An endpoint search method for a specific data cluster and its data processing system

Abstract

The present invention discloses an endpoint search method for a specific data cluster and a data processing system thereof. The endpoint search method for a specific data cluster comprises: a) a step of receiving, through a search system, respective values ​​of a plurality of individual data included in the data set; b) a step of generating histogram data, i.e., by using the respective values ​​received by the search system, dividing the possible value range of the values ​​into a plurality of groups with a predetermined group interval, and taking the number of individual data corresponding to each segmented group as a group value; and c) a step of the search system searching for a target group of terminals existing in a specific cluster based on the histogram data.

Description

Technical Field

[0001] The present invention relates to a method for searching for endpoints of a specific data cluster and a data processing system thereof, and more specifically, to a method and a system thereof that can effectively search for endpoints (e.g., an end point or a starting point, etc.) of a specific data cluster in a data set having at least one data cluster. Background Art

[0002] Large amounts of data are being analyzed and utilized in many technical or service fields.

[0003] For example, analyzing specific medical data to determine whether to use drugs for different patients or to adopt specific treatment methods for individuals is being actively applied.

[0004] Figure 1 An example of using medical data in a predetermined companion diagnostic tool, which is a means for selecting patients suitable for a predetermined target anti-cancer agent, is shown.

[0005] Figure 1 As an example, a test kit (e.g., GenesWell) for detecting the presence or absence of mutations in a specific gene (e.g., exons 18, 19, 20, 21) in EGFR, a representative biomarker in lung cancer, is shown. TM The medical data obtained by ddEGFR mutation test). By analyzing these medical data, patients who are effective for therapeutic agents can be selected before prescribing targeted anticancer drugs after surgery. However, the technical idea of ​​the present invention is not limited to the application of the embodiments, and can be used for various data analysis.

[0006] In addition, if Figure 1 As shown in FIG. 1 , each point in the figure represents individual medical data. Figure 1 The various medical data are clustered into at least one cluster (e.g., Figure 1 three clusters in ).

[0007] At this point, it may be necessary to determine the endpoint in a specific data cluster (for example, at least one individual medical data located at the top in the first cluster, i.e., the lowermost data cluster (for example, at least one of the data with the highest y-axis value) or the numerical value (y-axis value) of such individual medical data).

[0008] However, it is difficult to know which individual medical data is included in which individual cluster only from the numerical value of the data or the coordinate position displayed in the coordinate system, and the difficulty is further increased when there are multiple individual medical data between data clusters.

[0009] In fact, if Figure 1As shown, in the past, a human visually inspects each medical data displayed in a coordinate system and randomly draws a dividing line 10 for distinguishing endpoints (eg, the upper endpoint in the first (lowest) cluster).

[0010] However, in this case, there may be a problem that the endpoint varies depending on the person who performs the task and the accuracy decreases.

[0011] Detailed description of the invention

[0012] Technical issues

[0013] Therefore, the technical problem to be solved by the present invention is to provide a method and system for quickly and automatically searching for endpoints of a specific data cluster in a data set having at least one data cluster.

[0014] Technical solutions

[0015] According to one aspect of the present invention, a method for searching for endpoints of a specific cluster in a data set having at least one cluster includes: a) a step of receiving, through a search system, respective numerical values ​​of a plurality of individual data included in the data set, b) a step of generating histogram data, i.e., by using the respective numerical values ​​received by the search system, dividing the numerical range that the numerical value can have into a plurality of groups (bins) with predetermined group intervals, and taking the number of individual data corresponding to each divided group as the group value, c) a step of smoothing the histogram data, d) a step of differentiating the smoothed histogram data, and e) a step of searching, based on the differentiated histogram data, for a target group of terminals that both meet predetermined benchmark conditions and exist in a specific cluster.

[0016] Based on the differential histogram data, the step of searching for a target group of terminals that meet a predetermined benchmark condition and exist in a specific cluster may include: searching for group values ​​of each group in a certain direction in the differential histogram data, and searching for a group that meets the benchmark condition as the target group, wherein the benchmark condition is that the first group satisfies the group that is currently searched and the group value before is less than the group value after, the previous group value is equal to or less than 0, and the subsequent group value is equal to or greater than 0.

[0017] The endpoint search method for a specific data cluster also includes: f) when a target group satisfying the benchmark condition is not found using a predetermined group distance, reducing the group distance to a predetermined value, and performing steps b) to e) again using the reduced group distance.

[0018] According to another embodiment for solving the technical problem, a method for searching for endpoints of a specific cluster in a data set having at least one cluster includes: a) a step of receiving, through a search system, respective numerical values ​​of a plurality of individual data included in the data set, b) a step of generating histogram data, i.e., by using the respective numerical values ​​received by the search system, dividing the numerical range that the numerical value can have into a plurality of groups with predetermined group intervals, and taking the number of individual data corresponding to each divided group as the group value, c) a step of smoothing the histogram data, and d) a step of searching, based on the smoothed histogram data, for a target group of terminals that both meet a predetermined benchmark condition and exist in a specific cluster.

[0019] The step of searching for the target group that meets a predetermined benchmark condition based on the smoothed histogram data may include searching for the group value of each group in a certain direction in the smoothed histogram data, and searching for the group that meets the benchmark condition as the target group, wherein the benchmark condition is that the group value before the group currently being searched is not a cutoff value, the current group value is a cutoff value, and at least one subsequent group value is a cutoff value.

[0020] The method can be implemented by a computer program installed in a data processing system.

[0021] A system for solving the technical problem includes a processor and a storage device, in which software executed by the processor is stored, and the software receives individual numerical values ​​of multiple individual data included in a data set having at least one cluster, divides the numerical range that the numerical value can have into multiple groups with predetermined group intervals by using the received individual numerical values, and generates histogram data using the number of individual data corresponding to each divided group as a group value, and searches for a target group of terminals existing in a specific cluster based on the generated histogram data.

[0022] The software may smooth the histogram data, differentiate the smoothed histogram data, and search for a target group of terminals that both satisfy a predetermined reference condition and exist in a specific cluster based on the differentiated histogram data.

[0023] The software may smooth the histogram data and search for the target group satisfying a predetermined reference condition based on the smoothed histogram data.

[0024] When a target group that meets the benchmark condition is not found using a predetermined group distance, the software can reduce the group distance to a predetermined value, regenerate a histogram using the reduced group distance, and use the regenerated histogram to re-search the target group of terminals existing in the specific cluster.

[0025] Effects of the Invention

[0026] According to the technical idea of ​​the present invention, the present invention has the effect of quickly and automatically searching for the end point of a specific data cluster using the numerical value of the individual data without individually clustering a plurality of individual data.

[0027] This approach has the effect of enabling more consistent and accurate endpoint searches compared to existing manual operations.

[0028] BRIEF DESCRIPTION OF THE DRAWINGS

[0029] A brief description of each figure is provided to provide a more complete understanding of the figures referenced in the detailed description of the present invention.

[0030] Figure 1 is an example diagram for explaining a conventional method of searching for an endpoint of a specific data cluster.

[0031] Figure 2 is a diagram for explaining a schematic configuration of a search system according to an embodiment of the present invention.

[0032] Figure 3 FIG. 4 is a diagram for conceptually explaining a method for searching for an endpoint of a specific data cluster according to an embodiment of the present invention.

[0033] Figure 4 is a flowchart for illustrating a method for searching for endpoints of a specific data cluster according to an embodiment of the present invention.

[0034] Best Mode for Carrying Out the Invention

[0035] For a fuller understanding of the invention, its operating advantages and objectives attained by its implementation, reference should be made to the accompanying drawings which illustrate preferred embodiments of the invention and the matter described therein.

[0036] In addition, in this specification, when one component transmits data to another component, it means that the component can directly transmit the data to the other component or transmit the data to the other component through at least one other component.

[0037] In contrast, when one component transfers data "directly" to another component, this means that the data is transferred from the component to the other component without passing through other components.

[0038] Hereinafter, the present invention will be described in detail by describing preferred embodiments of the present invention with reference to the accompanying drawings. The same reference numerals in the drawings represent the same components.

[0039] Figure 2 is a diagram for explaining a schematic configuration of a search system according to an embodiment of the present invention.

[0040] Reference Figure 2 , the search system 100 according to the technical idea of ​​the present invention includes a processor 110 and a storage device 120.

[0041] The storage device 120 may store a computer program (software) for realizing the technical concept of the present invention.

[0042] The software may be driven by the processor 110 to execute a method for searching endpoints of a specific data cluster according to the technical concept of the present invention.

[0043] According to an embodiment, the search system 100 may include at least one predetermined peripheral device 130. The peripheral device may be various devices, such as a display device, a speaker, an audio / video processing module, an external memory, an input / output device, a communication device, and the like.

[0044] The search system 100 can be configured as any data processing system having data processing capabilities capable of searching for endpoints of a specific data cluster according to the technical concept of the present invention, such as a computer, a server, a mobile phone, etc.

[0045] The search system 100 may receive a predetermined data set. The data set may include a plurality of individual data. The plurality of individual data have predetermined values. The values ​​may be numerical values. In addition, the plurality of individual data may form at least one data cluster.

[0046] Examples of the technical idea according to the present invention may be as follows Figure 1 Medical data is shown, but the scope of the present invention is not limited thereto but can be applied to various data.

[0047] The search system 100 generates histogram data using the input data set. The histogram data has a range of values ​​that the individual data can have as a domain of a first axis (e.g., an x-axis) and includes information of each bin when the first axis is divided into a plurality of bins with a predetermined bin distance.

[0048] The numerical value that the individual data may have may be a positive real value, and the maximum value is determined in advance.

[0049] The information of each group may include the range of values ​​of the first axis of the corresponding group (or the index of the group indicating the group number) and the value of the second axis (e.g., the y axis) of the corresponding group. The value of the second axis of the group may be the number of individual data corresponding to the range of values ​​of the first axis (i.e., the numerical value of the individual data having the group interval range).

[0050] And based on the histogram data including the information of the group, the search system 100 can search for an endpoint (eg, an upper endpoint) of a specific cluster (eg, the first cluster) in at least one data cluster.

[0051] Searching for endpoints means that, among the individual data included in the specific cluster, the values ​​of the individual data are sorted from high to low (e.g., Figure 1 The upper end in the y-axis direction) to the value of the first individual data with the lowest value (or the value range immediately following the value of the first individual data), or the value range of the highest few individual data in order (or the value range immediately following the range).

[0052] Alternatively, the endpoint means that, among the individual data included in the specific cluster, the values ​​of the individual data are ranked from low (e.g., Figure 1 The range of values ​​includes the value of the first individual data (or the range of values ​​immediately before the value of the first individual data) (from the lower end in the y-axis direction) to the highest individual data (or the range of values ​​that is smaller than the value of the first individual data), or the range of values ​​that is the lowest predetermined number (such as 2 to 3) of individual data in order (or the range of values ​​that is immediately before the range).

[0053] exist Figure 1 In the described examples, the case of finding the upper endpoint of the first data cluster from multiple data clusters in order from low to high values ​​is used as an example for explanation, but the technical idea of ​​the present invention is not necessarily applied to the first data cluster or only applicable to searching for the upper endpoint of a specific data cluster. For example, as described below, when searching for the endpoint of a specific cluster using histogram data, the endpoint of any data cluster (for example, the second data cluster) can be found by searching for the number (for example, the second) endpoint according to the search direction (for example, the direction of the values ​​of the first axis from the low value group to the high value group). In addition, according to the search direction (for example, the direction of the values ​​of the first axis from the high value group to the low value group), the lower endpoint of a specific data cluster can also be found instead of the upper endpoint.

[0054] In the following, for the convenience of description, the case of finding the upper endpoint of the first data cluster from multiple data clusters in order of numerical values ​​from low to high is taken as an example for explanation, but the scope of the present invention is not limited to this.

[0055] According to the technical idea of ​​the present invention, the endpoints of a specific data cluster can be found using histogram data, in which case this can be defined as the problem of searching the histogram data for a terminal group corresponding to the corresponding data cluster, namely a target group.

[0056] In addition, according to an embodiment, based on the technical idea of ​​the present invention, instead of directly using the histogram data, histogram data obtained by smoothing the histogram data and / or differentially smoothing the histogram data may be used.

[0057] That is, in an embodiment of the present invention, the target group can be searched using the smoothed histogram data, or the target group can be searched using the differential histogram data by differentially smoothing the histogram data again. In addition, when performing differential differentiation, the inflection point of the histogram data can be determined more intuitively and more easily.

[0058] According to the technical idea of ​​the present invention, in the case of searching for a target group by simply converting the original individual data into histogram data through smoothing or smoothing and differencing, individual data corresponding to a specific group (i.e., a proximity range of a specific numerical value) temporarily does not exist, and then when the individual data re-exists, it is possible to prevent the specific group from being explored as a target group, i.e., as an endpoint of a specific data cluster to be searched.

[0059] In the following, reference will be made to Figure 3 and Figure 4 The technical idea of ​​the present invention is described in more detail.

[0060] Figure 3 is a diagram for conceptually illustrating a method for searching for an endpoint of a specific data cluster according to an embodiment of the present invention. Figure 4 is a flowchart for illustrating a method for searching for endpoints of a specific data cluster according to an embodiment of the present invention.

[0061] Figure 3 It represents example histogram data when smoothing and / or differencing is performed using a mask with a diameter (i.e., number of parameters) of 3 and parameter values ​​as simple as [-1.0.1], but obviously the diameter and parameter values ​​of the smoothing mask and / or differencing mask may be different.

[0062] Reference Figure 3 and Figure 4 , the search system 100 may sequentially receive original individual data O included in a data set ( S100 ).

[0063] like Figure 3 As shown, the original individual data O may have a positive real value (e.g., 1.23425, 2.13425, 4.23252, 3.13141, 1.14452, etc.). Figure 1 The values ​​corresponding to the second axis (e.g., y-axis) in the graph shown.

[0064] Then, the search system 100 may generate histogram data H based on the input original individual data 0 ( S110 ).

[0065] The histogram data H is data generated by dividing the possible numerical range of individual data into a plurality of groups (bins) 20 having a certain width and making the number of individual data having numerical values ​​corresponding to each divided group the group value of each group.

[0066] The histogram data H can be graphically represented as Figure 3 The histogram data H is the same.

[0067] exist Figure 3 The group 21 having the group value in the histogram data H may be a partial area of ​​the histogram corresponding to any one data cluster.

[0068] Therefore, the endpoints of the data cluster, that is, the target group 30 to be searched by the search system 100 can be as follows: Figure 3 That is, in Figure 3 In the figure, it is shown that the numerical range of the individual data corresponding to the group 21-1 immediately before the target group 30 is not searched for the last individual data (i.e., the range of the first axis value of the immediately preceding group 21-1), but the group immediately after the immediately preceding group 21-1 having a group value is searched as the target group 30. However, according to the embodiment, of course, the immediately preceding group 21-1 can also become the target group.

[0069] The search system 100 may not directly search for the target group 30 from the histogram data H, but may smooth the histogram data H S120.

[0070] In addition, the target group 30 may be searched using the smoothed histogram data S ( S150 ).

[0071] As described above, by using the smoothed histogram data S, when there is temporarily at least one group with a group value of 0 (called a blank group) between a series of groups 21 having the group value, that is, when there is a partial range of individual data in the numerical range corresponding to the data cluster to be searched, the problem of not knowing whether such a blank group is determined as a target group or a blank group can be solved. That is, in the case of using the smoothed histogram data S, even if it is a blank group in the original histogram H, it will have a certain value according to the group values ​​on the left and right, so that the group value can be non-0, so using the smoothed histogram data S may be more effective.

[0072] As an example, the result of smoothing the original histogram data H can be compared with Figure 3 The histogram S is the same.

[0073] Smoothing masks (filters) and / or differencing masks for smoothing histograms are well known.

[0074] In an embodiment of the present invention, a convolution mask may be used as a smoothing mask and / or a differencing mask, and a predetermined digital column x and a convolution mask h may be defined as follows.

[0075] [Formula 1]

[0076]

[0077] In one embodiment of the present invention, examples of the smoothing mask and the differencing mask use [1,1,1,1,1,1,1,1,1,1] and [-1,-1,-1,-1,0,1,1,1,1], respectively, and in another embodiment, [1,1,1,1,1,1,1,1,1] and [-1,-1,-1,-1,0,1,1,1,1] are used. However, the smoothing mask and the differencing mask can be set differently according to the characteristics of the data set, such as the number and clustering degree of individual data included in the data set.

[0078] Meanwhile, as described above, the search system 100 may search for the target group 30 using the smoothed histogram data S, but in other cases, the target group 30 may be searched more specifically by differentiating the smoothed histogram data S.

[0079] In what cases, the target group 30 is searched using the smoothed histogram data S, and in what cases, the target group 30 is searched using the differential histogram data D, which can be predetermined based on the characteristics of the data set. The characteristics of the data set can be determined based on the amount of data, the density of data, and the number of data clusters. Obviously, the benchmark can be determined by performing repeated experiments in advance, that is, in what range the characteristics can be used as the first case, that is, the target group 30 is searched using the smoothed histogram data S, and not as the second case, that is, the target group 30 is searched using the differential histogram data D.

[0080] Of course, according to an embodiment, one of the two methods may be randomly selected, and according to an example, both methods may be used simultaneously to search for the target group 30 and the results thereof may be compared.

[0081] If two methods are used simultaneously to search for the target group 30, when the position (1st axis value) of each searched target group is the same or within a predetermined position (1st axis value) range, the target group found by any method can be determined as the final target group.

[0082] Therefore, when the search system 100 determines that it is the first case based on the input original individual data 0 ( S130 ), the target group 30 may be searched by using the smoothed histogram data S ( S150 ).

[0083] In addition, when the search system 100 determines that it is the second case, the search system 100 may differentiate the smoothed histogram data again (S 140). Then, the search system 100 may search for the target group 30 by using the differentiated histogram data D (S 150).

[0084] An example of a case where the search system 100 searches for the target group 30 from the smoothed histogram data S is as follows.

[0085] For example, the search system 100 may search for group values ​​of each group in a predetermined direction (eg, a direction in which values ​​increase) in the smoothed histogram data S. As shown in FIG.

[0086] At the same time, a group with the following conditions, namely, the previous group value of the currently searched group is not the cutoff value (for example, 0), the current group value is the cutoff value (for example, 0), and a predetermined number (for example, 1 or more than 2) of the next group values ​​are the cutoff value (for example, 0) can be used as the target group 30.

[0087] In this case, Figure 3 In the figure, when the target group 30 is the current group currently being searched, the group value of the immediately preceding group 21-1 is not 0, and if the group value of the current group is 0, since the group values ​​of a predetermined number (e.g., 2) of immediately subsequent groups are 0, the current group can be determined as the target group 30.

[0088] The cutoff value may be 0, but in some embodiments may be set to have a small value, such as 1. In this case, the searched endpoint may be defined as an algorithm that finds a value for which only one individual data exists on the terminal side of the data cluster, and different cutoff values ​​may be set according to the embodiment.

[0089] Meanwhile, an example of searching the target group 30 from the differential histogram data D is as follows.

[0090] For example, the search system 100 may search for group values ​​of each group in a predetermined direction (eg, a direction in which values ​​increase) in the differential histogram data D. As shown in FIG.

[0091] At the same time, if the current group being searched is the target group 30, if the following conditions are met, namely, the group value of the group 21-1 immediately before the current group is less than the group value of the group 31 immediately after it, the group value of the immediately before group 21-1 is equal to or less than 0, and the group value of the immediately after group 31 is equal to or greater than 0, then the current group is the target group 30 to be searched.

[0092] That is, as the group value gradually increases from a negative value, a point corresponding to an area that becomes 0 is likely to become a target group 30 to be searched.

[0093] Meanwhile, as described above, depending on how the class interval is set when generating the histogram data, the target group 30 may not be searched. For example, if the class interval of the group is too large, there is a large amount of individual data between the data cluster to be searched and the next data cluster and it is relatively dense, then there may not be a group with a cutoff value.

[0094] However, if the class interval is set too narrow, there occurs a problem that many groups having the cutoff value may be detected even within one data cluster, or a problem that the search time becomes long due to an increase in the number of groups.

[0095] Therefore, it may be necessary to predetermine the appropriate group width through repeated experiments.

[0096] If it is difficult to predetermine such a group distance, a predetermined default group distance value is used to perform the search. If the target group is not found as described above (i.e., due to the excessively wide group distance, there is no group with a group value of 0 between the terminal group of the target data cluster to be searched and the terminal group located on one side of the target data cluster in the data cluster adjacent to the target data cluster), the group distance is reduced in turn according to the predetermined unit, and the histogram data is regenerated using the reduced group distance. Then, the target group search process as described above is performed using the regenerated histogram data (using the smoothed histogram data to search for the target group or using the differential histogram data to search for the target group).

[0097] In this case, there is an effect that a group distance having relatively excellent search efficiency can be searched for, and the target group as described above can be searched using the searched group distance.

[0098] Meanwhile, the search method for the endpoint of a specific data cluster according to an embodiment of the present invention can be implemented in the form of a computer-readable program command and stored in a computer-readable recording medium, and the control program and the target program according to an embodiment of the present invention can also be stored in a computer-readable recording medium. The computer-readable recording medium includes all types of recording devices that store data that can be read by a computer system.

[0099] The program instructions recorded on the recording medium may be specially designed and configured for the present invention, but may also be those known and usable by those skilled in the art of software.

[0100] Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magnetic-optical media such as floptical disks, and hardware devices specially configured to store and execute program instructions, such as read-only memories (ROMs), random access memories (RAMs), and flash memories. Computer-readable recording media can also be distributed on networked computer systems so that computer-readable codes can be stored and executed in a distributed manner.

[0101] Examples of the program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes executed by an apparatus that electronically processes information such as a computer using an interpreter or the like.

[0102] The hardware devices may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

[0103] It will be appreciated by those skilled in the art that the foregoing description of the present invention is for illustrative purposes only, and it will be readily appreciated by those skilled in the art that various changes and modifications may be made without changing the technical ideas or basic features of the present invention. Therefore, it will be appreciated that the above embodiments are exemplary and non-restrictive in all respects. For example, each component described as a single type may be implemented in a dispersed manner, and components described as dispersed may also be implemented in a combined form.

[0104] The scope of the present invention is determined by the following claims rather than the above detailed description. All changes and modifications derived from the meaning and scope of the claims and their equivalent concepts should be understood to be included in the scope of the present invention.

[0105] Possibility of industrial application

[0106] The present invention can be used in a method for searching endpoints of a specific data cluster and a data processing system thereof.

Claims

1. An endpoint search method for a specific medical data cluster, A method for searching for an endpoint of a specific cluster in a medical data set having at least one cluster, comprising: a) receiving, by a search system, respective values ​​of a plurality of individual medical data included in the medical data set; b) a step of generating histogram data, i.e., by using the respective values ​​received by the search system, dividing the possible value range of the values ​​into a plurality of groups with a predetermined group interval, and taking the number of individual medical data corresponding to each segmented group as a group value; c) a step of smoothing the histogram data; as well as d) a step of differentiating the smoothed histogram data; e) searching for a target group that satisfies a predetermined reference condition and exists at an end point of a specific cluster based on the differential histogram data; f) when a target group satisfying the reference condition is not found by searching using a predetermined group distance, reducing the group distance to a predetermined value; Performing steps b) to e) again using the reduced class distance; Based on the differential histogram data, the step of searching for a target group that meets a predetermined benchmark condition and exists at an endpoint of a specific cluster includes: searching for group values ​​of each group in a certain direction in the differential histogram data, and searching for a group that meets the benchmark condition as the target group, wherein the benchmark condition is that the first group satisfies the group that is currently being searched and the group value before the group that is currently being searched is less than the group value after the group, the previous group value is equal to or less than 0, and the subsequent group value is equal to or greater than 0.

2. An endpoint search method for a specific medical data cluster, A method for searching for an endpoint of a specific cluster in a medical data set having at least one cluster, comprising: a) receiving, by a search system, respective values ​​of a plurality of individual medical data included in the medical data set; b) a step of generating histogram data, i.e., by using the respective values ​​received by the search system, dividing the possible value range of the values ​​into a plurality of groups with a predetermined group interval, and taking the number of individual medical data corresponding to each segmented group as a group value; c) a step of smoothing the histogram data; as well as d) searching, based on the smoothed histogram data, for a target group that satisfies a predetermined benchmark condition and exists at an end point of a specific cluster; e) when a target group satisfying the reference condition is not found by searching using a predetermined group distance, reducing the group distance to a predetermined value; Performing steps b) to d) again using the reduced class distance; The step of searching the target group that meets the predetermined reference condition based on the smoothed histogram data comprises: The step of searching for the group value of each group in a certain direction in the smoothed histogram data, and searching for the group that meets the benchmark condition as the target group, wherein the benchmark condition is that the group value before the group currently being searched is not a cutoff value, the current group value is a cutoff value, and at least one subsequent group value is a cutoff value.

3. A computer program product, comprising a computer program, wherein the computer program is installed in a data processing system and records a program for executing the method described in any one of claims 1 to 2.

4. A data processing system comprising: processor; a storage device storing software executed by the processor, said software, receiving respective values ​​of a plurality of individual medical data included in a medical data set having at least one cluster, dividing a numerical range that the numerical value may have into a plurality of groups having a predetermined group interval by using the received respective numerical values, and generating histogram data using the number of individual medical data corresponding to each segmented group as a group value, and searching for a target group existing at an end point of a specific cluster based on the generated histogram data; smoothing the histogram data, differentiating the smoothed histogram data, and searching for a target group that satisfies a predetermined reference condition and exists at an end point of a specific cluster based on the differentiated histogram data; Alternatively, the histogram data is smoothed, and based on the smoothed histogram data, the target group satisfying a predetermined reference condition is searched; When a target group satisfying the reference condition is not found by searching using a predetermined group distance, the software reduces the group distance to a predetermined value, generates a histogram again using the reduced group distance, and uses the regenerated histogram to search again for a target group existing at an end point of the specific cluster; in, Based on the differential histogram data, searching for a target group that satisfies a predetermined benchmark condition and exists at an end point of a specific cluster comprises: searching for group values ​​of each group in a certain direction in the differential histogram data, and searching for a group that satisfies the benchmark condition as the target group, wherein the benchmark condition is that the first group that satisfies the group before the currently searched group is less than the group after it, the previous group value is equal to or less than 0, and the subsequent group value is equal to or greater than 0; Searching for the target group that meets a predetermined benchmark condition based on the smoothed histogram data includes: searching for the group value of each group in a certain direction in the smoothed histogram data, and simultaneously searching for a group that meets the benchmark condition as the target group, wherein the benchmark condition is that the group value before the group currently being searched is not a cutoff value, and the current group value is a cutoff value, and at least one subsequent group value is a cutoff value.

Images