KR20180126911A - Method and apparatus for mananing health - Google Patents

Abstract

The present invention relates to a health management method and a customized health management apparatus optimized for a user. The method can comprise the steps of: a recipe creation step of creating a recipe including at least one item implemented by a user for managing health of the user; and a compliance analysis step of analyzing compliance of the user by using user health information and the recipe.

Description

[0001] METHOD AND APPARATUS FOR MANANING HEALTH [0002]

The present invention relates to a health management system and method, and more particularly, to a health management system and method capable of providing individualized health management information by grasping an individual's health state based on the Internet of things.

Recently, wearable devices (wearable devices), which means 'electronic devices to be worn,' are emerging. These wearable devices are widely used for information processing / entertainment purposes and healthcare purposes.

With the development of wearable devices and communication technologies, services using the Internet of Things (IOT) technology are expanding. The above wearable devices are being launched in various industries including IT companies such as Samsung Electronics, Apple and Google, as well as athletic goods companies such as Nike and Adidas, and IT companies and small and medium sized IT companies Of wearable devices are mainly focused on health care products for health care and well-being purposes.

Meanwhile, the development of smart devices and sensor technology has enabled the quantification of personal activity data, and as the paradigm of medical services has shifted from the treatment center to the individual fitness / wellness center, It can be said that it is a fast performing area.

Meanwhile, the rapid growth of the quantified self market due to the expansion of the U-healthcare market is proceeding. In addition, recently, personal biometric data is collected in real time to be used for healthcare, and the self-quantification trend is promoting the market activation while creating an environment for easy self-care. In addition to the development of technologies for the Internet of things, products that implement conventional Internet services and sensor technologies have been introduced in various fields, but most technologies are still in the early stages.

In addition, as the disease progresses rapidly into an aging society, it is necessary to continuously manage chronic diseases such as diabetes and hypertension.

Particularly, it is very important to properly manage lifestyle such as diet management and exercise management as well as proper prescription of doctors.

However, the conventional health management apparatus has a problem in that it can not provide continuous management and personalized service tailored to each individual.

A problem to be solved by the present invention is to provide a customized health care apparatus and method optimized for a user.

It is another object of the present invention to provide a health management apparatus and method capable of analyzing a compliance of a user according to a recipe and providing a user-customized recipe according to compliance.

It is another object of the present invention to provide a health management apparatus and method that enables a user to perform a recipe well by informing instructions of a recipe at a predetermined time.

A health management method according to the present invention includes: a recipe generation step of generating a recipe including at least one item to be performed by a user in order to manage a health of a user; And

And analyzing the compliance of the user using the user health information and the recipe.

According to the embodiment of the present invention, it is possible to provide an optimized recipe according to the state change of the user by continuously monitoring and managing the state of the user.

In addition, according to the embodiment of the present invention, the compliance of use according to the recipe is analyzed to provide a customized recipe according to compliance, so that the user can easily manage the health.

Further, according to the embodiment of the present invention, the instructions of the recipe can be informed at a predetermined time so that the user can perform the recipe well.

In addition, the user according to the embodiment of the present invention can visually confirm that the health state is improved, and thus the user can actively participate.

According to an embodiment of the present invention, it is possible to provide an actual health group group position, an improvement goal, an improved state, etc., instead of simply measuring the health state of the user.

In addition, according to the embodiment of the present invention, when the user does not perform the recipe well and the health state deteriorates, the degree of deterioration can be confirmed, thereby inducing the user's alertness and inducing the user's recipe migration.

1 is a system configuration diagram to which a health management apparatus according to the present invention is applied.
2 is a block diagram showing the detailed configuration of the healthcare system 200. As shown in FIG.
3 shows a health care method according to an embodiment of the present invention.
4 illustrates a method of managing health according to another embodiment of the present invention.
5 shows an example of a learning pattern according to an embodiment of the present invention.
6 shows an embodiment of a UI displayed on the display unit 250 of the health care apparatus.
FIG. 7 shows an example of a health information pattern according to an embodiment of the present invention.
FIG. 8 shows an example of contents of message transmission / reception between the healthcare terminal 100 and the healthcare system 200. FIG.
FIG. 9 shows an example of a recipe database according to an embodiment of the present invention.
10 illustrates an actual service example according to an embodiment of the present invention.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the invention is not intended to be limited to the particular embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

Terms including ordinals, such as first, second, etc., may be used to describe various elements, but the elements are not limited to these terms.

The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the second component may be referred to as a first component, and similarly, the first component may also be referred to as a second component. And / or < / RTI > includes any combination of a plurality of related listed items or any of a plurality of related listed items.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted as either ideal or overly formal in the sense of the present application Do not.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings, wherein like or corresponding elements are denoted by the same reference numerals, and redundant description thereof will be omitted.

1 is a block diagram of a health management system to which a health management apparatus according to the present invention is applied.

Referring to FIG. 1, the health management system according to the present invention includes a health management terminal 100 for transmitting and receiving data through a communication network, a health management device 200, and a database 300.

The health management terminal 100 acquires health information of the user and transmits the acquired health information to the health management apparatus 200 via the communication network and the health management apparatus 200 transmits the health information received from the health management terminal 100 to the database 300 ), Checks the health state of the user, generates a recipe suitable for the user, and transmits the recipe to the user. In the present invention, the recipe means a comprehensive instruction for the user to determine a health information measurement rule, a diet, an exercise prescription, a sleeping time, a recommended health supplement food, a medication rule or instructions, The user can improve his or her health by implementing instructions according to the recipe. The recipe can be used for health information measurement rules, exercise amount, activity, diet, sleep time, medication rules and health supplements, stress management, weight, BMI, hospital screening, pharmacy visit, food related information (card, Activity information (SNS, telephone, etc.). In addition, the recipe may include one or more of the items, recommendations, prohibitions, and supplementary information.

The health information may include bio-information such as blood sugar, lifestyle information such as diet, exercise amount, and the like. Specifically, the lifestyle information means various lifestyle habits or patterns related to the user's health such as a health information measurement cycle, time, exercise amount, diet, sleep time, medication time and medication cycle, and health supplement.

In more detail, the health care terminal 100 may be attached to a user's body to sense various biometric information. The health care terminal 100 may include a biosensor. For example, the healthcare terminal 100 may sense various bio information such as blood pressure, blood glucose level, body weight, pulse, electrocardiogram, and the like, and may also check the exercise amount by sensing the movement of the user.

In addition, the health care terminal 100 can input information directly by the user through the input unit. The user can directly input information that can not be acquired by the biosensor through the input unit. For example, it is possible to input not only basic health information such as sex, age, height, weight, etc., but also medical record data, health examination data, etc. Also, the user can input various health information such as gene information, family history, disease, diet, exercise amount, lifestyle, etc. through the health care terminal 100. The health care terminal 100 may be formed integrally with a smart phone. For example, if a smartphone is appended with an app that checks health information or a sensor that detects bio information, the smartphone can be used as a health information terminal.

The input unit may be a touch pad, a keyboard, or a scanning device.

The healthcare terminal 100 transmits the various health information to the healthcare system 200 through a communication network.

The communication network may be a so-called machine-to-machine (M2M) communication, Internet of Things (IOT), or Internet network communication system.

The Lola network may be used as the IoT dedicated network and a WLAN (Wireless LAN), a Wi-Fi, a Wibro (Wireless broadband), a Wimax (World Interoperability for Microwave Access), a HSDPA Access) may be used.

The healthcare terminal 100 may be equipped with a communication module including a mobile communication module, a wireless communication module, and a short-range communication module using the communication network.

The mobile communication module transmits and receives radio signals to and from a base station and an external terminal on a mobile communication network. The wireless signal may include various types of data depending on a voice call signal, a video call signal or a text / multimedia message transmission / reception.

The health management apparatus 200 compares the bio information received from the health care terminal 100 and the health information received from the user terminal with the information stored in the database based on the information input by the user, And transmits the generated recipe to the healthcare management terminal 100. [ The detailed configuration of the healthcare system 200 will be described later.

The database 300 classifies various kinds of users through big data analysis, and stores recipes for age, sex, and diseases according to users. The database 300 may store the health information collected in real time through the public health information database and the medical record and health check data collected by the medical institutions or the wearable health collector. The database 300 can store the watch-keeping health information, which can show the change of the numerical value per health information within the range of about 10 years, rather than individual one-time acquired health information. In addition, the database 300 may collect various types of health care big data associated with the SNS. That is, the database 300 stores various personalized health information examples.

More specifically, the database 300 may classify and store the data according to similar patterns. That is, the database 300 may process the health information through filtering, grouping, or a combination thereof to group similar patterns of patients or symptoms and store recipes for each pattern. For example, information about a recipe for proper dieting, exercise, avoided food, and recommended dietary supplements can be stored for a diabetic male in his 60s and forty men in a diabetic who has 50 skinny men. The recipe generation unit can generate a recipe suitable for the user by referring to the recipe information for each pattern stored in the database.

Also, the database 300 stores the compliance of the patients according to the recipe. It can be used to predict user compliance by storing and learning recipe compliance of various patients.

For the real-time similar case search, the database 300 groups the unprocessed health data into classification elements such as age or sex, disease, and health status in advance. May be grouped by a unique classification element, or may be grouped by a combination of one or more classification elements. For example, it can be classified into sex (male, female), age (teenager, 20s, 30s,..., 80s or more), and hypertension ' 'Can be a specific disease. You may also divide the group into health status groups such as those with good health status, and groups with risk groups. The group may be subdivided into more than one condition, and as an example, one group may be created under a plurality of conditions, such as (sex = 'F' and age = '50' vs. disease = 'hypertension'). Also, some health cases may be stored in one or more groups. Or for real-time similar case retrieval, the database 300 is preferably configured using parallel storage. For example, the first database 310 stores information on diabetic patients, the second database 320 stores information on hypertensive patients, and the third database 330 stores diseases on arthritic patients . In addition, patients with the same disease can also store various information about the degree of improvement depending on diet and exercise. In the present embodiment, although three cases are illustrated by way of example, various patient groups can be grouped and stored. Each of the databases 310, 320, and 330 may store various data such as symptoms, symptoms, required exercise amount, food to be avoided, recommended food, etc. for each user group. Dietary therapy is especially important for chronic diseases such as diabetes and hypertension. Therefore, it is possible to store the recommended food for each patient group, the required calorie, additional side effects on the specific food, and the relationship of the food with the related food.

Classification of patients (clustering) can be classified based on the threshold value. For example, blood glucose data can be normalized between 0 and 1, with an average of 96.555 mg / dL (= 0.5) and a deviation of 24.815. When the fasting blood glucose level is higher than 126 mg / dL, it is diagnosed as diabetes. If the normalized value is higher than 0.697, it means abnormal group, so it can be classified as the diabetic user.

It can also be classified through pattern matching. At this time, similar patterns can be classified through K-means clustering. For example, pattern matching can be performed by classifying similar patterns using the Euclidean distance or by setting k to 5 in k-means clustering.

The configuration and operation of the health management system according to the present invention have been described above. Hereinafter, the detailed configuration and operation of the healthcare system 200 will be described in detail.

FIG. 2 is a block diagram showing a detailed configuration of the healthcare system 200. As shown in FIG.

The health management apparatus 200 may include a data transmission / reception unit 210, a recipe generation unit 220, an compliance analysis unit 230, a storage unit 240, and a display unit 250.

The data transmitting and receiving unit 210 periodically receives and updates the health information and the lifestyle information of the user from the health care terminal 100 and transmits an instruction according to the recipe to the health care terminal 100. [ The lifestyle information is information that a user inputs transition information according to a recipe such as the presence or absence of medicines, the diet, and the amount of exercise. The data transmission / reception unit 210 may not be used in some cases. The health information of the user may be stored in the storage unit 240 in advance.

The recipe generation unit 220 searches for a recipe corresponding to the health information received from the database and generates a recipe to be performed by the user. The recipe generation unit 220 learns existing patient information based on a database and generates a recipe suitable for a user. The recipe generation unit 220 may generate a recipe according to a set condition, and the recipe generation condition may be preset by a user. If there is no setting condition, a recipe can be created and managed based on user health information.

The recipe contains a variety of information such as the user's diet, exercise, medication time, and supplements. For example, information such as food to avoid, recommended food, daily exercise, recommended exercise, 30-minute post-meal medication may be included. When the additional health information and the lifestyle information are received after the recipe is generated, the recipe generation unit 220 corrects the recipe according to the compliance level described later.

The compliance analyzer 230 analyzes the compliance of the user using the user health information and the recipe. That is, the recipe item is compared with the user's transition information to analyze compliance. The user's transition information can be entered directly by the user or received from the data transmission / reception unit. Also, the compliance analyzer can determine the future compliance and health status of the user on the basis of the analyzed compliance.

The compliance analyzer 230 predicts that the compliance of the user is high if the user is highly compliant and the compliance of the user is low when the compliance is low. In addition, the compliance analyzer 230 learns compliance of a patient of a similar pattern stored in the database, and predicts future compliance.

The recipe generation unit 220 may reinforce or alleviate the existing recipe according to compliance. That is, the recipe generation unit 220 corrects the recipe by predicting whether or not the recipe of the user is performed according to the analyzed compliance. At this time, the recipe generation unit 220 may reinforce or relax a specific recipe item according to the user's compliance pattern. For example, in the recipe, one hour of exercise per day and a vegetable-oriented diet were indicated. Adherence analysis showed that exercise compliance was high. However, when diet compliance was low, the exercise part could be strengthened a little and the diet could be mitigated a little. This is because, if the portion with the very low compliance level is indicated equally, there is a high possibility that it will not be continuously maintained. If compliance is high, it is likely that the recipe will be implemented in the future. Therefore, the recipe of the item with high compliance can be strengthened, and in the opposite case, the recipe can be relaxed. When the likelihood of future recipe implementation is predicted, future compliance can be predicted by learning compliance of similar pattern patients stored in the database. In the case where the user's compliance is high or low and the case where the user's compliance is high or low is irregularly repeated and the health condition is worse, an optimal recipe is generated by predicting the future state by reflecting such situation. It enhances compliance by interacting with users and enhances specific recipes in a form specific to users. Recipes and changes in health status of patients are identified and weighted and optimized for each recipe item.

The storage unit 240 stores the received health information and lifestyle information.

The recipe generation unit 220 may analyze the data stored in the storage unit to generate a statistical change in the health state of the user and the change in the recipe accordingly.

The display unit 250 displays a recipe and various other items. The display unit 250 may be a display device such as a touch screen.

The recipe generation unit 220 extracts main health information from the user's health information, and analyzes what state the user is in and what type of the user's health state.

In addition, the recipe generation unit 220 searches for similar cases in the database 300 based on the analyzed user information, and searches for a doctor's prescription about the symptoms and symptoms of patients belonging to the group having similar health status, To generate a recipe. The received health information can be compared with the information stored in the database to analyze whether the health status of the user is in a normal range or an abnormal range. That is, referring to age, sex, blood sugar, etc., it is possible to determine whether the blood sugar value of the user is in the normal range or the abnormal range in the similar group. The information thus analyzed and the generated recipe can be stored in the storage unit.

Also, statistical data on the health information of the user of the recipe generation unit 220 is created. As described above, the data transmission / reception unit 210 periodically receives health information and lifestyle information of the user. Since the received health information and the recipe thus generated are stored in the storage unit 240, the data stored in the storage unit 240 can be analyzed to generate statistical data. And report-type reports can be generated based on previously analyzed health information and generated data (statistics, alarms, schedule).

In addition, the recipe generator 220 generates an alarm signal at a time set in the recipe, and transmits a transit instruction signal to the user through the data transceiver. For example, if the recipe contains 30 minutes after meals, an alarm signal is sent 30 minutes after meals. Meal times are somewhat variable, but meal times are generally constant. For example, when the user inputs the lunch time at 12 o'clock through the health information terminal, the user can transmit an alarm signal indicating the medication at 12:30.

In addition, the recipe generation unit 220 can predict whether the user's future instructions will be fulfilled based on the compliance level, and establish a schedule.

As described above, the compliance analyzer 230 analyzes the compliance degree for each item of the various instructions of the recipe. Therefore, the possibility of future implementation can be predicted based on compliance with each item. For example, it can be predicted that even if the compliance with the diet is high and the exercise compliance is very low, the exercise compliance will still be low even if the recipe is created with the same schedule. Therefore, it is possible to set a schedule for when and what exercise to take into account. Of course, it is possible to set the schedule so that the strength of the exercise is lowered and the time is shortened, considering the lower compliance.

In addition, the recipe generation unit 220 may encourage the user to increase the transition rate by assigning evaluation scores based on compliance. If the compliance is high, a high score is given, and a coupon or a point can be given as a reward for it, so that the instructions according to the recipe can be performed well.

3 shows a health care method according to an embodiment of the present invention.

Referring to FIG. 3, a health management method according to an embodiment includes a recipe creation step S310 for creating a recipe including at least one item to be performed by a user to manage health of a user, a user health information, and a recipe An acceptance analysis step S320 of analyzing the compliance of the user by using the user health information, and a step S330 of modifying the recipe according to the user health information and compliance.

The recipe generation step (S310) may include a step of generating a recipe (S310) in which the user health information is stored in a pattern that most closely matches the learned health pattern in a state where a preliminary step of learning a health pattern for various patient groups To create a recipe. That is, a pattern that most closely matches user health information in a database can be found and evaluated to generate a recipe. The evaluation method can evaluate the user's health condition and whether it is appropriate.

The recipe creation is preferably based on the user's health information, but the user's health information may or may not be received. In this case, a recipe can be created based on preset health information or reference health information. The health assistance information may include at least one of a behavior pattern, a food habit, and an environment information of the user. The reference health information is a standard for general health management, and can be set based on an average normal healthy person such as exercise for 30 minutes a day and meal calories for 2 kcal. It is desirable to periodically measure the health information of the body and to predict the current state and the future state of the patient based on the measurement. However, it is difficult to force the periodic blood glucose measurement to a healthy person, It may be difficult to input the measured value due to daily use.

As a method for solving such a problem, the first method is to pre-measure, record, analyze and store health information according to a variety of variables during a predetermined period, and predict based on the current input information. For example, after measuring the blood sugar change according to the patient's meal information, time information, and exercise information from various angles, the blood glucose value of the patient can be measured in the future only with some specific information such as the meal information, exercise information, Can be predicted.

In the second method, even if the patient does not have such prior information, the first method follows, but since there is no prior information, there is a method of finding a similar pattern first and applying a virtual pattern to the prediction.

The compliance analysis step 320 may predict a user's future compliance and health state based on the analyzed compliance. Compliance analysis and prediction methods can use one or more of linear regression analysis, regional weighting, nonlinear regression analysis, and machine learning algorithms.

The compliance analysis step S320 may include learning the compliance pattern of existing patients based on the database and predicting the compliance of the user according to the learned compliance pattern according to the user health information and the compliance pattern can do.

The items to be fulfilled by the user included in the recipe include health information measurement rules, exercise amount, activity, diet, sleep time, medication rules and health supplements, stress management, weight, BMI, hospital examination, pharmacy visit, (Card, food product barcode, etc.), and social activity information (SNS, telephone, etc.).

The step of modifying the recipe (S330) may add or modify specific recipe items according to the user's health state or compliance pattern. The health state of the user can be entered by a user, stored in advance, or received from an external device such as a health care terminal.

That is, the method further includes receiving user health information from an external device prior to the step of generating a recipe.

FIG. 4 shows a further step of learning the health pattern and the compliance pattern in the embodiment of FIG.

4, a health management method according to an embodiment includes a health pattern learning and optimization step S410, an adaptation pattern learning and optimization step S420, a user health information reception step S430, a recipe creation step S440, An adherence analysis step S450, a recipe modification step S460, and a recipe transfer step S470.

The health pattern learning and optimization step (S410) clusters existing patient information big data stored in the database or learns and optimizes the health pattern through a learning algorithm.

The compliance pattern and optimization step (S420) clusters existing patient information big data stored in the database or learns and optimizes the compliance pattern through a learning algorithm.

In the step of receiving the user health information (S430), the health information received from the external device includes information such as medication, meal, exercise, sleep, hospital examination, genetic information, document information, medicine, calorie consumption information, Time information, amount of information related to metabolism in blood, health examination information, allergy or side effects, complication information, nutritional information, auxiliary information, and the like.

The type of the received information may include at least one of a character, a number, an image, a photograph, a moving picture, and a bar code.

The recipe creation step (S440) compares user health information with the learned health pattern information to generate a recipe suitable for the user.

In the compliance analysis step S450, the user information is compared with the learned compliance pattern to analyze and predict the compliance of the user.

The step of modifying the recipe (S460) may add or modify specific recipe items according to the user's health state or compliance pattern.

The creation or modification of the recipe may include the implementation time, daily, weekly, monthly, yearly fulfillment of the item that the user has to fulfill, or may involve an event that can be performed immediately.

In addition, the recipe may include a prescription for the prevention of a specific disease or a complex disease including diabetes and blood pressure. The recipe may also include one or more of the requisites, recommendations, prohibitions, and supplemental information that the user must perform. The step of generating the recipe and the step of analyzing compliance are the same as the embodiment of Fig.

In the recipe transmission step S470, the recipe and the compliance pattern can be transmitted to the terminals registered in advance.

At this time, in addition to the information included in the recipe, an event, gift information, purchase information, health information, vaccination information, drug information, and food information that can improve or prevent the user's health care or increase the self- And may transmit information including one or more.

Because human beings do not execute specific information even if they know it, the willingness to execute it is also important. I understand all that smoking is harmful, but it is easy to understand if you think that many people do not smoke. Therefore, it is also important to increase the willingness of the user to perform the instructions according to the recipe.

The human brain is not a food of nature but is excessively attracted to dangerous stimuli such as refined carbohydrate, sugar, fat, salt, etc., and it turns into an abnormal body condition such as obesity, and it is conscious that it is difficult to escape from itself.

Likewise, humans instinctively try to minimize their movements, but by their urge to act, they are not willing to exercise themselves in order not to consume energy, but to television, games, and sports. The same is true for stimuli such as smoking, drinking, drugs, and addiction. Therefore, even though the physical condition has not been genetically improved compared to the past, dangerous stimulation by artifacts that have not been experienced in the past is continuously strengthened, which is a great problem for health. To solve these problems, it is necessary to minimize stimulation by artifacts in various pleasure centers such as the hypothalamus and limbic system, and control them to be able to escape by themselves through more enhanced delay compensation.

It is necessary to stimulate this role through the recipe gradually and reinforce it through other rewards.

Since the conditions and will of each user are different genetically and environmentally, it is also necessary to appropriately indicate the degree of enhancement over time or effective stimulation method considering individual differences.

IP networks, mobile networks, and non-mobile private networks can be used as transmission networks.

The pre-registered terminal may include a user, a family of a user, a primary care physician, a hospital, a pharmacy, a health care provider, and an insurance company terminal.

The subject receiving the information can manage the health state of the user based on the received health information.

Although not shown in the figure, the recipe and compliance can be stored in the storage unit before the recipe is transferred. By storing the results of the recipe and compliance analysis, it is possible to observe the change in the user's health status in a time-wise manner and to use it to create various statistics.

The learning algorithm may be one of k-means Clustering, Rule Based, Tree based, Machine Learning, Deep Learning (CNN, RNN), and Auto encoder.

The method of finding a pattern matching the user's health information in the learned pattern may use any one of Euclidian Distance, Hidden Markov Model, Dynamic Time Warping, Decision Tree, Rule Based, Machine Learning, and Deep Learning (CNN, RNN) .

Various algorithms for finding the learning algorithm and the pattern are well-known techniques, and a detailed description thereof will be omitted for brevity.

The k-means clustering calculates the Euclidean distance from each data to each cluster, finds the closest cluster in the data, allocates the data, and then re-sets the center of gravity of the data in each cluster. If it does not change, it stops repeating.

Ruled Base (rule base system) is to create several rules to fulfill the rule if the current state (memory) is satisfied.

Can be expressed as follows.

- IF <A> Then <B>,

Here A is the current state and B is the conclusion according to the current state. In other words, <A> is a condition and <B> is an action.

These rule base systems can be roughly classified into three types.

Working Memory: Memory that stores the current state

Production Memory: Memory that defines the state of the state

Rule Interpreter: A system that performs the actual work by matching the above two memories

Here, the execution of RM by the state of WM is called Fire.

Tree based (Decision Tree Learning Method) is a methodology commonly used in data mining. It aims to create a model that predicts the value of a target variable based on several input variables. The right figure shows an example of such a prediction model. In the tree structure of the figure, each internal node corresponds to one input variable, and the branches leading to the child nodes correspond to the possible values of the input variable. A leaf node corresponds to a target variable value when each input variable has a value corresponding to a path from a root node to a leaf node.

The decision tree learning method is one of the most useful techniques in map classification learning. In this article, we assume that all properties have a domain consisting of finite discrete values and that the classification has a single target property. Each element of the domain of the classification is called a class. All internal nodes of the decision tree or classification tree correspond to input attributes on a one-to-one basis. Branches connected to the inner node of the tree show the values that the attribute can have, and the leaf node shows the probability distribution of the class or class.

The 'learning' of a decision tree is the process of dividing the set of data used for learning into subsets according to the appropriate partitioning or partitioning test. This process is recursively repeated on each separate subset of data in a so-called cyclic partitioning scheme, where the partitioning no longer adds new predictors or continues until the subset of nodes has the same value as the target variable . This top-down induction of decision trees (TDIDT) is an example of a greedy algorithm and is the most common way to learn decision trees from data. In data mining, decision trees are described in mathematical terms to aid in the generalization and categorization of given data.

Machine Learning is a learning method that focuses on predictions based on known properties learned through training data.

Deep Learning (CNN, RNN) is a kind of neural network based on Deep Neural Network (DNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Restricted Boltzmann Machine (RBM) Deep Belief Network (DBN).

A neural network is an artificial neural network (ANN) composed of several hidden layers between an input layer and an output layer. In-depth neural networks can model complex non-linear relationships as well as general neural networks. For example, in a deep neural network structure for an object identification model, each object can be represented in a hierarchical configuration of image primitives. At this time, the additional layers can aggregate the features of the progressively gathered lower layers. This feature of the in-depth neural network allows to model complex data with fewer units (nodes) compared to a similar artificial neural network.

Convolutional Neural Network (CNN) is a kind of multilayer perceptrons designed to use minimal preprocessing. CNN is composed of one or several layers of convolutional hierarchy and general artificial neural networks layered on top of it, and utilizes additional weighting and pooling layers. This structure allows CNN to fully utilize the input data of the two-dimensional structure. Compared to other deep-run structures, CNN offers good performance in both video and audio applications. CNN can also be trained through standard reverse delivery. CNN is easier to train than other feedforward artificial neural network techniques and has the advantage of using fewer parameters. Recently Deep Learning has developed the Convolutional Deep Belief Network (CDBN), which is structurally very similar to the existing CNN, so that it can take advantage of the two-dimensional structure of the figure and at the same time, Deep Belief Network , DBN) can be taken advantage of. CDBN provides a generic structure that can be used for a variety of image and signal processing techniques and is used in several benchmark results for standard image data such as CIFAR.

The basic form of the auto encoder is as follows.

input vector: x ∈ [0,1] D x∈ and [0,1] D, hidden representation: y ∈ [0,1] d y∈ [0,1] d is then the deterministic mapping: y = fθ ( x) = s ( Wx + b ) y = f? (x) = s (Wx + b).

Here, θ denotes the parameter of W, b, W is the weight matrix of d x D, and b is the bias vector.

The latent representation: y value computed from the hidden layer is mapped back to the "Reconstructed" vector: x '∈ [0,1] D x' ∈ [0,1] D.

The formula to be calculated at this time is

, And? '= W', b '.

At this time, if the transpose matrix WT of the weight matrix of the input layer before mapping is equal to the remapping weight matrix W ', the corresponding auto encoder has tied weights.

The loss function is defined as follows.

In the early days, we used BP (Backpropagation) algorithm and SGD (Stochastic Gradient Descent) algorithm to learn the basic auto-encoder weight. For the value given by input (visible), the weight value between the hidden and visible units is calculated, put into the sigmoid function, and the weight connected with hidden and output once again to calculate the estimated value, It is Unsupervised Learning which continuously updates the error with visible data.

The gradient descent updates along a sequentially decreasing direction. However, for very large data sets, the time to converge to optimize is very slow. On the other hand, the SGD scheme is also considered a noisy descent, and each iteration rate is very fast. The training method is classified into a pattern mode and a batch mode method depending on how many samples are used for each training every time. In the case of the pattern mode, one sample is input, and the weight is updated by calculating the variation of the parameter for each sample. On the other hand, in the case of the batch mode, by inputting the samples as much as the batch size, the parameters of the samples are calculated and the weights are updated once for each batch.

If you train one by one like pattern mode, you will actually have the character of SGD. At this time, the objective function L can be well off the local optimal point, but the training time is long. On the other hand, the performance of the batch mode is lower than that of the SGD. However, there is an advantage in that the training time is reduced when a suitable size of the mini-batch is used. So some researchers suggest Mini-batch + SGD.

The Euclidian Distance is a method of calculating the distance between each document and the query term. The calculation method is the same as obtaining the dot product of a vector.

Outliers "refers to outliers that deviate significantly from the general characteristics or models of a given data, and outlier analysis refers to finding out these abnormal data and characterizing those characteristics. The types of outliers are as follows.

Global outlier - A point anomaly in which a single measurement is significantly out of the range.

Contextual outlier - "Current body temperature is 39 degrees." Whether this is abnormal or normal is interpreted differently depending on the situation. This means setting the situation and detecting abnormalities.

Collective outlier - the observed values are significantly out of balance.

Techniques for finding outliers include outlier detection by statistical techniques, outlier detection by proximity, and outlier detection by clustering.

The outlier detection by statistical technique assumes a probability model for given data and uses a discordance test. Of course, this requires knowledge of the key parameters of the given data. This includes the data distribution as assumed above (eg, normal distribution or Poisson distribution), expected value (ie, mean or variance), and so on.

If the distance to the nearest neighbor is greater than a given threshold value in the specific region R, the outlier detection method based on the proximity is used to determine the outliers.

First, the distance threshold r (r> 0) and π (0 <π ≤ ≤ 1) are specified. In this case, o object that satisfies the following conditions can be regarded as an outlier with DB (r, π) property.

Fig. 5 shows an example of a learning pattern.

(A) learning and normalization of existing patients' compliance based on big data, (b) evaluating based on the learned information when new data is input, and A suitable recipe can be generated (c).

In normalized data, an increase in slope implies improved health and compliance, while a horizontal absence of change in slope means no change in health status or compliance, while a slope in the horizontal portion indicates a change in health status It means that compliance is getting worse.

When new data (user health information) is inputted based on the normalized data, it is possible to generate an optimum recipe by predicting the user state based on the new data (user health information).

A specific recipe can be enhanced in a form that is user specific so that the greater the interaction with the user, the more likely it is to improve health.

The recipe, the statistics, the schedule, the evaluation information, and the like generated as described above may be transmitted to the health information terminal 100 through the data transmission / reception unit. Of course, if there is a terminal designated in advance by the user other than the health information terminal 100 at this time, the terminal can transmit the terminal together with the terminal.

The health information terminal 100 may receive the data and display the data on the display unit.

The health information terminal 100 can display various icons and a UI (user interface) on the display unit. For example, icons such as account management, statistical inquiry, history inquiry, report, attention / alarm, medication / exercise instruction, evaluation and the like can be displayed. When the user selects a desired icon from the health information terminal 100, the user can display the corresponding information.

Further, according to the present invention, the administrator can view the recipe and related information in addition to the user. That is, it can be confirmed through the contents displayed on the display unit of the healthcare apparatus 200. In this embodiment, the administrator may include both a system administrator, a hospital manager, and a healthcare manager. The administrator can check the instruction content displayed on the display unit, the medication of the user, or the like, and can quickly respond.

6 shows an embodiment of the UI displayed on the display unit 250 of the health care apparatus.

Referring to FIG. 6, the display unit 250 may display menus such as account management, customer-specific information, accumulated graph, recipe management, and operation settings. The administrator can monitor the displayed items on the display unit 250 and take necessary actions quickly.

The configuration of the health management apparatus according to the present invention has been described above.

7 illustrates an example of a health information pattern according to an embodiment of the present invention.

FIG. 7A shows an example of the change of the health information data by time and date in the data sheet, FIG. 7B shows the pattern of the blood glucose by time by date, FIG. 7C shows the average blood glucose pattern by time, And normalizing the nutrition information in the range of 0 to 1, and FIG. 7D shows the blood glucose pattern of each hour according to the day of the week. In the present embodiment, the blood glucose pattern is indicated by day of the week, but various parameter information other than the blood glucose pattern may be displayed on a day or a date basis.

7, data such as Blood Glucose Monitoring (BGM), food intake, exercise time, sleeping time, total cholesterol, triglyceride, and high density lipoprotein (cholesterol) It is also possible to display the blood glucose pattern as a graph on a time-by-hour basis, or to observe a change pattern of blood glucose on a day-by-day basis.

FIG. 8 shows an example of contents of message transmission / reception between the healthcare terminal 100 and the healthcare system 200. FIG. In this embodiment, the health care terminal includes a health care terminal.

As shown in FIG. 8, the healthcare terminal transmits the blood glucose trend, body weight, blood pressure, history, diet, lifestyle, family history, etc. to the health management device (S710).

The health care apparatus receives and analyzes the information, determines which group the user belongs to, and transmits the determination result (S720). As shown in the figure, the result of analyzing the information of Mr. A's blood glucose pattern is classified into the C group. The C group's recipe DB is applied and the 12-step management is progressed by one month. "Your goal for this month is to strengthen your basic physical strength. What you need to do this month is to walk 30 minutes after meals, diet, medication compliance, what you can do ... not to do ...".

The health care terminal checks this (S730), configures the exercise and the diet according to the instructions, and then periodically transmits the health information and information on whether or not to perform the instruction (S740). For example, a message such as "transfer of blood glucose value (150), transmission of meal (fitness), transmission of medication (yes)"

The healthcare apparatus generates and transmits a new recipe based on the information transmitted by the user (S750). For example, "Hello, Mr. A, please check your blood glucose before eating, please have a high blood sugar, please chew slowly in brown rice, and take fresh fruit (apple). If you have a meal, be sure to take the medicine and walk for about 30 minutes. "

The user continuously checks this and sends information about the snack or the strong supplements when he / she lives according to the changed recipe (S760). Information about snacks and health supplements can be transmitted by sending food bar codes.

The healthcare device continuously monitors and transmits a new indication according to the changed information (S770). For example, "Hello, A, please eat a little bit because there is a lot of saturated fat in the snack, and family history and genetic analysis result in hypertension, so reduce red meat and eat omega-3. So do not eat more than a certain amount of vitamin C. "

The user alters the living and changed health information and living information according to the changed diet and recipe (S780). For example, a message such as "Delivery with dinner (meal bar code, cake / bread), transfer of blood glucose value (160), automatic transfer of momentum, drug delivery (drug barcode, Tylenol) The momentum is expressed as the automatic transfer of momentum since the sensor that detects the user's movement can automatically transmit.

When the predetermined period of time elapses, the health care apparatus transmits the evaluation result (S790). For example, "Hello, you have completed all of your goals for this month, and your performance is as high as the top 10% in Group C. Congratulations, and send 5,000 points for your gift vouchers, I am maintaining a normal blood sugar level through improvement of life. I would like to have a good practice for the remaining 11 months. "

According to the above-described method, an optimum recipe is generated by comparing with the same user group group based on the health state information of the user, and the user can continuously monitor the diet, exercise amount, medication state, Can be provided.

FIG. 9 shows an example of a recipe database according to an embodiment of the present invention.

Referring to Fig. 9, the recipe includes a prescription, a food, and a supplements field, and various instructions are stored in each field.

For example, in the case of a prescription field, messages such as "lose weight when overweight", "avoid refined sugar and alcohol" and "eat only moderate salt" may be stored.

In addition, the database can store data for recommended, prohibited, or supplemental items.

For example, recommended items may include items such as "Take some of your carbohydrate intake with good unsaturated fat and omega-3 fatty acids," "Choose wholegrain, high fiber, hypoglycemic foods" Dietary items may include items such as "reduce fasting carbohydrate intake", "reduce high blood sugar", and supplementary items such as "chromium, magnesium, selenium are helpful when insulin resistance is low" "Cinnamon and ginseng increase glucose tolerance."

10 illustrates an actual service example according to an embodiment of the present invention.

Referring to FIG. 10, when the user transmits information such as presence / absence of dinner, exercise amount, medication, etc., the health management device can generate a recipe suitable for the user based on the health information of the user and transmit the recipe to the user.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined by the appended claims. Accordingly, the true scope of the present invention should be determined by the following claims.

100: Health management terminal
200: Health care device
300: Database

Claims (25)

A recipe generation step of generating a recipe including at least one item to be performed by the user based on the health information of the user; And
And analyzing the compliance of the user using the user health information and the recipe. The method according to claim 1,
Wherein the compliance analysis step predicts a user's future compliance and health state based on the analyzed compliance. 3. The method of claim 2,
And modifying the recipe according to the health information and compliance or predicted compliance. The method according to claim 1,
The recipe generation step includes:
The user's health information is firstly found in the most similar pattern to be matched in the learned health pattern, and then the evaluation is performed to evaluate the recipe Wherein the method comprises the steps of: The method according to claim 1,
In the compliance analysis step,
Learning an compliance pattern of existing patients based on the database;
And estimating the user health information and the compliance pattern according to the learned compliance pattern. 5. The method of claim 4,
Clustering the health information or optimizing health pattern learning through a learning algorithm that includes machine learning or neural networks. 6. The method of claim 5,
Clustering the learned compliance pattern information or optimizing compliance learning through a learning algorithm including machine learning or neural network. The method according to claim 1,
These items include social activity measures including health information measurement rules, exercise amount, activity, diet, sleeping time, medication rules and health supplements, stress management, weight, BMI, hospital screenings, pharmacy visits, And information about one or more of the information. The method of claim 3,
Wherein the step of modifying the recipe comprises modifying the recipe to at least one of adding, deleting, enhancing, and mitigating a specific recipe item according to a user's health status or compliance pattern. The method according to claim 1,
Storing the health information, the compliance degree, and the recipe in a storage unit; And
And analyzing data stored in the storage unit to statistically change a change in the user's health state and a change in the recipe accordingly. The method according to claim 1,
And transmitting the recipe information to the terminal that has been registered in advance when a time specified in the recipe occurs or an event occurs. The method according to claim 1,
And receiving the user health information from an external device. 12. The method of claim 11,
Together with the recipe information, further transmits information that can improve or prevent the user &apos; s health care or enhance the user's self-
The information may include at least one of an event, a gift offering, purchasing information, health information, vaccination information, drug information, and food information that may improve or prevent a user's health care or enhance the self-management commitment of the user How to manage. The method according to claim 1,
The recipe generation step includes:
A method for health care that generates a modified optimal recipe by further evaluating what is appropriate for a user from information of the pattern that most closely matches. 13. The method of claim 12,
The receiving information in the receiving step may include at least one of medication, meal, exercise, sleep, hospital screening, genetic information, document information, medication, calorie consumption information, metabolic reaction information in blood glucose or blood, A supplemental information including the presence, time, and amount associated with any one or more of the complication information, nutrition and adjuvant. 13. The method of claim 12,
Wherein the form of the received information in the receiving step includes at least one of a character, a number, an image, a photograph, a moving image, and a bar code. The method according to claim 1,
And transmitting or sharing the recipe and the compliance information to a predetermined external device. The method according to claim 1,
Wherein the recipe and the compliance information are managed by an external device in association with a predetermined external device. The method according to claim 1 or 3,
Wherein the recipe includes at least one of a time, day, week, month, and year transition of items to be performed by the user, or includes an event that can be performed immediately. The method according to claim 1,
Wherein the recipe includes an item for managing the health of a general person who does not have an individual disease, a complex disease, or a disease. The method according to claim 1,
Wherein the recipe comprises at least one of a requisite, a recommendation, a prohibition, and supplementary information. 13. The method according to claim 11 or 12,
Wherein the transmitting and receiving method is one of an IP network, a mobile network, and a non-mobile private network. The method according to claim 1,
The recipe generation step generates a recipe based on preset reference health information when the user does not have health information. A recipe generation unit for generating a recipe including at least one item to be performed by the user based on the user health information; And
And an compliance analyzer for analyzing the compliance of the user using the user health information and the recipe. 26. The method of claim 24,
If the health information of the user does not exist, the health care apparatus predicts future compliance and health state based on health supplement information or preset reference health information including at least one of the behavior pattern, eating habit, and environment information of the user A health care device that modifies the recipe.

Images