Pathological significance determination apparatus, pathological significance determination method, machine learning method, and learned model generation method

Abstract

To provide a technique that reduces a human burden to verify the validity of an estimation result of pathological significance.SOLUTION: A pathological significance determination apparatus includes: an input device which receives, as input, genetic mutation information indicating a genetic mutation and genetic mutation-related information related to the genetic mutation information; a processor which estimates a first score indicating the presence or absence of pathological significance of the genetic mutation and a second score indicating the strength or completion level of evidence related to the genetic mutation on the basis of the genetic mutation information and the genetic mutation-related information; and an output device which outputs the estimated first score and second score.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a pathogenic significance determination device, a pathogenic significance determination method, a machine learning method, and a method for generating a learned model.

Background Art

[0002] In cancer genomics medicine, an expert panel is conducted after the return of genomic test results. In the expert panel, the pathogenic significance (pathogenic / non-pathogenic / VUS: Variants for Unknown Significance) of a patient's gene mutation is discussed by experts such as doctors. The pathogenic significance is comprehensively determined based on various evidences, and the determination requires time, labor, and specialized knowledge. Furthermore, since there is no clear criterion for the determination of pathogenic significance, there are differences in the determination results among experts. Therefore, the determination of pathogenic significance becomes personal, and the burden on a specific expert increases.

[0003] On the other hand, in recent years, various systems and AIs for determining the pathogenic significance of gene mutations have been developed. For example, Patent Document 1 describes a technique for classifying DNA variants into five categories including "pathogenic", "highly likely to be pathogenic", "variant of unknown significance (VUS)", "highly likely to be benign", and "benign variant" according to a rule-based scoring system.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Even when the pathological significance of a gene mutation is determined by AI, it is assumed that the pathological significance will ultimately be confirmed by an expert. Since it is difficult to understand the accuracy of the determination result based solely on the determination of the pathological significance by AI, the burden on the expert is significant in confirming the determination result. The reasons for determining the pathological significance by AI are also diverse. In particular, there are various variations in the reasons for determining VUS. In addition, there are mutations in which the determination of the pathological significance in multiple pieces of evidence conflicts and is difficult to judge (the accuracy of the determination result is low), and mutations in which the determination of the pathological significance in multiple pieces of evidence is consistent and easy to judge (the accuracy of the determination result is high) mixed together.

[0006] In the technology of Patent Document 1, the estimation results for gene mutations (VUS) whose pathological significance is unknown include both VUS whose genetic significance is unknown simply due to insufficient evidence and VUS for which there is evidence but the presence or absence of pathological significance cannot be determined. Therefore, in order to extract VUS with evidence, it is necessary to manually check and judge the information related to the gene mutation corresponding to the gene information.

[0007] Therefore, the present disclosure provides a technique for reducing the human burden for confirming the validity of the estimation result of the pathological significance.

Means for Solving the Problem

[0008] In order to solve the above problems, the pathological significance determination device of the present disclosure includes an input device that receives an input of gene mutation information indicating a gene mutation and gene mutation-related information related to the gene mutation information, and based on the gene mutation information and the gene mutation-related information, a first score regarding the presence or absence of the pathological significance of the gene mutation and a second score regarding the strength or sufficiency of evidence related to the gene mutation are estimated by a processor, and an output device that outputs the estimated first score and second score.

[0009] Further features related to the present disclosure will become apparent from the description herein and the accompanying drawings. Also, aspects of the present disclosure are achieved and realized by elements and combinations of various elements and the aspects of the following detailed description and the appended claims. The description herein is merely exemplary and is not intended to limit the scope of the claims of the present disclosure or its applications in any way.

Advantages of the Invention

[0010] According to the technology of the present disclosure, the human burden for confirming the validity of the estimated results of pathological significance can be reduced. Other problems, configurations, and effects will be clarified by the description of the following embodiments.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2A

Figure 2B

Figure 2C

Figure 3

Figure 4

Figure 5

Figure 6A

Figure 6B

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Embodiments for Carrying Out the Invention

[0012] [First Embodiment] <Pathological Significance Determination Device> FIG. 1 is a configuration diagram of a pathological significance determination device 100 according to the first embodiment. The pathological significance determination device 100 is a device for determining the pathological significance (pathologically significant / not pathologically significant / VUS) of gene mutations in a patient's genomic test results. The pathological significance determination device 100 is composed of, for example, a personal computer, a server device, a smartphone, a tablet, an office computer, a general-purpose machine (mainframe), etc. As shown in FIG. 1, the pathological significance determination device 100 includes a processor 101, a memory 102, a storage device 103, a display device 104, an input device 105, and a bus 106 connecting these.

[0013] The processor 101 realizes the functions of the pathological significance determination device 100 by executing a program developed in the memory 102. As the processor 101, for example, a CPU, a GPU, etc. can be used. The number of processors 101 is not limited to one, and the functions of the pathological significance determination device 100 may be realized by a plurality of processors. The memory 102 includes a ROM and a RAM.

[0014] The memory device 103 stores learning data 11, correct answer data 12, test data 13, mutation scores 14, and a learning model 15. The learning data 11 is data for the learning model 15 to learn through machine learning or data that has already been learned. The correct answer data 12 is data that is the correct answer for the output of the learning model 15 and is associated with the learning data 11 by a common ID or the like. The test data 13 is data to be the object of estimation of pathogenic significance using the learning model 15 and includes information on gene mutations in the genomic test results of a patient. Details of the learning data 11 and the test data 13 will be described later.

[0015] The mutation scores 14 are the pathogenic significance scores and evidence scores of each gene mutation of the patient in the test data 13 estimated (output) by the learning model 15. The pathogenic significance score is a score indicating the estimation result of having pathogenic significance / not having pathogenic significance and its accuracy. The evidence score is a score indicating the strength and sufficiency of evidence. As will be described later, the pathogenic significance score and the evidence score are each represented by a numerical value, and (y 1 , y 2 ) = (pathogenic significance score, evidence score) is represented. Note that the correct answer data 12 is obtained by converting the correct pathogenic significance judged by an expert for the learning data 11 into a mutation score.

[0016] The learning model 15 is a machine learning model for estimating the pathogenic significance of each gene mutation in the genomic test results of a patient. The learning model 15 is a supervised learning model trained by the learning data 11 associated with the correct answer data 12. The learning model 15 is constructed to independently estimate and output the pathogenic significance score and the evidence score with the test data 13 as input. Instead of this, the learning model for estimating the pathogenic significance score and the learning model for estimating the evidence score may be separate. As the machine learning algorithm of the learning model 15, for example, any algorithm such as XGBoost or a neural network can be adopted.

[0017] Although illustration is omitted, the storage device 103 stores the basic information of the patient. The basic information of the patient includes, for example, patient ID, age, gender, cancer type, etc. The basic information of the patient is associated with the test data 13 and the mutation score 14 by, for example, a common ID.

[0018] The display device 104 is, for example, a liquid crystal display or the like. The input device 105 is, for example, a mouse and a keyboard or the like. The display device 104 and the input device 105 may be combined using a touch panel.

[0019] Figure 2A is a schematic diagram showing an example of the learning data 11. As shown in Figure 2A, the learning data 11 includes gene mutation information 16 and gene mutation-related information 17. The gene mutation information 16 indicates a gene mutation. In the present embodiment, the gene mutation information 16 is represented as "gene A / mutation a", "gene A / mutation b", "gene B / mutation c", etc. For each piece of gene mutation information 16, gene mutation-related information 17 is associated.

[0020] The gene mutation-related information 17 is represented in the form of a table including a plurality of determination items for determining the pathological significance and the determination content of each determination item. The determination items are, for example, known information regarding the gene mutation information 16 that can be obtained from external publicly known mutation information databases (such as Clinvar, COSMIC, etc.). Specific examples of the determination items include, for example, polymorphic allele frequency, determination result of pathological significance in the mutation information database, presence or absence of the possibility of carcinogenesis when a gene mutation exists, amino acid information, number of reports for each cancer type, position of the domain where the mutation exists, etc. Note that the determination items can also be a combination of clinical information (such as age, cancer type) in each case and information in the publicly known mutation information database. For example, the cancer type in the clinical information can be combined with information such as the number of reports for each cancer type.

[0021] The determination content can be represented in any data format. The determination content is represented by, for example, a numerical value corresponding to a determination item, True / False, a label, a determination result of pathogenic significance in a publicly known mutation information database, etc. For example, when the determination item is "polymorphic allele frequency", the determination content can be a numerical value such as "0.5". Also, when the determination item is "whether the pathogenic significance of the mutation has been reported", the determination content can be "True" or "False". When the determination item is "determination of pathogenic significance in a publicly known mutation information database", the determination content can be "pathogenic significance (Oncogenic)", "no pathogenic significance (Benign)", "VUS", "possibility of pathogenic significance (Likely Oncogenic)", "possibility of no pathogenic significance (Likely Benign)", etc. Each determination item and its determination content of the gene mutation-related information 17 are used in the estimation of the mutation score. For each determination item and its determination content, it is determined which pathogenic significance is supported, and by comprehensively considering these, it is determined which pathogenic significance the gene mutation has.

[0022] Figure 2B is a schematic diagram showing an example of the correct data 12. As shown in Figure 2B, the correct data 12 includes the value of the correct mutation score (pathogenic significance score, evidence score). The correct data 12 is created for each gene mutation information 16.

[0023] Below Figure 2B, the creation method of the correct data 12 is shown. The correct mutation score y = (y 1 , y 2 ) in the correct data 12 is determined by the determination result of the pathogenic significance of the gene mutation information 16 by an expert. When the determination result of the pathogenic significance of the gene mutation information 16 by an expert is "pathogenic significance", (pathogenic significance score, evidence score) = (1, 1). When the determination result of the pathogenic significance of the gene mutation information 16 by an expert is "no pathogenic significance", (pathogenic significance score, evidence score) = (-1, 1). When the determination result of the pathogenic significance of the gene mutation information 16 by an expert is "VUS", (pathogenic significance score, evidence score) = (0, -1).

[0024] FIG. 2C is a schematic diagram showing an example of test data 13. As shown in FIG. 2C, the format and content of the test data 13 are the same as those of the learning data 11. The gene mutation information 16 included in the test data 13 may be only the gene mutations detected in the genomic examination of the patient. As described above, the learning data 11 and the test data 13 include all the information used for calculating the mutation score 14.

[0025] FIG. 3 is a schematic diagram showing an example of the mutation score 14 estimated for the test data 13. The mutation score 14 (pathogenic significance score and evidence score) is estimated by the learning model 15 for each gene mutation information 16. The pathogenic significance score and the evidence score are each represented by a numerical value in the range of -1 to 1. The processor 101 determines the pathogenic significance of the gene mutation information 16 from the values of the pathogenic significance score and the evidence score.

[0026] An example of a method for determining pathogenic significance based on the pathogenic significance score will be described. When the pathogenic significance score is a positive value, the determination result is "pathogenic significance". When the pathogenic significance score is a negative value, the determination result is "no pathogenic significance". When the pathogenic significance score is 0, the determination result is "VUS". The greater the absolute value of the pathogenic significance score, the higher the accuracy of the determination result of the pathogenic significance. In this way, by setting the threshold value for determining pathogenic significance based on the pathogenic significance score to 0, it is possible to determine the presence or absence of pathogenic significance and VUS. Alternatively, the threshold value of the pathogenic significance score can be set to any value. Specifically, for example, two threshold values j and k (j > k) of the pathogenic significance score can be provided. At this time, when j < pathogenic significance score, the determination result is "pathogenic significance". When k < pathogenic significance score < j, the determination result is VUS. When pathogenic significance score < k, the determination result is "no pathogenic significance". For example, as the threshold values of the pathogenic significance score, j = 0.5 and k = -0.5 can be set. The threshold value of the pathogenic significance score can be set by the user on the setting screen described later.

[0027] The evidence score indicates the strength and sufficiency of the evidence that was the basis for the determination result of the pathological significance. When the evidence score is a positive value, the higher the evidence score, the larger the number of pieces of evidence that were the basis for the determination result of the pathological significance, or the greater the contribution of the evidence to the determination result of the pathological significance. "Having a large contribution to the estimated result of the pathological significance" refers to, for example, evidence that can clarify whether a gene mutation has pathological significance or not. When the evidence score is a negative value, the determination result is "VUS". Thus, based on the estimated mutation score 14, the pathological significance of the gene mutation information 16 can be determined.

[0028] The correct data of the pathological significance score and the evidence score, and the threshold value for the determination of the pathological significance are not limited to the above, and other mathematical formulas may be used, and any deformation is also possible.

[0029] FIG. 4 is a schematic diagram of the setting screen 200 according to the first embodiment. The setting screen 200 is used for setting input data for training the learning model 15, setting input data for which the pathological significance of gene mutations is to be estimated using the trained learning model 15, and setting internal parameters of the learning model 15.

[0030] As shown in FIG. 4, the setting screen 200 includes a box 201, a box 202, a setting button 203, a box 204, an edit button 205, and a setting button 206. In the box 201, the user can specify the file of the correct data 12 used for training the learning model 15. In the box 202, the user can specify the file of the training data 11 used for training the learning model 15 or the file of the test data 13 to be estimated by the learning model 15. By the setting button 203, the specified file can be input to the processor 101 (memory 102). When the user clicks on the box 201 and the box 202, the desired file can be selected from the files stored in the storage device 103, and thereby the path of the file can be specified.

[0031] In the box 204, the user can specify a file for setting the internal parameters of the learning model 15. At the edit button 205, the user can edit the parameters. At the setting button 206, the setting of the specified file or the edited parameters can be input. The parameters are, for example, hyperparameters, etc. They may also be parameters optimized by pre-training.

[0032] When the user edits the parameters via the edit button 205 and knows the judgment items important in the pathological significance judgment, the user may weight the internal parameters of the learning model 15 according to the magnitude of the contribution of the gene mutation-related information (judgment items) to the judgment.

[0033] FIG. 5 is a schematic diagram of the output screen 500 according to the first embodiment. The output screen 500 is displayed on the display device 104. The output screen 500 has patient basic information 501 and mutation information 502. The patient basic information 501 and the mutation information 502 are presented in a table format. The patient basic information 501 includes a patient ID, age, gender, cancer type, etc.

[0034] The mutation information 502 includes the determination result of the pathological significance (pathologically significant / not pathologically significant / VUS) using the learning model 15 by the processor 101, gene mutation information, VAF (Variant Allele Frequency: the ratio of cells in which the gene mutation was detected), the estimated mutation score, and the judgment items important for the estimation. In the column of the judgment items important for the estimation, the judgment items that strongly contributed to the estimation are ranked. Such ranking of the judgment items can be determined by a predetermined algorithm included in the learning model 15 such as XGBoost. With the mutation information 502, the user can confirm the estimated mutation score and the determination result of the pathological significance for each gene mutation information.

[0035] <Learning stage> FIG. 6A is a flowchart of a learning method of the learning model 15 according to the first embodiment. The learning method is executed by the processor 101 and includes the following steps S301 and S302. In step S301, the processor 101 receives input of learning data 11, correct data 12 of a mutation score, and setting of internal parameters via the setting screen 200 from the user. In step S302, the processor 101 performs learning of the learning model 15 using the learning data 11 with the correct data 12 attached as a correct label and the set internal parameters. The processor 101 stores the learned learning model 15 in the storage device 103.

[0036] <Test stage> FIG. 6B is a flowchart of a method for determining pathogenic significance using the learned learning model 15 according to the first embodiment. The method for determining pathogenic significance is executed by the processor 101 and includes the following steps S303 to S305.

[0037] In step S303, the processor 101 imports the learned learning model 15 from the storage device 103 to the memory 102.

[0038] In step S304, the processor 101 receives input of test data 13 via the setting screen 200 from the user. Then, the processor 101 inputs the test data 13 to the learning model 15 and obtains a mutation score 14 (pathogenic significance score and evidence score), which is the output (estimation result) of the learning model 15. Further, the processor 101 determines the pathogenic significance of the gene mutation information 16 based on the mutation score 14. The processor 101 stores the mutation score 14 and the determination result of the pathogenic significance in the storage device 103.

[0039] In step S305, the processor 101 generates an output screen 500 including the mutation score 14, the determination result of the pathogenic significance, and the basic information of the patient, and displays the output screen 500 on the display device 104.

[0040] Note that the processor 101 may further train the learning model 15 using the mutation score 14 obtained in step S304. The training of the learning model 15 using such an estimation result can also be performed every time the mutation score 14 is estimated.

[0041] <Estimation method by formula> As described above, the method for estimating the mutation score 14 by machine learning has been explained. Instead of this, the mutation score 14 can also be calculated by an evidence-based formula. In this case, the processor 101 determines which of the significance of disease, no significance of disease, and VUS is supported by each determination item and its determination content, and uses it for calculating the disease significance score and the evidence score.

[0042] Disease significance score y 1 is represented by the following formula (1).

[0043]

Number

[0044] In formula (1), σ represents the sigmoid function. x represents the evidence level. P represents the determination item and its determination content that support "significant disease". B represents the determination item and its determination content that support "no significance of disease". a(x) represents a constant of the evidence level and takes a numerical value from 5 (when x is high) to 1 (when x is low). a(x) is determined according to the rule-based guideline for each evidence. w 1 represents the weight. n represents the number. b 1 represents the baseline hyperparameter.

[0045] As in formula (1), the disease significance score y 1 is calculated as a value obtained by normalizing the difference between the sum of the evidence levels of the determination items that support "significant disease (P)" and the sum of the evidence levels of the determination items that support "no significance of disease (B)" by the sigmoid function (σ).

[0046] Evidence score y2 It is represented by the following formula (2).

[0047]

Number

[0048] In formula (2), σ represents the sigmoid function. x represents the evidence level. P represents the determination items that support "with pathological significance" and their determination contents. B represents the determination items that support "without pathological significance" and their determination contents. VUS represents the determination items that support "VUS" and their determination contents. a(x) represents a constant of the evidence level and takes numerical values from 5 (when x is high) to 1 (when x is low). a(x) is determined according to the rule-based guidelines for each piece of evidence. w 1 and w 2 represent weights. n represents the number. b 1 and b 2 represent the hyperparameters of the baseline.

[0049] The evidence score y as in formula (2) 2 is calculated as a value normalized by the sigmoid function (σ) by subtracting the sum of the evidence levels of the determination items that support "VUS" from the sum of the sum of the evidence levels of the determination items that support "with pathological significance (P)" and the sum of the evidence levels of the determination items that support "without pathological significance (B).

[0050] The mathematical formula for calculating the pathological significance score and the mathematical formula for calculating the evidence score are not limited to the above, and other mathematical formulas may be used or any deformation is possible.

[0051] Information described in known papers regarding the determination of the pathological significance of gene mutations can also be used for calculating the mutation score 14. Examples of information described in known papers include information in polymorphism databases such as gnomAD, evidence of the effect on carcinogenesis in vitro / in vivo, evidence of the pathogenicity of mutations, the number of case reports of the same amino acid mutation and the same position mutation in databases such as Cancer Hotspots and COSMIC, and the determination results of the pathological significance by computational tools. Examples of such known papers include P. Horak et al., Genetics in Medicine (2022) 24, 986-998.

[0052] When performing the determination of the pathological significance using the mathematical formula as described above, the ranking of the "important determination items for estimation" in the mutation information 502 of the output screen 500 can be performed according to the level of evidence.

[0053] <Summary of the First Embodiment> As described above, the pathological significance determination device 100 according to the first embodiment includes an input device 105 that receives, as inputs, gene mutation information 16 indicating a gene mutation and gene mutation-related information 17 related to the gene mutation information 16, and a processor 101 that estimates a pathological significance score (first score) regarding the presence or absence of the pathological significance of the gene mutation and an evidence score (second score) regarding the strength or sufficiency of the evidence related to the gene mutation based on the gene mutation information 16 and the gene mutation-related information 17, and a display device 104 (output device) that outputs the estimated pathological significance score and evidence score (first score and second score).

[0054] According to such a pathological significance determination device 100, the accuracy of the determination result of pathological significance is ensured by the pathological significance score (the first score) regarding the presence or absence of pathological significance. Also, the evidence score (the second score) indicating the sufficiency of evidence can be used to show the interpretation of gene mutations for which the presence or absence of pathological significance cannot be determined. Furthermore, by outputting and visualizing the pathological significance score and the evidence score, the user (expert) can easily identify the gene mutations that should be preferentially confirmed and discussed. As a result, the work for experts to check all gene information can be reduced.

[0055] When there is a therapeutic agent corresponding to a gene mutation, an expert can determine whether to recommend the treatment method. The certainty of "pathological significance exists" indicated by the pathological significance score can assist in that determination.

[0056] [Second Embodiment] In the above-described first embodiment, it has been explained that the mutation score 14 estimated by the learning model 15 is displayed on the output screen 500 as a numerical value. In addition to or instead of this, as described in the second embodiment, the estimated mutation score 14 may be plotted on a two-dimensional plane. Since the configuration of the pathological significance determination device according to the second embodiment is the same as that of the first embodiment, the description thereof is omitted.

[0057] FIG. 7 is a schematic diagram of the setting screen 200 according to the second embodiment. The setting screen of this embodiment has a selection button 207 and a setting button 208 in addition to the content of the setting screen of the first embodiment. By the selection button 207, the user can select whether to output the mutation score on the two-dimensional plane on the output screen 500. The selection button 207 can be checked for either "perform" or "not perform" two-dimensional output. By the setting button 208, the setting of whether to output the two-dimensional plane selected by the selection button 207 can be input to the processor 101 (memory 102).

[0058] FIG. 8 is a schematic diagram of the output screen 500 according to the second embodiment. The output screen 500 of this embodiment further includes a graph 503 (two-dimensional plane) of mutation scores in addition to the content of the output screen of the first embodiment. The graph 503 is a graph plotting the mutation scores corresponding to each gene mutation information calculated in the above-described step S304. The horizontal axis represents the pathogenic significance score (y 1 ), and the vertical axis represents the evidence score (y 2 ).

[0059] In the graph 503, the mutation scores (No. 1 to No. 7) for seven gene mutation informations are plotted. When the user clicks (selects) an arbitrary plot, information about the gene mutation of the plot (mutation score, pathogenic significance, determination items important for estimation, VAF) is displayed in the table of mutation information 502. In FIG. 8, the second plot is clicked. If none of the plots are clicked (selected), all the detected gene mutation informations and the estimated results of the pathogenic significance for those gene mutation informations are displayed as in the mutation information 502 shown in FIG. 5.

[0060] In the graph 503, it is also possible to plot the mutation scores of gene mutations estimated in the past. In this case, the user can refer to the determination of the pathogenic significance in the mutation scores of nearby plots by checking the mutation information 502 about the mutation scores estimated in the past.

[0061] <Test stage> FIG. 9 is a flowchart of a method for determining the pathological significance using the learned learning model 15 according to the second embodiment. The method for determining the pathological significance in the second embodiment is different from the first embodiment in that step S3051 is executed instead of step S305. In step S3051, the processor 101 generates an output screen 500 including a graph in which the mutation score calculated in step S304 is plotted on a two-dimensional plane. Then, the processor 101 causes the display device 104 to display the output screen 500. Note that the learning method of the learning model 15 is the same as that of the first embodiment.

[0062] <Summary of the Second Embodiment> As described above, in the pathological significance determination device according to the second embodiment, the display device 104 outputs a two-dimensional graph in which the pathological significance score and the evidence score (the first score and the second score) are plotted. Thereby, the user can easily visually confirm the determination result of the pathological significance of the gene mutation.

[0063] [Third Embodiment] In the above-described second embodiment, it has been described that a graph in which the estimated mutation score is plotted on a two-dimensional plane is output. Hereinafter, as described in the third embodiment, each region of "pathologically significant", "not pathologically significant", and "VUS" may be further illustrated on the graph.

[0064] FIG. 10 is a schematic diagram of a setting screen 200 according to the third embodiment. The setting screen of this embodiment further has a drawing condition button 209 for editing the drawing condition of the boundary curve in addition to the content of the setting screen of the second embodiment. As the drawing condition of the boundary curve, for example, it can be configured such that the function forms of the boundaries of "pathologically significant", "not pathologically significant", and "VUS" can be input respectively. Further, the setting screen 200 of this embodiment can be configured such that the drawing condition of the "region to be preferentially discussed" can be input.

[0065] The drawing conditions of the boundary curve may be set to draw an area where the correct prediction rate in the learning data 11 is 100%. At this time, the correct prediction rate in the learning data 11 is represented by the following formula (3).

[0066] [Number]

[0067] FIG. 11 is a schematic diagram of the output screen 500 according to the third embodiment. As shown in FIG. 11, with respect to the graph 503 of the mutation score shown in FIG. 8, the boundary curve of the region is superimposed and displayed. The regions separated by the boundary curve can be shown in different colors. Also, the portion near the boundary of the region is emphasized as the "region to be preferentially discussed". The mutations to be preferentially confirmed and discussed are mutations in which both the determination items supporting the pathogenic significance and the determination items not supporting the pathogenic significance exist, and there is no positive reason to determine as VUS. Further, when a plot is included in the "region to be preferentially discussed", it may be displayed via the output device that manual discussion is necessary. In this way, for mutations near the boundary line that have both evidences of "pathogenic significance / no pathogenic significance" and cause experts to be confused in judgment, experts can discuss them preferentially. On the other hand, for plots with a high accuracy of estimating the pathogenic significance, the estimation results of the pathogenic significance determination device 100 can be directly adopted, so that the labor and human burden of discussion can be reduced.

[0068] In addition to the region where 100% of the included data is correct, the boundary curve may also be illustrated for, for example, a region where 80% is correct, a region where 60% is correct, etc., and may be in the form of contour lines. In other words, among the correct prediction rates in the learning data 11, the boundary lines of a plurality of regions indicating a plurality of predetermined rates can be drawn.

[0069] [Test method] FIG. 12 is a flowchart of a method for determining pathological significance using the learned learning model 15 according to the third embodiment. The method for determining pathological significance in the third embodiment is different from the second embodiment in that it further includes step S3052. In step S3052, the processor 101 generates a boundary curve based on the drawing condition of the boundary curve set from the drawing condition button 209, and generates a figure superimposed on the graph generated in step S3051. Note that the learning method of the learning model 15 is the same as that in the first embodiment.

[0070] <Summary of the Third Embodiment> As described above, in the pathological significance determination apparatus according to the third embodiment, for the two-dimensional graph in which the pathological significance score and the evidence score are plotted, the boundary line of the region where the ratio of the correctly determined pathological significance in the learning data 11 is 100% (predetermined ratio) is drawn. Then, by displaying the plots near the boundary line (gene mutations having a predetermined relationship with the boundary line) as the "region to be preferentially discussed", the user can be prompted to confirm the determination of the pathological significance.

[0071] [Fourth Embodiment] In the above-described first embodiment, it has been described that the user selects the files of the learning data 11 and the test data 13 and uses them as the input of the learning model 15. Instead, as described in the fourth embodiment, the pathological significance determination apparatus may create the learning data 11 and the test data 13.

[0072] FIG. 13 is a configuration diagram of a pathological significance determination device 400 according to a fourth embodiment. The pathological significance determination device 400 further includes a network adapter 401 and is configured to be able to access an external public database 403 via a network 402. The public database 403 is a public known mutation information database such as Clinvar, COSMIC, etc. In the fourth embodiment, the processor 101 is configured to create learning data 11 and test data 13 using the information obtained by accessing the public database 403 and store them in the storage device 103. The method for creating the learning data 11 and the test data 13 will be described later.

[0073] FIG. 14 is a schematic diagram of a setting screen 200 according to the fourth embodiment. The setting screen of this embodiment has a table 210 and a setting button 211 for creating learning data 11 and test data 13 in addition to the contents of the setting screen of the first embodiment. The table 210 includes a check column for selecting the use or non-use of determination items, an editing column for determination items, and an editing column for determination rules. Note that the determination items and determination rules may be set by default or can be arbitrarily edited by the user.

[0074] The determination items set in the table 210 correspond to the determination items in the gene mutation-related information 17 described with reference to FIGS. 2A and 2C. Examples of determination items include, for example, polymorphic allele frequency, the number of case reports in the public database 403, and the like.

[0075] The determination rule is described as a rule for extracting information that is the determination content in the gene mutation-related information 17 described with reference to FIGS. 2A and 2C from the public database 403. As an example of the determination rule, for the polymorphic allele frequency, a rule such as "Search the public database and input the allele frequency of the mutation with the same mutation name" can be set. For the number of case reports, a rule such as "In the mutation with the same mutation name, extract and input the number of case registrations in the cancer type of that case for another item" can be set.

[0076] After the user inputs the necessary information in Table 210, the user clicks the setting button 211. When the setting button 211 is clicked, the processor 101 creates the training data 11 and the test data 13 based on the information input in Table 210 by executing a predetermined program. The training data 11 and the test data 13 are created in the formats described with reference to FIGS. 2A and 2C. Depending on which of the setting button 203 and the setting button 211 is clicked, the processor 101 can determine whether to use the file specified via the box 202 as the training data 11 and the test data 13, or to create the training data 11 and the test data 13 using the data input by the user via Table 210.

[0077] <Learning Method> FIG. 15 is a flowchart of a learning method of the learning model 15 according to the fourth embodiment. The learning method of the fourth embodiment is different from the first embodiment in that steps S307 and S308 are further included between step S301 and step S302.

[0078] In step S307, the processor 101 receives an input (check) of the determination item for which the mutation score is to be calculated from the user via the setting screen 200. Optionally, the processor 101 receives an edit of the determination rule from the user via the setting screen 200. In step S308, the processor 101 refers to and determines the public database 403 based on the determination rule, and creates the training data and the test data. Note that the method for determining the pathological significance using the trained learning model 15 is the same as that of the first embodiment.

[0079] <Modification Example of the Fourth Embodiment> In the fourth embodiment, it has been described that the processor 101 accesses an external public database 403 via the network 402 to obtain information necessary for creating the training data 11 and the test data 13. Alternatively, the vendor of the pathological significance determination device 400 may create a database similar to the public database 403 in advance and store it in the storage device 103 as a database 404 (see FIG. 13). In this case, in the above-described step S308, the processor 101 is configured to refer to the database 404. Also, in this case, the network adapter 401 becomes unnecessary.

[0080] Alternatively, the processor 101 may download information that can be obtained in the public database 403 and store it in the storage device 103 as a database 404.

[0081] <Summary of the Fourth Embodiment> As described above, in the pathological significance determination device 400 according to the fourth embodiment, the processor 101 creates, as gene mutation-related information 17, data including determination items for determining pathological significance and their determination contents based on information that can be obtained from an external public database 403 (public known mutation information database). In this way, by having the pathological significance determination device 400 be responsible for collecting information necessary for determining the pathological significance of gene mutations, the vendor or user can be relieved of the trouble of preparing the training data 11 and the test data 13.

[0082] [Modification Example] The present disclosure is not limited to the above-described embodiments and includes various modification examples. For example, the above-described embodiments have been described in detail for easy understanding of the present disclosure, and it is not necessary to have all the configurations described. Also, a part of one embodiment can be replaced with the configuration of another embodiment. Also, the configuration of another embodiment can be added to the configuration of one embodiment. Also, for a part of the configuration of each embodiment, a part of the configuration of another embodiment can be added, deleted, or replaced.

Description of Reference Numerals

[0083] 100, 400... Significance determination device for diseases 101... Processor 102... Memory 103... Storage device 104... Display device 105... Input device 106... Bus 11... Training data 12... Correct answer data 13... Test data 14... Mutation score 15... Learning model 16... Gene mutation information 17... Gene mutation-related information

Claims

1. An input device that receives an input of gene mutation information indicating a gene mutation and gene mutation-related information related to the gene mutation information; A processor that estimates a first score regarding the presence or absence of the pathological significance of the gene mutation and a second score regarding the strength or sufficiency of evidence related to the gene mutation based on the gene mutation information and the gene mutation-related information; An output device that outputs the estimated first score and second score; A pathological significance determination device comprising the above.

2. The pathological significance determination device according to claim 1, wherein the processor further determines the pathological significance of the gene mutation based on a combination of the first score and the second score.

3. The pathological significance determination device according to claim 1, wherein the output device outputs a two-dimensional graph plotting the first score and the second score.

4. The pathological significance determination device according to claim 1, wherein the processor further creates data including determination items for determining the pathological significance and their determination contents as the gene mutation-related information based on information obtainable from a publicly known mutation information database external to the pathological significance determination device.

5. The device further comprises a storage device that stores a learning model constructed to estimate the first score and the second score using the gene mutation information and the gene mutation-related information as inputs, The pathological significance determination device according to claim 1, wherein the processor inputs the gene mutation information and the gene mutation-related information to the learning model to output the first score and the second score.

6. The pathological significance determination device according to claim 5, wherein the processor further trains the learning model using learning data in which the gene mutation information and the gene mutation-related information are associated with the correct first score and second score.

7. The output device, draws a boundary line of a region indicating a predetermined ratio at which the pathological significance is correctly determined in the learning data for the two-dimensional graph plotting the first score and the second score. The pathological significance determination device according to claim 6.

8. The output device according to claim 7, wherein the output device highlights and displays a region including the gene mutation having a predetermined relationship with the boundary line.

9. The output device further The disease significance determination device according to claim 7, wherein boundary lines of a plurality of regions indicating the plurality of predetermined ratios are drawn on the two-dimensional graph.

10. The gene mutation-related information according to claim 1, wherein the gene mutation-related information includes the presence or absence of the possibility of canceration when the gene mutation is present, the number of reported cases, or amino acid information.

11. A disease significance determination method executed by a processor of a disease significance determination device, a process of receiving an input of gene mutation information indicating a gene mutation and gene mutation-related information related to the gene mutation information from an input device of the disease significance determination device; a process of estimating a first score regarding the presence or absence of the disease significance of the gene mutation and a second score regarding the strength or sufficiency of evidence related to the gene mutation based on the gene mutation information and the gene mutation-related information; a process of outputting the estimated first score and the second score to an output device; A disease significance determination method including

12. A machine learning method for a learning model applied to a disease significance determination device that determines the disease significance of a gene mutation, a process of acquiring, by a processor, gene mutation information indicating the gene mutation and gene mutation-related information related to the gene mutation information; a process of acquiring, by the processor, correct data of a first score regarding the presence or absence of the disease significance of the gene mutation and a second score regarding the strength or sufficiency of evidence related to the gene mutation; a process of training the learning model by using the gene mutation information, the gene mutation-related information, and the correct data as learning data by the processor; A machine learning method including

13. A method for generating a trained model for application to a disease significance determination device that determines the disease significance of a gene mutation, a process of acquiring, by a processor, gene mutation information indicating the gene mutation and gene mutation-related information related to the gene mutation information; a process of acquiring, by the processor, correct data of a first score regarding the presence or absence of the disease significance of the gene mutation and a second score regarding the strength or sufficiency of evidence related to the gene mutation; The processor constructs the learned model so as to estimate the first score and the second score using the gene mutation information, the gene mutation-related information, and the correct data as learning data and using the gene mutation information and the gene mutation-related information as inputs. A method for generating a learned model, including the above.

Images