KR102083995B1 - Disease Network Generating Method using Cohort and Computer Readable Record Medium thereof - Google Patents

Abstract

The present invention relates to a method for constructing a disease-causing network from a cohort database using a computer, wherein the control unit repeatedly performs steps for calculating a relative risk in response to different preset diseases, thereby performing the counterpart. Establishing a disease onset network that exhibits a correlation of morbidity between risks based on risk; And outputting the disease-prone network.

Description

Disease Network Generating Method using Cohort and Computer Readable Record Medium

The present invention relates to a method for establishing a disease onset network using a cohort, and a computer readable recording medium recording the same. Specifically, for example, a disease may be extracted by extracting a relative risk of each disease using clinical data of a long-term health insurance corporation. By constructing the invention network and making it easy to predict the possibility of the disease based on it, a method of establishing a disease onset network using a cohort that can lead to a healthy life of the patient and prevent the increase of medical costs of the people and the computer read the recorded It relates to a possible recording medium.

The research on life networks has been concentrated on molecular networks, which has raised many questions about actual clinical utility.

In other words, the molecular-based disease network model is limited to genetic diseases, and in some cases (10-20%) that are closely related to the genome, such as cancer, some validity is guaranteed, but most diseases in which patients visit hospitals (80 ~ 90%) are concerned that there is a limit in exploring correlations among all diseases through molecular-based disease networks, given that the disease is less related to heredity.

In addition, although the pattern of occurrence differs according to age, race, and gender, the specificity of the disease that is specifically related to Koreans is interpreted by grouping them into one network without considering these properties from the viewpoint of molecular genetics. Has a problem in that it cannot be identified according to characteristics such as age, gender, and socioeconomic index.

Although the importance of network medical research from the clinical medical point of view is increasing worldwide, most of the cases in Korea have been concentrated on traditional system biology research based on molecular genetics. Lack of converged human resources with knowledge is also a cause.

Since the existing network medical research has formed a disease network reflecting the characteristics of Westerners based on molecular genetics, it is virtually difficult to identify the relationship between diseases occurring in Koreans. In Korea, not only do health insurance corporations and reviewers have huge clinical big data for more than a decade, but also each hospital has a large number of electronic medical record (EMR) data, which can be used in research and disease networks. Modeling and constructing it can also identify the link between genetic and non-genetic diseases occurring in Koreans, and it is worth noting that the results of the research will be of great value and can be of practical help in establishing national health care policies and measures. do.

On the other hand, Korea's clinical clinical big data (Health Insurance Corporation, Simpyungwon, etc.) has exploded and can be used as an infrastructure for basic medical and policy research.In the early 2000s, the network of medical research fields using big data Research in the field of medicine begins to be studied in earnest, and this research complements the fundamental problem of reductionism, which is a traditional research methodology, and interprets life as a system to explore the emerging properties of life. I understand that you have contributed.

From this point of view, it is desirable to understand in detail how to predict which diseases will affect other diseases and how to visualize them using the clinical disease database of patients who visit the hospital sequentially.

[references]

Publication No. 10-2016-0043777 (published Apr. 22, 2016)

Disclosure of the Invention An object of the present invention is to construct a disease onset network using a cohort, which can efficiently and reliably derive correlations among all diseases based on diseases of patients using big data, a method of visualizing disease onset networks, and a computer recording the same. To provide a readable recording medium.

In addition, another object of the present invention, if the cohort well represented the population to identify the correlation between diseases occurring in Koreans disease disease network construction method using a cohort that can be used as a platform for precision medicine most suitable for Koreans, To provide a disease onset network visualization method and a computer-readable recording medium recording the same.

In addition, another object of the present invention is to show the characteristics of the disease in various forms, such as age, sex, socioeconomic index, nursing institutions, etc., using a cohort that can be used as a platform for precision medicine by processing simply and conveniently To provide a disease onset network construction method, a disease onset network visualization method and a computer-readable recording medium recording the same.

In addition, another object of the present invention, it is possible to determine the association between genetic and non-genetic diseases of patients, a disease onset network construction method using a cohort that can be utilized in health care policy, disease onset network visualization method and computer recording the same To provide a readable recording medium.

In addition, another object of the present invention is to create a statistical model for predicting the degree to which a particular disease mutually affects other diseases and to establish a disease onset network using a cohort that can effectively visualize it, a disease onset network visualization method and A computer readable recording medium having recorded thereon is provided.

In addition, another object of the present invention is to provide a method for predicting disease occurrence using a cohort capable of constructing and visualizing a network having a prognostic relationship of disease occurrence, a method for visualizing disease occurrence network, and a computer-readable recording medium recording the same. will be.

In a method according to an embodiment of the present invention for solving the above problems, in a method for constructing a disease onset network from a cohort database by using a computer, the control unit, from the cohort database, the target disease is a dangerous disease Classifying a person who does not develop a first target disease (D1) among the plurality of diseases in the group A during the first setting period set as a period affecting the group; The control unit may classify the number of persons having a second risk disease (R = D2) corresponding to a second target disease from the group A into the group E, and generate a second risk disease (R = D2). Classifying the unaffected personnel into group C; The control unit, after the first predetermined period, during the second set period in which the dangerous disease affects the target disease on the contrary, the number of people having developed the first dangerous disease D1 in the E group is the E1 group Classifying a person who does not develop a second risk disease (R = D2) in the C group into a C1 group; The control unit is configured to generate the first relative value calculated from the relative number of persons or relative frequency values of the E group and the E1 group, and the second relative value calculated from the relative number of persons or the relative frequency value of the C group and the C1 group. Calculating a relative risk between the risk disease (R = D2) and the target disease (D1); The control unit repeats the steps by sequentially setting different remaining preset diseases among the plurality of diseases as a target disease and a risk disease, and mutually correlating morbidity rates between the diseases based on the relative risk. Constructing an indicative disease onset network; And outputting the disease-prone network.

According to the present invention, big data can be used to derive correlations among all diseases more efficiently and reliably based on diseases of patients, and if the cohort is almost all citizens, the correlation between diseases occurring in Koreans A disease onset network construction method using a cohort that can be used as a platform for precision medicine that is most suitable for Koreans, a method of visualizing a disease onset network, and a computer readable recording medium recording the same can be provided.

In addition, the characteristics of the disease can be shown in various forms such as age, sex, socioeconomic index, and nursing institution, and can be processed simply and conveniently. Also, the relationship between the genetic and non-genetic diseases of patients can be identified. A disease onset network construction method using a cohort that can be used for a policy, a method of visualizing a disease onset network, and a computer-readable recording medium recording the same can be provided.

In addition, it is possible to create a statistical model to predict the extent to which a particular disease affects other diseases, and to visualize it effectively, and to establish a disease onset network using a cohort that can construct and visualize the network that is related to the occurrence of disease. A method, a disease onset network visualization method, and a computer readable recording medium recording the same can be provided.

1 is a flowchart illustrating an example of a disease prediction method by selecting a specific disease according to an embodiment of the present invention;
2 is an example of a visualization of an entire disease network;
3 is an example of visualizing disease networks by age
4 is an example of visualization of the gender disease network.

The disease onset prediction method and the disease onset network visualization method using the cohort according to the present invention will be described in detail with reference to FIGS. 1 to 4.

1 is a flowchart illustrating an example of a method for predicting disease occurrence according to an embodiment of the present invention, FIG. 2 is an example of a visualization of an entire disease network, and FIG. 3 is an example of a visualization of an age-specific disease network. 4 is an example of visualizing a gender disease network.

The disease occurrence prediction method using the cohort according to the present invention may be performed on a computer. Although not shown, the cohort data is received or collected through a communication unit or an input unit of a computer, and the control unit of the computer is shown in FIG. 1. At group A (at risk population) having no first target disease (D1) to analyze the incidence rate among a plurality of diseases (D1, D2, D3, ..., Dn) during the first set period set in the cohort. Obtaining (1); Second risk disease (R = D2) having any one risk factor corresponding to another second target disease D2 to be assumed to be a different risk factor from the first target disease D1 in the group A. (2) classifying the group E into (Exposed) and the group C without the second risk disease (R = D2) and controlling the number of persons ((n (E), n (C)) (3) obtaining a number n (E ′) of the group E having the first target disease D1 during the second setting period after the first setting period; (4) calculating the number of people (n (C1)) having the first target disease (D1) among the C group; and the number of people (n (E1), n obtained from the steps (3) and (4)). The second risk disease (R = D2) affects the first target disease (D1) based on the number of persons ((n (E), n (C)) obtained in step (2)) and (2). The relative risk (RR: Relative Risk, RR _{R-> D} below) is shown below. (5) to obtain the formula (X); and to establish a disease relationship network visualizing the results of performing the following sequential and repetitive steps to develop the disease characterized in that the output through the output of the computer It can be configured as a prediction method.

RR _{R- > D} = n (E1) / n (E) ÷ n (C1) / n (C) -------------- <expression x>

Here, the expression x is the control unit, the first relative value (n (E1) / n (E)) calculated from the relative number of people or relative frequency value of the E group and the E1 group, and the C group and the C1 group Relative risk RR that the second risk disease (R = D2) affects the target disease (D1) from the second relative value (n (C1) / n (C)) calculated from the relative number of people or the relative frequency value It can be described as calculating _{R-> D} by the division operation.

In the step (2), the second risk disease (R = D2) corresponds to the risk diseases (R = D3,...) Corresponding to each of the remaining target diseases (D3, ..., Dn) among a plurality of diseases. .., R = Dn) by sequentially changing to step (6) to perform the step (5) in step (3), respectively; after finishing, instead of the target disease (D1) for each remaining target disease ( D2, D3, ..., Dn) it is preferable to repeat the above step (1) to step (1) while sequentially changing the target disease. Accordingly, the steps are repeatedly performed in order to correspond to different preset diseases (D3, D4, ..., Dn) in order to show mutual correlation of morbidity rates between the diseases based on the relative risk. An outbreak network can be built.

More specifically, as shown in FIG. 1, in the step (2), the second risk disease (R = D2) corresponds to the risk disease (R = D3, corresponding to the remaining diseases (D3, ..., Dn). ..., R = Dn) by sequentially changing the step (3) to perform the step (5) in the step (3); after finishing (part enclosed by the dashed chain line 'L2' of Figure 1) Means D3, ..., Dn sequentially), the step (6) in the step (1) instead of the target disease (D1) the remaining diseases (D2, D3, ..., Dn) As a process for the step, performing the above steps repeatedly while sequentially changing the target disease (sequentially performing the portion surrounded by the red dotted line 'L2' in FIG. 1 for D2, D3, ..., Dn It is preferred that).

Here, it is preferable that the first predetermined period (X period) is set to be longer than one year, which is a period in which a person who has a disease is expected to visit a medical institution at least once for the disease. The second setting period (Y period) is preferably set as a period in which the effect of the dangerous disease (R) to be grasped out of the effects on the target disease (D) generated after the first setting period is reversed. For example, Y can be set to one year if you want to determine the short-term impact of a dangerous disease (R) on a target disease (D), and Y can be set to ten years if you want to identify a civil impact. The period setting of the degree is preferable because it is possible to collect the clinical data over the period to facilitate the correlation between the specific disease and other diseases formed during the set period or to predict whether the disease is caused by the drug.

Here, the cohort that is a sample of the population or statistics may include various databases, but it is preferable to include a database of the Health Insurance Corporation or a screening agency or a database of each university hospital.

Referring to Figure 1 in more detail as follows.

First, in terms of first period, the first setting period (X period) is applied in the first step S110 and the second step S130, and the second setting period (Y period, Y = X +). N) is partially applied to the third (3) step (S140), the (4) step (S150) and the (5) step and the (6) step beyond the (2) step.

First, when the computer performing the method according to the embodiment of the present invention determines that the first target disease (D1), which is the target of analyzing the incidence rate, occurs for each patient in the X period (S113), group A (Fig. Group A1, and if it is determined that the first target disease D1 has not occurred (S117), it is classified as group A (see group A in FIG. 1) (see step (1) above).

Next, the computer which performs the method according to an embodiment of the present invention may include risk factors corresponding to any other second target disease to be assumed as a risk factor different from the first target disease D1 in the group A. Group E (Exposed, see S133 of FIG. 1) having a second risk disease (R = D2) having a factor) and Group C (Control, 137 of FIG. 1) having no second risk disease (R = D2). ) And calculate each number of people ((n (E), n (C)) (see step (2) above).

Then, the computer performing the method according to the embodiment of the present invention, the group having the first target disease (D1) of the E group in the second set period after the first set period (E1 group, S143) And the number of people (n (E1)) of the group having the first target disease (D1) by classifying the group (E1 group, S147) without the first target disease (D1) (see step (3) above). ).

In addition, the computer for performing the method according to an embodiment of the present invention, the group (C1 group, S153) having the first target disease (D1) of the C group during the second set period and the first target disease ( The number of persons n (C1) of the group having the first target disease D1 is obtained by classifying the group (C1 group, S157) without D1) (see step (4) above).

Based on the number of people (n (E1), n (C1)) obtained from the step (3) and the step (4) and the number of people ((n (E), n (C)) obtained in the step (2), The relative risk (RR: Relative Risk, RR _{R-> D} below) that the second risk disease (R = D2) affects the target disease (D1) is obtained by the following formula (X).

RR _{R- > D} = n (E1) / n (E) ÷ n (C1) / n (C) -------------- <expression x>

In other words, the relative risk can be calculated using the above-described <Equation x>.

Here, the portion 'L2' surrounded by the thick red dashed line in FIG. 1 is applied to the case of the second risk disease (R = D2) in step (2) (S130). Then, the next step to be performed is the first target disease (D1) of step (1) is the same, but not the second risk disease (R = D2) in step (2), for example, the third disease (D3), 4th risk disease (D4), 5th disease (D5), ..., 3rd risk disease (R = D3), 4th risk disease (R = D4) sequentially assigned in response to the n-th disease (Dn), ..., n-th risk disease (R = Dn) is set, and the above-described step (2) to (5) for each risk disease should be performed. This step may be referred to as' (6) step).

That is, for example, a computer performing a method according to an embodiment of the present invention may perform an 'L2' operation surrounded by a thick red double-dotted line associated with a second risk disease (R = D2) for the first target disease (D1). The process of 'L3' enclosed by the thick red dashed line associated with the third risk disease (R = D3), and although not shown, the fourth risk disease (R = D4) and the fifth risk disease ( The same is true for R = D5), and the work process corresponding to the setting of the n-th risk disease (R = Dn) is also shown as the work process of 'Ln' surrounded by a red double-dot chain.

The computer performing the method according to an embodiment of the present invention, each of the second risk diseases (R = D2), the third risk diseases (R = D3), the fourth risk diseases (R = D4), the fifth risk diseases ( R = D5), ..., based on the basic data derived from the n-th risk disease (R = Dn), the relative risk of each risk disease affecting the target disease, It can be calculated through the process and <Equation x>.

The flow chart of the result of performing all these operations only for one specific disease (D1), which is one specific disease, is shown in FIG. 1 and the marking of the working ranges may correspond to the 'FC1' region surrounded by dotted lines. The computer which performs the method according to the embodiment of the present invention performs the same operation as the second target disease (D2), the third target disease (D3), ..., n-th target disease (Dn) Each of these flow diagrams is illustrated in FIG. 1 as 'FC2', 'FC3', ..., 'FCn'.

This method can be used to determine the prognosis of coronary disease that has a correlation between diseases by using a population database that is a sum of data inputted sequentially by each individual.

And, if the statistical test in each of the above-described process can be tested for statistical significance through a known chi-square test or Fisher accuracy test. In this process, the relative risk, for example, b-> P-value, which is the significance level of a, is less than the preset set significance level value including 0.05, 0.01, and 0.001, and the relative risk RR _R-> It is preferable to judge that only the pair whose _D is greater than the preset relative risk is significant and use it as analysis data.

In addition, the computer performing the method according to an embodiment of the present invention, further performing step (7) of representing the disease of each cohort in a posterior relationship matrix based on the data classified and obtained in each step, wherein (7) In the step, it is desirable to perform a method of obtaining a matrix using raw data as it is without processing, a method of simplifying the onset of a disease that is continuously occurring in the raw data, and a method of simplifying the raw data based only on the initial basis. . The computer performing the method according to an embodiment of the present invention extracts a disease pair of a preceding disease and a trailing disease for each cohort in consideration of the occurrence direction of the disease, and any one disease is different based on the disease pair. In order to calculate the relative risk affecting one disease, it is preferable to sequentially generate a contingency table and calculate the relative risk for each disease. A more detailed description of this process and a more detailed description of the above-described prediction method are as follows.

First, the first setting period is set to two years of 2002 and 2003, and the second setting period is set to ten years of 2004 to 2013. The method of constructing the disease prognosis matrix using the inputted patient database data (12 years) is as follows.

First, 12 types of diseases sequentially input for three patients (hereinafter referred to as 'patient 1', 'patient 2' and 'patient 3' respectively) are five types of D1, D2, D3, D4 and D5, and each patient A database of star relationships may be as follows. The disease that occurs first below is referred to as a 'predecessor disease' and the disease that occurs later is referred to as a 'following disease'.

* Patient 1: D1-D1-D2-D3-D1-D4

* Patient 2: D3-D5-D1-D3

* Patient 3: D2-D1-D2-D3-D4-D1-D1-D2

Here, if it is assumed that the preceding disease affects the following disease, the method of analyzing it for matrix modeling consisting of the preceding disease and the following disease is the three methods described above. Second, a method of simplifying a combination of diseases that occur continuously in the raw data as a single disease, and third, an example of a method of simplifying the process to omit subsequent repeated diseases from the data based on the initial data only. Can be mentioned. Compared to the first method, as the second and third methods are processed, the throughput of data can be reduced and the processing speed can be increased.

For example, in the second method, the raw data of the preceding and the following diseases of the patients 1, 2 and 3 are processed as follows.

* Patient 1: D1-D2-D3-D1-D4

* Patient 2: D3-D5-D1-D3

* Patient 3: D2-D1-D2-D3-D4-D1-D2

And, in the third method, if the raw data of the preceding and the following diseases of the patient 1, patient 2 and patient 3 is processed as follows.

* Patient 1: D1-D2-D3-D4

* Patient 2: D3-D5-D1

* Patient 3: D2-D1-D3-D4

It is preferable to extract a pair of mutual diseases in consideration of the patient's posterior direction by applying the third simplified method, and the results are as follows.

* Patient 1: (D1 → D2), (D1 → D3), (D1 → D4), (D2 → D3), (D2 → D4), (D3 → D4)

* Patient 2: (D3 → D5), (D3 → D1), (D5 → D1)

Patient 3: (D2 → D1), (D2 → D3), (D2 → D4), (D1 → D3), (D1 → D4), (D3 → D4)

In addition, the computer which performs the method according to the embodiment of the present invention configures the disease pair matrix based on the following table.

Referring to <Table 1>, the disease pair matrix sets each of the preceding diseases corresponding to each disease (D1, D2, D3, D4, D5) to a row, the following disease to a column, and analyzes each of the databases analyzed from the database. The number of cases of each subsequent disease after the preceding disease can be expressed by each matrix.

In addition, the computer which performs the method according to an embodiment of the present invention generates a contingency table as follows to create a statistical model based on the relative risk of the disease D1 affecting the disease D2, The partition table inserted in the calculation and inserted as shown in Table 2 can be constructed.

Here, the computer performing the method according to an embodiment of the present invention can obtain a relative risk of 'D1-> D2', in which D1 disease affects D2 disease, by applying <Formula x> as follows.

Relative risk RR _{R- > D} = (a / (a + b)) ÷ (c / (c + d))

As described above, statistical significance is tested by the chi-square test or Fisher accuracy test, and the P-value, which is the significance level in the relationship of actual RR _{R-> D} , is 0.05, 0.01, 0.001. It is desirable to determine that only pairs that are less than the pre-set set significance level value and whose relative risk RR _{R-> D} are greater than the preset set relative risk are significant.

The computer performing the method according to an embodiment of the present invention may be repeatedly performed in order for all diseases of the disease cohort database in the above order.

As described above, data, data, information, and the like for various diseases may be derived or extracted by a method for predicting disease onset, which is the same as the above-described example, and the computer performing the method according to an embodiment of the present invention may provide such data. Easier to see the back and to visualize the disease network can be built and output at a glance, the process is as follows.

First, the disease is expressed as a node. First, the node is represented by a rectangle, and the size of the rectangle represents the average prevalence of the 12-year data of the above example, and the vertical value of the rectangle is male prevalence and the horizontal value is female. The prevalence is shown respectively, and the color of the rectangle is preferably represented by a different color by disease class (eg, gynecological disease, heart disease, etc.). Secondly, it is desirable to represent the node in a circle, the size of the circle represents the average survival rate of the 12-year data, and the color in different colors by disease class (eg, gynecology, heart disease, etc.).

In addition, an arrow may represent an edge representing the relative risk between diseases. The thickness of the edge represents the magnitude of the relative risk, and it is preferable to mark it in blue when the morbidity is high in men, red when the morbidity is high in women, and light green when the morbidity is within a certain range in men and women. .

An example visualized by such a display method is illustrated in FIG. 2.

In addition, since the disease network constructed according to an embodiment of the present invention uses information obtained from a cohort database, the disease network may be imaged not only by the relative risk between diseases but also by age, socioeconomic indicator, region, and institution. Therefore, the age-specific disease network visualization method for each age set in order to explore the disease network change according to each age (see FIG. 3), the sex disease network visualization method represented by sex in order to explore the disease network characteristics for gender (see FIG. 4), Disease network visualization method according to socioeconomic indicators set by socioeconomic indicators set to explore disease network characteristics according to socioeconomic indicators, and regional disease represented by region set to explore disease network characteristics of patients visiting or living in a specific area. Net Various visualization processes such as the visualization method of the nursing institution, which is represented by the nursing institution set up to explore the disease network characteristics of patients visited or hospitalized by the nursing institution including primary, secondary, tertiary and nursing hospitals. Can provide.

The disease onset prediction method and disease onset network visualization method using such a cohort may be stored and utilized as a computer readable recording medium for executing the method on a computer.

Accordingly, according to the present invention, the correlation between all diseases can be derived more efficiently and reliably based on diseases of patients using big data, and if the cohort is almost all citizens, the correlation between diseases occurring in Koreans It is possible to provide a method for predicting disease onset using a cohort, a method for visualizing disease onset network, and a computer-readable recording medium recording the same, which can be used as a platform for precision medicine that is most suitable for Koreans.

In addition, the characteristics of the disease can be shown in various forms such as age, sex, socioeconomic index, and nursing institution, and can be processed simply and conveniently. Also, the relationship between the genetic and non-genetic diseases of patients can be identified. It is possible to provide a method of predicting disease onset using a cohort, a method of visualizing a disease onset network, and a computer-readable recording medium recording the same.

In addition, a statistical model for predicting the extent to which a particular disease affects other diseases can be created and effectively visualized, and a method of predicting disease occurrence using a cohort that can construct and visualize a network of prognostic relationships of disease occurrence The present invention provides a method for visualizing a disease-occurring network and a computer-readable recording medium recording the same.

Here, although an embodiment of the present invention has been illustrated and described, it will be appreciated by those skilled in the art that the present embodiment may be modified without departing from the spirit or spirit of the present invention. will be. It is intended that the scope of the invention be defined by the claims appended hereto and their equivalents.

Claims (11)

In a method for constructing a disease onset network from a cohort database using a computer,
The controller classifying, from the cohort database, a person who does not develop the first target disease D1 among the plurality of diseases in the group A during the first period of time set as the period in which the target disease affects the dangerous disease. ;
The control unit may classify the number of persons having a second risk disease (R = D2) corresponding to a second target disease from the group A into the group E and generate a second risk disease (R = D2). Categorizing the unaffected personnel into group C;
The control unit, after the first setting period, conversely, during the second setting period in which the risk disease affects the target disease, the number of people having developed the first dangerous disease D1 in the E group is group E1. Classifying a person who does not develop a second risk disease (R = D2) in the C group into a C1 group;
The control unit is configured to generate the first relative value calculated from the relative number or relative frequency values of the E group and the E1 group, and the second relative value calculated from the relative number or relative frequency value of the C and C1 groups. Calculating a relative risk between the risk disease (R = D2) and the target disease (D1);
The control unit repeats each of the steps by sequentially setting different remaining preset diseases among the plurality of diseases as a target disease and a risk disease, and mutually correlating morbidity rates between diseases based on the relative risk. Constructing an indicative disease onset network; And
Outputting the disease-prone network;
A method for constructing a disease onset network of a computer cohort database. The method of claim 1,
Building the disease onset network,
Obtaining, from the data obtained by repeatedly performing the sequential repetition, raw data in which the predecessor and the following disease occurring for each patient are listed according to each group classification;
Simplifying the raw data; And
Using the simplified raw data to calculate a posterior relationship matrix between diseases;
A method for constructing a disease onset network of a computer cohort database. The method of claim 2,
The simplification process,
A disease that is continuously developed in the raw data is simplified to one disease outbreak, or a simplified process of eliminating the subsequent repeated disease from the raw data based on the original data only.
A method for constructing a disease onset network of a computer cohort database. The method of claim 3,
Computing the posterior relationship matrix between the diseases,
Extracting a plurality of disease pairs having a direction between a preceding disease and a following disease from the simplified raw data; And
On the basis of the plurality of disease pairs, the preceding diseases corresponding to each disease are set to rows, and the following diseases are set to columns, and the number of cases of each subsequent disease occurring after each preceding disease is indicated for each matrix. Calculating a posterior relationship matrix between diseases.
A method for constructing a disease onset network of a computer cohort database. The method of claim 4, wherein
Computing the posterior relationship matrix between the diseases,
Calculating the relative risk corresponding to each disease pair in the posterior relationship matrix between the diseases using a contingency table.
A method for constructing a disease onset network of a computer cohort database. The method of claim 5,
Building the disease onset network,
Constructing the disease onset network based on the posterior relationship matrix between the diseases and the relative risk,
Outputting the disease onset network,
And visualizing and outputting the posterior relationship matrix and the relative risk of the diseases in the disease occurrence network.
A method for constructing a disease onset network of a computer cohort database. The method of claim 6,
The step of visualizing and outputting,
Forming each disease as a node, forming the relative risk with an edge between the diseases as an edge, and image-processing the disease-prone network;
A method for constructing a disease onset network of a computer cohort database. The method of claim 7, wherein
The disease formed by the node is represented by a color-coded square or a circle, and the average prevalence corresponding to the first and second set periods is displayed as the size, and when the disease is represented as a rectangle, the prevalence rate Is characterized in that it is displayed to be divided as a horizontal or vertical length depending on the gender
A method for constructing a disease onset network of a computer cohort database. The method of claim 7, wherein
The relative risk formed by the edge is indicated by the arrow, the thickness indicates the magnitude of the relative risk, the color represents the relative morbidity rate (gender morbidity rate)
A method for constructing a disease onset network of a computer cohort database. The method of claim 7, wherein
The step of visualizing and outputting,
And visualizing and outputting a disease occurrence network classified by age, socio-economic indicator, region or institution by using the information obtained from the cohort database.
A method for constructing a disease onset network of a computer cohort database. A computer-readable recording medium having recorded thereon a program for executing the method according to any one of claims 1 to 10 on a computer.

Images