Method for building APHV estimation model based on biometric data using artificial intelligence

Abstract

A method for building an age of peak height velocity (APHV) estimation model based on biometric data using artificial intelligence, according to exemplary embodiments of the present invention, may comprise the steps of: receiving time-series biometric data of a plurality of sample subjects; by using the time-series biometric data of the sample subjects, estimating a plurality of first growth rate curves indicating the rates of growth in height relative to the actual ages of the respective plurality of sample subjects; extracting, from each of the plurality of first growth rate curves, first APHV data indicating the age at which the growth rate reaches the maximum value; and inputting the first APHV data as target data into an artificial intelligence model and learning same.

Description

Method for building an APHV estimation model based on biometric data using artificial intelligence

[0001] The present invention relates to a method, device and computer program for classifying growth stages based on APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity reaches its maximum value extracted from the biometric data of children and adolescents in their growth period, and then predicting growth such as height, obesity and disease, and providing a customized solution for each growth stage.

[0002] With the recent development of artificial intelligence technology, it is being applied to various fields, and methods for extracting features inherent in data through neural network models and generating additional information are being developed and used as replacements for existing data processing methods.

[0003] Neural network models used for artificial intelligence can detect and recognize features within input data more quickly and accurately through learning than conventional data processing. Recently, AI technology has been applied beyond simple object tracking and detection to learning past history, predicting the future, and deriving current features that reflect time-series changes.

[0004] Among these, predictive analytics is a statistical and data mining technique that extracts information from data and uses it to predict trends and behavioral patterns. This predictive analytics can be applied to any area requiring decision-making based on information gleaned from data. The core of predictive analytics is understanding the relationships between variables and then predicting unknown variables.

[0005] For this purpose, various approaches are being used depending on the data characteristics and prediction target.

[0006] Among the various fields requiring predictive analysis, the area of ​​physical growth in adolescents is particularly noteworthy. Parents and adolescents alike are deeply interested in the timing and extent of height growth.

[0007] Conventional predictions of height growth suggest methods for predicting growth plates by X-ray or analyzing the relationship with genetic / environmental factors (Patent Publication No. 10-2075743, Patent Publication No. 10-1866208), or for converting the body data of sample subjects with different measurement periods or measurement counts into a form suitable for learning a growth prediction model (Patent Publication No. 10-2198302).

[0008] Although methods for predicting the physical growth of children and adolescents have been suggested, as mentioned above, research on methods for predicting and providing solutions for the risk of developing precocious puberty and obesity that may accompany the growth of children and adolescents is insufficient.

[0009] Additionally, since children and adolescents have different growth stages with different characteristics, taking this into account can increase the reliability of solutions provided through predicted data and analysis.

[0010] Meanwhile, in the case of conventional technologies disclosed to date, various methods for predicting growth by directly inputting physical or biometric data of children and adolescents into a predetermined growth prediction tool have been disclosed. However, according to these conventional technologies, growth is predicted only by general indicators such as past and present height, weight, and BMI without considering individual growth characteristics such as the growth rate of each child and adolescent or whether the growth stage was reached quickly compared to other sample subjects, so there is a disadvantage in that it is difficult to provide a personalized solution.

[0011] In addition, the SITAR (SuperImposition by Translation and Rotation) methodology has emerged in relation to height growth rate, but there is a problem in that only a posteriori estimation is possible because measurements from the entire growth period are required to estimate the growth rate curve using the SITAR methodology.

[0012] Meanwhile, the existing height growth prediction model is a model that predicts the height growth curve using the LGBM model as input variables such as height, weight, body composition, their quantiles, and the current quantile-based benchmark growth rate. However, there is a possibility that the estimation model may be upwardly biased in certain data sets with a high proportion of observations due to high height quantiles at later ages. In addition, the existing height growth prediction model has the disadvantage of low prediction accuracy because it utilizes only 8 variables centered on height, weight, and body composition mass despite having 52 body composition-related variables.

[0013] Consequently, in the existing height growth prediction model, although the growth rate tendency of the sample subject is important for height prediction, information on past measurements is not reflected in the estimation model, so it is necessary to measure the follow-up growth rate that reflects past information.

[0014] One of the various tasks of the present invention is to provide a method, device and computer program for classifying growth stages based on APHV (Age of Peak Height Velocity) data, which is the age at which the growth rate of children and adolescents reaches its maximum, and then predicting growth such as height, obesity and disease, and providing a customized solution for each growth stage.

[0015] One of the various tasks of the present invention is to provide a method, device and computer program for predicting growth, including height, obesity and disease, based on the growth stage of children and adolescents using an APHV estimation model constructed through artificial intelligence and providing a customized solution for each growth stage.

[0016] A method for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention may include the steps of: receiving time-series biometric data of a plurality of sample subjects; estimating a plurality of first growth velocity curves representing a growth velocity of height compared to actual age of each of the plurality of sample subjects using the time-series biometric data of the sample subjects; extracting first APHV (Age of Peak Height Velocity) data, which is an age at which the growth velocity reaches a maximum value, for each of the plurality of first growth velocity curves; and inputting the first APHV data into an artificial intelligence model using the model as target data to learn the model.

[0017] The step of estimating the first growth rate curve can be performed by extracting age data and height data from the time-series biometric data of the sample subjects and then applying the extracted age data and height data to a curve transformation model that uses the age data and height data as input variables.

[0018] The above curve transformation model may be a model constructed using a polynomial function.

[0019] The above curve transformation model may be a model constructed using mathematical formula 1,

[0020] [Mathematical Formula 1]

[0021]

[0022] Here, y it is α i , β i and γ i are random correction values ​​for adjusting the vertical shift, horizontal shift and slope of the curve, respectively, h is a specific function related to the height data, and t may be age data.

[0023] The method for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention is performed after the step of inputting time-series biometric data of the sample subjects, and may further include a step of preprocessing the time-series biometric data of the sample subjects; and the preprocessing step may be performed by determining all biometric data of the sample subjects for which there is no biometric data collected at a specific growth period as noise and removing them.

[0024] The above preprocessing step can be performed by dividing the growth period of the plurality of sample subjects into a normal growth period, a rapid growth period, and a decelerated growth period, and then removing as noise all data of the sample subjects for which there is no biometric data corresponding to each of the divided growth stages.

[0025] The above preprocessing step can be performed by further dividing the growth period of the plurality of sample subjects into a slow growth period and a growth plate period, and then removing as noise all data of the sample subjects for which there is no biometric data corresponding to each of the divided growth stages.

[0026] The above preprocessing step can be performed by dividing the age range of the plurality of sample subjects into 8 to 18 years, and then removing as noise all data of sample subjects whose input period difference between adjacent biometric data is 4 years or more.

[0027] The above preprocessing step can be performed by removing as noise all data of sample subjects for whom biometric data entered during the period between the ages of 9 and 14 among the multiple sample subjects exists.

[0028] According to exemplary embodiments of the present invention, the APHV estimation model based on biometric data using artificial intelligence applies a statistical methodology to a trace growth velocity extracted from past information to estimate a growth velocity curve, extracts APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity reaches the maximum value for the growth velocity curve, and then utilizes the APHV data to build various prediction models such as an integrated height growth prediction model, a precocious puberty prediction model, and an obesity prediction model, so that the reliability of various prediction models such as an integrated height growth prediction model, a precocious puberty prediction model, and an obesity prediction model can be maximized.

[0029] In addition, according to various embodiments of the present invention, it is possible to accurately predict height by considering the growth stage of children and adolescents through an APHV estimation model constructed through artificial intelligence learning, and to provide a solution necessary for height growth by considering each growth stage.

[0030] FIG. 1 is a schematic diagram illustrating the configuration of an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention.

[0031] Figure 2 is a diagram showing predicted height and target height for each growth stage according to exemplary embodiments of the present invention.

[0032] FIG. 3 is a flowchart illustrating a method for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention.

[0033] Figure 4 is a diagram showing multiple first growth rate curves that represent the growth rate of height compared to actual age using time-series biometric data of sample subjects.

[0034] Figures 5 and 6 are diagrams showing the APHV estimation error according to the age of the girl and the boy, respectively.

[0035] Figures 7 and 8 are diagrams showing the importance of predictive variables for APHV in women and men, respectively.

[0036] Figure 9 is a diagram showing variable importance in a conventional growth prediction model for women and men, and Figure 10 is a diagram showing variable importance in a growth prediction model of the present invention.

[0037] Hereinafter, specific embodiments of the present invention will be described. The following detailed description is provided to facilitate a comprehensive understanding of the methods, devices, and / or systems described herein. However, these are merely examples and the present invention is not limited thereto.

[0038] In describing embodiments of the present invention, if a detailed description of a known technology related to the present invention is judged to unnecessarily obscure the gist of the present invention, the detailed description will be omitted. In addition, the terms described below are terms defined in consideration of their functions in the present invention, and this may vary depending on the intention or custom of the user or operator. Therefore, the definitions should be made based on the contents throughout this specification. The terminology used in the detailed description is only for the purpose of describing embodiments of the present invention and should not be limited in any way. Unless clearly used otherwise, the singular form includes the plural form. In this description, expressions such as "comprises" or "having" are intended to indicate certain features, numbers, steps, operations, elements, parts or combinations thereof, and should not be construed to exclude the presence or possibility of one or more other features, numbers, steps, operations, elements, parts or combinations thereof other than those described.

[0039] Additionally, in describing components of embodiments of the present invention, terms such as first, second, A, B, (a), (b), etc. may be used. These terms are only intended to distinguish the components from other components, and the nature, order, or sequence of the components are not limited by the terms.

[0040] In exemplary embodiments of the present invention, "children and adolescents" can be understood as a concept encompassing the growth period of the human body. More specifically, "children and adolescents," referring to the sample subjects in exemplary embodiments of the present invention, are defined below through the meanings of infants, children, adolescents, toddlers, and children.

[0041] Infants are a continuation of the neonatal period, growing up biting their mother's nipple for up to two years after birth. The experiences of nutrition, caressing, and excretion during this period influence general tendencies thereafter. Children are generally people between the ages of 6 and 13, and in a broader sense, they can include infants (ages 1 to 5).

[0042] Adolescents are the transitional stage between childhood and adolescence, generally referring to individuals aged 13 to 19. Infants can refer to individuals from birth to 5 years of age. Children generally refer to children up to 15 years of age.

[0043] Therefore, the term “children and adolescents” referring to the above sample subjects may mean, in a narrow sense, the period from 5 years of age to 19 years of age, including children and adolescents, and in a broad sense, it may mean the period including all general periods of physical growth, from infancy to adolescence.

[0044] FIG. 1 is a schematic diagram illustrating the configuration of an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention.

[0045] Referring to FIG. 1, an APHV estimation model based on artificial intelligence using biometric data according to exemplary embodiments of the present invention may include a biometric data input unit (10), a preprocessing unit (20), a growth rate curve estimation unit (30), a first APHV data extraction unit (40), and a first APHV data learning unit (50).

[0046] The APHV estimation model based on artificial intelligence using biometric data according to exemplary embodiments of the present invention can receive time-series physical information of a sample subject through a biometric data input unit (10). The physical information of the sample subject may include not only basic information such as grade (or age), gender, and height, but also additional information such as body weight, protein, mineral content, body fat, body water, soft lean mass, fat free mass, bone tissue, skeletal muscle mass, body mass index (BMI), basal metabolic rate, neck circumference, chest circumference, abdominal circumference, thigh circumference, arm circumference, and hip circumference. Such physical information is merely an example to help understanding of the present invention, and the present embodiment is not limited thereto. It goes without saying that the types of information constituting the physical information may be varied depending on the embodiment.

[0047] The time-series physical information of the above sample subject may be continuous information or discontinuous information, but may be information included in at least one period corresponding to the growth stage of FIG. 2.

[0048] More specifically, the time-series physical information of the sample subjects may be collected at various times and at various collection frequencies. For example, the first sample subject may have physical information measured from the age of 8 to 12, which is part of the childhood and adolescence period, while the second sample subject may have physical information measured irregularly, such as at the ages of 8, 10 to 12, and 15. In addition, the third sample subject may have physical information measured multiple times within a certain period (within the period of one of the multiple growth stages in Figure 2), whereas the fourth sample subject may have physical information measured only once within a certain period (within the period of one of the multiple growth stages in Figure 2).

[0049] As described above, depending on the collection period and number of collections, the physical information of the sample subject may be included in two or more of the growth stages of Fig. 2 (the first sample subject, the second sample subject), but there may also be cases where this is not the case (the third sample subject, the fourth sample subject).

[0050] The preprocessing unit (20) can be controlled to perform preprocessing on the time-series biodata of the sample subjects.

[0051] In exemplary embodiments, the preprocessing unit (20) may be performed by determining all biometric data of a sample subject for which there is no biometric data collected during a specific growth period as noise and removing the same. This is to improve the prediction accuracy of the biometric data-based APHV estimation model according to the present invention, and may not be applied when data is sufficiently accumulated. In this case, preprocessing may be performed by supplementing or additionally generating some biometric data through various methods such as an artificial intelligence model and big data analysis, without determining all biometric data of a sample subject for which there is no biometric data collected during a specific growth period as noise and removing them.

[0052] The growth rate curve estimation unit (30) can be controlled to estimate a plurality of first growth rate curves representing the growth rate of height compared to actual age of each of the plurality of sample subjects using the time series biometric data of the sample subjects.

[0053] In exemplary embodiments, the estimation of the first growth rate curve can be performed by extracting age data and height data from the time-series biometric data of the sample subjects and then applying the extracted age data and height data to a curve transformation model that uses the age data and height data as input variables.

[0054] At this time, the type of the curve transformation model is not particularly limited. That is, the curve transformation model may be a model constructed using a polynomial function, and for example, the curve transformation model may be a model constructed using mathematical expression 1.

[0055] [Mathematical Formula 1]

[0056]

[0057] Here, y it is α i , β i and γ i are random correction values ​​for adjusting the vertical shift, horizontal shift and slope of the curve, respectively, h is a specific function related to the height data, and t may be age data.

[0058] However, the concept of the present invention is not necessarily limited to this, and may be a module constructed using a polynomial function other than the above mathematical expression 1.

[0059] The first APHV data extraction unit (40) can be controlled to extract first APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity reaches the maximum value, for each of the plurality of first growth velocity curves.

[0060] The above first APHV data can be used to build a growth stage diagnosis and estimation model and to derive growth age, and the above growth age and follow-up growth velocity can ultimately be used to build various prediction models such as an integrated height growth prediction model, a precocious puberty prediction model, and an obesity prediction model.

[0061] Meanwhile, the existing height growth prediction model is a model that predicts the height growth curve using the LGBM model as input variables such as height, weight, body composition, their quantiles, and the current quantile-based benchmark growth rate. However, there is a possibility that the estimation model may be upwardly biased in certain data sets with a high proportion of observations due to high height quantiles at later ages. In addition, the existing height growth prediction model has the disadvantage of low prediction accuracy because it utilizes only 8 variables centered on height, weight, and body composition mass despite having 52 body composition-related variables.

[0062] That is, in the conventional height growth prediction model, although the growth velocity tendency of the sample subject is important for height prediction, information on past measurements is not reflected in the estimation model, so it is necessary to measure the trace growth velocity reflecting the past information. Therefore, the APHV estimation model based on artificial intelligence according to exemplary embodiments of the present invention estimates a growth velocity curve by applying a statistical methodology to the trace growth velocity extracted from past information, and extracts APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity is the maximum, for the growth velocity curve, and then utilizes the APHV data to build various prediction models such as an integrated height growth prediction model, a precocious puberty prediction model, and an obesity prediction model.

[0063] The first APHV data learning unit (50) can be controlled to input the first APHV data as target data into an artificial intelligence model and perform learning.

[0064] In the above, the biometric data input unit (10) has been described based only on the fact that it receives time-series physical information about the sample subject, but the concept of the present invention is not necessarily limited to this. That is, if there is a measurement subject who wants to estimate APHV through the APHV estimation model, the biometric data input unit (10) may be controlled to receive additional biometric data of the measurement subject.

[0065] In this case, the APHV estimation model based on artificial intelligence-based bio-data according to exemplary embodiments of the present invention may further include a predicted APHV data extraction unit (60), and the predicted APHV data extraction unit (60) may be controlled to extract predicted APHV (Age of Peak Height Velocity) data, which is an age at which the growth velocity of the plurality of sample subjects reaches a maximum, by applying the bio-data of the subject to the learned artificial intelligence model using the first APHV (Age of Peak Height Velocity) data as target data.

[0066] Figure 2 is a diagram showing predicted height and target height for each growth stage according to exemplary embodiments of the present invention.

[0067] Looking at the growth stages with reference to Figure 2, the childhood and adolescence period may include a general growth period (301), a rapid growth period (303), a decelerated growth period (305), a slow growth period, and a growth plate period (309).

[0068] Each growth stage can be categorized by growth rate, and the height grown each year varies depending on each growth stage, and even within the same growth stage, the actual height grown can vary depending on the growth type.

[0069] The general growth period (301) usually refers to the period before puberty when secondary sexual characteristics appear, and children and adolescents during this period generally have open growth plates, so they generally grow 4 to 5 cm per year in the case of a short growth type and 6 to 7 cm per year in the case of a tall growth type, depending on the growth environment. In exemplary embodiments, the general growth period (301) may be defined as a period before the onset of rapid pubertal growth in children and adolescents, and may be a growth stage that ends at a point in time that is earlier than the predicted APHV by the first growth period. In one embodiment, the first growth period may be a period of between 1 and 2 years, for example, about 1.5 years.

[0070] The rapid growth phase (303) is the period when secondary sexual characteristics begin to appear, in women, the breasts swell and pubic hair develops, and in men, the testicles enlarge and pubic hair begins to grow, and a change called the pubertal stage occurs. The rapid growth phase (303) generally lasts about 2 to 3 years after the normal growth phase (301), and grows at an average of 7 to 10 cm per year. In exemplary embodiments, the rapid growth phase (303) may be defined as a period when the rapid growth of puberty begins and the growth rate increases before APHV, and may be a growth stage that starts at a time prior to the predicted APHV by the first growth period and ends at the predicted APHV. In one embodiment, the first growth period may be a period of 1 to 2 years, for example, about 1.5 years.

[0071] The deceleration period (305) refers to the period when secondary sexual characteristics are fully developed. During this period, women can be distinguished by the onset of menstruation, and men can clearly see the changes in pubic hair, voice change, and armpit hair. When the deceleration period (305) begins, the growth rate rapidly decreases compared to the rapid growth period (303), and it generally lasts for about 2 to 3 years, and natural growth stops after growing at an average of 5 to 6 cm per year. The growth plate begins to close little by little after the rapid growth period (304), and about 50% of it is closed about 6 months after entering the deceleration period (305). In exemplary embodiments, the deceleration phase (305) may be defined as a period in which the growth rate gradually decreases from the APHV until the end of the adolescent growth spurt, and may be a growth phase that starts at the predicted APHV and ends at a time that is a second growth period after the predicted APHV. In one embodiment, the second growth period may be a period of between one and two years, for example, about 1.5 years.

[0072] The slow growth period (307) may refer to a period in which the growth velocity approaches 0 before entering the growth plate phase after the rapid growth of puberty. In exemplary embodiments, the slow growth period (307) may be defined as a period in which the growth velocity gradually converges to 0 after the end of the rapid growth of puberty in children and adolescents, and may be a growth stage that starts at a time point that is the second growth period longer than the predicted APHV and ends at a time point that is the third growth period longer than the predicted APHV. In one embodiment, the third growth period may be a period of between 2 and 4 years, for example, about 3 years.

[0073] The growth plate phase (309) refers to the period when the growth period is not completely over, but natural height growth becomes difficult, and the growth plate is closed. Generally, women enter the growth plate phase (307) about 1 year and 6 months to 2 years after menarche, and men enter the growth plate phase (307) about 1 year and 6 months to 2 years after the time when hair begins to grow in the armpits. During the growth plate phase (307), the growth plate closes and natural growth stops, but growth can be achieved in the range of about 1 to 3 cm by changing bad lifestyle habits and improving physical functions through customized exercise, posture correction, and nutrient intake. In exemplary embodiments, the growth plate phase (309) can be defined as a period when the growth rate is close to 0 after the growth plate closes, and can be a growth stage starting at a point in time that is the third growth period longer than the predicted APHV. In one embodiment, the third growth period may be a period of between two and four years, for example, about three years.

[0074] FIG. 3 is a flowchart illustrating a method for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention.

[0075] Referring to FIG. 3, a method for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention may include a step (S1) of receiving time-series biometric data of a plurality of sample subjects, a step (S2) of estimating a plurality of first growth velocity curves representing a growth velocity of height compared to actual age of each of the plurality of sample subjects using the time-series biometric data of the sample subjects, a step (S3) of extracting first APHV (Age of Peak Height Velocity) data, which is an age at which the growth velocity reaches a maximum value, for each of the plurality of first growth velocity curves, and a step (S4) of inputting the first APHV data into an artificial intelligence model using the target data to learn.

[0076] The biometric data input step (S1) can be performed by inputting time-series physical information about the sample subject through the biometric data input unit (10).

[0077] The estimation step (S2) of the first growth rate curve can be performed by extracting age data and height data from the time-series biometric data of the sample subjects and then applying the extracted age data and height data to a curve transformation model that uses the age data and height data as input variables.

[0078] In exemplary embodiments, the curve transformation model may be a model constructed using a polynomial function.

[0079] In one embodiment, the curve transformation model may be a model constructed using mathematical expression 1.

[0080] [Mathematical Formula 1]

[0081]

[0082] In the above mathematical expression 1, y it is α i , β i and γ i are random correction values ​​for adjusting the vertical shift, horizontal shift and slope of the curve, respectively, h is a specific function related to the height data, and t may be age data.

[0083] The method for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention is performed after the step of inputting time-series biometric data of the sample subjects, and may further include a step (not shown) of preprocessing the time-series biometric data of the sample subjects.

[0084] The above preprocessing step can be performed by determining as noise all biometric data of sample subjects for which biometric data collected at a specific growth period do not exist, and removing them.

[0085] In exemplary embodiments, the preprocessing step may be performed by dividing the growth phases of the plurality of sample subjects into a normal growth phase, a rapid growth phase, and a decelerated growth phase, and then removing as noise all data of the sample subjects for which there is no biometric data corresponding to each of the divided growth phases.

[0086] In one embodiment, the preprocessing step may be performed by further dividing the growth phase of the plurality of sample subjects into a slow growth phase and a growth plate phase, and then removing as noise all data of the sample subjects for which there is no biometric data corresponding to each of the divided growth phases.

[0087] For example, the above preprocessing step can be performed by dividing the age range of the plurality of sample subjects into 8 to 18 years, and then removing as noise all data of sample subjects whose input period difference between adjacent biometric data is 4 years or more.

[0088] Alternatively, the preprocessing step may be performed by removing as noise all data of sample subjects for whom biometric data entered during the period between the ages of 9 and 14 among the multiple sample subjects exists.

[0089] Figure 4 is a diagram showing multiple first growth rate curves that represent the growth rate of height compared to actual age using time-series biometric data of sample subjects.

[0090] FIG. 4 illustrates a plurality of first growth velocity curves generated through a first growth velocity curve estimation unit (30), and a first APHV data extraction unit (40) can be controlled to extract first APHV (Age of Peak Height Velocity) data, which is an age at which the growth velocity reaches a maximum value, for each of the plurality of first growth velocity curves. The first APHV data may mean data indicating a point at which the growth velocity reaches a maximum during rapid pubertal growth, i.e., an age at which the growth acceleration is 0.

[0091] Meanwhile, the first APHV data extraction unit (40) may be controlled to extract additional data in addition to the first APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity reaches the maximum value for each of the plurality of first growth velocity curves.

[0092] For example, the first APHV data extraction unit (40) can be controlled to further extract AOGS (Age of onset of growth spurt) data, which is the point at which pubertal growth spurt begins, AEGS (Age of end of growth spurt) data, which is the point at which pubertal growth spurt ends, etc.

[0093] The above AOGS data may refer to a point where the growth rate slope increases rapidly and the growth acceleration reaches a maximum, and may refer to data indicating age approximately 1.5 years prior to the first APHV data.

[0094] The above AEGS data may refer to the point at which the adolescent growth spurt ends, that is, the point at which the growth velocity after APHV becomes equal to the growth velocity (Onset of Growth Spurt Velocity) in AOGS, and may refer to data indicating the age approximately 1.5 years after the first APHV data.

[0095] Figures 5 and 6 are diagrams showing the APHV estimation error according to the age of the girl and the boy, respectively.

[0096] Referring to FIGS. 5 and 6, predicted APHV data of subjects of various ages were extracted using an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention, and then the predicted APHV data and the actual APHV data of the subjects were compared.

[0097] Specifically, through the above experimental results, it was confirmed that the average error between the predicted APHV data value and the actual APHV data value was 0.35 years for girls aged 10 years or older, and that the average error between the predicted APHV data value and the actual APHV data value was 0.39 years for boys aged 12 years or older. That is, as a result, in the case of the APHV estimation model based on biometric data using artificial intelligence according to the exemplary embodiments of the present invention, it was confirmed that the average error between the predicted data value and the actual data value was within 0.5 years in the average age at which secondary sexual characteristics appear for both girls and boys, and thus the APHV prediction accuracy was very high.

[0098] In addition, through the above experiment, it was confirmed that the prediction accuracy of the APHV estimation model improves as the actual age of the measurement subjects approaches the actual APHV data value.

[0099] Figures 7 and 8 are diagrams showing the importance of predictive variables for APHV in women and men, respectively.

[0100] Referring to FIGS. 7 and 8, various variables for constructing an APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention are classified according to importance.

[0101] Specifically, as shown in Fig. 7, the variable with the highest importance in predicting APHV in women was Previously Velocity, whereas as shown in Fig. 8, the variable with the highest importance in predicting APHV in men was Height Quantile.

[0102] Meanwhile, in predicting APHV in women, current growth velocity was the fourth most important variable, and in predicting APHV in men, previous growth velocity and current growth velocity were the second and sixth most important variables, respectively.

[0103] That is, it was confirmed that the previous growth velocity (Previously Velocity) and the current growth velocity (Velocity) were very important variables for predicting the APHV of women and men, respectively. In particular, it was confirmed that the previous growth velocity (Previously Velocity) was the most important variable for women and the second most important variable for men. Consequently, it was confirmed that the accuracy of the predicted APHV data extracted using the APHV estimation model can be greatly improved when there are two or more biometric data measured at different times of the subject so that the previous growth velocity can be confirmed.

[0104] In addition, in order to increase the accuracy of the predicted APHV data extracted through the APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention, it was confirmed that additional variables such as height quantile, waist to hip ratio (WHR), and body fat mass are required.

[0105] Figure 9 is a diagram showing variable importance in a conventional growth prediction model for women and men, and Figure 10 is a diagram showing variable importance in a growth prediction model of the present invention.

[0106] Referring to FIGS. 9 and 10, the most important variable in both the conventional female and male growth prediction models and the growth prediction model of the present invention was the growth rate, whereas in the conventional growth prediction model, the second most important variable was the height quantile, and in the growth prediction model of the present invention, the second most important variable was the age of peak height velocity (APHV).

[0107] That is, in the conventional growth prediction model before the concept of APHV was applied, the height quantile was shown to be the second most important variable after the growth rate, but in the growth prediction model according to the present invention after the concept of APHV (Age of Peak Height Velocity) was applied, the age of peak height velocity (APHV) was calculated to be higher in importance than the height quantile, and was shown to be the second most important variable after the growth rate, and accordingly, it was confirmed that the importance of APHV was very high among the variables affecting the predicted final height.

[0108] As described above, the APHV estimation model based on biometric data using artificial intelligence according to exemplary embodiments of the present invention estimates a growth velocity curve by applying a statistical methodology to a trace growth velocity extracted from past information, and extracts APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity reaches the maximum value for the growth velocity curve, and then utilizes the APHV data to build various prediction models such as an integrated height growth prediction model, a precocious puberty prediction model, and an obesity prediction model, so that the reliability of various prediction models such as an integrated height growth prediction model, a precocious puberty prediction model, and an obesity prediction model can be maximized.

[0109] In addition, according to various embodiments of the present invention, it is possible to accurately predict height by considering the growth stage of children and adolescents through an APHV estimation model constructed through artificial intelligence learning, and to provide a solution necessary for height growth by considering each growth stage.

[0110] However, the concept of the present invention is not necessarily limited thereto, and the device / method / system according to the exemplary embodiments of the present invention can be applied to various products / technology fields in addition to the products / technology fields described above.

[0111] While various embodiments of the present invention have been described in detail above, those skilled in the art will appreciate that various modifications to the above-described embodiments are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined not only by the claims set forth below but also by equivalents thereof.

Claims

1. A step of receiving time-series biometric data of multiple sample subjects; A step of estimating a plurality of first growth rate curves representing the growth rate of height compared to actual age of each of the plurality of sample subjects using time-series biometric data of the above sample subjects; A step of extracting first APHV (Age of Peak Height Velocity) data, which is the age at which the growth velocity reaches the maximum value for each of the plurality of first growth velocity curves; and A method for constructing an APHV estimation model based on bio-data using artificial intelligence, comprising: a step of inputting the first APHV data as target data into an artificial intelligence model for learning; 2. In paragraph 1, A method for constructing an APHV estimation model based on bio-data using artificial intelligence, characterized in that the step of estimating the first growth rate curve is performed by extracting age data and height data from the time-series bio-data of the sample subjects and then applying the same to a curve transformation model that uses the age data and height data as input variables.

3. In paragraph 2, A method for constructing an APHV estimation model based on artificial intelligence and bio-data, characterized in that the above curve transformation model is a model constructed using a polynomial function.

4. In paragraph 3, The above curve transformation model is a model constructed using mathematical formula 1. [Mathematical Formula 1] Here, y it is α i , β i and γ i A method for building an APHV estimation model based on biometric data using artificial intelligence, wherein are random correction values ​​for adjusting the vertical movement, horizontal movement and slope of the curve, respectively, h is a specific function related to height data, and t is age data.

5. In paragraph 1, It is performed after the step of inputting time-series bio-data of the above sample subjects, and further includes a step of preprocessing the time-series bio-data of the above sample subjects; A method for building an APHV estimation model based on bio-data using artificial intelligence, characterized in that the above preprocessing step is performed by determining as noise all bio-data of sample subjects for whom bio-data collected at a specific growth period do not exist.

6. In paragraph 5, A method for constructing an APHV estimation model based on bio-data using artificial intelligence, characterized in that the above preprocessing step is performed by dividing the growth period of the plurality of sample subjects into a general growth period, a rapid growth period, and a decelerated growth period, and then removing as noise all data of sample subjects for which there is no bio-data corresponding to each of the divided growth stages.

7. In paragraph 6, A method for constructing an APHV estimation model based on bio-data using artificial intelligence, characterized in that the above preprocessing step is performed by further dividing the growth period of the plurality of sample subjects into a slow growth period and a growth plate period, and then removing as noise all data of the sample subjects for which there is no bio-data corresponding to each of the divided growth stages.

8. In paragraph 5, A method for constructing an APHV estimation model based on biometric data using artificial intelligence, characterized in that the above preprocessing step is performed by dividing the age groups of the plurality of sample subjects into 8 to 18 years old, and then removing as noise all data of sample subjects with a difference in input period of 4 years or more between adjacent biometric data.

9. In paragraph 8, A method for building an APHV estimation model based on biometric data using artificial intelligence, characterized in that the above preprocessing step is performed by removing as noise all data of sample subjects for whom biometric data entered during a period between 9 and 14 years of age among the multiple sample subjects exists.

10. In paragraph 6, The above general growth period is defined as the period before the onset of rapid pubertal growth in children and adolescents. A method for constructing an APHV estimation model based on artificial intelligence and biometric data, characterized in that the general growth period is a growth stage whose end point is a point in time preceding the first APHV by the first growth period.

11. In paragraph 10, A method for constructing an APHV estimation model based on artificial intelligence and bio-data, characterized in that the first growth period is a period of between 1 and 2 years.

12. In paragraph 11, The above rapid growth period is defined as the period when secondary sexual characteristics begin to appear in children and adolescents. A method for constructing an APHV estimation model based on artificial intelligence and bio-data, characterized in that the rapid growth period is a growth stage that starts at a point in time preceding the first APHV by the first growth period and ends at the first APHV.

13. In paragraph 11, The above growth deceleration period is defined as the period when secondary sexual characteristics of children and adolescents are completed. A method for constructing an APHV estimation model based on artificial intelligence and bio-data, characterized in that the above-mentioned decelerating growth period is a growth stage that starts with the first APHV and ends at a point in time that is equal to the second growth period after the first APHV.

14. In paragraph 13, A method for constructing an APHV estimation model based on artificial intelligence and bio-data, characterized in that the second growth period is a period of between 1 and 2 years.

15. In paragraph 14, A method for constructing an APHV estimation model based on artificial intelligence and bio-data, characterized in that the first growth period and the second growth period are set to different periods.

Images