WO2022211385A1

WO2022211385A1 - Health care consultation system using distribution of disease prediction values

Info

Publication number: WO2022211385A1
Application number: PCT/KR2022/004222
Authority: WO
Inventors: 김세연; 윤준영; 김상수; 송승재
Original assignee: 주식회사 라이프시맨틱스
Priority date: 2021-03-30
Filing date: 2022-03-25
Publication date: 2022-10-06
Also published as: KR102342770B1

Abstract

The present invention relates to a health care consultation system using the distribution of disease prediction values, the system comprising: a sample data collection unit for collecting health information data about past patients as sample data; a health checkup collection unit for collecting health information data about customers; a disease prediction unit for predicting diseases by using the health information data; a risk group classification unit for obtaining the disease prediction value of each patient from each patient datum of the sample data through the disease prediction unit, and classifying each patient into a plurality of risk groups according to the size of the disease prediction value; an exponential distribution generation unit for generating a distribution with the biometric index values of patients belonging to each risk group for each biometric index; a biometric index evaluation unit for calculating the influence of each biometric index on each disease by calculating the degree of difference in biometric index distribution between risk groups of a corresponding disease; and a management index extraction unit which calculates customer disease onset prediction values through the disease prediction unit to extract diseases with a high risk, and which extracts management indexes according to the magnitude of influence of a corresponding biometric index from among the biometric indexes of the extracted diseases, wherein the diseases with a high onset prediction value are selected and management indexes having a large influence on a corresponding disease are extracted through big data, and thus the health management indexes that are most important and necessary for customers can be more scientifically and accurately extracted.

Description

Health care consultation system using distribution of disease prediction values

The present invention groups the patients in the sample data into disease onset prediction values, calculates the distribution of each group for the biometric index, and uses the calculated distribution to convert the biometric index that affects the reduction of the possibility of disease occurrence of the customer as a management index. It relates to a health care counseling system using the distribution of disease prediction values to be extracted.

The present invention estimates a customer's disease onset prediction value, which can be seen as an important indicator of health deterioration and health status, and based on the estimated disease onset prediction value, the health to be managed for each customer by considering the influence of each bio-index on each disease prediction It relates to a health care counseling system using the distribution of disease prediction values that extracts the management index.

The present invention is a research conducted with the support of the Information and Communication Industry Promotion Agency 　AI Precision Medical Solution (Doctor Answer 2.0) Development Project in 2022 with the funding of the Republic of Korea Government (Ministry of Science and Technology Information and Communication) (No. S0252-21-1001).

In general, in order to receive medical services, a patient must directly visit a hospital or public health center where medical services are performed to receive medical services. However, recently, through rapid development of network technology, health management system services through the Internet have been diversified, making it possible to easily manage one's own health condition. In particular, in addition to the generalization of U-Health, which enables prevention, diagnosis, treatment, follow-up management and health care services anytime, anywhere by linking information and communication with health care, health management programs for various diseases such as stress, high blood pressure, and diabetes This has been developed and technologies to be provided through the Internet or a portable terminal have been proposed.

As an example, the user can directly check the periodically measured blood flow and oxygen levels through the mobile terminal so that he or she is aware of the current health condition, inquire about his or her health condition to a medical institution, and respond to the inquiry in real time in a chat format [Patent Document 1] has been proposed. In addition, a technique for providing health management information using integrated or convergence information in nursing, food nutrition, and physical education fields to older people such as the elderly using a mobile device has been proposed [Patent Document 2]. In addition, a technology has been proposed that allows the user to directly select a service provider for each service item, and generates and provides data on the user's exercise program compliance status and diet program compliance status based on the information provided by the service providers. [Patent Document 3]

However, the prior art generates and provides a lot of health management information for identifying the user's condition by the variously measured biometrics and optimizing each condition.

However, these prior technologies make the user inconvenient because the process of collecting the user's status is complicated, such as the need to measure bio-signals in real time. In addition, since these prior arts suggest a health program in preparation for all diseases of the user, there are too many programs to be observed from the user's point of view. Therefore, the prior art makes it difficult for the user to perform all the corresponding programs, and the user may not be able to properly perform the corresponding health program.

On the other hand, in terms of health management, the health status and factors of health deterioration are all different for each individual. For example, it would be undesirable to primarily provide a heart-friendly health care regimen to users with a low risk of heart disease but a high risk of diabetes. That is, in this case, it is good to focus on providing the user with a health management program good for diabetic disease.

In addition, if the influence of each variable (management index) on each predicted disease can be grasped, it is possible to grasp the variables to be managed intensively for health management. That is, it is very important to find an index or variable to be managed as important for each user.

(Patent Document 1) Korean Patent Application Laid-Open No. 10-2006-0037123

(Patent Document 2) Korean Patent Application Laid-Open No. 10-2016-0145244

(Patent Document 3) Korean Patent Publication No. 10-2017-0131067

An object of the present invention is to solve the above-mentioned problems, group the patients of the sample data into disease onset prediction values, calculate the distribution of each group for the biometric index, and use the calculated distribution to determine the disease of the customer. It is to provide a health care counseling system using the distribution of disease prediction values that extracts the biometric index that affects the reduction of the incidence as a management index.

In addition, it is an object of the present invention to estimate a customer's disease onset prediction value, which can be seen as an important indicator of health deterioration and health status, and consider the influence of each bio-index on each disease prediction based on the estimated disease onset prediction value. It is to provide a health care counseling system using the distribution of disease prediction values that extracts the health care index to be managed separately.

In order to achieve the above object, the present invention relates to a health care counseling system using the distribution of disease prediction values, comprising: a sample data collection unit for collecting health information data of past patients as sample data; a health checkup collection unit that collects customer health information data; a disease prediction unit that predicts a disease using health information data; a risk group classification unit for obtaining a disease prediction value of each patient from each patient data of the sample data, through the disease prediction unit, and classifying each patient into a plurality of risk groups according to the size of the disease prediction value; an exponential distribution generating unit for generating a distribution with the bioindices values of patients belonging to each risk group for each bioindices; For each disease, a bio-index evaluation unit for calculating the influence of each bio-index by calculating the degree of difference in the distribution of the bio-index between the risk groups of the disease; And, a management index extraction unit for extracting a disease with a risk of onset by calculating the disease onset prediction value of the customer through the disease prediction unit, and extracting it as a management index according to the magnitude of the influence of the biometric index among the extracted disease bioindices characterized by including.

In addition, the present invention provides a health care consultation system using the distribution of disease prediction values, wherein the disease prediction unit predicts a customer's disease using a disease prediction model, and the disease prediction model inputs input values of predetermined input variables. Upon receiving, it outputs a predetermined probability of occurrence of each disease variable, and the disease prediction model is characterized in that it is composed of a neural network in which internal variables are learned by learning data.

In addition, in the health care consultation system using the distribution of disease prediction values, the present invention is characterized in that the biometric index is created as an item of the health information data.

In addition, the present invention provides a health care consultation system using the distribution of disease prediction values, wherein the management index extraction unit determines which risk group the acquired disease prediction value belongs to, and if the disease prediction value belongs to a predetermined risk group as a high risk group It is characterized as a disease at risk.

In addition, the present invention is a health care consultation system using the distribution of disease prediction values, wherein the management index extraction unit selects a biometric index with a high influence among the biometric indexes of the disease at risk of the customer, but in the order of the highest influence A predetermined number is selected, or a biometric index that is greater than or equal to the threshold of a predetermined influence is selected, and with respect to the selected biometric index, the customer's biometric index is calculated, and the calculated customer's biometric index and the normality of the disease are selected. It is determined whether the bio-index of the risk group is outside the normal quantile of the distribution of the bio-index of the normal risk group, and if it is outside the quantile, the bio-index is finally extracted as a management index.

In addition, in the health care counseling system using the distribution of disease prediction values, the present invention calculates the maximum residual of the cumulative distribution function by using the cumulative distribution function of the distribution for the difference in the distribution, and calculates the calculated It is characterized in that the difference of the distribution is calculated using the maximum residual.

In addition, the present invention is characterized in that in the health care consultation system using the distribution of disease prediction values, the health information data of the customer or the patient is composed of demographic information and health checkup data.

In addition, in the present invention, in the health care consultation system using the distribution of disease prediction values, the demographic information includes gender, age, residential area, insurance subscription type, income decile, disability, height, and weight, and the health Screening data includes waist circumference, systolic blood pressure, diastolic blood pressure, fasting blood sugar, total cholesterol, high-density cholesterol, low-density cholesterol, triglycide, hemoglobin, urine protein, serum creatinine, serum GOT, serum GPT, and gamma GTP. characterized in that

In addition, the present invention provides a health care consultation system using the distribution of disease prediction values, wherein the system includes a result output unit for displaying the distribution of risk groups for a specific biometric index of a specific disease, and outputting the customer's biometric index on the distribution It is characterized in that it further comprises.

As described above, according to the health care counseling system using the distribution of disease prediction values according to the present invention, by selecting a disease with a high onset prediction value and extracting a management index that has a large influence on the disease through big data, the most The effect of extracting important and necessary health care index more scientifically and accurately is obtained.

In addition, according to the health management consulting system using the distribution of disease prediction values according to the present invention, by extracting the management factor index of health management by calculating the disease incidence prediction value from the customer's health examination data, medical records at a medical institution, etc., additional customers The effect of presenting a more accurate customer's health management program is obtained without biometric measurement.

1A and 1B are block diagrams of an entire system for implementing the present invention.

2 is a block diagram of the configuration of a health care consultation system using the distribution of disease prediction values according to an embodiment of the present invention.

3 is a table showing input variables of a disease prediction model according to an embodiment of the present invention.

4A and 4B are distribution diagrams of BP_HIGH and HMG variables for each hypertension prediction group according to an embodiment of the present invention.

5 is a graph showing the influence/importance of the bio-index of hypertension by group according to an embodiment of the present invention.

6 is a table showing the severity of BMI when predicting hypertension according to an embodiment of the present invention.

7A and 7B are graphs showing the quantiles of BMI and systolic blood pressure variables of patient A in the normal group according to an embodiment of the present invention.

8A and 8B are graphs showing quantiles of BMI and systolic blood pressure variables of customer A in a high-risk group according to an embodiment of the present invention.

9 is a table illustrating the degree of management necessity according to BMI and systolic blood pressure level according to an embodiment of the present invention.

10 is a table illustrating the degree of management necessity according to BMI and systolic blood pressure quantile according to an embodiment of the present invention.

11 is a graph showing the management variable (bio-index of the super-risk group) of the prediction of five major cancers according to an embodiment of the present invention.

12A and 12B are graphs illustrating distributions of GAMMA_GTP and TRICLYCERIDE indices (distributions corresponding to 5 major cancer predictions) according to an embodiment of the present invention.

Hereinafter, specific contents for carrying out the present invention will be described with reference to the drawings.

In addition, in demonstrating this invention, the same part is attached|subjected by the same code|symbol, and the repetition description is abbreviate|omitted.

First, the configuration of the entire system for carrying out the present invention will be described with reference to FIGS. 1A and 1B.

As shown in FIG. 1A , the health care consultation system (hereinafter referred to as the consultation system) using the distribution of disease prediction values using predictive data according to the present invention estimates the customer/user's disease onset prediction value and uses the distribution to estimate the customer's/user's Extracting the health management index, it can be implemented as a program system on the computer terminal (10).

That is, the consultation system 30 may be implemented as a program system on the computer terminal 10 such as a PC, a smartphone, or a tablet PC. In particular, the consultation system may be configured as a program system or a mobile application (or an application, an app), and may be installed and executed in the computer terminal 10 . The counseling system 30 provides a service of receiving health data, estimating a disease outbreak prediction value, and extracting health care factors by using the hardware or software resources of the computer terminal 10 .

Also, as another embodiment, as shown in FIG. 1B , the counseling system may be configured and executed as a server-client system composed of a counseling client 30a and a counseling server 30b on the computer terminal 10 .

Meanwhile, the counseling client 30a and the counseling server 30b may be implemented according to a typical method of configuring a client and a server. That is, the functions of the entire system can be divided according to the performance of the client or the amount of communication between the server and the server. Hereinafter, it will be described as a consultation system, but it may be implemented in various forms according to the configuration method of the server-client.

Alternatively, the counseling server 30b may be a server that provides a health care counseling service on the web, and the counseling client 30a may be a web browser that accesses the counseling server 30b and uses the corresponding service.

In addition, the consultation server 30b may additionally include a database 40 that stores statistical information of a user's or customer's health checkup results, disease prediction information, and the like.

Specifically, the database 40 includes a sample data DB 41 for storing sample data such as demographic information and health examination results of patients, a risk group DB 42 for storing risk group information for each disease, and a biometric index for each risk group. and an exponential distribution DB 43 for storing the distribution. However, the configuration of the database 40 is only a preferred embodiment, and in developing a specific device, it may be configured in a different structure according to the database construction theory in consideration of the ease and efficiency of access and search.

Next, a health care consultation system 30 using the distribution of disease prediction values according to an embodiment of the present invention will be described with reference to FIG. 2 .

As shown in FIG. 2 , the disease prediction service system 30 according to an embodiment of the present invention includes a sample data collection unit 31 that collects demographic information and health checkup information of past patients as sample data, a customer A health information collection unit 32 that collects sociodemographic information and health examination results of Risk group classification unit 34, exponential distribution generating unit 35 for obtaining the distribution of the biometric index for each risk group, the biometric index evaluation unit 36 for evaluating each biometric index by comparing the distribution between risk groups, and customer management It consists of a management index extraction unit 37 for extracting the index. Additionally, it may be configured to further include a result output unit 38 for outputting the distribution and the management index of the customer.

First, the sample data collection unit 31 collects demographic information and health checkup information of past patients as sample data.

Demographic information is data representing the sociodemographic characteristics of a patient, and consists of age, gender, height, weight, presence or absence of a disability, lifestyle, and the like.

In addition, the health checkup information is the patient's health checkup data, and is data measured (checked) when performing a health checkup, such as blood pressure, cholesterol level, hemoglobin level, and urine protein level.

In addition, the onset data is data on the disease onset of the patient, and indicates whether the patient has the disease.

On the other hand, sample data is collected by being classified or identified by demographic information. That is, personal information that identifies the patient, such as the patient's name, is excluded, and medical data is collected based on health status information such as age, gender, height, weight, disability, and lifestyle.

In particular, preferably, the sample data uses a sample cohort DB. The total data that is building the sample cohort DB refers to the data of 1 million people. The 1 million subjects were stratified based on the gender and age of the citizens and the distribution of residential areas, so it can be said that the results derived from this data are representative of the whole nation.

Next, the health information collection unit 32 collects demographic information and health checkup data of the customer.

First, the health information collection unit 32 receives the customer's demographic information. In this case, demographic information of the customer may be input through a questionnaire or questionnaire.

As described above, the customer's demographic information consists of age, gender, height, weight, presence or absence of a disability, lifestyle, income quintile, past medical history, family medical history, and the like.

Preferably, the questionnaire or questionnaire consists of 26 items. In other words, the questionnaire data includes gender, age, region of residence, insurance subscription type, income quintile, disability, type of examination institution, height, weight, person (stroke, heart disease, high blood pressure, diabetes, dyslipidemia, pulmonary tuberculosis, and other diseases including cancer) ), family history (stroke, heart disease, high blood pressure, diabetes, liver disease, cancer) history, smoking status, smoking period, smoking amount per day, drinking habits, alcohol consumption per day, exercise amount per week, etc.

In addition, the health information collection unit 32 collects the customer's health checkup data.

The customer's health checkup data consists of data measured during the health checkup, such as systolic blood pressure, diastolic blood pressure, pre-meal blood sugar, total cholesterol, high-density cholesterol, low-density cholesterol, triglycerides, and urine protein. Screening data that are usually measured during a health checkup include waist circumference, systolic blood pressure, diastolic blood pressure, fasting blood sugar, total cholesterol, HDL cholesterol, LDL cholesterol triglycide, hemoglobin, urine protein, serum creatinine, serum GioT, serum GPT, Gamma GTP and the like.

At this time, the customer's health checkup data is directly input by the customer, or the most recent health checkup data is obtained from the Health Insurance Corporation, a medical data institution (Health Insight), etc. through the customer's accredited authentication process.

In general, when the National Health Insurance Corporation general health checkup is conducted, the items collected by the questionnaire before the checkup are usually called questionnaire data. However, when importing the customer's health checkup data, the questionnaire data during the health checkup process cannot be imported. Therefore, the corresponding data can be input through a separate questionnaire. In addition, the customer's 'height' and 'weight' are included in the health checkup items, but they are entered through a separate direct questionnaire.

Next, the disease prediction unit 33 predicts the disease of the patient or customer using demographic information and health examination data of the patient or customer.

Preferably, the disease prediction unit 33 predicts the customer's disease using the disease prediction model. When the disease prediction model receives an input value of a predetermined input variable, the disease prediction model outputs an onset probability of each predetermined disease variable.

In particular, the disease prediction model is composed of a neural network, etc., and internal variables are learned by learning data. And when the disease prediction model is trained, when the values of the input variables are input, the disease prediction model outputs the probability of occurrence of each disease.

The disease prediction model is the result of an artificial intelligence neural network that machine-learned millions of health check-up results from thousands to tens of thousands of people selected to represent domestic patients for each disease, demographic factors, and lifestyles. The calculation result may vary freely due to the user's steady improvement in health behavior, etc. 3 shows examples of input variables of the disease prediction model. 3 illustrates a case in which all 44 input variables are configured.

Also, as an embodiment, the output variable consists of the incidence probabilities for 12 diseases (or 12 major diseases). In particular, the diseases are breast cancer, five major cancers, cancer integration, cerebrovascular disease, osteoporosis, cataract, hypertension, obesity, diabetes, COPD (chronic obstructive pulmonary disease), joint disease, dyslipidemia, and the like.

That is, the output variable is an onset probability of each disease, or an onset prediction value.

Next, the risk group classification unit 34 obtains a disease prediction value of each patient from each patient data of the sample data, and classifies each patient into a plurality of risk group groups according to the size of the disease prediction value.

First, the risk group classification unit 34 obtains a disease prediction value of each patient from the data of each patient through the disease prediction unit 33 . Disease predictors are obtained for each disease. That is, as described above, the disease prediction unit 33 may calculate a disease onset prediction value by inputting the patient's demographic information and health checkup information.

In addition, the risk group classification unit 34 divides the acquired disease prediction value into quantiles to form a risk group group for each quantile, and groups patients having disease prediction values belonging to the quantile into the corresponding risk group.

That is, a plurality of risk group groups are divided into a plurality of quantiles (intervals) according to the size of the overall disease prediction value. Preferably, the risk group is classified into a low-risk group (0-25%), a medium-risk group (25-50%), a high-risk group (50-75%), and a very high-risk group (75-100%). That is, if the disease prediction values of all patients are all sorted by size, the low-risk group is a group of patients with disease prediction values ranging from 0% to 25% of the disease prediction value.

In addition, the risk group is divided by each disease. For example, one patient may belong to a high-risk group for gastric cancer but a low-risk group for liver cancer.

Next, the exponential distribution generating unit 35 generates a distribution (bioindex distribution or exponential distribution) as the biometric index values of patients belonging to each risk group for each biometric index.

Preferably, the biometric index is composed of items (or variables) of demographic data or health examination data, or a combination of these items. For example, the biometric index may be BMI and blood pressure. BMI (Body Mass Index) is obtained by combining the weight and height of the health examination data. Also, blood pressure is one item of health checkup data. Alternatively, when the health checkup data includes various blood pressure data, the corresponding blood pressure index may be a biometric index by combining the various blood pressure data and setting it as one blood pressure index.

That is, since the biometric index is calculated as an item of demographic data or health checkup data, it is possible to calculate the biometric index for all patients of standard data.

The exponential distribution generating unit 35 calculates the bioindices of all patients belonging to each risk group for the risk groups of a specific disease, and generates a distribution using the calculated bioindices. This will be referred to as the bioindex distribution (or exponential distribution) of the risk group.

An exponential distribution is generated for each risk group. For example, it is obtained from the exponential distribution of the low-risk group, the exponential distribution of the medium-risk group, the exponential distribution of the high-risk group, and the exponential distribution of the very high-risk group. If the risk groups are divided into 4 groups, 4 exponential distributions are generated.

In addition, the exponential distribution of all risk groups is obtained and generated for each bio-index and each disease.

Next, for each disease, the bio-index evaluation unit 36 calculates the degree of difference in the index distribution between risk groups of the disease. That is, the influence of the corresponding bio-index on the disease is evaluated using the degree of distribution difference.

For a specific bioindex of a specific disease, the fact that there is a significant difference in the distribution of the bioindex between risk groups means that the bioindex affects the risk of the disease. In other words, if the distribution of the low-risk group and the distribution of the high-risk group are very different from each other, if the bio-index value of the high-risk group is managed to be moved to the average of the bio-index distribution of the low-risk group, the patient is the predicted disease value of the disease. can be lowered.

Preferably, the bio-index evaluation unit 36 calculates the influence of the bio-index by using the degree of difference in the distribution.

That is, the influence of the biometric index (ex. BMI, blood pressure, etc.) is calculated as follows.

After determining the distribution of the index for each risk group, the importance (or influence) of the index in predicting the disease is checked according to the degree of distribution difference between the indices.

Preferably, the degree of distribution difference uses the residual of the cumulative distribution function of each distribution map. That is, the cumulative distribution function of the distribution of each group is derived, and the cumulative distribution function of each group and the normal risk group (people predicted to develop hypertension with a probability of 0-25% in the low-risk group in the example of the previous four groups) Calculate the maximum residual of the cumulative distribution function. The residual of the cumulative distribution function is the difference between the values of the cumulative distribution function. That is, the difference with the maximum value among the differences between the function values of the cumulative distribution function by each exponential value is calculated as the maximum residual.

An index with a large maximum residual value (0 to 1) of the cumulative distributions between groups can be considered that the value of the variable has a large correlation with the disease prediction probability, so that the index has a large influence on the disease prediction value. .

Therefore, the importance of the biometric index is calculated using the maximum residual value.

Referring to the distribution map shown in the upper diagram of FIG. 4A , the difference in the distribution of the BP_HIGH (systolic blood pressure) index between the groups divided by the predicted onset of hypertension is clearly revealed. When the maximum residual of each group is calculated from the cumulative distribution shown in the lower figure of FIG. 4A , the maximum residual value of the ultra-high-risk group and the low-risk group is 0.776. As the risk of disease development increases, the maximum residual value increases significantly, indicating that the systolic blood pressure value has a great influence on the prediction of hypertension. Conversely, since the maximum residual values of HMG (hemoglobin) do not change significantly even with an increase in risk, it is concluded that they are insignificant variables.

Referring to FIG. 5 , the importance of variables in predicting hypertension can be confirmed for each of the medium-risk group, the high-risk group, and the ultra-high-risk group. There is a difference in the ranking of the importance (influence) of the indices between groups, and the absolute importance (influence) of each index decreases as the level of risk decreases.

Next, the management index extraction unit 37 extracts a disease with a risk of onset by calculating the customer's disease onset prediction value, and extracts a biometric index with a high influence from the biometric index of the disease as a management index. That is, the corresponding biometric index is selected as the management index according to the size of the influence.

First, the management index extraction unit 37 obtains the disease prediction value of the customer from the customer's data through the disease prediction unit 33 . Disease predictors are obtained for each disease. That is, the disease prediction unit 33 may input the customer's demographic information and health checkup information to calculate the customer's disease onset prediction value.

Next, the management index extraction unit 37 determines which risk group the acquired disease prediction value belongs to, and if it belongs to the risk group to be managed, it is determined as a disease to be managed (disease with risk). At this time, the risk group to be managed is determined in advance. For example, it may be set as a risk group to be managed in a high-risk group, such as a high-risk group and an ultra-high-risk group, among the above four risk groups.

Next, the management index extraction unit 37 selects a biometric index having a high influence among the biometric index of the disease to be managed by the customer. In this case, a predetermined number is selected in the order of the highest influence, or a biometric index that is greater than or equal to the reference value of the predetermined influence is selected.

Next, the management index extraction unit 37 calculates the customer's biological index with respect to the selected biological index, and the calculated customer's biological index is the normal quantile of the distribution of the biological index of the normal risk group of the disease, or, in advance to determine if it is outside the specified range.

Here, the normal risk group (or normal group) is a risk group group with the lowest predicted incidence and is predetermined. In the previous example, when divided into four risk groups, the low-risk group corresponds to the normal-risk group.

If the difference is greater than the standard value, the corresponding biometric index is finally extracted as the management index.

Preferably, the management level is determined according to which range (quantile) the biometric index belongs to the distribution of the normal risk group.

Next, the result output unit 38 outputs the biometric index distribution of the risk group and the customer's biometric index or management index on the screen.

That is, for a specific biometric index of a specific disease, the distribution of risk groups is displayed, and the customer's biometric index is output on the distribution.

Through this, the customer can check his/her condition with the naked eye, thereby accurately grasping his/her health condition, so that he/she can be alert to health care.

Next, an example of extracting the customized management index of a specific customer (A) according to an embodiment of the present invention will be described in detail with reference to FIGS. 6 to 12 .

After confirming the distribution in the low-risk group and the distribution in the group to which the customer belongs (EX. ultra-high-risk group) of each of the previously selected management indices, the severity of the result is checked, and a health plan is established. .

First, the quantiles of customers are identified in the exponential distribution of the low-risk group. When compared with the normal group, the severity is confirmed by the measured value (bio-index) of customer A, and the need for management is determined.

For each index, a range of quantiles is determined to determine the severity of the measurement and the need for management. For example, as shown in the table of Figure 6, the higher the BMI, the higher the risk of hypertension. Therefore, when the BMI value of a specific customer is within 25% of the top/bottom of the normal group, it is "severe", and when it is within 10%, it is "very high". determined to be serious.

As an example, quantiles within the normal group distribution of BMI and systolic blood pressure values of customers predicted to develop hypertension by 97% are shown in FIGS. 7A and 7B . At this time, the customer's BMI value is 24 and the systolic blood pressure value is 159. 7A shows the quantiles of BMI of customer A in the normal group, and FIG. 7B shows the quantiles of the systolic blood pressure variable.

Customer A's BMI value is in the top 22.6% of the distribution of the normal group. This can be seen as requiring proper management and attention. In addition, customer A's systolic blood pressure is in the top 0.01% compared to the distribution of the normal group, which requires intensive management and serious attention.

Next, check the decile in the group to which the customer belongs.

For each index, determine the range of quantiles that determine the severity of the measurement and the need for management. For example, it is determined when the number is above the top 20%.

Check the upper quantile (position) of the measured value in the exponential distribution of the group to which customer A belongs. The quantile check determines the severity and management need of customer A's measurements compared to the group to which customer A belongs.

The case where the BMI and systolic blood pressure values of a customer with a high blood pressure prediction rate of 97% are quantiles within the super-risk group distribution is exemplified. That is, FIGS. 8A and 8B show the quantiles of BMI and systolic blood pressure variables of customer A in the high-risk group. At this time, the management standard as shown in FIG. 9 is established.

Referring to FIGS. 8A, 8B, and 9 , the BMI value of customer A is 62.8% higher than the risk group to which customer A belongs, and is an average value within the risk group. In addition, customer A's blood pressure level was in the top 13.4%, confirming that it was very serious even within the risk group. In other words, customer A's blood pressure level requires intensive management.

Meanwhile, health counseling is possible using the identified management indices.

The following example exemplifies that the disease is 5 major cancers, and the probability of occurrence of 5 major cancers of customer A is 0.8, as shown in FIG. 10 . 11 shows the bioindices of the prediction of five major cancers in the super-risk group. In addition, FIGS. 12A and 12B illustrate distributions of GAMMA_GTP and TRICLYCERIDE indices (distributions corresponding to the prediction of 5 major cancers).

As shown in FIG. 10 , when the influence/importance of the biometric index is calculated in the manner described above, the GAMMA_GTP (gamma GTP) value of a customer with a very high risk of 5 major cancers is a very important index for predicting 5 major cancers. Nevertheless, customer A's gamma GTP value is in the upper 79% (lowest 21%) level of the distribution of the normal group, so it can be seen as a variable that requires relatively little management. In the high-risk group to which customer A belongs, the quantile of the gamma GTP value is also in the top 89%, so there is no seriousness.

On the other hand, in the case of TRIGLYCERIDE (triglyceride), although the importance of this variable to the variable high-risk group is relatively low (8th place), it is not the lowest importance, and the value corresponds to the value of the top 8% of the distribution of the normal group, Very serious. So it needs to be managed.

Next, the effects of the present invention will be described in more detail.

Identify the indices that have a high influence on the predictive value of each disease for each individual, and identify the indices that need to be managed intensively by checking the distribution quantiles for each index.

Calculate the index importance (impact) for the risk levels of low-risk, medium-risk, high-risk, and ultra-high-risk groups.

By checking the distribution of the low-risk group for a specific index and the index quantile of the group determined by individual predictions, the relative distribution and location of the index can be identified, and the importance of the index can be grasped without prior background knowledge.

By concentrating on the identified important indices to manage health, it is possible to more accurately estimate an individual's appropriate health level, and through this, it is possible to more effectively proceed with health care to be carried out later.

There are the following differences compared to the existing method of identifying important indices. In other words, the existing method of determining the importance of each index was limited to the algorithm used (ex. the method of determining the importance of each index of the random forest), and it was not intuitive (it is difficult to grasp the criteria for determining the importance of each index, Not intuitively understandable). In addition, one index importance value for all customer groups is being calculated.

However, the exponential importance/influence calculation method of the present invention can be commonly used for all models, and can be intuitively understood. In addition, since different index importance is calculated for each risk group such as low-risk group, medium-risk group, high-risk group, and very high-risk group, the importance of the index suitable for the customer can be identified.

In the above, the invention made by the present inventors has been described in detail according to the above embodiments, but the present invention is not limited to the above embodiments, and various modifications can be made without departing from the gist of the present invention.

The present invention selects a disease with high predictive value and extracts the management index that has a lot of influence on the disease through big data to more scientifically and accurately extracts the most important and necessary health management index for customers, A health care consultation system technology using the distribution of disease prediction values that provides a more accurate customer health management program without additional customer biometric measurements by extracting the management factor index of health management by calculating the disease outbreak prediction value from medical records at medical institutions can be applied to

Claims

In the health care consultation system using the distribution of disease prediction values,

a sample data collection unit that collects health information data of past patients as sample data;

a health checkup collection unit that collects customer health information data;

a disease prediction unit that predicts a disease using health information data;

a risk group classification unit for obtaining a disease prediction value of each patient from each patient data of the sample data, through the disease prediction unit, and classifying each patient into a plurality of risk groups according to the size of the disease prediction value;

for each patient of the sample data, an exponential distribution generating unit for generating a distribution with the bio-index values of patients belonging to each risk group for each bio-index;

For each disease, a bio-index evaluation unit for calculating the influence of each bio-index by calculating the degree of difference in the distribution of the bio-index between the risk groups of the disease; and,

It includes a management index extraction unit for extracting a disease with a risk of developing a disease by calculating the disease onset prediction value of the customer through the disease prediction unit, and extracting it as a management index according to the magnitude of the influence of the biometric index among the extracted disease bioindices, ,

The disease prediction unit predicts a customer's disease using the disease prediction model, and the disease prediction model receives an input value of a predetermined input variable, outputs a probability of occurrence of each predetermined disease variable, and predicts the disease The model consists of a model in which internal variables are learned by training data,

The biometric index is made of items of the health information data,

The management index extraction unit determines which risk group the acquired disease prediction value belongs to, and if it belongs to a predetermined risk group as a high risk group, it is determined as a disease with a risk of onset,

The bio-index evaluation unit calculates the maximum residual of the corresponding cumulative distribution function by using the cumulative distribution function of the distribution for the difference in the distribution, and calculates the difference in the distribution using the calculated maximum residual Distribution of disease prediction values A health care consultation system using
The method of claim 1,

The management index extraction unit selects a high-influence bio-index from among the bio-indices of a disease that is at risk of developing the customer, and selects a predetermined number in the order of the highest influence, or selects a predetermined number of bio-indexes that are greater than or equal to the threshold value of the pre-determined influence. Select an index, calculate the customer's bio-index for the selected bio-index, and check whether the calculated customer's bio-index and the bio-index of the normal risk group of the disease are outside the normal quantile of the distribution of the bio-index of the normal risk group A health care counseling system using the distribution of disease prediction values, characterized in that the determination and final extraction of the biometric index as a management index when it is outside the quantile.
The method of claim 1,

The health information data of the customer or the patient is a health care counseling system using a distribution of disease prediction values, characterized in that it consists of demographic information and health checkup data.
The method of claim 1,

The system, with respect to a specific biometric index of a specific disease, displays the distribution of the risk group, and a health care counseling system using the distribution of disease prediction value characterized in that it further comprises a result output unit for outputting the customer's biometric index on the distribution.