When the acquired exercise energy consumption label value includes energy consumption, the trained machine learning model can output the energy consumption of preschool children. When the acquired exercise energy consumption label value includes metabolic equivalents, the trained machine learning model can monitor and output the metabolic equivalents of preschool children. In other words, different samples (exercise energy consumption label values) can be used to train different machine learning models to monitor different exercise energy consumption data.

[0092] Preschool children's movement states can include a variety of different states to increase sample diversity. For example, these states can include sitting still, walking slowly, walking fast, running, jumping, and other movement states. By collecting triaxial acceleration signals and movement energy consumption labels of preschool children in different movement states, these are used as independent variables and target variables for training machine learning models.

[0093] A2: Determine sample characteristics based on personalized indicators and three-axis acceleration signals.

[0094] After obtaining personalized indicators and three-axis acceleration signals of preschool children, sample characteristics for training can be determined.

[0095] Because the data collected about preschool children may contain errors or omissions, in order to simplify the model's data processing, one possible implementation method is to preprocess the collected data to obtain preprocessed personalized indicators and preprocessed triaxial acceleration signals. Sample features are then determined based on these preprocessed personalized indicators and preprocessed triaxial acceleration signals.

[0096] In practice, missing values ​​in the data can be interpolated. Specifically, when missing values ​​are present in the personalized metrics or triaxial acceleration signals of preschool children, these missing values ​​can be interpolated to avoid impacting subsequent data processing. For example, missing values ​​in categorical data (e.g., male or female) are interpolated with "NaN" values, while missing values ​​in numerical data (e.g., age) are interpolated with the mean.

[0097] Since most models cannot directly process categorical data, one-hot encoding is required to convert categorical data into numerical data. One-hot encoding, also known as single-bit encoding, involves using an N-bit state register to encode N states. Each state has a separate register bit, taking a value of 0 or 1, and only one bit is active at any given time. For example, to encode the categorical data "boy" and "girl," since there are only two types in this category, N=2, and "01" and "10" can be used to represent "boy" and "girl."

[0098] For continuous data, normalization can be performed to simplify data processing. Alternatively, the normalized continuous data can be converted into data with a near-normal distribution to improve data processing efficiency.

[0099] In a possible implementation, the personalized index and the three-axis acceleration signal can be fused as sample features.

[0100] In one possible implementation, the three-axis acceleration signal includes an x-axis acceleration signal, a y-axis acceleration signal, and a z-axis acceleration signal. In order to improve the diversity and richness of the samples and enhance the ability of the training model, a one-dimensional vector sum signal can be determined based on the x-axis acceleration signal, the y-axis acceleration signal, and the z-axis acceleration signal. That is, the square sum of the x-axis acceleration signal, the y-axis acceleration signal, and the z-axis acceleration signal is calculated, and then the arithmetic square root of the square sum is calculated as the one-dimensional vector sum signal. After obtaining the one-dimensional vector sum signal, since the three-axis acceleration signal is a three-axis acceleration signal collected from preschool children over a period of time, the statistical feature index of the x-axis acceleration signal, the statistical feature index of the y-axis acceleration signal, the statistical feature index of the z-axis acceleration signal, and the statistical feature index of the one-dimensional vector sum signal can be calculated based on multiple x-axis acceleration signals, multiple y-axis acceleration signals, multiple z-axis acceleration signals, and multiple one-dimensional vector sum signals over a period of time.

[0101] The statistical characteristic indicators may represent mathematical statistical characteristics, for example, they may include maximum value, minimum value, mean value, standard deviation, variance, percentile, coefficient of variation, etc.

[0102] In one possible implementation, the statistical characteristic indicators of the x-axis acceleration signal include at least the maximum values ​​of multiple x-axis acceleration signals. The statistical characteristic indicators of the y-axis acceleration signal include at least the peak-to-peak values ​​of multiple y-axis acceleration signals, where the peak-to-peak value represents the difference between the maximum value and the minimum value. The statistical characteristic indicators of the z-axis acceleration signal include at least the minimum values ​​of multiple z-axis acceleration signals. The statistical characteristic indicators of the one-dimensional vector sum signal include at least the mean and variance of the multiple one-dimensional vector sum signals.

[0103] Then, based on the statistical feature indicators and the personalized indicators, the sample characteristics are determined. For example, the statistical feature indicators and the personalized indicators can be integrated to determine the sample characteristics.

[0104] It should be noted that the specific types and numbers of statistical characteristic indicators are not limited in the embodiments of the present application and can be set according to actual needs. In one possible application scenario, for the x-axis acceleration signal, y-axis acceleration signal, z-axis acceleration signal, one-dimensional vector and signal, a total of 91 statistical characteristic indicators including maximum value, minimum value, mean, standard deviation, variance, percentile, coefficient of variation, etc. can be determined. Then, combined with five personalized indicators of gender, age, height, weight, and body mass index, the statistical characteristic indicators and personalized indicators are integrated to obtain 96 indicators to determine the sample characteristics.

[0105] While including more indicators in sample features can increase sample diversity, it also increases the complexity of data processing and the difficulty of training machine learning models, potentially leading to overfitting and affecting the accuracy of trained machine learning models. Based on this, in one possible implementation, a feature selection algorithm can be used to select a preset number of indicators from the statistical feature indicators and personalized indicators determined above, and then determine sample features based on the selected preset number of indicators. In other words, some of the statistical feature indicators and personalized indicators can be selected as sample features, which can both ensure sample diversity and appropriately reduce the amount of sample data (the number of indicators), reducing the complexity of data processing and speeding up model training.

[0106] For example, the feature selection algorithm can be the minimum redundancy maximum correlation method, which primarily selects features from the dataset that are highly correlated with the target variable but have low redundancy with each other. Specifically, from statistical feature indicators and personalized indicators, indicators that are highly correlated with exercise energy consumption are selected, while also having low redundancy with each other.

[0107] Optionally, the preset number of selected indicators can be determined through multiple experiments. That is, multiple different numbers can be predetermined, and a sample feature can be obtained after selecting multiple indicators for each number. After the machine learning model is trained using the sample features, it is determined which number of sample features trained to obtain a machine learning model with higher accuracy can be determined as the preset number. Subsequently, when the machine learning model is applied to monitor preschool children to determine exercise energy consumption data, a preset number of indicators can be selected to determine the sample features, so that the exercise energy consumption data predicted by the machine learning model is more accurate.

[0108] In one possible implementation, the collected three-axis acceleration signals of preschool children may correspond to motion states in multiple time periods, and the duration of each time period may be different. To unify the format of sample features and facilitate sample feature processing, the acquired three-axis acceleration signals can be divided according to preset time windows to obtain first acceleration signals corresponding to multiple preset time windows. In other words, each preset time window corresponds to the first acceleration signal obtained by division. For each preset time window, the first sample feature corresponding to the preset time window can be determined based on the first acceleration signal of the preset time window and the personalized indicator. Then, based on the first sample feature corresponding to each preset time window, a machine learning model is trained.

[0109] Among them, the process of determining the first sample feature can refer to the implementation process of determining the sample feature in the above embodiment. For any preset time window, based on multiple x-axis acceleration signals, multiple y-axis acceleration signals, multiple z-axis acceleration signals and corresponding multiple one-dimensional vector sum signals within the preset time window, the statistical feature indicators of the x-axis acceleration signal, the statistical feature indicators of the y-axis acceleration signal, the statistical feature indicators of the z-axis acceleration signal and the statistical feature indicators of the one-dimensional vector sum signal are calculated respectively. Then, based on the statistical feature indicators and personalized indicators corresponding to the preset time window, the first sample feature corresponding to the preset time window is determined. For example, the various statistical feature indicators and personalized indicators can be fused to determine the first sample feature.

[0110] In one possible implementation, a feature selection algorithm can be used to select a preset number of indicators from the statistical feature indicators and personalized indicators determined above, and then the first sample feature can be determined based on the selected preset number of indicators. In other words, some of the statistical feature indicators and personalized indicators can be selected as the first sample feature.

[0111] Based on this, after determining the first sample feature corresponding to each preset time window, since the multiple triaxial acceleration signals within the preset time window correspond to multiple motion energy consumption label values, the sample true value corresponding to the first sample feature can be determined to be the average of the multiple motion energy consumption label values ​​within the preset time window. In other words, for each preset time window, the multiple motion energy consumption label values ​​within the preset time window are obtained, and the average motion energy consumption label value of the multiple motion energy consumption label values ​​is calculated as the sample true value corresponding to the first sample feature.

[0112] A3: Input the sample features and the corresponding exercise energy consumption label values ​​into the machine learning algorithm to obtain the predicted exercise energy consumption value.

[0113] After determining the sample features for training, the sample features can be input into a machine learning algorithm, which then outputs a predicted value of exercise energy consumption.

[0114] In a possible implementation, the machine learning algorithm may include an extreme random tree regression algorithm, a random forest regression algorithm, a light gradient boosting machine (LGBM) algorithm, a CatBoost algorithm, or a linear regression algorithm.

[0115] A4: Train the machine learning algorithm based on the predicted energy consumption value and the label value until the training cutoff condition is met to obtain a trained machine learning model.

[0116] Once the predicted energy consumption value is obtained, since the three-axis acceleration signal in the sample features corresponds to the energy consumption label value (i.e., the true value of the sample), and the predicted energy consumption value represents the predicted value of the machine learning algorithm, the machine learning algorithm can be trained based on the predicted energy consumption value and the energy consumption label value. The training is iterative until the training cutoff condition is met. The goal of the training is to reduce the error between the predicted energy consumption value and the energy consumption label value.

[0117] In a specific implementation, an estimation error can be determined based on the predicted value of exercise energy consumption and the label value of exercise energy consumption, and the estimation error is used to represent the error between the predicted value of exercise energy consumption and the label value of exercise energy consumption. When the estimation error is larger, it indicates that the error between the predicted value of exercise energy consumption and the label value of exercise energy consumption is also larger, and the accuracy of the machine learning algorithm is lower. When the estimation error is greater than or equal to the preset value, it indicates that the error between the predicted value of exercise energy consumption and the label value of exercise energy consumption is large and does not meet the requirements. It is necessary to readjust the parameters of the machine learning algorithm and re-execute steps A3 and A4, that is, re-execute the training process of inputting the sample features into the machine learning algorithm after adjusting the parameters and the subsequent determination of the estimation error until the training cutoff condition is met.

[0118] In one possible implementation, the training cutoff condition may be that the estimated error is less than a preset value, or the number of iterative training reaches a preset number (the number of times the parameters of the machine learning algorithm are adjusted reaches a preset number). At this time, the training process can be stopped to obtain a trained machine learning model.

[0119] In one possible implementation, the k-fold cross-validation method can be used to train a machine learning model. The main principle of the k-fold cross-validation method is to randomly divide the sample data into k parts, of which k-1 parts are used as training sets, and the remaining 1 part is used as a test set. During training and testing, the training set and the test set corresponding to the training set are selected in turn. For example, the ten-fold cross-validation method is to divide the sample data into 10 sub-samples, a single sub-sample is used as the test set for verifying the test model, and the remaining 9 sub-samples are used for training. The cross-validation is repeated 10 times, and each sub-sample is verified once as a test set. In other words, the sample features can be divided into a training set and a test set, and the training set is used to train the machine learning model, and the test set is used to test the machine learning model. The results obtained from the 10 tests are then averaged as the final test result.

[0120] When testing the machine learning model, the test set is input into the machine learning model to obtain the exercise energy consumption test value. Based on the exercise energy consumption test value and the exercise energy consumption label value corresponding to the test set, the test error is determined. Optionally, the root mean squared error (RMSE) of the exercise energy consumption test value and the exercise energy consumption label value can be calculated as the test error. For example, when using the ten-fold cross-validation method for training, after the exercise energy consumption test value is output in each test process, the difference between the exercise energy consumption test value and the exercise energy consumption label value can be calculated. Then the mean of the sum of squares of the ten differences is calculated, and the arithmetic square root of the mean of the sum of squares is calculated as the test error. The ten-fold cross-validation method can obtain 10 test errors, and the average of the 10 test errors is calculated as the final test result of the machine learning model.

[0121] According to the above embodiment, in order to enable the trained machine learning model to more accurately monitor the exercise energy consumption of preschool children, multiple machine learning algorithms can be used to train the machine learning model. For example, the multiple machine learning algorithms may include an extreme random tree regression algorithm, a random forest regression algorithm, LGBM, a CatBoost algorithm, or a linear regression algorithm. By testing the accuracy of multiple machine learning models, the most accurate machine learning model is selected.

[0122] In a specific implementation, the sample features and the exercise energy consumption label values ​​corresponding to the sample features can be fused to obtain a data set, and the data set can be divided into a training set and a test set. The training set includes multiple sample features and the exercise energy consumption label values ​​corresponding to each sample feature, and the test set also includes multiple sample features and the exercise energy consumption label values ​​corresponding to each sample feature. For each machine learning algorithm, the training set is used to train the machine learning algorithm to obtain a trained machine learning model. The specific training process can be found in the above embodiment and will not be repeated here. After obtaining the trained machine learning model, the test set is input into the machine learning model to obtain the exercise energy consumption test value. Based on the exercise energy consumption test value and the corresponding exercise energy consumption label value in the test set, the test error is determined. Since each machine learning model corresponds to a test error, as the accuracy test result of the machine learning model, the minimum test error can be determined based on the test errors corresponding to multiple machine learning models, and the machine learning model corresponding to the minimum test error is determined as the preferred machine learning model. That is, the machine learning model corresponding to the minimum test error is used as the machine learning model for subsequent monitoring of preschool children.

[0123] Among them, the method for determining the test error can be to calculate the root mean square error between the exercise energy consumption test value and the exercise energy consumption label value, or to calculate the mean absolute error between the exercise energy consumption test value and the exercise energy consumption label value, etc., and the embodiments of the present application do not limit this.

[0124] In the examples of this application, the five machine learning algorithms mentioned above were trained and tested. It was found that the machine learning model trained with the extreme randomized tree regression algorithm had the highest test accuracy among the five machine learning algorithms. Therefore, when subsequently monitoring the exercise energy consumption of preschool children, the machine learning model trained with the extreme randomized tree regression algorithm can be selected for monitoring.

[0125] Based on the embodiment of determining sample features introduced in step A2 above, it can be known that the acquired three-axis acceleration signal can be divided according to the preset time window to obtain the first acceleration signals corresponding to the multiple preset time windows. For each preset time window, the first sample feature corresponding to the preset time window can be determined based on the first acceleration signal and personalized indicators of the preset time window. And obtain multiple motion energy consumption label values ​​within the preset time window, and calculate the average motion energy consumption label value of the multiple motion energy consumption label values ​​as the sample true value corresponding to the first sample feature. Then, the first sample feature corresponding to each preset time window and the average motion energy consumption label value corresponding to the first sample feature of each preset time window are input into the machine learning algorithm to obtain the first motion energy consumption prediction value corresponding to each preset time window. The machine learning algorithm is trained based on the first motion energy consumption prediction value and the average motion energy consumption label value corresponding to each preset time window to obtain a trained machine learning model.

[0126] The length of the preset time windows may affect the accuracy of the trained machine learning model. Therefore, multiple different time windows can be pre-determined. The machine learning model is trained once for each time window, and the accuracy of the machine learning model is tested. The time window corresponding to the machine learning model with the highest accuracy is determined as the preset time window. When the machine learning model is subsequently applied to monitor the movement of preschool children, the three-axis acceleration signal can be divided according to the preset time windows.

[0127] In specific implementation, multiple different time windows are first determined. The different lengths of the multiple time windows can be divided based on experience and actual needs. For example, the length of the time window can be set to 15 seconds, 30 seconds, 60 seconds, etc. For the first time window, the acquired three-axis acceleration signal is divided according to the first time window, thereby obtaining second acceleration signals corresponding to multiple first time windows. The first time window represents any time window among the multiple different time windows. Based on the second acceleration signal and the personalized indicator, the second sample feature corresponding to the first time window is determined. The second sample features corresponding to the multiple first time windows and the average motion energy consumption label value corresponding to the second sample feature of each first time window are fused to determine the training set and the test set. For example, the second sample features of all first time windows and the average motion energy consumption label value corresponding to the second sample feature of each first time window can be fused to obtain a data set, and then the data set can be divided into a training set and a test set according to a preset ratio. For example, the fused data set can be divided into a training set and a test set according to an 8:2 ratio. The training set includes multiple second sample features and the average motion energy consumption label value corresponding to each second sample feature, and the test set also includes multiple second sample features and the average motion energy consumption label value corresponding to each second sample feature.

[0128] The training set is then input into the machine learning algorithm to obtain the predicted energy consumption value output by the machine learning algorithm. The machine learning algorithm is trained based on the predicted energy consumption value and the average energy consumption label value corresponding to the first time window until the training cutoff condition is met, thereby obtaining a trained machine learning model. The process of determining the average energy consumption label value corresponding to the first time window can be found in the above embodiment and will not be further described here.

[0129] After obtaining a trained machine learning model, the accuracy of the machine learning model can be verified using a test set. Specifically, a test set is determined based on the second sample features corresponding to multiple first time windows. For example, a portion of the second sample features from multiple first time windows can be allocated as a test set. The test set is input into the machine learning model to obtain an exercise energy consumption test value. Based on the exercise energy consumption test value and the exercise energy consumption label value corresponding to the test set, a test error is determined.

[0130] By performing the above process for multiple different time windows, the test error corresponding to each time window can be determined. Based on the test errors corresponding to multiple different time windows, the minimum test error is determined, and the time window corresponding to the minimum test error can be determined as the preset time window. When using machine learning models to monitor the exercise energy consumption of preschool children, the three-axis acceleration signal can be divided according to the preset time windows.

[0131] After multiple training and testing processes, when the preset time window is determined to be 15 seconds in the embodiment of the present application, the accuracy of the machine learning model obtained by training is relatively high.

[0132] According to the above embodiment, the three-axis acceleration signal can include an x-axis acceleration signal, a y-axis acceleration signal, and a z-axis acceleration signal, and their one-dimensional vector sum signal is determined. In a specific implementation, after the three-axis acceleration signal is divided according to a preset time window, the statistical characteristic index of the x-axis acceleration signal, the statistical characteristic index of the y-axis acceleration signal, the statistical characteristic index of the z-axis acceleration signal, and the statistical characteristic index of the one-dimensional vector sum signal are determined for the x-axis acceleration signal, the y-axis acceleration signal, the z-axis acceleration signal, and the one-dimensional vector sum signal in each preset time window. When determining the sample characteristics through various statistical characteristic indicators and personalized indicators, since the sample characteristics include a large number of indicators, the complexity of data processing will increase, affecting the accuracy of training the machine learning model. Therefore, some indicators can be selected from various statistical characteristic indicators and personalized indicators to train the machine learning model. The accuracy of the machine learning model obtained by training will also be different depending on the number of selected indicators. Based on this, a variety of different numbers can be determined in advance to select indicators, and the process of training to obtain a machine learning model can be completed for each number of indicators. The accuracy of the machine learning model can be tested, and the number of indicators corresponding to the machine learning model with the highest accuracy can be selected as the preset number.

[0133] In specific implementation, a feature selection algorithm is used to select a variety of indicators of different numbers from statistical feature indicators and personalized indicators, and determine the sample features corresponding to the multiple different numbers of indicators. A training set and a test set are determined for the third sample feature corresponding to the first number of indicators. The first number represents any one of the multiple different numbers. That is, for any first number, a feature selection algorithm can be used to select the first number of indicators from statistical feature indicators and personalized indicators, and the selected indicators are fused to determine the third sample feature. The third sample feature and the exercise energy consumption label value corresponding to the third sample feature are fused, and then the fused data set is divided into a training set and a test set. The training set is used to train the machine learning algorithm to obtain a machine learning model. The specific training process will not be described in detail.

[0134] The test set is then input into the trained machine learning model to obtain the exercise energy consumption test value. The test error is determined based on the exercise energy consumption test value and the exercise energy consumption label value corresponding to the test set. The minimum test error is determined based on the test errors corresponding to a variety of different numbers of indicators. The number corresponding to the minimum test error is determined to be a preset number. Subsequently, when applying the machine learning model to monitor exercise energy consumption, the preset number of indicators can be selected from the statistical feature indicators and personalized indicators as the features input into the machine learning model.

[0135] The exercise energy consumption label value in the embodiment of the present application includes energy consumption or metabolic equivalent, and two machine learning models can be trained for energy consumption and metabolic equivalent. In order to obtain a machine learning model for monitoring energy consumption, after obtaining the personalized indicators, three-axis acceleration signals and energy consumption label values ​​of preschool children, it is possible to determine a plurality of numbers of indicators to form sample features to train and obtain a machine learning model. After completing the training and testing process using a plurality of numbers of indicators, the embodiment of the present application determines that when the sample features include 80 indicators, the accuracy of the obtained machine learning model in monitoring energy consumption is higher.

[0136] Similarly, in order to obtain a machine learning model for monitoring metabolic equivalents, after obtaining the personalized indicators, triaxial acceleration signals, and metabolic equivalent label values ​​of preschool children, a variety of indicators can be determined to form sample features to train the corresponding machine learning model. After completing the training and testing process using a variety of indicators, the embodiment of the present application determined that when the sample features include 96 indicators, the resulting machine learning model has a higher accuracy in monitoring metabolic equivalents.

[0137] For details, see Table 1, which shows an example of determining sample features. The total number of samples indicates the number of sample features. Whether feature selection was performed indicates whether only some of the statistical and personalized features were selected to form the sample features.

[0138] Table 1

[0139]

[0140] By collecting the acceleration signals, personalized indicators and exercise energy consumption data of preschool children, and obtaining a trained machine learning model, the machine learning model can be used to determine the exercise energy consumption data of the preschool children to be monitored in subsequent application scenarios. Therefore, the average daily moderate-to-strong activity index can be determined based on the relationship between the exercise energy consumption data and the first moderate-intensity threshold and the second moderate-intensity threshold. That is, a moderate activity intensity assessment model for preschool children is constructed, which realizes the exercise energy consumption monitoring of preschool children and improves the accuracy of exercise energy consumption monitoring.

[0141] In one possible implementation, after obtaining the personalized indicators, three-axis acceleration signals, and one-dimensional vector sum signals of the three-axis acceleration signals of preschool children, the three-axis acceleration signals and the one-dimensional vector sum signals can be divided according to preset time windows of 15 seconds. For each preset time window, the statistical characteristic indicators of the three-axis acceleration signals and the one-dimensional vector sum signals within the preset time window are extracted.

[0142] In one possible implementation, the determined statistical characteristic indicators of the three-axis acceleration signal and the one-dimensional vector sum signal include a total of 91 indicators. The personalized indicators include five indicators: gender, age, height, weight, and body mass index.

[0143] In one possible implementation, when the desired exercise energy consumption data is energy expenditure, 80 indicators from statistical feature indicators and personalized indicators can be extracted to construct a feature set. The feature set is then input into a trained extreme randomized tree regression model to calculate energy expenditure.

[0144] In one possible implementation, when the desired exercise energy expenditure data is metabolic equivalents, 96 indicators from the statistical feature indicators and personalized indicators can be extracted to construct a feature set. The feature set is then input into a trained extreme randomized tree regression model to calculate the metabolic equivalents. Based on the relationship between the metabolic equivalents and the first and second moderate-intensity thresholds, the average daily moderate-to-vigorous activity index is determined.

[0145] Based on the method described in the above embodiment, the present application also provides a method for monitoring the energy consumption of preschool children. Figure 2 , which is a flow chart of a method for monitoring exercise energy consumption of preschool children provided in an embodiment of the present application.

[0146] The method may include the following steps:

[0147] S201: Obtaining personalized indicators of the preschool child to be monitored and acceleration signals in the state to be monitored.

[0148] S202: Inputting the personalized index and the acceleration signal into the exercise energy consumption monitoring model to obtain the exercise energy consumption index of the preschool child to be monitored.

[0149] The exercise energy consumption monitoring model can be represented by a machine learning model trained by the above method embodiment. The exercise energy consumption index can include energy consumption or metabolic equivalent.

[0150] In specific implementations, input features can be determined based on personalized indicators and acceleration signals. These input features are then fed into a sports energy consumption monitoring model to obtain sports energy consumption indicators. These indicators can include energy consumption or metabolic equivalents.

[0151] Among them, after obtaining the personalized indicators and acceleration signals, the process of determining the input features based on the personalized indicators and acceleration signals can refer to the process of determining the sample features during the training process of the above embodiment. The principles of the two are the same and will not be repeated here.

[0152] Since the exercise energy consumption monitoring model is obtained by collecting acceleration signals, personalized indicators and exercise energy consumption data of preschool children, using this exercise energy consumption monitoring model to monitor the exercise energy consumption of preschool children can improve the accuracy of exercise energy consumption monitoring.

[0153] In one possible implementation, when the exercise energy consumption indicator includes metabolic equivalents, the average daily moderate-intensity activity index can be determined based on the relationship between the metabolic equivalents and the first moderate-intensity threshold and the second moderate-intensity threshold.

[0154] Based on the above method embodiment, the present application embodiment also provides a device for constructing a moderate activity intensity assessment model for preschool children. Figure 3 As shown, Figure 3 A schematic diagram of a device for constructing a moderate activity intensity assessment model for preschool children provided in an embodiment of the present application.

[0155] The apparatus 300 comprises:

[0156] The data acquisition unit 301 is used to obtain personalized indicators of preschool children, three-axis acceleration signals, and one-dimensional vector sum signals of the three-axis acceleration signals;

[0157] a feature extraction unit 302 configured to extract statistical feature indicators of the three-axis acceleration signal and the one-dimensional vector sum signal within a preset time window, wherein the multiple preset time windows are non-overlapping time windows;

[0158] A feature construction unit 303 is configured to construct a feature set based on the extracted statistical feature indicators and the personalized indicators;

[0159] The prediction unit 304 is configured to calculate and obtain exercise energy consumption data based on the relationship between the feature set and exercise energy consumption;

[0160] The judgment unit 305 is used to determine the daily average moderate-intensity activity index based on the relationship between the exercise energy consumption data and the first moderate-intensity threshold and the second moderate-intensity threshold.

[0161] In a possible implementation, the daily average moderate-strong activity index , t is the length of the preset time window, N is the total number of preset time windows that meet the requirements; the initial value of N is 0. When the exercise energy consumption data corresponding to the preset time window is greater than or equal to the first moderate intensity threshold and less than or equal to the second moderate intensity threshold, N is updated by adding 1.

[0162] In one possible implementation, the three-axis acceleration signal includes an x-axis acceleration signal, a y-axis acceleration signal, and a z-axis acceleration signal. The statistical characteristic indicators of the x-axis acceleration signal include at least the maximum value of multiple x-axis acceleration signals within the preset time window; the statistical characteristic indicators of the y-axis acceleration signal include at least the peak value of multiple y-axis acceleration signals within the preset time window; the statistical characteristic indicators of the z-axis acceleration signal include at least the minimum value of multiple z-axis acceleration signals; and the statistical characteristic indicators of the one-dimensional vector sum signal include at least the average value and variance of multiple one-dimensional vector sum signals within the preset time window.

[0163] In one possible implementation, the relationship between the feature set and exercise energy consumption is established through a trained machine learning model.

[0164] In one possible implementation, the trained machine learning model includes an extreme random tree regression model.

[0165] In a possible implementation, the exercise energy consumption data includes at least one of an energy consumption mean and a metabolic equivalent mean.

[0166] In a possible implementation, the personalized indicator includes at least one of gender, age, height, weight, and body mass index.

[0167] In one possible implementation, the first medium intensity threshold is 3.2 Mets, and the second medium intensity threshold is 5.3 Mets.

[0168] In one possible implementation, the first moderate intensity threshold for a 3-year-old preschooler is 3.2 Mets, and the second moderate intensity threshold for a 3-year-old preschooler is 4.4 Mets. This means that the range of moderate activity intensity for a 3-year-old preschooler is [3.2, 4.4]. The overall standard deviation is plus or minus 0.2, with the standard deviation ranging from [0, 0.3]. The specific implementation of this standard deviation depends on factors such as the height range of the measured sample.

[0169] In one possible implementation, the first moderate intensity threshold for a 4-year-old preschooler is 3.4 Mets, and the second moderate intensity threshold for a 4-year-old preschooler is 4.6 Mets. This means that the range of moderate activity intensity for a 4-year-old preschooler is [3.4, 4.6], with an overall standard deviation of plus or minus 0.2. The standard deviation ranges from [0, 0.3]. The specific implementation of this standard deviation depends on factors such as the height range of the measured sample.

[0170] In one possible implementation, the first moderate intensity threshold for a 5-year-old preschooler is 3.7 Mets, and the second moderate intensity threshold for a 5-year-old preschooler is 4.8 Mets. This means that the range of moderate activity intensity for a 5-year-old preschooler is [3.7, 4.8], with an overall standard deviation of plus or minus 0.2. The standard deviation ranges from [0, 0.4]. The specific implementation of this standard deviation depends on factors such as the height range of the sample being measured.

[0171] In one possible implementation, the first moderate-intensity threshold for a 6-year-old preschooler is 3.8 Mets, and the second moderate-intensity threshold for a 6-year-old preschooler is 5.1 Mets. This means that the range of moderate activity intensity for a 6-year-old preschooler is [3.8, 5.1]. The overall standard deviation is plus or minus 0.2, with the standard deviation ranging from [0, 0.6]. The specific implementation of this standard deviation depends on factors such as the height range of the measured sample.

[0172] In one possible implementation, the age of a preschool child is calculated based on the age corresponding to the actual date of birth.

[0173] In addition, the present application also provides a device for monitoring the energy consumption of preschool children. Figure 4 As shown, Figure 4 This is a schematic diagram of an exercise energy consumption monitoring device for preschool children provided in an embodiment of the present application.

[0174] The apparatus 400 comprises:

[0175] An acquisition unit 401 is configured to acquire personalized indicators of the preschool child to be monitored and an acceleration signal in the state to be monitored;

[0176] The monitoring unit 402 is configured to input the personalized index and the acceleration signal into the exercise energy consumption monitoring model to obtain the exercise energy consumption index of the preschool child to be monitored;

[0177] Among them, the sports energy consumption monitoring model can be represented as a machine learning model trained by the above method embodiment.

[0178] Based on the above method embodiment and device embodiment, the present application also provides an electronic device, which will be described below with reference to the accompanying drawings.

[0179] See also Figure 5 , Figure 5 A schematic diagram of an electronic device provided in an embodiment of the present application.

[0180] The device 500 includes: a memory 501 and a processor 502;

[0181] The memory 501 is used to store relevant program codes;

[0182] The processor 502 is used to call the program code to execute the method for constructing a moderate activity intensity assessment model for preschool children or the method for monitoring exercise energy consumption for preschool children described in the above method embodiments.

[0183] In addition, an embodiment of the present application also provides a computer-readable storage medium, which is used to store a computer program, and the computer program is used to execute the method for constructing a moderate activity intensity assessment model for preschool children or the method for monitoring exercise energy consumption for preschool children described in the above method embodiment.

[0184] An embodiment of the present application also provides a computer program product, which includes a computer program / instructions. When the computer program / instructions are executed by a processor, they implement the method for constructing a moderate activity intensity assessment model for preschool children or the method for monitoring exercise energy consumption for preschool children as described in the above method embodiment.

[0185] In addition, the present application also provides a method for determining an exercise energy consumption monitoring model, the method comprising:

[0186] Obtaining personalized indicators, acceleration signals, and motion energy consumption label values ​​corresponding to the acceleration signals for preschool children;

[0187] Determining a sample feature based on the personalized indicator and the acceleration signal;

[0188] Inputting the sample features and the exercise energy consumption label values ​​corresponding to the sample features into a machine learning algorithm to obtain an exercise energy consumption prediction value;

[0189] The machine learning algorithm is trained based on the exercise energy consumption prediction value and the exercise energy consumption label value until a training cutoff condition is met, thereby obtaining a trained exercise energy consumption monitoring model.

[0190] In a possible implementation, determining the sample feature based on the personalized indicator and the acceleration signal includes:

[0191] Dividing the acceleration signal according to preset time windows to obtain first acceleration signals corresponding to a plurality of preset time windows respectively;

[0192] Based on the first acceleration signal of each preset time window and the personalized indicator, a first sample feature corresponding to each preset time window is determined.

[0193] In one possible implementation, inputting the sample features into a machine learning algorithm to obtain a predicted exercise energy consumption value includes:

[0194] Inputting the first sample feature corresponding to each preset time window and the exercise energy consumption label value corresponding to the first sample feature of each preset time window into the machine learning algorithm to obtain the first exercise energy consumption prediction value corresponding to each preset time window;

[0195] The training of the machine learning algorithm based on the exercise energy consumption prediction value and the exercise energy consumption label value includes:

[0196] For each preset time window, obtaining multiple exercise energy consumption label values ​​within the preset time window, and calculating an average exercise energy consumption label value of the multiple exercise energy consumption label values;

[0197] The machine learning algorithm is trained based on the first exercise energy consumption prediction value and the average exercise energy consumption label value corresponding to each preset time window.

[0198] In a possible implementation, determining the sample feature based on the personalized indicator and the acceleration signal includes:

[0199] Identify multiple different time windows;

[0200] For a first time window, dividing the acceleration signal according to the first time window to obtain a plurality of second acceleration signals corresponding to the first time windows respectively, where the first time window represents any time window among the plurality of different time windows;

[0201] Determining a second sample feature corresponding to the first time window based on the second acceleration signal and the personalized indicator;

[0202] The step of inputting the sample features and the exercise energy consumption label values ​​corresponding to the sample features into a machine learning algorithm to obtain an exercise energy consumption prediction value includes:

[0203] Determining a training set based on a plurality of second sample features corresponding to the first time windows and exercise energy consumption label values ​​corresponding to the second sample features of the first time windows;

[0204] Inputting the training set into the machine learning algorithm to obtain the predicted value of exercise energy consumption;

[0205] The method further comprises:

[0206] Determining a test set based on the second sample features corresponding to the plurality of first time windows and the exercise energy consumption label values ​​corresponding to the second sample features of the first time windows;

[0207] Inputting the test set into the exercise energy consumption monitoring model to obtain an exercise energy consumption test value;

[0208] Determining a test error based on the exercise energy consumption test value and the exercise energy consumption label value corresponding to the test set;

[0209] Determine the minimum test error based on the test errors corresponding to the multiple different time windows;

[0210] The time window corresponding to the minimum test error is determined to be the preset time window.

[0211] In one possible implementation, the machine learning algorithm includes a plurality of different machine learning models, each machine learning model corresponds to a trained exercise energy consumption monitoring model, and the method further includes:

[0212] Determine a test set based on the sample features and the exercise energy consumption label values ​​corresponding to the sample features;

[0213] For any machine learning model, input the test set into the exercise energy consumption monitoring model corresponding to the machine learning model to obtain the exercise energy consumption test value;

[0214] Determining a test error based on the exercise energy consumption test value and the exercise energy consumption label value corresponding to the test set;

[0215] Determine the minimum test error based on the test errors corresponding to multiple machine learning models;

[0216] Determine that the machine learning model corresponding to the minimum test error is the preferred machine learning algorithm.

[0217] In one possible implementation, the acceleration signal includes an x-axis acceleration signal, a y-axis acceleration signal, and a z-axis acceleration signal, and determining the sample feature based on the personalized indicator and the acceleration signal includes:

[0218] Determining a vector sum signal of the x-axis acceleration signal, the y-axis acceleration signal, and the z-axis acceleration signal;

[0219] respectively determining statistical characteristic indicators of the x-axis acceleration signal, the y-axis acceleration signal, the z-axis acceleration signal, and the vector sum signal;

[0220] The sample characteristics are determined based on the statistical characteristic indicators and the personalized indicators.

[0221] In a possible implementation, determining the sample feature based on the statistical feature indicator and the personalized indicator includes:

[0222] Using a feature selection algorithm, selecting a preset number of indicators from the statistical feature indicators and the personalized indicators;

[0223] The sample characteristics are determined based on the preset number of indicators.

[0224] In one possible implementation, the process of determining the preset number includes:

[0225] Using a feature selection algorithm, selecting a plurality of different numbers of indicators from the statistical feature indicators and the personalized indicators, and determining sample features corresponding to the plurality of different numbers of indicators;

[0226] Determine a training set and a test set for a third sample feature corresponding to an indicator of a first number and a sports energy consumption label value corresponding to the third sample feature, wherein the first number represents any one of a plurality of different numbers;

[0227] Using the training set to train the machine learning algorithm to obtain the exercise energy consumption monitoring model;

[0228] Inputting the test set into the exercise energy consumption monitoring model to obtain an exercise energy consumption test value;

[0229] Determining a test error based on the exercise energy consumption test value and the exercise energy consumption label value corresponding to the test set;

[0230] Determine the minimum test error based on the test errors corresponding to various numbers of indicators;

[0231] The number corresponding to the minimum test error is determined to be the preset number.

[0232] In a possible implementation, determining the sample feature based on the personalized indicator and the acceleration signal includes:

[0233] Preprocessing the personalized index and the acceleration signal to obtain a preprocessed personalized index and a preprocessed acceleration signal;

[0234] determining the sample feature based on the preprocessed personalized index and the preprocessed acceleration signal;

[0235] The pretreatment process includes one or more of the following:

[0236] impute missing values; or,

[0237] One-hot encoding of categorical data; or,

[0238] Normalize continuous data; or,

[0239] Normalize the continuous data and convert the normalized data into data with an approximate normal distribution.

[0240] In one possible implementation, the machine learning algorithm is trained using the sample features and a k-fold cross validation method;

[0241] The machine learning algorithm includes an extreme randomized trees regression algorithm.

[0242] In a possible implementation, the personalized indicator includes: at least one of gender, age, height, weight, and body mass index;

[0243] The exercise energy consumption label value includes at least one of energy consumption and metabolic equivalent.

[0244] In a possible implementation, the acceleration signal is measured by an accelerometer worn by the preschool child;

[0245] The exercise energy consumption label value is measured by a gas metabolism analyzer worn by the preschool child.

[0246] It should be noted that the computer-readable medium mentioned above in this application may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0247] The computer program product may be written in any combination of one or more programming languages ​​to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0248] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other. In particular, for system or device embodiments, since they are basically similar to method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiments described above are merely illustrative, wherein the units or modules described as separate components may or may not be physically separated, and the components shown as units or modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units or modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Ordinary technicians in this field can understand and implement it without expending creative work.

[0249] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations that may be implemented according to the methods, devices and equipment of various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0250] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or plural.

[0251] It should also be noted that, in this application, relational terms such as first and second, etc. are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0252] The steps of the methods or algorithms described in conjunction with the embodiments disclosed in this application can be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0253] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to encompass the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a moderate activity intensity assessment model for preschool children, characterized by: The method comprises: Acquiring personalized indicators of preschool children, three-axis acceleration signals, and a one-dimensional vector sum signal of the three-axis acceleration signals; Extracting statistical characteristic indicators of the three-axis acceleration signal and the one-dimensional vector sum signal within a preset time window, wherein the multiple preset time windows are non-overlapping time windows; Constructing a feature set based on the extracted statistical feature indicators and the personalized indicators; Calculating exercise energy consumption data based on a relationship between the feature set and exercise energy consumption, wherein the exercise energy consumption data includes at least one of energy consumption and metabolic equivalents; When the running energy consumption data includes metabolic equivalents, determining the daily average moderate-to-vigorous activity index based on the relationship between the metabolic equivalents and the first moderate-intensity threshold and the second moderate-intensity threshold; The daily average moderate-to-vigorous activity index T = t*N, where t is the length of the preset time window and N is the total number of preset time windows that meet the requirements. The initial value of N is 0. When the metabolic equivalent corresponding to the preset time window is greater than or equal to the first moderate-intensity threshold and less than or equal to the second moderate-intensity threshold, N is updated by adding 1. The relationship between the feature set and exercise energy consumption is established through a trained machine learning model; The trained machine learning model includes an extreme random tree regression model; The personalized indicators include: at least one of gender, age, height, weight and body mass index; The first medium-intensity threshold is 3.2 Mets, and the second medium-intensity threshold is 5.3 Mets.

2. The method according to claim 1, characterized in that The age of preschool children is calculated based on the age corresponding to their actual date of birth.

3. A device for constructing a moderate activity intensity assessment model for preschool children, characterized in that: The device comprises: A data acquisition unit, configured to acquire personalized indicators of preschool children, a three-axis acceleration signal, and a one-dimensional vector sum signal of the three-axis acceleration signal; a feature extraction unit, configured to extract statistical feature indicators of the three-axis acceleration signal and the one-dimensional vector sum signal within a preset time window, wherein the multiple preset time windows are non-overlapping time windows; A feature construction unit, configured to construct a feature set based on the extracted statistical feature indicators and the personalized indicators; a prediction unit, configured to calculate and obtain exercise energy consumption data based on a relationship between the feature set and exercise energy consumption, wherein the exercise energy consumption data includes at least one of energy consumption and metabolic equivalents; a judgment unit, configured to, when the running energy consumption data includes metabolic equivalents, determine an average daily moderate-to-vigorous activity index based on a relationship between the metabolic equivalents and a first moderate-intensity threshold and a second moderate-intensity threshold; wherein the average daily moderate-to-vigorous activity index T = t*N, where t is the length of the preset time window and N is the total number of preset time windows that meet the requirements; the initial value of N is 0, and when the metabolic equivalent corresponding to the preset time window is greater than or equal to the first moderate-intensity threshold and less than or equal to the second moderate-intensity threshold, N is updated by adding 1; The relationship between the feature set and exercise energy consumption is established through a trained machine learning model; The trained machine learning model includes an extreme random tree regression model; The personalized indicators include: at least one of gender, age, height, weight and body mass index; The first moderate intensity threshold is 3.2 Mets, and the second moderate intensity threshold is 5.3 Mets; The device is used to execute the method for constructing a moderate activity intensity assessment model for preschool children as described in claim 1 or 2.

4. An electronic device, characterized in that: The device includes: a memory and a processor; The memory is used to store relevant program codes; The processor is used to call the program code to execute the method for constructing a moderate activity intensity assessment model for preschool children as described in claim 1 or 2.

5. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store a computer program, and the computer program is used to execute the method for constructing a moderate activity intensity assessment model for preschool children according to claim 1 or 2.

Images