KR102560491B1 - Method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight - Google Patents

Abstract

A method of providing personalized health lifestyle information through artificial intelligence based on a user's life log in order to induce or maintain a user's appropriate weight is related to a method for managing the user's weight, which is appropriate for each user. In proposing a more realistic, accurate and practicable specific life pattern about the user's weight through the user's lifelog data, the user service app and life It is provided through the log database and health management service server, life log data analysis step (S10), weight status determination step (S20), and life pattern information leading to normal weight in the case of underweight according to the user's weight condition is provided. And, if the user's weight status is obese, lifestyle pattern information leading to overweight is provided, and if the user's weight status is overweight, lifestyle pattern information leading to normal weight is provided, and if the user's weight status is normal, It is characterized in that it includes a step (S30) of providing personalized healthy lifestyle information that provides lifestyle pattern information that can maintain a normal weight, and is composed of a lifestyle pattern that can be sustained based on the user's lifestyle information. There is an effect that the weight of the user is effectively and practically managed by providing.

Description

Method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight to induce or maintain the user's appropriate weight}

The present invention relates to a method for providing information on a user's life pattern to improve a user's lifestyle based on life log data of the user, and in particular, a lifestyle pattern for inducing or maintaining an appropriate weight for each individual user. As a method provided through artificial intelligence, it proposes a more realistic, accurate, and practicable specific life pattern regarding the user's weight through the user's lifelog data, thereby inducing the user's motivation and behavior, resulting in an appropriate weight due to improved lifestyle. Induction or maintenance can be achieved, and additionally collected lifelogs are gradually learned to further improve the classification performance of the user's life pattern through the advancement of the artificial intelligence model, and more accurate and practicable through various prediction paths. It is about how to provide a lifestyle pattern.

The development of medical technology is changing the way of life, changing the approach, thinking and definition of health, and accordingly, the paradigm of medical service is also changing to focus on prevention, prediction, personalized customization, and participation. Through the development of artificial intelligence technology and the development of healthcare smart devices, the industry in which technology and medical technology are converged is developing significantly around the world, and in particular, the smart healthcare field is attracting attention as an industry that will create new added value.

On the other hand, as storage devices for storing data and sensors for collecting data are more precisely developed and related devices are diversified, more people are producing various types of data.

For example, information produced by a wearable device that is worn on the user's body to measure the user's activity level, nutritional information about the food consumed by the user, and information produced by various measuring devices such as blood pressure monitors, weight scales, and blood glucose meters used by the user Various types of health-related data, such as clinical information, are being produced, which means all information that records and stores users' daily lives. It is a compound word of 'life' meaning daily life and 'log' meaning record. The concept of 'life log' is formed, and especially in the field of smart healthcare, life log is used in various forms.

As an example of using lifelog data, prior art Application No. 10-2020-01143031 "Remote health management system using artificial intelligence based on lifelog data" has been disclosed.

The prior art is a remote health management system using artificial intelligence, which includes a user terminal that collects lifelog data from a user, analyzes the lifelog data to generate life pattern information of the user, and generates life pattern information of the user. A server that provides corresponding health management content, wherein the server performs machine learning based on a data collection unit that receives the lifelog data from the user terminal and previously input learning data, and the lifelog data An artificial intelligence analysis unit that generates life pattern information of the user using a machine learning model using as an input variable, and a content provider that transmits health management contents corresponding to the user's life pattern information to the user terminal. It is characterized by doing.

The prior art uses the user's lifelog for the purpose of real-time monitoring of the user's health condition anytime, anywhere without time and space constraints, and providing related content corresponding to the monitoring result, using the user's lifelog to prevent time and space constraints Real-time monitoring of the user's health status anytime, anywhere, and provision of relevant content in response to the monitoring results, enabling early detection and treatment of user's disease through real-time monitoring and provision of optimized content to the user, thereby improving health It is expected that the efficiency of management can be improved, the manager or person in charge at the work site can monitor the health status of employees in real time, and the effect of prompt response in case of an emergency is expected.

However, the prior art using artificial intelligence based on lifelog data as above, as shown in FIG. 1 schematically illustrating a method according to a conventional personalized health management system, collects users' lifelog data and collects their gender or age. It forms physical clusters based on simple criteria such as the above, and provides health management information such as personalized exercise curriculum information for each individual through statistical analysis of each physical cluster, making it difficult to accurately grasp the lifestyle of each user There is a problem, and due to this problem, there is a problem in that it is not possible to accurately present a life pattern that requires improvement for each user.

In addition, due to these problems, the prior art has a problem in that it does not induce a user's motivation or action, and does not lead a user's steady participation, thereby inducing a change in each user's practical lifestyle.

In other words, the conventional methods of providing health-related information for each user failed to present a specific life pattern to be improved to improve the lifestyle of each user, so that each user had a lifestyle that could maintain a healthy life. There is a problem in that it does not serve as a guide to be formed.

In particular, the prior art as described above is based on very quantitatively limited factors in the ability to analyze the user's life pattern through the user's lifelog data, so that an accurate analysis of how much weight is appropriate for each user is made. As a result, it is not possible to set an accurate goal corresponding to a healthy weight for each user, and accordingly, it is not possible to present an accurate life pattern that can maintain the user's proper weight, so that the user cannot induce a healthy lifestyle. there is

In addition, it is not possible to guide the life pattern so that the user's current weight can gradually change from the user's current weight to the user's appropriate weight, so that the user's weight management, which is the root of chronic disease, is more accurately and actively managed. In preventing or improving users' chronic disease problems due to problems, etc., the user's execution power and continuity are not elicited, and special effects are not exerted.

The present invention method for providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight was invented to solve the above problems of the prior art, A healthy life consisting of lifestyle patterns that allow users to reach an appropriate weight for each user and to maintain various life patterns that require improvement to maintain the reached appropriate weight based on the user's life pattern while continuing at the same time. It is an object of the present invention to provide a method for guiding habit information so that a user's weight eventually becomes a target proper weight and such an appropriate weight can be maintained.

In addition, an object of the present invention is to provide a method capable of preventing and managing chronic diseases through proper weight management of the user.

In addition, an object of the present invention is to provide a method for forming a desirable lifestyle for a user through proper weight management of the user.

In addition, the present invention is to provide a method capable of presenting a desirable life pattern to reach an appropriate weight based on the user's life log data through gradually advanced artificial intelligence by learning through the life pattern of a healthy user. do.

In addition, an object of the present invention is to provide a method capable of more accurately presenting a life pattern that can be practiced by each user by searching for and providing a life pattern of a healthy user similar to the life log data of the user.

Another object of the present invention is to provide a method capable of guiding life patterns through activity variables of lifelog data and improving performance of clustering and classification of life patterns through inactivity variables of lifelog data.

In addition, the present invention easily induces the interpretation of the learning model through the organic combination of the ensemble learning algorithm and the prototype selection algorithm by the transcendental quadrangle, and minimizes the prediction time by relatively reducing the number and depth of the tree to artificial intelligence. The purpose of this study is to provide a method for making the advancement of the user's lifelog data relatively easy and reducing the data processing amount and time for analyzing the user's lifelog data.

In addition, the present invention accurately analyzes the user's life pattern through the user's lifelog data through the number and depth of trees, and presents the life pattern that requires improvement extracted through accurate analysis, thereby increasing the user's reliability in the present invention. It is an object of the present invention to provide a method for improving performance, inducing motivation to participate in the service according to the present invention according to reliability improvement, and maintaining the user's executive power.

In addition, the present invention provides a lifestyle pattern for each user to form a healthy lifestyle that can maintain an appropriate weight in the form of a goal or program that can be practiced by each user, and provides a reward according to whether or not the achievement is achieved, thereby continuously providing lifelog data for each user. The purpose of this is to provide a way to advance the service by inducing it to occur, and through this, the life pattern provided to the user can be relatively specified and diversified.

Furthermore, the present invention suggests that each user practice a specific life pattern that requires improvement for inducing or maintaining an appropriate weight for each user so that a healthy lifestyle is formed, and through this, the user's risk due to chronic disease is detected and Its purpose is to provide a method for predicting various chronic diseases.

Other objects of the present invention can be understood through the characteristics of the present invention, can be more clearly known through the embodiments of the present invention, and can be realized by means and combinations shown in the claims.

The present invention method for providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight has the following technical features in order to achieve the problem to be solved by the present invention have

According to the present invention, a method for providing personalized health lifestyle information through artificial intelligence based on a user's lifelog is installed in a smart terminal, receives lifelog data and transmits it to a health management service server, and the health management service server A user service app (10) that provides users with lifestyle patterns proposed in; a lifelog database 20 that stores lifelog data of a user through the user service app; And converting the collected lifelog data into training data, generating and storing new training data with meaningful data selected through prototype selection, and creating a learning model using the new training data using a tree-based ensemble learning algorithm Personal health lifestyle pattern information is created using the user's daily life log and visualized through the service app, and the performance of the learning model used to calculate life pattern guide information and life pattern information requests are continuously managed. Including, the health care service server 30 provides an environment in which the user can check personalized health life pattern information in real time through the user service app 10, It analyzes the lifelog data generated by the user the previous day at a certain time every day to generate a plurality of pieces of health life pattern information suitable for the user and provide them by category. The machine learning model database (310 ); And, the user's daily lifelog data is received from the lifelog database 20 or the life pattern guide server 330, converted into training data, and machine learning features are extracted from the converted training data to perform the life log data. A machine learning feature generation server 320 that provides machine learning features to the pattern guide server 330; and receives and receives the machine learning features of the daily lifelog data of the user from the machine learning feature generation server 320. The machine learning model corresponding to the machine learning feature is provided from the machine learning model database 310, and health life pattern information corresponding to the user's daily lifelog data is generated and provided to the health management service server 30 Life pattern guide server 330; Through the personalized health lifestyle guide system through artificial intelligence based on the user's life log, characterized in that it further includes, in order to induce or maintain the user's appropriate weight, the user's In the method of providing personalized health lifestyle information through lifelog-based artificial intelligence, the machine learning feature generation server 320 implements a tree-based ensemble learning algorithm based on user's lifelog data Lifelog data analysis step (S10) of analyzing a user's life pattern through an artificial intelligence program; According to the analysis result in the lifelog data analysis step (S10), the life pattern guide server 330 receives a machine learning model from the machine learning model database 310 and determines the weight of the user as underweight, normal, overweight, or obese. Weight status determination step (S20) to determine any one of; And according to the weight condition of the user determined in the weight condition determination step (S20), the life pattern guide server 330 analyzes branch nodes through a tree-based ensemble learning algorithm to provide personalized information according to the user's specialized life pattern. Provides healthy lifestyle information, but if the user is underweight, lifestyle pattern information that leads to normal weight, if the user's weight status is obese, lifestyle pattern information that leads to overweight, and if the user's weight status is overweight, normal weight Personalized healthy lifestyle information providing step (S30) of providing lifestyle pattern information leading to the user's weight, and lifestyle pattern information that can maintain a normal weight when the user's weight is normal, to the health management service server 30; It is characterized by including.

In addition, in the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight, the weight status determination step (S20) is performed by the user It is characterized by including; a predicted path extraction step of extracting a predicted path according to the user's lifelog data through a tree-based ensemble learning algorithm based on the lifelog data generated during the day.

In addition, in the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate body weight of the present invention, the providing of the personalized healthy lifestyle information step (S30 ) is formed by searching for a similar path close to the predicted path predicted in the predicted path extraction step.

In addition, in the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate body weight of the present invention, the providing of the personalized healthy lifestyle information step (S30 ) extracts a branch node that distinguishes the predicted path from the similar path, and determines the user's life according to the variables included in the extracted branch node, the threshold value, the Gini coefficient, the number of predictions for each class, and the depth of the tree where the branch node is located. It is characterized by generating a lifestyle pattern for inducing or maintaining an appropriate weight based on the log data.

delete

delete

In addition, in the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight, the branch node has a value according to the user's behavior. A distinction is made between activity variables that change (e.g. total number of steps, average step speed, optimal bedtime, etc.) and inactivity variables whose values do not change (e.g. gender, age, percentage of REM sleep stages, etc.). In the step of providing customized healthy lifestyle information (S30), a step guide, a sleep guide, and a weight guide are provided based on the activity variables, and the step guide provides the number of steps, step speed, and distance walked for each time period. The sleep guide is provided with recommended bedtime and wake-up time, and the weight guide is provided with calorie consumption, muscle strength increase, water intake, and number of weight measurements.

In addition, in the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog to induce or maintain the user's proper weight, the user's lifelog data includes user information, steps information, sleep information, and weight information, wherein the user information includes information about gender, age, and height, and the step information includes total number of steps, average step speed, total distance walked, total calories consumed, and dawn Time Step count, morning step speed, dawn step distance, dawn step calories burned, morning step count, morning step speed, morning step distance, morning step step calories, lunch step count, lunch break Step speed, distance walked at lunchtime, calories burned at lunchtime, number of steps taken at dinner time, step rate at evening time, distance walked at evening time, calories burned at dinnertime, time walked over 20 minutes, number of steps walked over 20 minutes, total Variability in step count, variability in average step speed, variability in total distance walked, variability in total calories burned, variability in number of steps taken at dawn, variability in step speed at dawn, variability in distance walked at dawn, variability in total calories burned at dawn Variability in step calories at morning time, variability in step speed at morning time, variability in distance walked at morning time, variability in calorie burned at breakfast time, variability in number of steps at lunch time, variability in step speed at lunch time, lunch Include information on the variability of hourly walking distance, the lunchtime variability of walking calories burned, the evening hour step count variability, the evening hour walking speed variability, the evening hour walking distance variability and the evening hour walking calorie burn variability, . Includes information about total sleep time variability, total sleep efficiency variability, awake sleep phase ratio variability, REM sleep phase ratio variability, light sleep phase ratio variability, and deep sleep phase ratio variability, wherein the weight information includes weight It is characterized by including information on the number of measurements, the amount of muscle mass, the amount of water, the amount of body fat, and the number of measurements of body weight.

In addition, in the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate body weight of the present invention, the providing of the personalized healthy lifestyle information step (S30 ) ranks the personalized lifestyle information to be provided to the user based on the depth information of the tree where the branch nodes extracted from the plurality of trees included in the learning model are located, and ranks according to the user's grade or situation. It is characterized by providing personalized lifestyle information.

In addition, in the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate body weight of the present invention, the providing of the personalized healthy lifestyle information step (S30 ) Statisticalizes threshold information of branch nodes extracted from a plurality of trees included in the learning model, and calculates the central, average, maximum, and minimum threshold values for each lifestyle information, and provides them according to the user's obesity level. and adjusting the level of difficulty of personalized health lifestyle information.

In addition, in the present invention method for providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain a user's proper weight, the health management service server collects the user's life The log data is periodically trained on the existing model to gradually increase the classification performance of the learning model, and the user who has performed the healthy lifestyle information provided in the personalized healthy lifestyle information providing step (S30) is added using the lifelog data. Through learning, the variables and values of the branch nodes included in the tree are specified to advance the prediction path that expresses the user's lifestyle, and through additional learning using the lifelog data of users who have not performed the health lifestyle information, new It is characterized by creating a life pattern having a number of cases and diversifying the predicted path.

In addition, in the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight, the prototype selection is performed on the received training data. Using the similarity between data and class information, a plurality of transcendental rectangles including only the same class data and defined by the minimum vertex (min) and maximum vertex (max) that divide the multi-dimensional training data space are constructed, and each transcendental rectangle constructed It is characterized by determining a prototype representing a class in and generating new training data.

In addition, in the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate body weight of the present invention, the providing of the personalized healthy lifestyle information step (S30 ) analyzes the user's lifelog data for the past 7 days and provides a customized mission program to the user, and the customized mission program provides health lifestyle information by mission item (step, sleep, exercise, etc.) on a daily basis. , At this time, the items and target values of the mission are continuously changed and provided;

In addition, in the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate body weight of the present invention, the providing of the personalized healthy lifestyle information step (S30 ) provides in-kind rewards according to the results in order to induce the performance of a healthy lifestyle, which has a relatively high importance in the plurality of healthy lifestyle information calculated from the lifelog data generated by the user, and provides continuous lifelog data of the user. To collect; characterized by.

The present invention method for providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight solves the problems of the prior art through the above technical features, Health consisting of lifestyle patterns that enable users to carry out and continue at the same time based on the user's life pattern by analyzing the life pattern that requires improvement so that the user can reach an appropriate weight and maintain the reached appropriate weight. There is an effect of providing lifestyle information so that the weight of the user eventually becomes a target appropriate weight and such an appropriate weight can be maintained.

In addition, the present invention has the effect of providing a method capable of preventing and managing chronic diseases through proper weight management of the user.

In addition, the present invention has the effect of forming a desirable lifestyle for the user through proper weight management of the user.

In addition, the present invention provides healthy lifestyle information consisting of detailed and accurate life patterns to reach an appropriate weight based on the user's life log data through advanced artificial intelligence learned using the healthy user's life pattern. It works.

In addition, the present invention has an effect of more accurately presenting a life pattern that can be practiced by each user by searching for and providing a healthy user's life pattern similar to the user's life log data.

In addition, the present invention has an effect of guiding life patterns through activity variables of lifelog data and improving performance of clustering and classifying life patterns through inactivity variables of lifelog data.

In addition, the present invention easily induces the interpretation of the learning model through the organic combination of the ensemble learning algorithm and the prototype selection algorithm by the transcendental quadrangle, and minimizes the prediction time by relatively reducing the number and depth of the tree to artificial intelligence. Sophistication of the user can be performed more easily, and there is an effect of relatively reducing the data processing amount and time of analyzing the user's lifelog data.

In addition, the present invention accurately analyzes the user's life pattern through the user's lifelog data through the number and depth of trees, and presents the life pattern that requires improvement extracted through accurate analysis, thereby increasing the user's reliability in the present invention. There is an effect of improving the reliability, inducing motivation to participate in the service according to the present invention according to the improvement of reliability, and maintaining the user's executive power.

In addition, the present invention provides a lifestyle pattern for each user to form a healthy lifestyle that can maintain an appropriate weight in the form of a goal or program that can be practiced by each user, and provides a reward according to whether or not the achievement is achieved, thereby continuously providing lifelog data for each user. , and through this, it is possible to relatively specify and diversify the life patterns provided to users, so that the service can be advanced.

Furthermore, the present invention suggests that each user practice a specific life pattern that requires improvement for inducing or maintaining an appropriate weight for each user so that a healthy lifestyle is formed, and through this, the user's risk due to chronic disease is detected and It has the effect of predicting various chronic diseases.

Other effects of the present invention can be understood through the characteristics of the present invention, can be more clearly seen through the examples of the present invention, and can be exhibited by means and combinations shown in the claims.

1 is a schematic diagram of a method according to a conventional personalized health management system;
2 is a diagram showing conceptual features of a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user according to the present invention;
3 is a diagram conceptually showing the basic configuration of a system in which a method of providing personalized health lifestyle information through artificial intelligence based on a user's life log is implemented in order to induce or maintain an appropriate weight for a user according to the present invention. ,
4 is a health care service of a personalized healthy lifestyle guide system according to a method of providing personalized healthy lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for the user of the present invention. A diagram conceptually showing the characteristics of the server;
5 is a health care service of a personalized healthy lifestyle guide system according to a method of providing personalized healthy lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for the user of the present invention. A drawing showing an embodiment of the server,
6 is a diagram showing a modeling example of a lifelog data analysis step according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate body weight for a user according to the present invention. ,
7 is a view for explaining multi-dimensional variables according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's life log in order to induce or maintain an appropriate weight for a user according to the present invention;
8 is a diagram showing an example of a predicted path according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user according to the present invention;
Figure 9 explains the characteristics of providing personalized healthy lifestyle information according to the method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight according to the present invention. drawing to do,
10 is a view showing an example of a similar path according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user according to the present invention;
11 is a diagram showing an example of branch node analysis according to the method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user according to the present invention;
12 shows the specification and diversification of life patterns by branching nodes according to the method of providing personalized health lifestyle information through artificial intelligence based on the user's life log in order to induce or maintain the user's appropriate body weight according to the present invention. A drawing to explain,
13 is a view for explaining selection of a transcendental rectangle-based prototype according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate body weight for a user according to the present invention. ,
14 is a diagram showing an example of lifestyle pattern guide information according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's life log in order to induce or maintain an appropriate weight for a user according to the present invention;
15 is a diagram showing an example of a service according to the present invention method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain a user's proper body weight.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The detailed description of the present invention which follows refers to the accompanying drawings which illustrate, by way of example, specific embodiments in which the present invention may be practiced. These embodiments are described in sufficient detail to enable any person skilled in the art to practice the present invention. It should be understood that the various embodiments of the present invention are different, but need not be mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the present invention with respect to one embodiment. Also, it should be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the present invention. Accordingly, the detailed description set forth below is not to be taken in a limiting sense, and the scope of the present invention is limited only by the appended claims, along with all equivalents as claimed by those claims. Like reference numbers in the drawings indicate the same or similar function throughout the various aspects.

2, as shown in the diagram showing the conceptual characteristics of a method for providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for the user of the present invention, The method for providing personalized healthy lifestyle information according to the invention extracts complex patterns based on the user's lifelog data, presents important habits through selection of key variables, and explores the lifestyle patterns of healthy users. It is characterized by providing healthy lifestyle information consisting of a lifestyle pattern in which the user's weight can be induced or maintained to an appropriate weight by suggesting a lifestyle pattern that requires improvement of the user.

As a characteristic difference, as shown in FIG. 1, the prior art extracts a selective cluster only with the user's gender and age, and provides information to the user through statistical analysis for each selected cluster. As shown in Fig. 2, complex patterns are extracted using activity variables and inactivity variables such as gender, age, number of steps, step speed, muscle mass, and sleep time, and life patterns of users with healthy lifestyles are explored. Through pattern and comparison, it has the characteristics of analysis of lifestyle suitable for users and provision of healthy lifestyle information.

Hereinafter, a method for providing personalized health lifestyle information according to the present invention having the above technical features will be described in detail through an embodiment.

3 is a diagram conceptually showing the basic configuration of a system in which a method of providing personalized health lifestyle information through artificial intelligence based on a user's life log is implemented in order to induce or maintain an appropriate weight for a user according to the present invention. As shown, a method for providing personalized health lifestyle information according to the present invention through a system including a user service app 10, a lifelog database 20, and a health management service server 30, the health management The lifelog data analysis step (S10) of analyzing the user's life pattern through an artificial intelligence program that implements a tree-based ensemble learning algorithm based on the user's lifelog data in the service server (S10) and the lifelog data analysis step (S10) According to the results analyzed in the weight status determination step (S20) of determining the user's weight as one of underweight, normal, overweight, and obese, and the user's weight status determined in the weight status determination step (S20), if the user is underweight Provide lifestyle pattern information leading to normal weight, provide lifestyle pattern information leading to overweight if the user's weight status is obese, and provide lifestyle pattern information leading to normal weight if the user's weight status is overweight and, when the user's weight is normal, a personalized health lifestyle information provision step (S30) is carried out to provide lifestyle pattern information capable of maintaining a normal weight.

As shown in FIG. 3, the service app 10 is a program installed and driven in a smart terminal 1 such as a smartphone or tablet of the Android operating system or the iOS operating system, and generates various lifelog data. Receives lifelog data from various types of healthcare smart devices (2) such as watches, smart weight scales, smart blood glucose meters, and smart blood pressure meters, and transmits them to the health management service server 30, and the health management service server 30 It plays a role in providing users with the life patterns suggested by

That is, the service app 10 performs user registration processing and registration processing of the healthcare smart device 2 for receiving lifelog data.

For example, after downloading the service app 10 through the smart terminal 1, installing the service app 10 on the smart terminal, driving the installed service app 10, and entering the requested information through the driven service app 10. After that, service user authentication and registration are performed.

The lifelog database 20 stores various lifelog data of users received through the service app 10 .

At this time, the user's lifelog data stored in the lifelog database 20 includes user information, step information, sleep information, and weight information.

The user information includes information about gender, age, and height, and the step information includes the total number of steps, average walking speed, total distance walked, total calories burned, number of steps at dawn, walking speed at dawn, and time at dawn. Distance walked, calories burned at dawn, number of steps at morning time, step speed at morning time, distance walked at morning time, calories burned at morning time, number of steps at lunchtime, step speed at lunchtime, distance walked at lunchtime, lunchtime consumption Calorie steps taken, number of steps during evening time, step speed during evening time, distance walked during evening time, calories burned during evening time, time walked over 20 minutes, number of steps over 20 minutes, variability in total number of steps, variability in average walking speed, Variability in total distance walked, variability in total calories burned, variability in number of steps at dawn, variability in variability in walking speed at dawn, variability in variability in distance walked at dawn, variability in variability in calories burned at dawn, variability in number of steps in morning , variability in walking speed at morning time, variability in distance walked at morning time, variability in calorie burned steps at morning time, variability in number of steps taken at lunchtime, variability in walking speed at lunchtime, variability in distance walked at lunchtime, variability in steps burned at lunchtime It includes information about variability in calories, variability in the number of steps taken at night, variability in walking speed at night, variability in walking distance at night, and variability in walking calories consumed at night, wherein the sleep information includes total sleep time and sleep efficiency , ratio of Awake sleep phase, ratio of REM sleep phase, ratio of Light sleep phase, ratio of Deep sleep phase, overlapping sleep, optimal bedtime, optimal wake-up time, total sleep time variability, total sleep efficiency variability, It includes information on the variability of the Awake sleep phase ratio, the REM sleep phase ratio, the variability of the Light sleep phase ratio, and the variability of the Deep sleep phase ratio, and the weight information includes the number of measurements of weight, muscle mass, water content, body fat mass, and weight measurement. Contains information about the number of times.

As shown in FIG. 3, the health management service server 30 converts the lifelog data collected through the service app 10 and stored in the lifelog database 20 into training data to form a prototype. New training data selected from meaningful data through selection is created and stored, and a learning model is created and stored using the tree-based ensemble learning algorithm.

In addition, by using the daily life log of the user, personalized health life pattern information is created and visualized through the service app 10, and the performance of the learning model used in calculating the life pattern guide information and the life pattern information request are continuously updated. manage with

That is, the health management service server 30 according to the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight according to the method of FIG. 4 according to the present invention As shown in the drawing conceptually showing the characteristics of the health management service server of the healthy lifestyle guide system, search for a lifestyle pattern to induce or maintain the user's weight at an appropriate weight based on the user's daily lifelog data. and has a feature of recommending the user's healthy life pattern corresponding to the search result, and artificial intelligence through activity and inactivity variables stored in the lifelog database 20 described above to search and recommend such a healthy life pattern. A learning model is created, and through the learning model, user-customized life patterns are extracted and analyzed.

At this time, the health care service server 30 is configured according to the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight according to the method of FIG. 5. As shown in (a) of the drawing showing an embodiment of the health management service server of the customized healthy lifestyle guide system, the health management service server 30 analyzes the user's life pattern requiring improvement, and the user's In order to present a desirable life pattern to the user according to the life pattern that needs improvement for weight management, the machine learning model database 310 in which the machine learning model related to the life pattern according to the various variables described above is stored is stored, and the life log A machine learning feature generation server 320 that receives daily lifelog data of a user from the database 20 or life pattern guide server 330, converts it into training data, and extracts machine learning features from the converted training data; , The machine learning model according to the machine learning feature received from the machine learning feature generation server 320 is provided from the machine learning model database, and health life pattern information corresponding to the daily lifelog data of the user is generated to manage the health It includes a life pattern guide server 330 provided to the service server 30 .

Furthermore, as shown in (b) of 5 above, the health management service server 30 including the machine learning model database 310, the machine learning feature generating server 320, and the life pattern guide server 330 is generated by the prototype selection server 331 and the prototype selection server 331 that selects only meaningful data from the training data received from the machine learning feature generation server 320 and generates the selected prototype as new training data. Based on the machine learning feature database 332 storing training data and the training data received from the machine learning feature database 332, a machine learning model is generated through a tree-based ensemble learning algorithm, and the machine learning model database ( 310) may further include a machine learning server 333 for transmission.

At this time, the prototype selection server 331 applies a method of selecting only meaningful data from training data through a prototype selection algorithm to generate a relatively small number of new training data used for learning the artificial intelligence model, and the new The training data is preferably composed of representative data in a given class area.

On the other hand, the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight carried out in the above system is the lifelog data analysis step (S10) and It includes a weight condition determination step (S20) and a personalized health lifestyle information provision step (S30), and the lifelog data analysis step (S10) is the lifelog data of a specific user stored in the lifelog database 20. is loaded from the health management service server 30, and the user's life pattern is analyzed through modeling through an artificial intelligence program that implements a tree-based ensemble learning algorithm based on the user's lifelog data.

That is, the lifelog data analysis step (S10) is performed according to the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight according to the present invention of FIG. 6 As shown in the drawing showing an example of modeling in the lifelog data analysis step, through the selection of important variables and the calculation of thresholds, important variables are classified into the number of evening steps (male) and the number of weight measurements (female) according to gender, According to the number of men's evening steps, total sleep time is selected as an important variable in the case of more than 3,000 steps, and walking speed is selected as an important variable in the case of less than 3,000 steps.

In addition, according to the number of weight measurements of the woman, the total number of steps for 20 minutes or more is selected as an important variable for two or more times, and the number of morning steps is selected as an important variable for less than two times to form modeling as shown in FIG. 6 do.

As such, a method for providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user of the present invention is shown in FIG. 7 to induce or maintain an appropriate weight for a user of the present invention As shown in the drawing for explaining the multidimensional variables according to the method of providing personalized health lifestyle information through artificial intelligence based on the user's life log, the user variable, the step variable, the sleep variable, and the weight variable By improving the method of statistical analysis in the prediction class, the modeling as shown in FIG. 6 is formed through the correlation between the prediction class and the multi-dimensional variable pattern as an interpretable machine learning analysis, and through this, accurate analysis based on the user's lifelog data analysis can be made.

On the other hand, in the lifelog data analysis step (S10), through modeling as shown in FIG. 6, in the weight state determination step (S20), the user's weight is classified as underweight, normal, or overweight based on the user's lifelog data. , is determined as one of obesity.

On the other hand, the step of determining the weight state (S10) is the prediction according to the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight in FIG. 8 As shown in the diagram showing an example of a path, a predicted path extraction step of extracting a predicted path according to the user's lifelog data through a tree-based ensemble learning algorithm based on the lifelog data generated by the user during the day is included. can be done by

That is, based on the user's lifelog data, a predicted path for men/less than 3,000 evening steps/walking speed of less than 2.4 km/h is extracted, and the user's weight condition is determined according to the extracted predicted path.

The personalized healthy lifestyle information providing step (S30) is based on the user's lifelog in order to induce or maintain the user's proper weight according to the user's weight status determined in the weight status determination step (S20). As shown in the drawing for explaining the characteristics of providing personalized healthy lifestyle information according to the method of providing personalized healthy lifestyle information through artificial intelligence, in the case of underweight, a lifestyle pattern leading to normal weight is included. If the user's weight status is obese, provide healthy lifestyle information including lifestyle patterns that lead to overweight, and if the user's weight status is overweight, lifestyle patterns that lead to normal weight If the user's body weight is normal, health lifestyle information including a lifestyle that can maintain a normal weight is provided.

On the other hand, in the step of providing personalized healthy lifestyle information (S30), artificial intelligence based on the user's lifelog is used to induce or maintain an appropriate weight for the user of the present invention in FIG. 10 in order to provide personalized healthy lifestyle information. As shown in the diagram showing an example of a similar path according to the method of providing personalized health lifestyle information through, the lifestyle pattern of other users who have lifelog data similar to the user's lifelog data but maintain a normal weight Analysis, that is, it is preferable to search for a similar path (normal) close to the user's predicted path (overweight) predicted in the prediction path extraction step.

That is, by searching for a similar path (normal) that is close to the user's predicted path (overweight), a similar path as a man/evening step less than 3,000 steps/walking speed of 2.4 km/h or more as the life pattern of other users with normal weight is found. 11 is a diagram showing an example of branch node analysis according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user according to the present invention. As shown, it is analyzed that there is a difference in walking speed in the case of other normal weight users who have similar lifelog data to the user through the search results, and according to the analysis, the user's overweight is a lifestyle pattern to induce normal weight. As a method for improving the life pattern of the current user walking at a walking speed of less than 2.4 km/h with less than 3,000 steps in the evening, to improve the life pattern so that the user walks only at a walking speed of 2.4 km/h or more in the current life pattern of the user. Provides personalized lifestyle information that is suggested.

As such, the present invention, in managing the user's weight, searches for a similar path based on the predicted path of the life pattern of other users who have the weight to induce the user's weight, and presents a lifestyle different from the user's, so that the user can be easily presented. Life patterns can be implemented, and life patterns to be improved can be maintained.

On the other hand, the branch node analysis extracts branch nodes that distinguish the predicted path from the similar path, and the variables included in the extracted branch nodes, the threshold value, the Gini coefficient, the number of predictions for each class, and the depth of the tree where the branch node is located Accordingly, it is preferable to generate a life pattern for inducing or maintaining an appropriate weight based on the user's life log data.

On the other hand, the branching node is an activity variable whose value changes according to the user's action (eg, total number of steps, average walking speed, optimal bedtime, etc.) and an inactive variable whose value does not change (eg, gender, age, ratio of REM sleep stages, etc.), and the step of providing personalized healthy lifestyle information (S30) consists of a step guide, a sleep guide, and a weight guide based on the activity variable, and provides the step guide. The guide provides the number of steps, step speed, and distance walked by time period, the sleep guide provides recommended bedtime and wake-up time, and the weight guide provides calorie consumption, muscle strength increase, water intake, and weight measurement frequency it is desirable

In addition, such a branch node is a life log according to the branch node according to the method of providing personalized health lifestyle information through artificial intelligence based on the life log of the user in order to induce or maintain the user's appropriate weight in the present invention of FIG. 12 As shown in the drawing for explaining the materialization and diversification of patterns, as the variable (branching node) representing the life pattern increases, the materialization and diversification of the life pattern is made, and in the case of materialization in which the depth of the tree deepens, accurate It becomes possible to guide healthy life patterns, and in the case of diversification in which the number of trees increases, as the number of cases representing the life patterns of users increases, relatively more similar life patterns can be searched.

Furthermore, in the step of providing the personalized healthy lifestyle information (S30), threshold value information of branch nodes extracted from a plurality of trees included in the learning model is statistically provided through materialization and diversification of the tree as above, and each The central, average, maximum, and minimum thresholds are calculated for each lifestyle information, and the level of difficulty of the personalized healthy lifestyle information provided according to the user's obesity level is characterized.

In other words, as various lifestyle patterns of healthy users similar to those of users are explored, the difficulty of lifestyle patterns included in the healthy lifestyle information provided to users through statistics is adjusted so that users' execution can be more effectively induced. .

Therefore, the health care service server 30 according to the method for providing personalized health life information according to the present invention periodically learns the collected lifelog data of the user from the existing model to gradually increase the classification performance of the learning model, Through additional learning using the lifelog data of users who have performed the healthy lifestyle information provided in the personalized healthy lifestyle information provision step (S30), the variables and values of the branch nodes included in the tree are materialized to determine the user's lifestyle and diversify the prediction path by creating a life pattern having a number of new cases through additional learning using lifelog data of users who have not performed the health lifestyle information.

13 is a diagram for explaining selection of a transcendental rectangle-based prototype according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's lifelog in order to induce or maintain an appropriate weight for a user according to the present invention. As shown, the prototype selection algorithm includes only the same class data using similarity and class information between training data in the received training data, and the minimum vertex (min) and maximum vertex that divides the multi-dimensional training data space. It may be to construct a plurality of transcendental rectangles defined by (max), and to generate new training data by determining a prototype representing a class in each transcendental rectangle.

Specifically, the prototype selection algorithm using the plurality of transcendental rectangles includes only the same class data using similarity and class information between the entire training data and the minimum vertex (min) that divides the multidimensional training data space, and A plurality of transcendence rectangles defined by the maximum vertex (max) are formed, and in adjusting the maximum size of the transcendence rectangle, it is adjusted by reflecting the data distribution of the class area, or whether the data is included within the area of the transcendence rectangle and the size of the transcendence rectangle. The diagonal length calculation calculates the distance (dist) through the difference between the center (mid) of the transcendental rectangle and the diagonal length (r) of the transcendental rectangle and coordinates appearing in each dimension.

The transcendental rectangle (hr) dividing the multi-dimensional training data space is defined by a minimum vertex (min) and a maximum vertex (max), and includes only data of the same class.

Whether or not the data x=(x ₁ ,...,x _d )∈R ^d is included in the region of the transcendental rectangle (hr) and the diagonal length of the transcendental rectangle are calculated using the formula of Equation 1 below.

Since Equation 1 calculates the distance through the difference between the center of the transcendental rectangle mid, the diagonal length r of the transcendental rectangle, and the coordinates appearing in each dimension, the time complexity is reduced compared to the distance calculation of multiplication operations such as the Euclidean distance equation. .

The determination of the prototype determines a prototype representing a class in each transcendental rectangle configured in the transcendental rectangle configuration.

For example, a variable number (a=(a ₁ , a ₂ ,...a _n ) for determining the median value of data in the selected transcendental rectangle as a prototype or finding the minimum prototype set including the entire training data. ) is set, and if it belongs to the prototype, it is set to '1' and if it does not belong, it is set to '0' to select the smallest prototype that includes the entire training data.

That is, the determination of the prototype sets dummy variables a=(a ₁ , a ₂ ,...a _n ) to find the minimum set of prototypes including the training data x _i ∈X .

If data x _i is included in prototype P, the value of a _i is 1, otherwise it is 0.

thus, is the transcendental rectangle This is the number of data dml included in the area of . The minimum prototype including all training data is selected through Equation 2 below.

Equation 2 corresponds to the set cover optimization problem and is an NP-hard problem having a time complexity of polynomial time.

Since this problem has high computational complexity, it takes a lot of time to find a solution as the number of data increases.

transcendent square Since does not contain data of different classes, it is possible to reduce time through parallel processing by using the class-based segregated set covering optimization problem strategy.

is the transcendental rectangle representing the class Data included in the area of becomes the number of

Therefore, prototypes representing training data of each class are independently selected through Equation 3 below.

The solution according to Equation 3 uses a greedy algorithm as long as relatively low computation is required. The greedy algorithm selects a prototype through the ratio of newly included data of the same class, and the selected prototype includes all data of each class.

As shown in Equation 5 below, When this prototype is selected, it is the number of data that includes data of the same class that do not belong to other prototypes in the area.

On the other hand, the representative training data is generated by only the training data determined as the prototype, and the new training data P is composed of a prototype set P _l representing each class.

Therefore, the selected prototype is is the maximum value of , and is included in the new training data.

Through the above machine learning method, the personalized healthy lifestyle guide system according to the present invention can provide the healthy lifestyle pattern information provided to the user through artificial intelligence (AI) learned relatively quickly and with high generalization performance.

That is, in generating a learning model using the tree-based ensemble learning algorithm, the prototype selection server 331 selects only meaningful data from training data through the above transcendental rectangle-based prototype selection algorithm. By generating a relatively small number of new training data used for learning the intelligent model, the relatively new training data easily induces the interpretation of the learning model as the significance of the data increases and the number of data decreases, and the tree according to the ensemble learning algorithm The prediction time can be minimized by reducing the number and depth of .

In addition, the step of providing personalized healthy lifestyle information (S30) is to provide personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain an appropriate weight for the user of the present invention in FIG. 14 As shown in the drawing showing an example of life pattern guide information according to the method, a customized mission program is provided to the user by analyzing the user's lifelog data for the past 7 days, and the customized mission program is provided for each mission item (step, Sleep, exercise, etc.), health life pattern information is provided on a daily basis, and at this time, mission items and target values can be continuously changed and provided.

In addition, as shown in FIG. 15, an example of a service according to a method of providing personalized health lifestyle information through artificial intelligence based on a user's life log in order to induce or maintain an appropriate weight for a user according to the present invention, as shown in FIG. In order to induce the performance of a healthy life pattern with relatively high importance in a plurality of healthy life pattern information calculated from the lifelog data generated by the user, it is desirable to provide in-kind rewards according to the result, through which the user can It is desirable to make it possible to collect continuous lifelog data of users by making an action for improvement.

In the above, the present invention has been described based on the preferred embodiments, but the technical spirit of the present invention is not limited thereto, and modifications or changes can be made within the scope described in the claims. Those skilled in the art to which the present invention belongs is obvious, and such modifications or alterations will fall within the scope of the appended claims.

1: Smart terminal
2: Healthcare smart device
10: Service app
20: Lifelog database
30: Health management service server
310: machine learning model database
320: machine learning feature generation server
330: Life pattern guide server
331: prototype selection server
332: machine learning feature database
333: machine learning server

Claims (13)

A user service app (10) installed in a smart terminal, receiving lifelog data, transmitting it to a health management service server, and providing a user with a life pattern suggested by the health management service server; a lifelog database 20 that stores lifelog data of a user through the user service app; And converting the collected lifelog data into training data, generating and storing new training data with meaningful data selected through prototype selection, and creating a learning model using the new training data using a tree-based ensemble learning algorithm Personal health lifestyle pattern information is created using the user's daily life log and visualized through the service app, and the performance of the learning model used to calculate life pattern guide information and life pattern information requests are continuously managed. A health care service server 30 to do; including,
The health management service server 30 provides the user with an environment in which personalized health life pattern information can be checked in real time through the user service app 10, and lifelog data generated by the user the previous day is stored at a certain time every day. A machine learning model database 310 in which a machine learning model according to a life pattern is stored; and, the life log database 20 or life log database 20 or life log database 20 The pattern guide server 330 receives the user's daily lifelog data, converts it into training data, extracts machine learning features from the converted training data, and provides the machine learning features to the life pattern guide server 330 A machine learning feature generating server 320; and receiving machine learning features of daily lifelog data of a user from the machine learning feature generating server 320, and generating a machine learning model corresponding to the received machine learning features to the machine learning feature generating server 320. A life pattern guide server 330 receiving from the learning model database 310 and generating health life pattern information corresponding to the user's daily life log data and providing it to the health management service server 30; Through a personalized health lifestyle guide system through artificial intelligence based on the user's life log,
In the method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog to induce or maintain the user's appropriate weight,
Lifelog data analysis step (S10) of analyzing the user's life pattern through an artificial intelligence program that implements a tree-based ensemble learning algorithm based on the user's lifelog data in the machine learning feature generation server 320;
According to the analysis result in the lifelog data analysis step (S10), the life pattern guide server 330 receives a machine learning model from the machine learning model database 310 and determines the weight of the user as underweight, normal, overweight, or obese. Weight status determination step (S20) to determine any one of; and
According to the user's weight status determined in the weight status determination step (S20), the life pattern guide server 330 analyzes branch nodes through a tree-based ensemble learning algorithm to provide personalized health life according to the user's specialized life pattern. Habit information is provided, but if the user is underweight, lifestyle pattern information that leads to normal weight, if the user's weight status is obese, lifestyle pattern information that leads to overweight, and if the user's weight status is overweight, leads to normal weight Personalized health lifestyle information providing step (S30) of providing lifestyle pattern information that can maintain a normal weight when the user's weight status is normal to the health management service server 30; A method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight. The method of claim 1, wherein the weight status determination step (S20)
A predicted path extraction step of extracting a predicted path according to the user's lifelog data through a tree-based ensemble learning algorithm based on the lifelog data generated by the user during the day; A method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain it. The method of claim 2, wherein the step of providing personalized health lifestyle information (S30)
Personalized health life through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight, characterized by: searching for a similar route predicted in the prediction route extraction step How to provide habit information. delete The method of claim 3, wherein the step of providing personalized health lifestyle information (S30)
Branching nodes that distinguish the predicted path from the similar path are extracted, and the user's lifelog data is obtained according to the variable included in the extracted branching node, the threshold value, the Gini coefficient, the number of predictions for each class, and the depth of the tree where the branching node is located. Personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight, characterized by creating a lifestyle pattern for inducing or maintaining an appropriate weight based on How to provide. The method of claim 1, wherein the branch node
Activity variables whose values change based on the user's actions (e.g. total number of steps, average walking speed, optimal bedtime, etc.) and inactive variables whose values do not change (e.g. gender, age, percentage of REM sleep stages, etc.) ), and
In the step of providing personalized healthy lifestyle information (S30), a step guide, a sleep guide, and a weight guide are provided based on the activity variables,
The step guide provides the number of steps, step speed, and distance walked by time period;
The sleep guide provides the recommended bedtime and wake-up time,
The weight guide is provided as calorie consumption, muscle strength increase, water intake, and weight measurement frequency; personalized health through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight, characterized in that How to provide lifestyle information. The method of claim 1, wherein the user's lifelog data is
It consists of user information, step information, sleep information and weight information,
The user information includes information about gender, age and height,
The step information includes the total number of steps, average step speed, total distance walked, total calories burned, number of steps taken at dawn, step speed at dawn, distance walked at dawn, calories burned at dawn, number of steps taken in the morning, and morning steps. Time step speed, distance walked during morning time, calories burned at morning time, number of steps taken at lunch time, step speed at lunch time, distance walked during lunch time, step calories burned during lunch time, number of steps taken at evening time, step speed at evening time, evening time Distance walked, calories burned in the evening, time walked over 20 minutes, number of times walked over 20 minutes, variability in total number of steps, variability in average walking speed, variability in total distance walked, variability in total calories burned, dawn time Variability in the number of steps, variability in step speed at dawn, variability in distance walked at dawn, variability in calories burned at dawn, variability in number of steps in the morning, variability in step speed in the morning, variability in distance walked in the morning, morning Time variability in steps calories burned, lunchtime step variability, lunchtime step speed variability, lunchtime distance walked variability, lunchtime step variability, evening step step variability, evening step step variability Include information about variability, variability in evening walking distance and evening walking calories burned;
The sleep information includes total sleep time, sleep efficiency, ratio of awake sleep phase, ratio of REM sleep phase, ratio of light sleep phase, ratio of deep sleep phase, overlapping sleep, optimal bedtime, optimal wake-up time, total Include information about sleep time variability, total sleep efficiency variability, awake sleep phase ratio variability, REM sleep phase ratio variability, light sleep phase ratio variability and deep sleep phase ratio variability,
The weight information includes information on the number of times of weight measurement, muscle mass, water content, body fat amount and number of weight measurements; How to provide personalized health lifestyle information through The method of claim 1, wherein the step of providing personalized health lifestyle information (S30)
Based on the depth information of the tree where the branch nodes extracted from the plurality of trees included in the learning model are located, the personalized lifestyle information to be provided to the user is ranked, and personalized according to the ranking according to the user's grade or situation A method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's proper weight, characterized by providing lifestyle information. The method of claim 1, wherein the step of providing personalized health lifestyle information (S30)
The threshold information of the branch nodes extracted from the plurality of trees included in the learning model is statistically calculated, and the central, average, maximum, and minimum threshold values are calculated for each lifestyle information, and provided according to the user's obesity level. A method of providing personalized healthy lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight, characterized by adjusting the level of difficulty of the customized healthy lifestyle information. The method of claim 1, wherein the health management service server
The collected lifelog data of the user is periodically trained on the existing model to gradually increase the classification performance of the learning model,
Through additional learning using the lifelog data of the user who performed the healthy lifestyle information provided in the step of providing personalized healthy lifestyle information (S30), the user's life pattern is determined by specifying the variables and numerical values of the branch nodes included in the tree. It is characterized by enhancing the prediction path that expresses and diversifying the prediction path by creating a life pattern having a number of new cases through additional learning using lifelog data of users who have not performed the health lifestyle information. A method of providing personalized health lifestyle information through artificial intelligence based on the user's life log in order to induce or maintain the user's appropriate weight. The method of claim 1, wherein the prototype selection,
In the received training data, a plurality of transcendental rectangles defined by the minimum vertex (min) and the maximum vertex (max) are constructed including only the same class data by using the similarity and class information between the training data and divide the multi-dimensional training data space. and generating new training data by determining a prototype representing a class in each of the configured transcendence squares; through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight. How to provide personalized health lifestyle information. The method of claim 1, wherein the step of providing personalized health lifestyle information (S30)
By analyzing the user's lifelog data for the past 7 days, a customized mission program is provided to the user, and the customized mission program provides health lifestyle information by mission item (step, sleep, exercise, etc.) on a daily basis. Provides personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight, characterized by continuously changing and providing mission items and target values method. The method of claim 1, wherein the step of providing personalized health lifestyle information (S30)
In order to induce the performance of a healthy lifestyle, which has a relatively high importance in a plurality of healthy lifestyle information calculated from the lifelog data generated by the user, in-kind rewards are provided according to the result, and the user's continuous lifelog data is collected A method of providing personalized health lifestyle information through artificial intelligence based on the user's lifelog in order to induce or maintain the user's appropriate weight.