Method for predicting obesity and device for providing solution using ensemble model

Abstract

Disclosed, according to one exemplary embodiment of the present invention, is a method for predicting obesity by obtaining an ensemble of result values derived by predicting obesity by means of mutually different methods, in order to increase obesity prediction accuracy for children and adolescents. The method for predicting obesity using an ensemble technique of the present embodiment may comprise the steps of: receiving, as input, body information of a subject to be assessed; generating a plurality of obesity prediction values from the inputted body information through mutually different trained neural networks; predicting obesity by comparing the plurality of obesity prediction values; and generating mutually different solutions on the basis of the difference between the plurality of obesity prediction values.

Description

Obesity prediction method and solution providing device utilizing ensemble models

[0001] The present invention relates to a method for predicting obesity in children and adolescents through results of predicting obesity by different methods, and an apparatus for providing a solution based thereon.

[0002] With the recent advancement of artificial intelligence technology, AI technologies are being applied across various fields, and methods are being developed and used to generate additional information by extracting inherent features from data through neural network models, replacing conventional data processing methods.

[0003] Neural network models used for artificial intelligence can detect and recognize features within input data more quickly and accurately compared to general data processing through learning. Recently, AI technology has moved beyond simply tracking and detecting objects to being applied to learn from past history, predict the future, or derive current features that reflect time-series changes.

[0004] Among these, predictive analytics is a technology within the fields of statistics and data mining that extracts information from data and uses it to predict trends and behavioral patterns. Such predictive analytics can be applied to all areas requiring decision-making based on information obtained from data. The core of predictive analytics lies in predicting unknown variables after understanding the relationships between variables.

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

[0006] Among the various fields requiring predictive analysis, there is the field of physical growth in children and adolescents. There is significant interest among parents, children, and adolescents regarding when height growth will occur and how much they will grow, as well as whether there is a risk of complications such as precocious puberty or obesity that may accompany growth.

[0007] Conventional methods for predicting height growth involve predicting growth plates by X-ray imaging or analyzing the relationship with genetic and environmental factors (Registered Patent Publication No. 10-2075743, Registered Patent Publication No. 10-1866208), or they also propose methods for converting physical data from sample subjects with different measurement times or frequencies into a format suitable for training a growth prediction model (Registered Patent Publication No. 10-2198302).

[0008] While conventional methods for predicting physical growth in children and adolescents have been presented, research on methods for predicting the risk of occurrence and providing solutions for issues such as precocious puberty and obesity that may accompany growth in children and adolescents, as mentioned above, is currently insufficient.

[0009] One of the various objectives of the present invention is to provide a method for predicting obesity by ensembling the results of obesity predictions made by different methods in order to increase the accuracy of obesity prediction in children and adolescents.

[0010] In addition, we aim to provide optimal solutions for children and adolescents by utilizing the results of obesity prediction using different methods.

[0011] Various embodiments for solving the problem of the present invention may provide an obesity prediction method utilizing an ensemble technique, comprising the steps of receiving body information of an evaluation subject, generating a plurality of obesity prediction values ​​through different neural networks learned based on the input body information, predicting obesity by comparing the plurality of obesity prediction values, and generating different solutions according to the difference between the plurality of obesity prediction values.

[0012] The plurality of obesity prediction values ​​may be characterized by including a first obesity prediction value generated based on the weight and height information of the subject being evaluated, and a second obesity prediction value generated based on information including the growth stage information of the subject being evaluated.

[0013] The body information of the subject being evaluated may be characterized by being received as input, respectively, to generate the first obesity prediction value and the second obesity prediction value.

[0014] The above second obesity prediction value may be characterized by being generated based on information that further includes the weight and height information of the subject being evaluated.

[0015] The above second obesity prediction value may be characterized by being generated based on information including the weight and height information of the subject being evaluated, which was input to generate the above first obesity prediction value.

[0016] The solution generation step may be characterized by including the step of receiving the first obesity prediction value and the second obesity prediction value, and the step of comparing the difference between the first obesity prediction value and the second obesity prediction value with a preset standard.

[0017] The solution generation step may further include a first solution generation step for generating a solution for weight loss of the subject being evaluated when the difference between the first obesity prediction value and the second obesity prediction value in the comparison step is less than or equal to a preset value.

[0018] The solution generation step may further include a step of generating a growth graph based on the second obesity prediction value when the difference between the first obesity prediction value and the second obesity prediction value exceeds a preset value in the comparison step.

[0019] The solution generation step may further include a second solution generation step for generating a solution by comparing information on a growth graph generated based on the second obesity prediction value with the input body information of the subject being evaluated.

[0020] The second solution generation step may be characterized by comparing the body information of the subject being evaluated with the growth graph generated based on the second obesity prediction value at the time when the subject's body information is entered.

[0021] The second solution generation step may be characterized by providing a solution for increasing the growth prediction value during the rapid growth stage of the growth stage when the difference between the information on the growth graph generated based on the second obesity prediction value and the input body information of the subject being evaluated exceeds a preset value.

[0022] The second solution generation step may be characterized by providing a solution for increasing the duration of the rapid growth phase during the growth phase when the difference between the information on the growth graph generated based on the second obesity prediction value and the input body information of the subject being evaluated is less than or equal to a preset value.

[0023] The body information of the subject being compared in the second solution generation step may be characterized as the body mass index.

[0024] An exemplary embodiment of the present invention may provide a program stored on a computer-readable recording medium that includes program code for executing an obesity prediction method utilizing the ensemble technique described above.

[0025] An exemplary embodiment of the present invention may provide a computer-readable recording medium comprising program code for executing an obesity prediction method utilizing the ensemble technique described above.

[0026] An exemplary embodiment of the present invention may provide an obesity prediction and solution providing device utilizing an ensemble technique, comprising an input unit for receiving body information of an evaluation subject, an obesity prediction unit for generating a plurality of obesity prediction values ​​based on the body information of the evaluation subject input through the input unit, and a solution generation unit for generating an obesity management solution based on the body information of the evaluation subject when the evaluation subject is obese, wherein the solution generation unit generates different solutions according to the difference between the plurality of obesity prediction values.

[0027] The above obesity prediction unit may be characterized by including a first prediction unit that generates a first obesity prediction value based on the weight and height information of the subject being evaluated, and a second prediction unit that generates a second obesity prediction value based on information including the growth stage information of the subject being evaluated.

[0028] The above obesity prediction unit may be characterized by further including a third prediction unit that generates an obesity prediction value based on the ensemble of the first prediction unit and the second prediction unit.

[0029] The third prediction unit may be characterized by calculating the difference between the first obesity prediction value and the second obesity prediction value.

[0030] The solution generation unit may be characterized by generating different solutions by comparing the difference between the first obesity prediction value and the second obesity prediction value with a preset standard.

[0031] Each of the features of the above-described embodiments may be implemented in combination in other embodiments, provided that such features do not contradict or are not exclusive of other embodiments.

[0032] According to various embodiments of the present invention, accurate obesity can be predicted based on the growth stage and body composition information of children and adolescents by comparing different obesity prediction values.

[0033] In addition, appropriate solutions to prevent obesity can be provided by considering the subject's body composition information and growth stage.

[0034] The effects of the present invention are not limited to those described above, and other unmentioned effects will be clearly recognized by a person skilled in the art from the description below.

[0035] FIG. 1 is a drawing showing an ensemble model according to an exemplary embodiment of the present invention.

[0036] FIGS. 2 and 3 are diagrams showing the configurations of an obesity prediction method and a solution providing device according to an exemplary embodiment of the present invention.

[0037] FIG. 4 is a diagram showing predicted elongation and target elongation at each growth stage according to an exemplary embodiment of the present invention.

[0038] FIG. 5 is a diagram illustrating an obesity prediction method according to an exemplary embodiment of the present invention.

[0039] FIG. 6 is a diagram showing the flow of generating a plurality of obesity prediction values ​​according to an exemplary embodiment of the present invention.

[0040] FIG. 7 is a diagram showing identification information and body composition information of an evaluation subject according to an exemplary embodiment of the present invention.

[0041] FIG. 8 is a diagram illustrating a method for providing a solution according to an exemplary embodiment of the present invention.

[0042] FIGS. 9 to 11 are graphs illustrating a method for predicting obesity and providing a solution for an evaluation subject according to an exemplary embodiment of the present invention.

[0043] FIG. 12 is a diagram illustrating an obesity prediction method according to an exemplary embodiment of the present invention.

[0044] FIGS. 13 and FIGS. 14 are drawings showing weight changes according to gender according to exemplary embodiments of the present invention.

[0045] FIG. 15 is a diagram showing the configuration of a neural network according to an exemplary embodiment of the present invention.

[0046] FIG. 16 is a diagram showing a first neural network model according to an exemplary embodiment of the present invention.

[0047] FIG. 17 is a diagram showing a second neural network model according to an exemplary embodiment of the present invention.

[0048] Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. The following detailed description is provided to facilitate a comprehensive understanding of the methods, apparatuses, and / or systems described herein. However, this is merely illustrative and the present invention is not limited thereto.

[0049] In describing the embodiments of the present invention, detailed descriptions of known technologies related to the present invention are omitted if it is determined that such descriptions may unnecessarily obscure the essence of the invention. Furthermore, the terms described below are defined considering their functions in the present invention, and these definitions may vary depending on the intentions or practices of the user or operator. Therefore, their definitions should be based on the content throughout this specification.

[0050] The terms used in the detailed description are merely for describing embodiments of the invention and should not be limiting in any way. Unless explicitly stated otherwise, expressions in the singular form include the meaning of the plural form.

[0051] In this description, expressions such as “include” or “equipped” are intended to refer to certain characteristics, numbers, steps, actions, elements, parts or combinations thereof, and should not be interpreted to exclude the existence or possibility of one or more other characteristics, numbers, steps, actions, elements, parts or combinations thereof other than those described.

[0052] In addition, terms such as first, second, A, B, (a), (b), etc. may be used when describing the components of the embodiments of the present invention. These terms are used merely to distinguish the components from other components, and the essence, order, or sequence of the components is not limited by the terms.

[0053] In the exemplary embodiments of the present invention, children and adolescents may be understood as concepts encompassing the period of human growth. More specifically, children and adolescents, referring to the subjects of evaluation in the exemplary embodiments of the present invention, are defined below through the meanings of infant, child, adolescent, toddler, and child.

[0054] Infants are a continuation of the neonatal period, a period during which they grow by sucking on their mother's nipple until two years after birth, and the experiences of nutrition, caress, and excretion during this period influence their general tendencies thereafter. Children are generally defined as individuals in the age group of 6 years or older but under 13 years, and in a broader sense, they may include toddlers (ages 1 to 5).

[0055] Adolescents refer to the intermediate stage between childhood and youth, generally meaning individuals aged 13 to under 19 years based on full age. Infants may refer to individuals from one year of age to five years of full age. Children may generally refer to children up to the age of 15.

[0056] Accordingly, the term "children and adolescents" referring to the subjects of evaluation mentioned above may mean, in a narrow sense, the period from age 5 to age 19, including children and adolescents, and in a broad sense, the period that includes all general periods of physical growth, from infancy to adolescence.

[0057]

[0058] FIG. 1 is a drawing showing an ensemble model according to an exemplary embodiment of the present invention.

[0059] The following explanation refers to Fig. 1.

[0060] The first obesity prediction model (161) is a first model for predicting obesity and can operate primarily based on quantitative data such as body measurement data (e.g., weight, height, BMI). The first obesity prediction model (161) predicts the likelihood of obesity using algorithms such as linear regression or a decision tree, and can finally output a quantitative value in the form of a first predicted value.

[0061] The second obesity prediction model (162) can be defined as a model that performs sophisticated predictions by reflecting complex data compared to the first obesity prediction model (161). The second obesity prediction model (162) utilizes unstructured and multivariate data such as lifestyle (dietary habits, amount of exercise) and genetic information, and can be trained through non-linear algorithms such as Random Forest and SVM (Support Vector Machine).

[0062] The second obesity prediction model (162) predicts whether or not a person is obese from a different perspective than the first obesity prediction model (161) and finally outputs a second prediction value. The second prediction value can complementarily evaluate the risk of obesity through additional data analysis.

[0063] The third obesity prediction model (163) can be defined as a model that performs more sophisticated obesity prediction by utilizing deep learning technology. The third obesity prediction model (163) can use changes in health status over time (time series data) and data collected from wearable devices as input data.

[0064] In particular, deep learning models such as LSTM (Long Short-Term Memory Network) or CNN (Convolutional Neural Network) are applied, which allows for the analysis of even complex and non-linear patterns. The third obesity prediction model (163) ultimately generates a third prediction value, which can predict obesity risk with higher accuracy based on advanced data analysis.

[0065] The ensemble model (18) can perform the role of generating an optimal prediction result by combining the prediction values ​​derived from the first, second, and third obesity prediction models (161, 162, 163). The ensemble model (18) can receive each prediction value as input and derive a final prediction value through various ensemble techniques.

[0066] The ensemble method of this embodiment may apply a method of assigning weights in proportion to the reliability of each model through a weighted average method, a method of combining using a meta-model (logistic regression or neural network) through a stacking method, or a method of combining predicted values ​​based on the principle of majority rule through a voting-based method, and finally, the ensemble model outputs a final obesity predicted value based on comprehensive analysis results, and this value can be transmitted to the solution generation unit (19).

[0067] The solution generation unit (19) is configured to generate a customized solution to prevent obesity based on the final obesity prediction value derived from the ensemble model (18). The solution generation unit (19) can analyze the final prediction value to evaluate the obesity risk and then design a personalized solution.

[0068] The above solution may include, for example, a solution regarding the subject's nutrition, a solution regarding exercise, a solution regarding habit improvement, a solution from an observational perspective, etc.

[0069] The above-mentioned nutrition solution can provide an optimized diet and nutritional intake plan based on an individual's obesity risk and dietary data, the exercise solution can design an exercise program tailored to an individual's physical condition and lifestyle patterns, the habit improvement solution can suggest lifestyle improvement measures such as sleep management and stress management, and the observational solution can improve the solution by continuously collecting data from the subject and evaluating the effects after applying at least one of the above-mentioned solutions.

[0070] The process of providing obesity prediction and solutions through the aforementioned configurations is explained exemplarily as follows.

[0071] First, when physical data, lifestyle habits, genetic information, and time series data are received from the subject being evaluated, the first obesity prediction model (161), the second obesity prediction model (162), and the third obesity prediction model (163) can independently generate the first prediction value, the second prediction value, and the third prediction value by utilizing their respective data.

[0072] And, the first to third predicted values ​​derived above are input into an ensemble model (18), and the ensemble model can output an optimized final obesity predicted value through various ensemble techniques. The final predicted value is transmitted to a solution generation unit (19), and the solution generation unit (19) can provide a personalized obesity prevention solution based on this.

[0073]

[0074] FIGS. 2 and 3 are diagrams showing the configurations of an obesity prediction method and a solution providing device according to an exemplary embodiment of the present invention, and FIG. 4 is a diagram showing predicted height and target height at different growth stages according to an exemplary embodiment of the present invention.

[0075] The following explanation will be explained with reference to FIGS. 2 to 4.

[0076] An obesity prediction and solution provision device (10) utilizing an ensemble model according to an exemplary embodiment of the present invention may include an input unit (11), a display unit (12), an obesity prediction unit (13), a processor (14), a gender determination unit (15), a growth stage determination unit (17), and a solution generation unit (19). The obesity prediction unit (13) may include a first prediction unit (131), a second prediction unit (132), and a third prediction unit (133). The growth stage determination unit (17) may include a growth stage classification unit (171) and a body information extraction unit (172).

[0077] Through the input unit (10), the obesity prediction and solution provision device (10) utilizing an ensemble model can receive time-series body information of the subject being evaluated. The body information of the subject being evaluated may include not only basic information such as grade (or age), gender, and height, but also additional information such as 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 body information is merely an example to aid in understanding the present invention, and the present embodiment is not limited thereto; it is understood that the types of information constituting the body information can be varied according to the embodiment.

[0078] The time-series physical information of the subject being evaluated may be continuous information or discontinuous information. That is, the physical information of the subject being evaluated may be information included in at least one of the periods corresponding to the growth stages of FIG. 12.

[0079] More specifically, the time-series physical information of the subjects being evaluated varies in terms of collection timing and frequency. For example, in the case of the first subject, physical information measured up to 8 to 12 years of age, which is part of the childhood and adolescent period, exists, and in the case of the second subject, physical information measured irregularly, such as at ages 8, 10 to 12, and 15, exists. Additionally, in the case of the third subject, physical information measured multiple times within a certain period (within the period of any one of the multiple growth stages in Fig. 12), whereas in the case of the fourth subject, physical information measured only once within a certain period (within the period of any one of the multiple growth stages in Fig. 12).

[0080] As described above, depending on the collection time and frequency, the physical information of the subject being evaluated may be included in two or more of the growth stages of FIG. 12 (the first subject being evaluated, the second subject being evaluated), but may not be (the third subject being evaluated, the fourth subject being evaluated).

[0081] In cases where there is body information that has been measured multiple times within a period of one of the multiple growth stages, such as the third evaluation subject mentioned above, the growth stage determination unit (17) can classify the growth stage corresponding to the body information of the third evaluation subject through the growth stage classification unit (171) and extract body information through the body information extraction unit (172), and the body information extracted here may generally include all of the body information of the evaluation subject input through the input unit (11).

[0082] However, in the case where the body information of the subject being evaluated corresponds to only one of multiple growth stages, such as the fourth subject being evaluated, and the body information is measured only once within that period, additional corresponding body information may be generated within any period before the body information of the subject being evaluated is entered into the growth stage determination unit (17).

[0083] More specifically, time-series body information of a subject can be generated within an arbitrary period based on the input body information of the subject and the time-series body growth information of multiple previously stored sample subjects.

[0084] For example, body information can be generated based on a distribution model (similarity) between the body information of an input subject and the time-series body growth information of multiple previously stored sample subjects, or body information can be generated based on a Bayesian inference model (conditional probability).

[0085] The growth stage determination unit (171) can classify the growth stage into one of the multiple growth stages based on the body information of the subject being evaluated input through the input unit (11) in the growth stage classification unit (171) and extract body information corresponding to the classified growth stage in the body information extraction unit (172).

[0086] Referring to FIG. 4, the growth stages may include a general growth period (301), a rapid growth period (303), a decelerated growth period (305), and a growth plate-free period (307).

[0087] Each growth stage can be classified according to the degree of growth, and the height gained each year varies depending on the stage; even within the same growth stage, the actual height gained can differ depending on the growth type.

[0088] The general growth period (301) usually refers to the period before puberty when secondary sexual characteristics appear. Children and adolescents in this period generally have open growth plates, and depending on the growth environment, they generally grow 4 to 5 cm per year in the case of short growth type, and grow 6 to 7 cm per year in the case of tall growth type.

[0089] The rapid growth period (303) is the time when secondary sexual characteristics begin to appear. In women, breasts swell and buds form, and in men, testicles enlarge and pubic hair begins to grow, and voice changes occur. The rapid growth period (303) generally lasts for about 2 to 3 years after the general growth period (301) and grows in the range of about 7 to 10 cm per year on average.

[0090] The deceleration growth period (305) refers to the period when secondary sexual characteristics are being completed. During this period, women can be distinguished by the onset of menstruation, while men can clearly see the changes through pubic hair, voice change, and underarm hair. When the deceleration growth period (305) begins, the rate of growth drops rapidly compared to the rapid growth period (303). It generally lasts for about 2 to 3 years, and natural growth stops as growth averages within a range of about 5 to 6 cm per year. The growth plates begin to close gradually after the rapid growth period (304), and about 50% close after about 6 months have passed since entering the deceleration growth period (305).

[0091] The growth plate period (307) is a time when the growth period has not completely ended, but natural height growth has become difficult, and it refers to the time when the growth plates are closed. Generally, for women, the growth plate period (307) is entered about 1 year and 6 months to 2 years after the onset of menstruation, and for men, the growth plate period (307) is entered about 1 year and 6 months to 2 years after the time when body hair grows in the armpits. During the growth plate period (307), the growth plates close and natural growth stops, but growth of about 1 to 3 cm may be possible by changing bad lifestyle habits and improving physical function through customized exercise, posture correction, and nutrient intake.

[0092] In Figure 4, the x-axis represents age in months, and the y-axis represents height (cm). The lower dotted line (P) represents the predicted growth rate of the subject being evaluated, and the upper solid line (G) represents the target growth rate of the subject being evaluated.

[0093] Obesity is not merely an increase in body weight, but a disease accompanied by overweight resulting from the excessive accumulation of fat tissue in the body or metabolic disorders caused by it. Medically, childhood and adolescent obesity usually refers to cases where a person's weight is 20% or more above the standard weight for their height in the age group from infancy to puberty.

[0094] In most cases, obesity in infancy disappears after the first birthday as the child's movement and activity become more vigorous. However, in some cases, obesity persists, and it is not uncommon for it to recur during the school-age years after returning to a normal state.

[0095] 75% to 80% of childhood and adolescent obesity progresses into adult obesity. Furthermore, it inhibits the secretion of growth hormone, which, particularly in girls, accelerates puberty and shortens the period of potential growth, thereby hindering growth or causing precocious puberty.

[0096] Therefore, for school-aged children and adolescents who are prone to obesity, it is necessary to provide systematic solutions to predict obesity and prevent it if it is predicted.

[0097] Meanwhile, childhood and adolescent obesity can be divided into simple obesity, for which no clear cause is identified, and syndromic obesity, which is caused by specific underlying diseases, and more than 99% of childhood obesity is simple obesity. Children with simple obesity are generally characterized by both boys and girls having average or slightly taller height compared to their peers (multiple sample subjects) during the normal growth period (301), but after the rapid growth period (303), they tend to be shorter or have lower growth rates compared to their peers (multiple sample subjects).

[0098] In summary, childhood and adolescent obesity can be understood as a group of diseases accompanied by overweight or metabolic disorders caused by a wide variety of causes. Generally, both boys and girls tend to have average or slightly higher growth rates compared to multiple sample subjects of the same age group during the normal growth period (301), but after the rapid growth period (303), they tend to have lower growth rates compared to multiple sample subjects of the same age group.

[0099] Accordingly, in order to predict obesity more accurately and generate a solution, the present embodiment classifies the gender through the gender determination unit (15) based on the body information of the subject being evaluated input through the input unit (11), classifies the growth stage through the growth stage determination unit (17) based on the classified gender, extracts the body information, and generates a solution through the solution generation unit (19) by considering the gender and growth stage of the subject being evaluated.

[0100] The obesity prediction unit (13) can be implemented as an artificial intelligence with a Recursive Neural Network (RNN) structure that can utilize not only current values ​​but also time-series values ​​as a type of prediction model. For example, the prediction model can be implemented with an architecture such as a Long Short Term Memory (LSTM) or Gated Recurrent Units (GRU), which are recurrent neural networks. Of course, various other conventional artificial intelligence architectures can also be applied to the prediction model of this embodiment, and as described above, each prediction value can be derived through a plurality of prediction units (131, 132, 133), and based on the derived prediction values, a solution can be generated through a processor (14) and a solution generation unit (19) and displayed on a display unit (12).

[0101] The solution generation unit (19) can generate a growth management solution based on the physical information of the subject being evaluated, corresponding to the classified growth stage.

[0102] More specifically, if the subject being evaluated is in a general growth stage (301), a solution for increasing the predicted growth value of the subject being evaluated can be provided. The predicted growth value is a value corresponding to the y-axis in FIG. 4, and the solution for increasing the predicted growth value can provide various solutions for increasing the target value of the expected y-axis to the subject being evaluated through the display unit (12).

[0103] In addition, the solution generation unit (19) may generate different solutions depending on the situation by comparing multiple predicted values ​​derived from the obesity prediction unit (13). This will be explained in more detail through the obesity prediction method described later.

[0104] Meanwhile, examples of solutions provided through the display unit (12) may include current height, predicted height, degree of obesity, body fat mass, skeletal muscle mass, protein mass, mineral mass, amount of sleep, amount of exercise, nutritional information, lifestyle habits, posture, etc. Each indicator may be displayed in stages such as caution, normal, good, etc. based on a preset range, or may be displayed as a level.

[0105] Additionally, the current status, customized solutions, and precautions for each indicator can be displayed. The current status can be displayed in steps or levels based on the target value, and the customized solutions can include details on adjusting protein, mineral content, body fat, body water, soft lean mass, fat-free mass, bone tissue, skeletal muscle mass, Body Mass Index (BMI), and basal metabolic rate to reach the current target value based on the entered body information.

[0106] In the case of precautions, they may include information on adjusting currently deficient protein, mineral content, body fat, body water, muscle mass (soft lean mass), body fat mass (fat-free mass), bone tissue, skeletal muscle mass, Body Mass Index (BMI), basal metabolic rate, etc., based on the entered body information.

[0107] In addition, if the subject being evaluated is in a rapid growth stage (303), a solution for increasing the growth prediction value of the subject being evaluated in the rapid growth stage (303) can be provided. The growth prediction value is a value corresponding to the y-axis in FIG. 4, and the solution for increasing the growth prediction value can provide various solutions for increasing the target value of the expected y-axis to the subject being evaluated through the display unit (12).

[0108] With reference to FIG. 3, the connection relationships of the components of the obesity prediction and solution provision device (10) utilizing an ensemble model according to an exemplary embodiment of the present invention will be explained below.

[0109] The first prediction unit (131) can generate a first obesity prediction value by receiving weight and height information of the subject being evaluated from the input unit (11), and, for example, the first obesity prediction value can be expressed as a Body Mass Index.

[0110] The second prediction unit (132) can generate a second obesity prediction value by receiving time-series body information of the subject being evaluated (including growth stage information, gender information, body composition information, etc.) from the input unit (11), and, for example, the second obesity prediction value can be expressed as a body mass index. However, it goes without saying that the second obesity prediction value can be expressed not only as the aforementioned body mass index but also as various forms of values ​​based on various input information such as gender, growth stage, and body composition information.

[0111] Meanwhile, the first prediction unit (131) and the second prediction unit (132) may each independently receive the information described above, or the second prediction unit (132) may receive the weight and height information of the subject being evaluated through the first prediction unit (131).

[0112] For example, the first prediction unit (131) may generate a first prediction value by receiving height and weight information of the subject being evaluated from the input unit (11), and the second prediction unit (132) may generate a second prediction value by receiving height and weight information, growth stage information, gender information, and body composition information of the subject being evaluated. As described above, the second prediction unit (132) may generate time-series body information of the subject being evaluated within an arbitrary period based on time-series body growth information of a plurality of sample subjects, based on the information of the subject being evaluated received from the input unit (11).

[0113] For example, the first prediction unit (131) can generate a first prediction value by receiving height and weight information of the subject being evaluated from the input unit (11), and the second prediction unit (132) can generate a second prediction value by receiving the remaining information from the input unit (11), excluding the height and weight information of the subject being evaluated that was entered into the first prediction unit. The second prediction unit (132) may generate time-series physical information of the subject being evaluated within an arbitrary period based on time-series physical growth information of a plurality of sample subjects, based on the information of the subject being evaluated received from the input unit (11) as described above.

[0114] The second prediction unit (132) can classify the gender and growth stage of the subject's body information through the gender determination unit (15) and the growth stage determination unit (17), extract body information corresponding to the classified growth stage, and generate a second prediction value based on this information.

[0115] The third prediction unit (133) can generate a third prediction value based on the first prediction value and the second prediction value generated from the first prediction unit (131) and the second prediction unit (132), and the solution generation unit (19) can generate a solution based on the third prediction value, taking into account the current state and predicted state of the subject being evaluated, and display the generated solution through the display unit (12).

[0116]

[0117] FIG. 5 is a diagram showing an obesity prediction method according to an exemplary embodiment of the present invention, FIG. 6 is a diagram showing a flow for generating a plurality of obesity prediction values ​​according to an exemplary embodiment of the present invention, and FIG. 7 is a diagram showing identification information and body composition information of an evaluation subject according to an exemplary embodiment of the present invention.

[0118] The following explanation refers to FIGS. 5 to 7.

[0119] The obesity prediction and solution provision method (S10) of the present embodiment may include the step of receiving body information of an evaluation subject (S11), the step of generating a first obesity prediction value and a second obesity prediction value (S13), the step of predicting obesity (S15), and the step of generating a solution (S17).

[0120] The above step (S11) is a step of receiving physical information of the subject being evaluated. As described above, the first prediction unit (131) and the second prediction unit (132) may each independently receive physical information of the subject being evaluated. The second prediction unit (132) may receive physical information entered into the first prediction unit (131) and, if necessary, receive the remaining physical information from the input unit (11), or generate time-series physical information of the subject being evaluated within an arbitrary period based on time-series physical growth information of a plurality of sample subjects.

[0121] The above step (S13) is a step of generating multiple obesity prediction values ​​through different neural networks trained on input body information, and can generate a first prediction value and a second prediction value through a first prediction unit (131) and a second prediction unit (132).

[0122] The above step (S15) can predict obesity of the subject being evaluated by comparing and judging (or generating a third prediction value) each prediction value generated in the above step (S13) through the third prediction unit (133). For example, in the above step (S13), the first prediction unit (131) can predict obesity through a body mass index based on the subject's weight and height information, and the second prediction unit (132) can predict obesity through a body mass index based on a height prediction model or a weight prediction model.

[0123] The above step (S17) can generate a solution based on the result determined in the above step (S15).

[0124] First, the generation of the first predicted value is explained.

[0125] An obesity prediction and solution provision device (10) utilizing an ensemble model can receive weight and height information of a subject being evaluated through an input unit (11) (S111), and based on the input weight and height information of the subject being evaluated, a first prediction unit (131) can predict the subject's BMI (S131) ​​and generate a first prediction value (S133). It is preferable that the prediction time point in the above step (S131) ​​be set to a time point after the rapid growth stage among the growth stages. This is because during the rapid growth stage, the growth of children and adolescents occurs within a large range.

[0126] And the generation of the second predicted value is explained below.

[0127] The obesity prediction and solution provision device (10) utilizing an ensemble model can receive information for growth prediction, such as weight, height, gender, age in months, protein amount, and growth stage, through the input unit (11) (S112).

[0128] As described above, weight and height information may be received by inputting information into the first prediction unit (131), and the remaining information may generate time-series physical information of the subject being evaluated within an arbitrary period based on the time-series physical growth information of multiple sample subjects (S113). Furthermore, it goes without saying that gender, identification information, etc., may be received through the input unit (11) as needed.

[0129] Additionally, information necessary for generating a second prediction value can be received (S112) independently of the information input for generating a first prediction value through the input unit (11), and time-series physical information of an evaluation subject corresponding to an arbitrary period can be generated (S113) based on time-series physical growth information of multiple sample subjects for growth stage classification (S115), and other physical information can be received or generated (S116) in accordance with the matrix of the generated information. The information generated in this way is exemplarily shown in FIG. 7.

[0130] The physical information of the subject being evaluated may include information capable of identifying the subject (identification information) and information regarding the subject's biological components.

[0131] For example, the above identification information may include information for identifying a subject, such as grade (or age), gender, and height, and the above biological composition information may include information such as body weight, protein, mineral content, body fat, body water, muscle mass, body fat mass, bone tissue, skeletal muscle mass, Body Mass Index (BMI), basal metabolic rate, neck circumference, chest circumference, abdominal circumference, thigh circumference, arm circumference, hip circumference, etc.

[0132] More specifically, referring to FIG. 7, the subject's identification information (1201) may include date of birth (1201-1), gender (1201-2), age in months (1201-3), and date of measurement (1201-4).

[0133] Additionally, the subject's biological composition information (1101) may include data number (1101-1), height (1101-2), body weight (1101-3), protein mass (1101-4), mineral mass (1101-4), body fat mass (1101-6, Body Fat Mass), muscle mass (1101-7, Soft Lean Mass), bone mineral mass (1101-8, Osseous mineral), and skeletal muscle mass (1101-9).

[0134] Such body information is merely an example to aid in understanding the present invention, and the present embodiment is not limited thereto; it goes without saying that the types of information constituting the body information can be varied according to the embodiment.

[0135] Meanwhile, based on the input information or generated information, an obesity (growth) prediction value can be generated (S135, second prediction value).

[0136] The first prediction value and the second prediction value generated through the first prediction unit (131) and the second prediction unit (132) can derive an ensemble-based obesity prediction result (S151) in the third prediction unit (133), and will be explained in detail below with reference to the drawings.

[0137]

[0138] FIG. 8 is a diagram showing a method for providing a solution according to an exemplary embodiment of the present invention, and FIGS. 9 to 11 are diagrams showing a method for predicting obesity of an evaluation subject and providing a solution according to an exemplary embodiment of the present invention on a graph.

[0139] The following explanation will be explained with reference to FIGS. 8 to 11.

[0140] An obesity prediction and solution providing device (10) utilizing an ensemble model according to an exemplary embodiment of the present invention can compare and determine the first prediction value and the second prediction value generated in FIG. 6 above through a third prediction unit (133), and generate different solutions according to the result of the comparison between the second prediction value and the second prediction value through a solution generating unit (19).

[0141] More specifically, the obesity prediction and solution providing device (10) utilizing an ensemble model receives the generated first prediction value and second prediction value as input (S152), and can compare the difference between the first prediction value and the second prediction value with a preset value (S153).

[0142] If the difference between the first predicted value and the second predicted value is less than or equal to a preset value (S152: Yes), the first solution can be generated (S171). For example, if the difference between the first predicted value and the second predicted value is less than or equal to a preset value (S152: Yes), it can be considered that the difference between the first predicted value and the second predicted value falls within the error range, so the subject being evaluated may be in a case where there is a very high probability of becoming obese at the time of prediction (e.g., at some point after the rapid growth stage).

[0143] Accordingly, the first solution generated in the above step (S171) can generate a solution for improving factors directly related to obesity, such as improving dietary habits, promoting physical activity, and improving lifestyle habits, for the subject being evaluated.

[0144] If the difference between the first predicted value and the second predicted value exceeds a preset value (S152: No), the third prediction unit (133) or the processor (14) can generate a growth prediction graph based on the second predicted value (S154), and generate a solution by comparing the current body information and the predicted body information based on a preset time point (S155).

[0145] If the difference between the first and second predicted values ​​exceeds a preset value (S152: No), a judgment can be made based on the second predicted value, which takes into account various factors, in order to increase the accuracy of the obesity prediction rate.

[0146] For example, referring to FIG. 9, the body mass index (m1) of the subject measured at the time of evaluation, the first predicted value (p1), and the second predicted value (p2) can be generated as age and body mass index. As described above, if the difference (p1-p2) between the first predicted value and the second predicted value exceeds a preset value, a growth prediction curve (gc) based on the second predicted value (p2) can be generated (S154) as shown in FIG. 10.

[0147] The growth prediction curve (gc) may be generated based on time-series physical information of multiple sample subjects, or may be generated based on similar information extracted based on the physical information of the subject being evaluated (S154).

[0148] Meanwhile, the third prediction unit (133) or processor (14) may derive an arbitrary prediction value (p21) at the point where the generated growth prediction curve (gc) intersects with a preset time point, and compare the derived arbitrary prediction value (p21) with the body information of the subject being evaluated based on the preset time point (S155). The arbitrary prediction value (p21) may represent the body mass index at the preset time point on the growth prediction curve (gc) generated based on the second prediction value (p2).

[0149] The previously set point in time above can be set as an evaluation point in time, for example.

[0150] In addition, the aforementioned pre-set point in time may be set as the boundary point between the normal growth stage and the rapid growth stage. In this case, the measured body mass index (m1) of the subject being evaluated can be used to generate a growth prediction curve (not shown) based on a growth graph of children and adolescents or the body mass index (m1) at the time of evaluation, thereby allowing for the comparison and determination of arbitrary predicted values ​​based on the aforementioned pre-set point in time.

[0151] If the difference between an arbitrary predicted value (p21) and the physical information (m1) of the subject being evaluated is less than or equal to a preset value based on the preset time point (S156: Yes), the solution generation unit (19) can generate a second solution (S172), and if the difference between an arbitrary predicted value (p21) and the physical information (m1) of the subject being evaluated exceeds a preset value (S156: No), the solution generation unit (19) can generate a third solution (S173).

[0152] More specifically, if the difference between an arbitrary predicted value (p21) and the subject's physical information (m1) based on the above-mentioned preset time point is less than or equal to a preset value (S156: Yes), the subject is likely to grow showing a trend similar to the growth prediction curve (gc) generated based on the second predicted value (p2).

[0153] Accordingly, the second solution generated (S172) through the solution generation unit (19) may include content for increasing the duration of the rapid growth stage during the growth stage. That is, various solutions for expanding the range of the x-axis corresponding to the rapid growth stage (303) in FIG. 4 may be generated and provided to the subject of evaluation through the display unit (12).

[0154] If the difference between an arbitrary predicted value (p21) and the subject's physical information (m1) based on the above-mentioned preset point in time exceeds a preset value (S156: No), the subject is highly likely to grow exhibiting characteristics different from the growth prediction curve (gc) generated based on the second predicted value (p2). These characteristics may refer to a growth peak appearing during the rapid growth phase, and if the difference between an arbitrary predicted value (p21) and the subject's physical information (m1) based on the above-mentioned preset point in time exceeds a preset value (S156: No), a difference in the growth peak may occur.

[0155] Accordingly, the third solution generated (S173) through the solution generation unit (19) may include content for increasing the growth prediction value (growth peak) during the rapid growth stage of the growth stage. That is, various solutions for increasing the target value of the y-axis corresponding to the rapid growth stage (303) in FIG. 4 may be generated and provided to the subject of evaluation through the display unit (12).

[0156]

[0157] FIG. 12 is a diagram showing an obesity prediction method according to an exemplary embodiment of the present invention, FIG. 13 is a diagram showing weight change according to gender according to an exemplary embodiment of the present invention, and FIG. 14 is a diagram showing the weight change rate according to gender for the same evaluation subject as FIG. 13.

[0158] The following explanation will be explained with reference to FIGS. 12 to 14.

[0159] The obesity prediction and solution provision method (S20) of the present embodiment may include the steps of inputting a prediction value (S21), determining gender (S23), setting an obesity analysis point (S25), predicting obesity (S27), and generating a solution (S29).

[0160] In this embodiment, the first and second predicted values ​​generated in FIG. 6 described above are received through the third prediction unit (133), or the second predicted value is received (S21), and an obesity analysis point according to the gender of the subject being evaluated is set to predict obesity and generate a solution.

[0161] More specifically, in FIGS. 13 and 14, the curve represented by the dotted line represents the growth curve (c1) of a boy, and the curve represented by the solid line represents the growth curve (c2) of a girl. It can be seen that both the boy and the girl gain weight rapidly during the rapid growth phase (303).

[0162] However, the entry point into the rapid growth stage (303) is different for boys and girls. Referring to FIG. 14, the rapid growth stage begins at time x1 for girls and at time x2 for boys.

[0163] Accordingly, the obesity prediction and solution provision method (S20) of the present embodiment may determine the gender of the subject being evaluated (S23) and set an obesity analysis point (S25), and in step (S25), the analysis point may be set to the point (x1, x2) at which the rapid growth phase expressed in FIG. 14 begins. In this case, as described above in FIG. 8, the first predicted value and the second predicted value may be compared and judged to set an obesity analysis point (S25) under a pre-set standard and perform obesity prediction (S27), or only the second predicted value may be input, and then obesity may be predicted (S27) by comparing and judging the current subject's body information with the second predicted value as described above in FIG. 8.

[0164] The comparison judgment between the physical information of the subject being evaluated and the second predicted value can be made by generating growth prediction curves based on the measured physical information of the subject being evaluated and the second predicted value, as described in detail in FIG. 8, and then comparing them at different reference points according to gender to generate a solution (S29). The growth prediction curve generated based on the physical information of the subject being evaluated may utilize a growth graph of children and adolescents, or may utilize a method similar to the growth prediction curve based on the second predicted value.

[0165] And the solution generation unit (19) can generate a solution (S29) based on the result of a comparison judgment between the physical information of the subject being evaluated and the second predicted value, and it goes without saying that a different solution based on the result of the comparison judgment can be generated through a method similar to the method described above.

[0166]

[0167] FIG. 15 is a diagram showing the configuration of a neural network according to an exemplary embodiment of the present invention, FIG. 16 is a diagram showing a first neural network model according to an exemplary embodiment of the present invention, and FIG. 17 is a diagram showing a second neural network model according to an exemplary embodiment of the present invention.

[0168] The following explanation will be explained with reference to FIGS. 15 to 17.

[0169] An exemplary embodiment of the present invention may include a first model (50) and a second model (3), and may construct a pipeline in which at least a portion of the output of the second model (13) is input to the first model (50).

[0170] More specifically, the first model (50) is a model that learns body information corresponding to at least one of a plurality of growth stages as learning data based on time-series body information of a plurality of sample subjects.

[0171] The above growth stage may adopt a rapid growth stage (303), which is the point at which growth retardation due to obesity occurs, but as described above, it is obvious that one or more of the multiple growth stages may be adopted to predict obesity more accurately.

[0172] The first model (50) is equipped with an LSTM neural network (50-1, 50-2, 50-3, 50-4) for time-series data learning and trains the LSTM neural network (50-1, 50-2, 50-3, 50-4) using past physical information of a plurality of sample subjects. Then, the physical information of the current subject is input into the trained LSTM neural network (50-1, 50-2, 50-3, 50-4) to output a predicted growth rate and a solution considering the growth rate.

[0173] The physical information of the above subject may refer to refined data that has undergone an error judgment process through a preprocessing unit.

[0174] An LSTM neural network (50-1, 50-2, 50-3, 50-4) is trained using at least one of the physical information of multiple sample subjects as a default value. For example, regarding height, it is trained using yearly height information for an arbitrary period or a specific growth stage, and the next year is predicted and compared with actual data. Through such comparison, the training set is trained by moving forward in units of an arbitrary period or a specific growth stage.

[0175] Additionally, the LSTM neural network (50-1, 50-2, 50-3, 50-4) may be trained for each growth stage. Thus, it can be trained with past body information of the corresponding growth stage for each of the normal growth stage (301), rapid growth stage (303), deceleration growth stage (305), and non-growth stage (307).

[0176] The physical information of the subject for the above learning may refer to refined data that has undergone an error judgment process through a preprocessing unit.

[0177] Meanwhile, as an example, in this embodiment, time-series physical information of multiple sample subjects is input sequentially according to age or an arbitrary period as training data, and the calculation result of the predicted value for a past point in time or growth level can be transmitted to the prediction of growth level at the next age or arbitrary period.

[0178] Therefore, the LSTM neural network (50-1, 50-2, 50-3, 50-4) can learn not only the growth prediction based on current body information but also the extent to which prediction results for various indicators from past time points (50-1, 50-2, 50-3, 50-4) influence the current growth prediction, and through this, it is possible to extract items among the indicators that have a significant influence on the change in growth according to age or an arbitrary period and reflect them in the growth prediction.

[0179] Furthermore, for time-series learning, it is necessary to obtain physical information over a regular period for multiple sample subjects. However, as mentioned above, since it may be difficult to regularly acquire physical information for multiple sample subjects based on age or arbitrary periods, the data may be used after being temporally normalized by removing outliers or non-continuous physical information per unit period.

[0180] Meanwhile, the second model (13) can derive bone maturity (age) from a hand bone image using a convolutional neural network trained with bone maturity information of the subject as training data.

[0181] More specifically, the convolutional neural network includes multiple convolution layers that generate feature maps of features within the analysis target image among the hand bone images, and pooling layers in which sub-sampling is performed between the multiple convolution layers, thereby allowing different levels of features for the analysis target area to be extracted, and the features can be probabilistically inferred through an activation function, or bone maturity can be derived through weight learning between nodes via regression analysis.

[0182] The bone maturity extracted through the second model (13) is input into the LSTM neural network (50-1, 50-2, 50-3, 50-4) along with at least some of the subject's body information, which can increase the accuracy of predicting the subject's growth, thereby increasing the accuracy of predicting obesity.

[0183]

[0184] The present invention has been described so far with reference to its preferred embodiments. All embodiments and conditional examples disclosed herein are intended to help readers of the art who have ordinary knowledge in the technical field of the present invention to understand the principles and concepts of the present invention, and those of the art will understand that the present invention may be implemented in modified forms without departing from the essential characteristics of the present invention.

[0185] Therefore, the disclosed embodiments should be considered in an illustrative rather than a limiting sense. The scope of the invention is defined by the claims, not by the foregoing description, and all variations within the scope of the claims should be interpreted as being included in the invention.

[0186] Meanwhile, the method according to the various embodiments of the present invention described above may be implemented as a program and provided to servers or devices. Accordingly, each device may connect to a server or device where the program is stored and download the program.

[0187] In addition, the method according to the various embodiments of the present invention described above may be implemented as a program and provided by being stored on various non-transitory computer-readable media. A non-transitory computer-readable medium refers to a medium that stores data semi-permanently and is readable by a device, rather than a medium that stores data for a short period of time, such as a register, cache, or memory. Specifically, the various applications or programs described above may be provided by being stored on non-transitory computer-readable media such as CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, ROMs, etc.

[0188] Furthermore, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention.

Claims

Step of receiving the physical information of the subject of evaluation; A step of generating multiple obesity prediction values ​​through different neural networks trained on input body information; A step of predicting obesity by comparing the above plurality of obesity prediction values; and A method for predicting obesity using an ensemble technique, comprising the step of generating different solutions based on the difference between the plurality of obesity prediction values. In paragraph 1, The above plurality of obesity prediction values ​​are, A first obesity prediction value generated based on the weight and height information of the subject being evaluated; and A method for predicting obesity using an ensemble technique, characterized by including a second obesity prediction value generated based on information including growth stage information of the subject being evaluated. In paragraph 2, The physical information of the above-mentioned evaluation subject is, Obesity prediction method using an ensemble technique, characterized by receiving inputs to generate the first obesity prediction value and the second obesity prediction value, respectively. In paragraph 2, The above second obesity prediction value is, Obesity prediction method using an ensemble technique, characterized by being generated based on information that further includes the weight and height information of the subject being evaluated. In paragraph 4, The above second obesity prediction value is, Obesity prediction method using an ensemble technique, characterized by being generated based on information including weight and height information of the subject being evaluated, which is input to generate the first obesity prediction value. In paragraph 2, The above solution generation step is, A step of receiving the first obesity prediction value and the second obesity prediction value; and A method for predicting obesity using an ensemble technique, characterized by including the step of comparing the difference between the first obesity prediction value and the second obesity prediction value with a preset standard. In paragraph 6, The above solution generation step is, A method for predicting obesity using an ensemble technique, further comprising: a first solution generation step for generating a solution for weight loss of the subject being evaluated when, in the comparison step above, the difference between the first obesity prediction value and the second obesity prediction value is less than or equal to a preset value. In paragraph 6, The above solution generation step is, A method for predicting obesity using an ensemble technique, further comprising the step of generating a growth graph based on the second obesity prediction value when the difference between the first obesity prediction value and the second obesity prediction value exceeds a preset value in the comparison step above. In paragraph 8, The above solution generation step is, A method for predicting obesity using an ensemble technique, further comprising: a second solution generation step of generating a solution by comparing information on a growth graph generated based on the second obesity prediction value with the body information of the input subject. In Paragraph 9, The above second solution generation step is, Obesity prediction method using an ensemble technique, characterized by comparing a growth graph generated based on the second obesity prediction value at the time when the body information of the subject is entered with the body information of the subject. In Paragraph 10, The above second solution generation step is, Obesity prediction method using an ensemble technique, characterized by providing a solution for increasing the growth prediction value during the rapid growth stage among growth stages when the difference between the information on the growth graph generated based on the second obesity prediction value and the body information of the input subject exceeds a preset value. In Paragraph 10, The above second solution generation step is, A method for predicting obesity using an ensemble technique, characterized by providing a solution to increase the duration of the rapid growth phase during the growth phase when the difference between the information on the growth graph generated based on the second obesity prediction value and the body information of the input subject is less than or equal to a preset value. In Paragraph 9, A method for predicting obesity using an ensemble technique, characterized in that the body information of the subject being compared in the second solution generation step is the body mass index. A program stored on a computer-readable recording medium comprising program code for executing an obesity prediction method utilizing an ensemble technique described in any one of claims 1 to 13. A computer-readable recording medium comprising program code for executing an obesity prediction method utilizing an ensemble technique described in any one of claims 1 to 13. Input unit for receiving physical information about the subject of evaluation; An obesity prediction unit that generates multiple obesity prediction values ​​based on the body information of the subject being evaluated input through the above input unit; and A solution generation unit that generates an obesity management solution based on the body information of the subject being evaluated when the subject being evaluated is obese; The above solution generation unit is, Obesity prediction and solution provision device utilizing an ensemble technique characterized by generating different solutions based on the difference between the plurality of obesity prediction values. In Paragraph 16, The above obesity prediction unit is, A first prediction unit that generates a first obesity prediction value based on the weight and height information of the subject being evaluated; and Obesity prediction and solution provision device utilizing an ensemble technique, characterized by including: a second prediction unit that generates a second obesity prediction value generated based on information including growth stage information of the evaluation subject. In Paragraph 17, The above obesity prediction unit is, Obesity prediction and solution providing device utilizing an ensemble technique, characterized by further including a third prediction unit that generates an obesity prediction value based on the ensemble of the first prediction unit and the second prediction unit. In Paragraph 18, The above third prediction unit is, Obesity prediction and solution providing device utilizing an ensemble technique characterized by calculating the difference between the first obesity prediction value and the second obesity prediction value. In Paragraph 17, The above solution generation unit is, Obesity prediction and solution provision device utilizing an ensemble technique, characterized by generating different solutions by comparing the difference between the first obesity prediction value and the second obesity prediction value with a preset standard.

Images