KR101546041B1 - Method and apparatus for converting data, and method for verifying data conversion - Google Patents

Abstract

The present invention relates to a data conversion method and apparatus for converting binary data of each of a plurality of binary data sets into real data, and a data conversion verification method. According to the present invention, It is possible to obtain real-valued data that can express a variety of states, for example, the process of moving from normal to abnormal or the degree of abnormality based on binary data that can represent one of the abnormal states.

Description

METHOD AND APPARATUS FOR DATA CONVERSION DATA, AND METHOD FOR CONVERTING DATA CONVERSION

The present invention relates to a data conversion method and apparatus, and a data conversion verification method, and more particularly, to a data conversion method and apparatus for converting binary data of each of a plurality of binary data sets into real data, and a data conversion verification method .

As the state data representing the state represented by the biometric data for analysis of disease association, it is common to use binary data that can be classified into only one disease or normal one. Particularly, in the case of a disease state, since the disease state is classified only after completion of the diagnosis by the disease among the stages progressing to the disease, in the case of the patient who is in the process of changing from the normal state to the disease state, It is difficult to express the patient and it is difficult to separately extract the patient in the process of changing from the normal state to the disease state when biometric data is searched based on the current state data.

To solve this problem, it is necessary to convert binary data, which can be distinguished only to either disease or normal, into real data, which can indicate the degree of specific association with the disease.

SUMMARY OF THE INVENTION The object of the present invention is to provide a method and apparatus for representing binary data of each of a plurality of binary data sets including at least one measurement data and binary data representing two states in the state of such measurement data, A data conversion method and apparatus for converting data into real data, and a data conversion verification method.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not intended to limit the invention to the particular embodiments that are described. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention, There will be.

According to an aspect of the present invention, there is provided a method for transforming data, the method including converting binary data representing one of a first state and a second state into at least one measurement data and a state of the measurement data, A binary data set input step of inputting a plurality of sets of binary data; A real data type permutation generation step of generating a plurality of real data type permutation data in which different real data types are allocated to each of the plurality of binary data sets based on binary data of each of the plurality of binary data sets; Performing a multiple regression analysis on each of the generated real data type permutations to derive a regression equation corresponding to each of the real data type permutations; An optimal regression equation selecting step of selecting one optimal regression equation based on at least one of a decision coefficient of the derived regression equations, an intersection verification correlation coefficient of the decision coefficient, and a Pearson correlation coefficient; And providing real data types corresponding to the plurality of binary data sets based on the selected optimal regression equation.

According to another aspect of the present invention, there is provided a data conversion verification method for converting binary data representing one of a first state and a second state into at least one measurement data and a state of the measurement data, Categorizing a portion of a plurality of binary data sets including a plurality of binary data sets for learning into a plurality of binary data sets for learning, and classifying the remaining portions into a plurality of binary data sets for verification; A real data type permutation generation step of generating a plurality of real data type permutations in which different real data types are allocated to each of the learning binary data sets based on the binary data of each of the learning binary data sets; Performing a multiple regression analysis on each of the generated real data type permutations to derive a regression equation corresponding to each of the real data type permutations; An optimal regression equation selecting step of selecting one optimal regression equation based on at least one of a decision coefficient of the derived regression equations, an intersection verification correlation coefficient of the decision coefficient, and a Pearson correlation coefficient; Calculating real data types corresponding to the verification binary data sets by applying the selected optimal regression equation to the verification binary data set; And verifying the optimal regression equation based on the real data corresponding to the verification binary data sets.

According to another aspect of the present invention, there is provided a data conversion apparatus for converting binary data representing one of a first state and a second state into a state of at least one measurement data and a measurement data, A binary data set loading unit loading at least a portion of the plurality of binary data sets; A real data type permutation generation unit for generating a plurality of real data type permutation data in which different real data types are allocated to each of the plurality of binary data sets based on binary data of each of the loaded binary data sets; A regression derivation unit for performing a multiple regression analysis on each of the generated real data type permutations to derive regression equations corresponding to the real data type permutations; A regression equation selection unit for selecting one optimal regression equation based on at least one of a decision coefficient of the derived regression equations, an intersection verification correlation coefficient of the decision coefficient, and a Pearson correlation coefficient; And a real-valued data calculation unit for calculating real-valued data corresponding to each of the plurality of binary data sets based on the selected optimal regression equation.

According to another aspect of the present invention, there is provided a computer readable recording medium recording a program for causing a computer to execute the data conversion method or the data conversion verification method.

According to the present invention, it is possible to obtain real-valued data that can express more various states, for example, a process of moving from normal to abnormal or a degree of abnormality based on binary data that can represent one of normal and abnormal states with respect to measurement data .

Particularly, when the data conversion method according to the present invention is applied to a process of biometric data of a patient who has been classified into one of existing diseases and normal ones, the patient's state changes from normal to disease through real- It is possible to indicate the presence of a disease or a degree close to a disease, so that the biometric data can be more effectively provided for predicting and diagnosing a disease.

1 is a block diagram illustrating a data conversion apparatus according to an embodiment of the present invention.
2 is a flowchart illustrating a data conversion method according to an exemplary embodiment of the present invention.
3 is a diagram illustrating a process of generating a plurality of real data type permutations using a brute force search method in a data conversion method according to an embodiment of the present invention.
4 is a diagram illustrating a data conversion verification process for verifying an optimal regression equation selected in a data conversion method according to an exemplary embodiment of the present invention.
FIG. 5 is a diagram illustrating a result of generating real-valued data representing a patient's state by applying a data conversion method according to a preferred embodiment of the present invention to biometric data processing of a patient.

The following merely illustrates the principles of the invention. Thus, those skilled in the art will be able to devise various apparatuses which, although not explicitly described or shown herein, embody the principles of the invention and are included in the concept and scope of the invention. Furthermore, all of the conditional terms and embodiments listed herein are, in principle, intended only for the purpose of enabling understanding of the concepts of the present invention, and are not intended to be limiting in any way to the specifically listed embodiments and conditions . It is also to be understood that the detailed description, as well as the principles, aspects and embodiments of the invention, as well as specific embodiments thereof, are intended to cover structural and functional equivalents thereof. It is also to be understood that such equivalents include all elements contemplated to perform the same function irrespective of currently known equivalents as well as equivalents to be developed in the future.

Thus, the functions of the various elements shown in the drawings, including the functional blocks shown in the figures or similar concepts, may be provided by use of dedicated hardware as well as hardware capable of executing software in connection with appropriate software. When provided by a processor, the functions may be provided by a single dedicated processor, a single shared processor, or a plurality of individual processors, some of which may be shared. Also, the use of terms such as processor, control, or similar concepts should not be construed as exclusive reference to hardware capable of executing software, but may include, without limitation, digital signal processor (DSP) hardware, (ROM), random access memory (RAM), and non-volatile memory. Other hardware may also be included.

The above objects, features and advantages will become more apparent from the following detailed description in conjunction with the accompanying drawings. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

On the other hand, when an element is referred to as "including " an element, it does not exclude other elements unless specifically stated to the contrary.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating a configuration of a data conversion apparatus 100 according to a preferred embodiment of the present invention.

Referring to FIG. 1, a data conversion apparatus 100 according to the present invention includes a storage unit 110, a binary data loading unit 120, a real data permutation generation unit 130, a regression formula derivation unit 140, A regression formula selecting unit 150, an optimal regression formula verifying unit 160, and a display unit 170. [ It is apparent that other components other than the above-described components can be included in the data conversion apparatus 100 according to the present embodiment.

The binary data loading unit 120, the real data permutation generator 130, the regression equation derivation unit 140, the optimal regression equation selection unit 150, May be included in the data conversion apparatus 100 as an operating system, an application program module, and other program modules, and may be physically various Type of known memory device. These program modules may also be stored in a remote storage device capable of communicating with the data conversion apparatus 100. [ These program modules, on the other hand, encompass routines, subroutines, programs, objects, components, data structures, etc., that perform particular tasks or perform particular abstract data types as described below in accordance with the present invention, Do not.

The storage unit 110 is a component for storing a plurality of sets of binary data, and may be a storage medium in the data conversion apparatus 100 according to the present embodiment or a DBMS on the storage medium.

The binary data set according to the present embodiment may be configured to include at least one measurement data and status data indicating the status of such measurement data. Generally, such status data includes one of two states such as normal and abnormal And binary data that can be used.

A representative binary data set may be a set of diagnostic data of a patient used in a medical information database or the like.

The patient's diagnostic data set includes biometric data such as hormone measurements of the patient and also indicates the status or diagnosis result of the patient represented by such biometric data as one of disease and normal status, As the data type of the state data representing the result, binary data having a value of 0 or 1 is mainly used.

The binary data set loading unit 120 loads data sets corresponding to at least some of the plurality of binary data sets stored in the storage unit 110 for data conversion.

The real data type permutation generation unit 130 assigns different real data types to each of the binary data sets based on the binary data of each of the binary data sets loaded in the binary data set loading unit 120, And generates a plurality of real data type permutations having a size as many as the number of data sets.

In this case, the real data in each of the binary data sets is arbitrarily assigned with a different real number value, and generates a real data type permutation for all possible arrays of real values within the predetermined range for the brute force search.

At this time, when real data is allocated, binary data sets having a binary data value of "1" and binary data sets having a binary data value of "0" are classified into each other, and real data .

For example, different binary data sets having a binary data value of " 1 " are assigned different real data types having a value of 1 or more, and binary data sets having a binary data value of " Different real data can be allocated.

Thus, if there are a total of 14 binary data sets loaded in the binary data set loading unit 120 and the binary data set consists of 8 training sets and 6 test sets, the data set of all converted orders is 8! * 6! = 29,030,400.

Assuming that there are six binary data sets with binary data values of "1" and eight binary data sets among the total of 14 binary data sets, and the binary data value is "1", the real data type permutation generation unit 130, The real-number data permutation generator 130 generates a permutation of 8! * 6! = 29,030,400 data permutations.

The regression equation derivation unit 140 may perform a multiple regression analysis on each of the real data type permutations generated by the real data type permutation generation unit 130 to derive regression equations corresponding to each of the real data type permutations have.

The multiple regression analysis according to the present embodiment can be applied to a multiple regression analysis model previously known by the inventors of the present application (refer to Moon, T., Chi, MH, Kim, DH, Yoon, CN, activity relationship (QSAR) study of flavonoid derivatives for inhibition of cytochrome P450 1A2. Quant Struct-Act Rel 19 (3), 257-263, 2000). In this process, Method can be used. The multi-collinearity among the variables of the multiple regression analysis model can use a variance inflation factor (see, eg, Myers, RH, Classical and modern regression withapplications, PWS / KENT: Boston, 1990). However, it is to be understood that the present invention is not limited to the above embodiment, and various other regression analysis models may be applied to the present invention, It can be done.

The regression equation selection unit 150 selects at least one of the determination coefficient r ² of the regression equations derived from the regression equation derivation unit 140, the cross correlation verification coefficient q ² of the determination coefficient, and the Pearson correlation coefficient PCC , One of the regression equations derived from the regression equation derivation unit 140 is selected.

The decision coefficient (r ² or r square) of the regression equation according to the multiple regression analysis is a coefficient that can evaluate the predictive ability of the regression equation, and can be defined as Equation 1 below.

[Equation 1]

r ² = 1.0 -? (y _pred- y _actual ) ² /? (y _pred- y _mean ) ²

Where y _pred means the y value predicted by the regression equation, y _actual means the _actual value of the y value corresponding to y _pred , and y _mean means the _mean value of these real values.

In addition to the coefficient of determination (r ² ) of this regression equation, the cross-validated correlation coefficient r ^{2 (} or q ² or q square) of the coefficient of determination by the above-described cross- It can be used as a criterion to decide.

In addition, Pearson correlation coefficients (PCC) (Pearson, K., Mathematical contributions to the theory of evolution, III. Regression, heredity and panmixia, Philos. (187, 253-318, 1896) can be used as a criterion for determining the optimal regression equation.

The optimal regression equation verifying unit 160 may output a binary data set that is different from the binary data set used to derive the optimal regression equation previously loaded from the binary data sets stored in the storage unit 110, (120), and the optimal regression equation selected by the regression equation selection unit (150) based on the other set of the binary data sets is used to derive real data representing the state of at least one measurement data Verify that it is suitable for your environment.

For example, the optimal regression equation verifying unit 160 applies a set of other loaded binary data to the optimal regression equation selected by the regression equation selecting unit 150, and calculates a real number corresponding to each of the other partial binary data sets Type data can be derived and the optimal regression equation can be verified based on the derived result.

The display unit 170 displays the binary data sets loaded through the binary data loading unit 120 in the optimal regression formula selected by the regression formula selecting unit 150, Data, or various states of the measurement data of the corresponding binary data set, which each of the real-valued data means.

In addition, the display unit 170 can display the optimal regression equation selected by the regression equation selector 150, the verification result of the optimal regression equation by the optimal regression equation verification unit 160, and the like.

The display unit 170 may provide visual or auditory information to the user of the data conversion apparatus according to the present embodiment. In order to provide visual information, the display unit 170 may include a liquid crystal display (LCD) , A TFT (Thin Film Transistor), and an organic EL (Organic Electroluminescence) device.

FIG. 2 is a flowchart illustrating a data conversion method according to an embodiment of the present invention. The data conversion method according to the present embodiment may be performed in the data conversion apparatus 100 shown in FIG. Therefore, the same matters as those in the data conversion apparatus 100 shown in Fig. 1 are referred to.

First, a plurality of sets of binary data to be converted into real data are input into the binary data included in each of the data sets (S201).

In step S202, a plurality of real data type permutation data in which different real data types are allocated to each of the plurality of binary data sets is generated based on the binary data of each of the binary data sets received in step S201. At this time, it is desirable to randomly assign real-valued data to each of the binary data sets with different real-valued values and to generate a real-valued data permutation for all possible arrays of real values within the predetermined range for the brute force search Do.

A multiple regression analysis is performed on each of the real data type permutations generated in step S202 to derive regression equations corresponding to real data type permutations (S203).

Determining coefficients of a regression equation derived from the S203 step (r ^2), and selects the one best regression equation on the basis of at least one of a cross-validation correlation coefficient (q ²⁾ and Pearson's correlation coefficient (PCC) in the coefficient of determination (S204 ).

For example, when selecting the optimal regression equation in step S204, a regression equation with an r2 value of 0.6 or more, a q2 value of 0.6 or more, or a Pearson correlation coefficient value of 0.6 or less among the regression equations derived in step S203 is selected as an optimal regression equation . At this time, the r2 and q2 values show the fitness and predictive power of the regression equation, and it is preferable that they have a value of 0.6 or more. In addition, the PCC value is a value showing the correlation between variables. For example, in the case of a regression equation in which the PCC value is 0.6 or more, it means that the variables of the regression equation are not independent variables. It can be judged.

In step S205, real-type data corresponding to each of the binary data sets is provided based on the optimal regression equation selected in step S204.

The real-valued data provided in the step S205 includes various information, for example, normal to abnormal, in addition to the fact that the state of the measurement data (s) of the binary data set corresponding to the corresponding real- It is possible to express the degree of movement or the degree of abnormality.

FIG. 3 is a diagram illustrating a process of generating a plurality of real data type permutations using a brute force search method in a data conversion method according to an embodiment of the present invention. In the real data type permutation generation process according to the present embodiment, May be performed in the real data type permutation generation unit 130 shown in FIG. 1 and / or the real data type permutation generation step shown in FIG. 2 (S203). Therefore, the same matters as those in the data conversion apparatus 100 shown in Fig. 1 and / or the data conversion method shown in Fig. 2 are referred to.

The plurality of binary data sets are classified into the first group corresponding to the first state and the second group corresponding to the second state based on the binary data of each of the plurality of binary data sets to be converted into the real data type data ). Here, the first state and the second state are the states of the measurement data (s) included in the binary data set, respectively, and may be " normal " and " abnormal " The binary data values represented may be " 1 " and " 0 ".

Different real data having one or more values (e.g., 1.1, 1.2, 1.3, etc.) may be assigned to the binary data sets belonging to the first group corresponding to the first state (" normal "

It is also possible to assign different real data types having values of 0 to less than 1 (e.g., 0.1, 0.2, 0.3, etc.) to the binary data sets belonging to the second group corresponding to the second state (S330).

It is checked whether different real data types are allocated to all of the plurality of binary data sets to be converted into real data (S340). Then, one of the plurality of binary data sets having different real data types A real data type permutation is generated (S350).

In step S360, it is checked whether a real-number data permutation of another array is available for searching for brute force. If there are other possible arrays, the real data permutation generation process of steps S320 to S340 is repeatedly performed Generates real data permutations for all arrays.

For example, seven sets of binary data, each consisting of actual data of hormone concentrations for each of the seven patients and their binary data classified as normal and abnormal, respectively, (A) and (B) abnormal (disease) soldiers belonging to four patients.

1.2, and 1.3 different real data types were assigned to the binary data sets of the three patients in the normal army group A and 0.1, 0.2, 0.3, and 0.4 were assigned to the binary data sets of the four patients in the abnormal group B group, respectively 3! * 3, when converting binary data of these seven sets of binary data into real data under the assumption of assigning different real data types of the binary data sets. = 144 real data sets can be generated.

FIG. 4 illustrates a data conversion verification process for verifying an optimal regression equation selected in the data conversion method according to an exemplary embodiment of the present invention. The data conversion verification process may be performed in the data conversion apparatus 100 shown in FIG. Therefore, the same matters as those in the data conversion apparatus 100 shown in Fig. 1 are referred to.

A part of the plurality of pre-stored binary data sets is classified into a learning binary data set (S410), and the remaining part is classified into a binary data set for verification (S420).

The candidate regression equations of the optimal regression equation are derived on the basis of the learning binary data set classified in step S410. The process of deriving the candidate regression equations refers to the process of deriving the regression equations corresponding to the real data type permutations described in the previous description of FIG. 1 to FIG.

For example, if 144 real-valued data sets are generated, as in the example described in the description of FIG. 3, then 144 multiple candidate regression equations can be derived by performing multiple regression analysis on each of them.

For example, one of the candidate regression equations may be Equation 2 below.

&Quot; (2) "

Index = 2.7 V ₁ - 3.2 V ₂ + 2.5 V ₃ (r ² = 0.92, q ² = 0.67, MaxPCC = 0.5)

Here, V ₁ , V ₂ , and V ₃ are variables of the regression equation (measurement data included in the binary data set), and Index is real data derived by the regression equation.

At this time, as a criterion for selecting the optimal regression equation from the candidate regression equations, the first threshold condition r ² > 0.6), a second threshold condition (q ² > = 0.6), and a third threshold condition (Max (PCC) <= 0.6).

Here, since the Pearson correlation coefficient is a correlation between the variables making up the regression equation, there are three Pearson correlation coefficients of the equation (2): PCC _v1v2 , PCC _v2v3 , and PCC _v1v3 . The third threshold condition is applied.

In this embodiment, the first threshold condition r ² The optimal regression equation is selected on the basis of whether there is a regression equation that satisfies at least one of the second critical condition (q ² > = 0.6), the second critical condition (q ² > = 0.6), and the third critical condition (Max (PCC) <= 0.6).

For example, the first threshold condition r ² > = 0.6), the second threshold condition (q ^2> = 0.6), the third threshold condition (Max (PCC) is the highest, if present more than one regression equation that satisfies at least one of <= 0.6) (S450) If the r ² value is represented by Hr ² , the regression equation having the highest q ² among the regression equations satisfying the fourth threshold condition having the r ² value between Max (0.6, 0.9Hr ² ) and Hr ² is optimized And can be selected by a regression equation (S460).

If the first threshold condition r ² If there is only one regression equation satisfying at least one of the second threshold condition (q ² > = 0.6), the third threshold condition (Max (PCC) <= 0.6) One regression equation can be selected as the optimal regression equation (S480).

If the first threshold condition r ² , The second threshold condition (q ² > = 0.6), and the third threshold condition (Max (PCC) <= 0.6), there is no regression equation satisfying at least one of the first threshold condition Another part of the data sets is re-classified into a binary data set for learning (S410).

The optimal regression equation selected in step S460 or step S480 may be applied to the verification binary data set classified in step S420 to calculate real data corresponding to the verification binary data sets in step S490.

The optimal regression equation is verified based on the real data of each of the verification binary data sets calculated in S490 (S500).

FIG. 5 is a diagram illustrating a result of generating real-valued data representing a patient's state by applying a data conversion method according to a preferred embodiment of the present invention to biometric data processing of a patient.

In order to apply the data conversion method according to the present invention to the analysis of biomedical data of thyroid diseases, the urine of the subjects (patients) including thyroid cancer patients and normal persons is analyzed by Gas Chromatography-Mass Spectrometry-Selected Ion-Monitoring system (Hewlett-Packard GC-MS: 5890A GC model, 5970B mass-selective detector, HP 59970C MS chemstation) for 18 metabolite hormones on androgen and estrogen metabolic pathways To obtain profile data and to identify risk factors for thyroid disease among these hormones, t-test analysis was performed on the corresponding hormone profile data. As a result, the total of 18 metabolic hormones, 2-hydroxyestrone (2- OH-E1), 2-hydroxyestradiol (2-OH-E2), 2-methoxyestrone (2-MeO-E1), 2-methoxyestradiol Ethoxyestradiol-3-methyl ether (2-MeO-E2-3- Butyl ether) hormone was found to be a significant risk factor for determining the diagnosis and progression of thyroid cancer.

In the present embodiment, the state data of eight normal persons are set to 0, and the state data of six thyroid cancer patients are set to 1 to perform real data conversion on a total of 14 binary data sets.

In order to convert binary data of each of the binary data sets according to the present embodiment into real data, real data of 0.2 to 0.7 is allocated to normal persons and real data of 1.2 to 1.9 are allocated to patients of thyroid cancer To generate all possible real-valued data permutations. As a result, 8! = 29,030,400 real data type permutations were obtained.

The regression equations corresponding to all of these real data type permutations are obtained and a regression equation such as the following equation (3) can be obtained as the regression equation having the best result among them, for example, having the largest determination coefficient (r ² ) .

Hereinafter, the real-valued data obtained through the regression equation of Equation (3) is referred to as the thyroid disease index (TDI), which means the degree of progression to thyroid disease.

&Quot; (3) &quot;

Thyroid Disease Index (TDI) = 0.0153 * 2-MeO-E2-3-methyl ether + 0.0874 * 2-MeO-E2 / 2-OH-E2-2.2571 * 2-MeO- -E1 + 0.5649

Referring to FIG. 5, biometric data 501 of each subject included in the binary data set and the real data 502 calculated by applying the biometric data to Equation (3) are illustrated.

As a result of comparison with actual patient information (patient information), the thyroid disease index ranged from 0 to 0.8 in normal subjects, and a thyroid disease index of 1.0 or more was found in patients with thyroid cancer. Four subjects located in the border area below the index of 1.0 were classified as thyroid mass (0.8055), goiter (0.8806), thyroid mass (preoperative thyroid cancer, 0.8951), thyroid mass (preoperative thyroid cancer, 0.9112) Respectively. According to these results, since the thyroid disease index, which is the best regression equation obtained according to the present embodiment, was found to increase in accordance with the progression of thyroid cancer, the thyroid disease was diagnosed according to the regression formula of Equation 3, It is possible to classify the degree.

Based on the range of the thyroid disease index obtained according to the thyroid disease index obtained according to the present embodiment, it is possible to determine the presence or absence of thyroid cancer and the progress of the thyroid cancer. As described above, it can be seen that the data conversion method according to the present embodiment can be particularly useful for biomedical data for analyzing the association of diseases.

According to the present invention, it is possible to obtain real-valued data that can express more various states, for example, a process of moving from normal to abnormal or a degree of abnormality based on binary data that can represent one of normal and abnormal states with respect to measurement data .

Particularly, when the data conversion method according to the present invention is applied to a process of biometric data of a patient who has been classified into one of existing diseases and normal ones, the patient's state changes from normal to disease through real- It is possible to indicate the presence of a disease or a degree close to a disease, so that the biometric data can be more effectively provided for predicting and diagnosing a disease.

In addition, according to the present invention, a data conversion method for converting binary data of a binary data set into real data and a data conversion verification method for verifying such data conversion are also computer readable codes on a computer readable recording medium It is possible to implement. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like. In addition, the computer-readable recording medium may be distributed over network-connected computer systems so that computer readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily inferred by programmers of the technical field to which the present invention belongs.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, You will understand. Accordingly, the true scope of protection of the present invention should be determined only by the appended claims.

100: Data conversion device
110:
120: loading section
130: Real-type data permutation generator
140: regression equation derivation unit
150: Regression Expression Selection
160: Optimal Regression Equation Verification Unit
170:

Claims (17)

A binary data set input step of inputting a plurality of sets of binary data including at least one measurement data and binary data representing one of a first state and a second state in a state of the measurement data;
A real data type permutation generation step of generating a plurality of real data type permutation data in which different real data types are allocated to each of the plurality of binary data sets based on binary data of each of the plurality of binary data sets;
Performing a multiple regression analysis on each of the generated real data type permutations to derive a regression equation corresponding to each of the real data type permutations;
An optimal regression equation selecting step of selecting one optimal regression equation based on at least one of a decision coefficient of the derived regression equations, an intersection verification correlation coefficient of the decision coefficient, and a Pearson correlation coefficient; And
And providing real data types corresponding to the plurality of binary data sets based on the selected optimal regression equation. The method according to claim 1,
The real data type permutation generation step
Classifying the plurality of binary data sets into a first group corresponding to the first state and a second group corresponding to the second state based on the binary data of each of the plurality of binary data sets Further included,
Wherein each of the real data type permutations assigns real data of a section that does not overlap with each other between binary data sets belonging to the first group and binary data sets belonging to the second group. 3. The method of claim 2,
Wherein each of the real data type permutations assigns different real data types having one or more values to binary data sets belonging to the first group and assigns a value of 0 or more and less than 1 to the binary data sets belonging to the second group And assigning different real-valued data having the same value to each other. The method according to claim 1,
The step of inputting the binary data set may include classifying a portion of the plurality of binary data sets into a plurality of sets of binary data for learning and classifying the remaining portions of the plurality of binary data sets into a plurality of sets of binary data for verification Further included,
Wherein the real data type permutation generation step generates the real data type permutations based on the binary data sets for learning among the plurality of binary data sets,
Wherein the optimal regression equation selected in the optimal regression equation selection step is an optimal regression equation for the learning binary data sets. 5. The method of claim 4,
Calculating real data types corresponding to the verification binary data sets by applying an optimal regression equation for the learning binary data sets to the verification binary data sets; And
And verifying the optimal regression equation based on real-valued data corresponding respectively to the verification binary data sets. The method according to claim 1,
The optimal regression equation selection step
At least one of a predetermined first threshold condition with respect to the decision coefficient among the derived regression equations, a predetermined second threshold condition with respect to the cross validation correlation coefficient of the decision coefficient, and predetermined threshold conditions with respect to the Pearson correlation coefficient And extracting a regression equation that satisfies the threshold condition. The method according to claim 6,
The optimal regression equation selection step
When the extracted regression equations are a plurality of regression equations having the highest cross correlation coefficient among regression equations having a decision coefficient satisfying a fourth threshold condition set based on a maximum value among the decision coefficients of the extracted plurality of regression equations Wherein the regression equation is selected by an optimal regression equation. The method according to claim 1,
Wherein the measurement data is biometric data for diagnosis of a specific disease,
The binary data is disease occurrence data indicating whether or not a specific disease has occurred,
Wherein the real type data is disease index data indicating a degree of progress of a specific disease. Classifying a part of a plurality of binary data sets including at least one measurement data and binary data representing one of a first state and a second state into a state of the measurement data into a plurality of binary data sets for learning, A binary data extracting step of classifying the binary data sets into a plurality of verification binary data sets;
A real data type permutation generation step of generating a plurality of real data type permutations in which different real data types are allocated to each of the learning binary data sets based on the binary data of each of the learning binary data sets;
Performing a multiple regression analysis on each of the generated real data type permutations to derive a regression equation corresponding to each of the real data type permutations;
An optimal regression equation selecting step of selecting one optimal regression equation based on at least one of a decision coefficient of the derived regression equations, an intersection verification correlation coefficient of the decision coefficient, and a Pearson correlation coefficient;
Calculating real data types corresponding to the verification binary data sets by applying the selected optimal regression equation to the verification binary data set; And
And verifying the optimal regression equation based on real-valued data corresponding to the verification binary data sets. 10. The method of claim 9,
The real data type permutation generation step
Classifying the plurality of binary data sets into a first group corresponding to the first state and a second group corresponding to the second state based on the binary data of each of the plurality of binary data sets Further included,
Wherein each of the real data type permutations is allocated real data of a section that does not overlap with each other between binary data sets belonging to the first group and binary data sets belonging to the second group. . 11. The method of claim 10,
Wherein each of the real data type permutations assigns different real data types having one or more values to binary data sets belonging to the first group and assigns a value of 0 or more and less than 1 to the binary data sets belonging to the second group And a plurality of different real-valued data having the plurality of real-valued data are allocated. 10. The method of claim 9,
The step of inputting the binary data set may include classifying a portion of the plurality of binary data sets into a plurality of sets of binary data for learning and classifying the remaining portions of the plurality of binary data sets into a plurality of sets of binary data for verification Further included,
Wherein the real data type permutation generation step generates the real data type permutations based on the binary data sets for learning among the plurality of binary data sets,
Wherein the optimal regression equation selected in the optimal regression equation selection step is an optimal regression equation for the learning binary data sets. 10. The method of claim 9,
The optimal regression equation selection step
A second threshold condition preset for the cross correlation correlation coefficient of the decision coefficient, and a third threshold condition preset for the Pearson correlation coefficient with respect to the decision coefficient among the derived regression equations And extracting a regression equation that satisfies one or more threshold conditions. 14. The method of claim 13,
Wherein the optimum regression equation selecting step includes a step of calculating a best fit among the regression equations having a decision coefficient satisfying a fourth threshold condition set on the basis of a maximum value among decision coefficients of the extracted plurality of regression equations when there is a plurality of the extracted regression equations And selecting a regression equation having a high cross-validation correlation coefficient as an optimal regression equation. 14. The method of claim 13,
The optimal regression equation selecting step may include returning to the binary data extracting step if all of the derived regression equations do not satisfy both of the first threshold condition, the second threshold condition, and the third threshold condition The data conversion verification method comprising: A binary data set loading unit loading at least a part of a plurality of binary data sets including at least one previously stored measurement data and binary data representing one of a first state and a second state in a state of the measurement data;
A real data type permutation generation unit for generating a plurality of real data type permutation data in which different real data types are allocated to each of the plurality of binary data sets based on binary data of each of the loaded binary data sets;
A regression derivation unit for performing a multiple regression analysis on each of the generated real data type permutations to derive regression equations corresponding to the real data type permutations;
A regression equation selection unit for selecting one optimal regression equation based on at least one of a decision coefficient of the derived regression equations, an intersection verification correlation coefficient of the decision coefficient, and a Pearson correlation coefficient; And
And a real-valued data calculation unit for calculating real-valued data corresponding to each of the plurality of binary data sets based on the selected optimal regression equation. A computer-readable recording medium recording a program for causing a computer to execute the method according to any one of claims 1 to 15.

Images