Assessment method, assessment device, assessment system, assessment program, recording medium, and method for generating trained model for degree of hemolysis

Abstract

In this assessment method, small RNA data that represents the results of measuring the expression levels of a plurality of small RNAs in blood of a subject is used to assess the degree of hemolysis of the blood collected from the subject. The assessment method has an assessment step for performing an assessment of the degree of hemolysis by inputting the small RNA data into a trained model for degree of hemolysis, obtained by performing machine learning of a correlation between: measurement data that represents the results of measuring the expression levels of a plurality of small RNAs in blood collected from a plurality of subjects, and degree of hemolysis data that represents the degree of hemolysis of the blood collected from the plurality of subjects.

Description

Inspection method, inspection device, inspection system, inspection program, recording medium, and method for generating hemolysis level trained model

[0001] The present disclosure relates to an inspection method, an inspection device, an inspection system, an inspection program, a recording medium, and a method for generating a trained model of the degree of hemolysis.

[0002] Japanese Patent No. 7021097 discloses a disease prevalence assessment device including a sample data acquisition unit and a prevalence assessment unit. The sample data acquisition unit acquires sample data including expression levels of multiple types of miRNA in a biological sample. The prevalence assessment unit uses a trained model to output prevalence assessment results for the multiple diseases in the multiple body parts for the acquired sample data. The trained model is a trained model that is capable of determining the prevalence of each of the multiple diseases, including multiple malignant diseases or multiple benign diseases, including cases where the subject is affected by multiple diseases, obtained in advance by machine learning using training data including multiple sample data having items for identifying the presence or absence of the multiple diseases in the multiple body parts.

[0003] WO 2021 / 132547 discloses a testing method for testing for a disease using a disease marker, the testing method including a specimen data acquisition step and a discrimination step. The specimen data acquisition step acquires marker data indicating the results of measuring a disease marker in a body fluid sample collected from a subject, and preparation data indicating the preparation conditions of the body fluid sample. The discrimination step determines the presence or absence of a disease in the subject by inputting the marker data and preparation data acquired in the specimen data acquisition step into a trained model that has been machine-learned to determine the correlation between a set of marker data indicating the results of measuring a disease marker in a body fluid sample and preparation data indicating the preparation conditions of the body fluid sample, and the presence or absence of a disease in the subject from whom the body fluid sample was collected.

[0004] WO 2018 / 056432 discloses a method for detecting hemolysis, which includes a measurement step of measuring the abundance of at least one miRNA selected from the group consisting of miR-191-5p, miR-22-3p, miR-3158-5p, miR-320a, miR-484, miR-486-5p, miR-6073, miR-6131, miR-642b-3p, miR-92a-3p, and miR-92b-3p in a blood sample.

[0005] Here, when diagnosing a disease using blood collected from a subject, the expression level of small RNA changes depending on the degree of hemolysis of the blood, and therefore the degree of hemolysis of the blood affects the diagnosis of the disease.

[0006] An object of the present disclosure is to provide a testing method, testing device, testing system, testing program, and recording medium that are capable of testing the degree of hemolysis of blood collected from a subject with high accuracy.

[0007] A testing method according to one aspect of the present disclosure is a testing method for testing the degree of hemolysis of blood collected from a subject using small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the blood of the subject, and includes a testing step of testing the degree of hemolysis by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data indicating the degree of hemolysis of the blood collected from the multiple subjects.

[0008] According to the present disclosure, the degree of hemolysis of blood collected from a subject can be tested with high accuracy.

[0009] It is a flowchart showing each step of the inspection method according to the present embodiment.It is a schematic block diagram of an example of a computer that functions as the inspection apparatus of the present embodiment.It is a block diagram showing the functional configuration of the inspection apparatus of the present embodiment.

[0010] An example of an embodiment of the technology of the present disclosure will be described below with reference to the drawings. Note that components and processes that perform the same operations, actions, and functions are given the same reference numerals throughout the drawings, and redundant explanations may be omitted as appropriate. Each drawing is merely a schematic illustration to allow a sufficient understanding of the technology of the present disclosure. Therefore, the technology of the present disclosure is not limited to the illustrated examples. Furthermore, in this embodiment, explanations of configurations that are not directly related to the present disclosure or well-known configurations may be omitted.

[0011] <Inspection Method 10> First, an explanation will be given of an inspection method 10 according to this embodiment. Fig. 1 is a schematic diagram showing each step of the inspection method 10 according to this embodiment.

[0012] Test method 10 is a test method that uses small RNA data to test the degree of hemolysis of blood collected from a subject. The term "blood" may refer not only to blood itself but also to blood components (e.g., serum and plasma) that are part of blood. A subject is a concept that refers to a subject from which blood is collected. The subject from which blood is collected may be a human or a non-human animal. Examples of non-human animals include non-human mammals (monkeys, dogs, cats, mice, rats, rabbits, cows, horses, pigs, sheep, etc.), birds (chickens, quails, etc.), etc.

[0013] Examples of small RNAs include microRNAs. Note that the small RNA may also be small RNAs other than microRNAs (for example, piRNAs and tsRNAs).

[0014] 1, the testing method 10 specifically includes a collection step 11, a measurement step 12, a testing step 14, and a disease determination step 16. In the present embodiment, the collection step 11, the measurement step 12, the testing step 14, and the disease determination step 16 are performed in this order, for example.

[0015] The testing method 10 includes a sampling step 11, a measurement step 12, and a disease assessment step 16, and can therefore also be called a sampling method, a measurement method, and a disease assessment method. Furthermore, if analysis is performed based on the assessment in the disease assessment step 16, the testing method can also be called an analysis method. Each step of the testing method 10 will be described below.

[0016] <Collecting step 11> The collecting step 11 is a step of collecting blood from a subject. The collecting step 11 is performed before the measuring step 12. When collecting serum, which is a blood component, in the collecting step 11, the collection process is performed, for example, as follows.

[0017] First, blood is collected using a vacuum blood collection tube containing a serum separator (blood collection process). Next, immediately after the blood collection process, the blood is mixed by inversion. Next, the blood is left at room temperature for 30 minutes or more to allow coagulation. Next, the blood is centrifuged and the serum is separated (centrifugation process). The serum is then separated and stored at -80°C (storage process).

[0018] In the collection step 11, the steps from collecting blood from the subject to completing the centrifugation procedure (specifically, the steps from the blood collection process to the preservation process) are performed within a predetermined time (e.g., 2 hours). The centrifugation procedure is, for example, the steps from the centrifugation process to the preservation process.

[0019] <Measurement Step 12> Measurement Step 12 is a step of measuring the expression levels of multiple small RNAs in the subject's blood. Specifically, in Measurement Step 12, a next-generation sequencer (NGS) is used as a measurement device to measure multiple small RNAs contained in the subject's blood (specifically, serum) and identify the base sequence of each small RNA. Next, the number of identified small RNAs is counted for each base sequence to determine the number of small RNA reads in the NGS. This number of small RNA reads corresponds to the expression level (specifically, the absolute expression level) of the small RNA. In other words, this number of small RNA reads represents small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the subject's blood. However, although the expression levels of multiple small RNAs (e.g., microRNAs) in blood are absolute values ​​derived from the living body, it is difficult to quantify the expression levels in blood as absolute values ​​because the expression levels must be quantified through a process involving measurement devices and reagent treatment. Therefore, the expression levels of small RNAs may be expressed as relative values ​​by processing the measurement results of NGS, and such relative values ​​also represent small RNA data.

[0020] Furthermore, to measure absolute expression levels, the number of reads of small RNAs output by NGS may be normalized to obtain relative expression levels (i.e., relative expression levels). For example, normalization methods such as RPM (Read Per Million) normalization and normalization using an internal standard small RNA may be used. In this way, small RNA data may indicate normalized relative expression levels. Small RNA data may also be absolute quantitative values ​​quantified as absolute values.

[0021] Furthermore, NGS allows the expression levels of multiple small RNAs to be measured simultaneously in blood samples collected from multiple subjects, i.e., in measurement step 12, the expression levels of multiple small RNAs are measured for multiple samples (specifically, blood) in the same process.

[0022] In addition to next-generation sequencers, other measuring devices such as DNA chips, quantitative PCR, and flow cytometers can also be used as long as they can measure the expression levels of multiple small RNAs. As described above, if the expression levels of multiple small RNAs can be measured and small RNAs that represent the measurement results can be obtained, various methods, including known methods, can be used to measure the expression levels of multiple small RNAs.

[0023] <Testing step 14> The testing step 14 is a step of testing the degree of hemolysis of blood collected from a subject. Specifically, in the testing step 14, the small RNA data obtained in the measurement step 12 is input into a trained model for the degree of hemolysis, thereby testing whether the degree of hemolysis is appropriate.

[0024] The test criteria for determining whether the degree of hemolysis is appropriate are set, for example, based on whether a correct determination can be made in the disease determination step 16. If a correct determination can be made even when a low level of hemolysis has occurred in the disease determination step 16, the test criteria will be set to allow a low level of hemolysis. For example, in the testing step 14, it is possible to determine whether no hemolysis has occurred or whether a low level of hemolysis has occurred as appropriate (good), and whether a moderate level of hemolysis has occurred or whether a high level of hemolysis has occurred as inappropriate (poor). The acceptable level of hemolysis varies depending on the disease to be determined in the disease determination step 16, and the test criteria in the testing step 14 also vary accordingly. Furthermore, the determination step is not limited to disease determination, and various other determinations can be applied. The acceptable level of hemolysis varies depending on the target of determination, and the test criteria in the testing step 14 also vary accordingly.

[0025] <Method for generating trained model of degree of hemolysis> Here, the trained model of degree of hemolysis used in the testing process 14 is generated as follows: That is, the trained model of degree of hemolysis is generated by machine learning the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis degree data indicating the degree of hemolysis in the blood collected from the multiple subjects.

[0026] Specifically, the measurement data is the result of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects. The multiple subjects may include subjects with a disease and subjects without a disease (i.e., healthy subjects), or may include only subjects with a disease and subjects without a disease (i.e., healthy subjects).

[0027] The measurement data is obtained by the same measurement method as the small RNA data described above. Specifically, a next-generation sequencer (NGS) is used as a measurement device to measure the expression levels of multiple small RNAs contained in the subject's blood and identify the base sequence of each small RNA. Next, the number of identified small RNAs is counted for each base sequence to determine the number of small RNA reads in the NGS. This number of small RNA reads corresponds to the expression level (specifically, the absolute expression level) of the small RNA. In other words, this number of small RNA reads becomes the measurement data.

[0028] Specifically, the hemolysis level data is data indicating whether the hemolysis level of the blood related to the measurement data is appropriate. The hemolysis level in the hemolysis level data is based on whether the hemolysis level is appropriate for determination in the disease-trained model described below used in the disease determination step 16. That is, the hemolysis level data is data that defines a hemolysis level at which a correct determination can be made in the disease determination step 16 as appropriate (good), and a hemolysis level at which a correct determination cannot be made in the disease determination step 16 as inappropriate (bad).

[0029] Specifically, a trained model for the degree of hemolysis is generated using measurement data (hereinafter sometimes referred to as training data) to be used for machine learning, including measurement data obtained from blood collected from multiple subjects under conditions where the degree of hemolysis is appropriate, and measurement data obtained from blood collected from multiple subjects under conditions where the degree of hemolysis is inappropriate.

[0030] The disease of the subject that is machine-learned by the hemolysis degree trained model used in the testing step 14 may include the disease that is determined in the disease assessment step 16. That is, if the disease that is determined in the disease assessment step 16 is a cancer, the measurement data of a sample from a subject with the cancer is included. Note that the type of cancer determined in the disease assessment step 16 and the type of cancer in the measurement data may be the same or different. Therefore, for example, if the disease that is determined in the disease assessment step 16 is pancreatic cancer, the measurement data of a sample from a subject with lung cancer may be used.

[0031] In addition, in the testing step 14, measurement data of the top 100 small RNAs expressed in the blood of a plurality of subjects is used as training data.

[0032] Thus, in this embodiment, measurement data is used that excludes small RNAs with relatively low expression levels. Small RNAs with relatively low expression levels have the advantage of functioning easily as explanatory variables because even small fluctuations in expression level result in large fold changes. However, they are susceptible to fluctuations due to factors other than the element of interest (specifically, the degree of hemolysis) and fluctuations due to measurement errors, and therefore have the disadvantage of being less robust. Therefore, using only small RNAs with relatively high expression levels as training data can have the effect of enhancing robustness.

[0033] Furthermore, measurement data excluding small RNAs whose expression level was zero in the blood of any of the subjects is used as training data in the testing step 14. Since fluctuations from zero result in an infinite fold change and have a large impact, small RNAs whose expression level was zero in any of the samples used as training data are excluded from the training data.

[0034] As described above, in this embodiment, the trained model of the degree of hemolysis is generated by having the learning model qualitatively learn whether the degree of hemolysis is appropriate. Then, the trained model of the degree of hemolysis determines whether the degree of hemolysis is appropriate and outputs the appropriateness, thereby testing the appropriateness of the degree of hemolysis in the testing step 14. Therefore, the testing step 14 displays a test result indicating that the degree of hemolysis is appropriate or inappropriate.

[0035] In this embodiment, the trained model for the degree of hemolysis is generated by having the learning model learn the appropriateness of the degree of hemolysis qualitatively, but this is not limiting. For example, the trained model for the degree of hemolysis may be generated so that the degree of hemolysis is determined as one of three values: appropriate (good), warning, and inappropriate (bad).

[0036] Furthermore, for example, a trained model for the degree of hemolysis may be generated by quantitatively training a learning model on the degree of hemolysis. In this case, the degree of hemolysis is determined by the trained model for the degree of hemolysis, and the determination result of the degree of hemolysis is output, for example, as a judgment value. The judgment value here is not a binary value of 0 or 1, but a value having a predetermined numerical range. Then, in the inspection process 14, the appropriateness of the degree of hemolysis is inspected based on the judgment result of the judgment unit, which determines whether the degree of hemolysis is appropriate based on a comparison between the output judgment value and a threshold value. That is, in the inspection process 14, if the judgment unit determines that the output judgment value is equal to or greater than the threshold value, the inspection result indicates that the degree of hemolysis is inappropriate, and if the judgment unit determines that the output judgment value is less than the threshold value, the inspection result indicates that the degree of hemolysis is appropriate.

[0037] <Presence determination step 16> Presence determination step 16 is a step of determining the presence or absence of a disease. Specifically, in the presence determination step 16, the small RNA data obtained in the measurement step 12 is input into a disease-trained model that has been machine-learned to determine the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood and prevalence data indicating the presence or absence of a disease in the subject from whom the blood was collected, thereby determining the presence or absence of a disease.

[0038] In the disease determination step 16, if the test result in the testing step 14 indicates that the degree of hemolysis is appropriate, the presence or absence of a disease is determined. In other words, if the test result in the testing step 14 indicates that the degree of hemolysis is inappropriate, the disease determination step 16 is not executed. In this case, for example, the test result indicating that the degree of hemolysis is inappropriate may be presented to the user (i.e., the person who performs the testing method).

[0039] It is desirable that the disease determined by the disease-trained model be the same disease as the disease used for training in the hemolysis level trained model. For example, if the disease used for training in the hemolysis level trained model is a cancer, the disease determined by the disease-trained model is also a cancer. Specifically, it is desirable that the measurement data used in the disease-trained model be the same data as the measurement data used in the hemolysis level trained model.

[0040] As described above, if the test result in the testing step 14 indicates that the level of hemolysis is inappropriate, the disease determination step 16 is not executed; however, this is not limited to this. For example, even if the test result in the testing step 14 indicates that the level of hemolysis is inappropriate, the disease determination step 16 may be executed if the determination result is to be obtained as reference data, for example. In this case, for example, the fact that the determination was made may be presented to the user after the test result indicating that the level of hemolysis is inappropriate is obtained. Furthermore, although the disease determination step 16 is executed after the testing step 14 in this embodiment, it may also be executed before the testing step 14. In this case, if the test result in the testing step 14 indicates that the level of hemolysis is inappropriate, the determination result in the disease determination step 16 is treated as reference data, for example.

[0041] <Inspection System 20> Next, a description will be given of an inspection system 20 that performs the above-described inspection method. As shown in FIG.

[0042] <Measurement device 21> The measurement device 21 is an example of a measurement unit, and is a device that performs the above-mentioned measurement step 12. That is, the measurement device 21 measures the expression levels of multiple small RNAs contained in each of multiple samples. As the measurement device 21, for example, an NGS is used.

[0043] <Testing Device 30> The testing device 30 is an example of a testing unit. The testing device 30 is a device that executes the above-mentioned testing step 14. That is, the testing device 30 acquires small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the blood of the subject, and inputs the acquired small RNA data into a trained model for the degree of hemolysis, thereby testing the degree of hemolysis of the blood collected from the subject.

[0044] Furthermore, the testing device 30 executes the aforementioned disease determination step 16. That is, the testing device 30 determines the presence or absence of a disease by inputting the small RNA data obtained in the measurement step 12 into a disease-trained model that has been machine-learned to determine the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in the blood and disease prevalence data indicating the presence or absence of a disease in the subject from whom the blood was collected.

[0045] 2, the inspection device 30 has a function as a computer, and includes a CPU (Central Processing Unit) 31, a ROM (Read Only Memory) 32, a RAM (Random Access Memory) 33, a storage 34, an input unit 35, a display unit 36, and a communication interface (I / F) 37. The components are connected to each other via a bus 39 so as to be able to communicate with each other.

[0046] The CPU 31 (an example of a processor) is a central processing unit that executes various programs and controls each part. That is, the CPU 31 reads a program from the ROM 32 or the storage 34 and executes the program using the RAM 33 as a work area. The CPU 31 controls each of the above components and performs various arithmetic processing in accordance with the program stored in the ROM 32 or the storage 34. The CPU 31 is an example of a processor.

[0047] The ROM 32 stores various programs and various data. The RAM 33 temporarily stores programs or data as a working area. The storage 34 is configured by an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs including the operating system and various data.

[0048] In this embodiment, for example, an inspection program for executing an inspection process that performs the above-described inspection method is recorded in the storage 34. The inspection program may be a single program, or a group of programs configured from multiple programs or modules. The inspection program may be recorded in the ROM 32. The ROM 32 and the storage 34 function as an example of a non-transitory recording medium.

[0049] An example of a processor is not limited to the above-mentioned CPU, which is a general-purpose processor, but may be, for example, a dedicated processor configured with a circuit designed specifically for executing a specific process. Also, an example of a processor is not limited to a single processor, but may be a processor configured by multiple processors located in physically separate locations working together.

[0050] The input unit 35 includes a pointing device such as a mouse and a keyboard, and is used to input various information. The input unit 35 also receives information on the expression levels of multiple small RNAs measured by the measurement device 21 as input.

[0051] The display unit 36 ​​is, for example, a liquid crystal display, and displays various information. The testing device 30 can present the test results of the degree of hemolysis and the determination result of the presence or absence of a disease to the user through the display unit 36. The display unit 36 ​​may be a touch panel type and function as the input unit 35.

[0052] The communication interface 37 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI (Fiber Distributed Data Interface), and Wi-Fi (registered trademark).

[0053] As shown in FIG. 3 , in the testing device 30 , the CPU 31 executes the testing program to function as a testing function unit 160 and a disease determination unit 170 .

[0054] The inspection function unit 160 executes the above-mentioned inspection step 14. That is, in the inspection step 14, the inspection function unit 160 inputs the small RNA data obtained by the measurement device 21 into the hemolysis degree trained model, thereby inspecting whether the degree of hemolysis of the blood collected from the subject is appropriate (see the above-mentioned inspection step 14).

[0055] The morbidity determination unit 170 executes the morbidity determination step 16. That is, the morbidity determination unit 170 determines the presence or absence of a disease by inputting the small RNA data obtained by the measurement device 21 into a disease-trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in the blood and morbidity data showing the presence or absence of a disease in the subject from whom the blood was collected (see the morbidity determination step 16 described above).

[0056] In this embodiment, the inspection system 20 includes the measuring device 21 and the inspection device 30, but the inspection system 20 may be configured with a single device. In this case, the single device functions as an example of a measurement unit and an inspection unit.

[0057] Furthermore, the testing device 30 may be composed of a plurality of devices. For example, the testing device 30 may be composed of a plurality of devices (e.g., two devices) that share the tasks of performing the testing step 14 and the disease determination step 16.

[0058] <Effects of the Present Embodiment> In the present embodiment, as described above, in the testing step 14, the degree of hemolysis of the blood collected from the subjects is tested by inputting the small RNA data obtained in the measurement step 12 into a hemolysis level trained model that has been machine-learned to determine the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data indicating the hemolysis level of the blood collected from the multiple subjects.

[0059] In this way, the degree of hemolysis of blood is tested by inputting small RNA data into a hemolysis degree trained model that has learned the correlation between measurement data and hemolysis degree data through machine learning. This makes it possible to test the degree of hemolysis of blood with a high degree of accuracy compared to testing by visual inspection or measuring Hb concentration using a measurement kit, for example.

[0060] In this way, in this embodiment, the degree of hemolysis can be tested with high accuracy, so that highly accurate information on the degree of hemolysis can be obtained, and for example, in the disease determination step 16, there is no need to obtain information on the degree of hemolysis by inputting it by the user.

[0061] Furthermore, in this embodiment, the degree of hemolysis in the hemolysis degree data is based on whether the degree of hemolysis is appropriate for determination in the disease-trained model used in the disease determination step 16. That is, the hemolysis degree data is data that defines a hemolysis degree that allows correct determination in the disease determination step 16 as appropriate (good), and a hemolysis degree that does not allow correct determination in the disease determination step 16 as inappropriate (bad).

[0062] This prevents unnecessary exclusion of hemolyzed samples from being assessed in the disease assessment step 16, enabling efficient operation.

[0063] <Examples> Next, examples will be described. Note that the examples are intended to illustrate examples of the technology of the present disclosure, and the technology of the present disclosure is not limited to the contents of the examples.

[0064] <Preparation of Hemolyzed Samples> In this example, serum samples were obtained from 10 healthy individuals by performing the collection step 11 within 2 hours. Then, a dilution series of blood clots was mixed with each of the sera from the 10 healthy individuals to artificially prepare hemolyzed samples. The degree of hemolysis was set to 10 levels, from 0 to 9. This resulted in hemolyzed samples (10 healthy individuals x 10 levels of hemolysis) (i.e., 100 types of hemolyzed samples). Level 0 represents a hemolysis level where no hemolysis occurred. Levels 1 to 9 represent hemolysis levels prepared so that the concentration of blood clots in the serum gradually increases. As described above, when the degree of hemolysis was set to 10 levels, levels 6 and below were hemolysis levels that were difficult to determine visually. Furthermore, levels 4 and below were hemolysis levels that were difficult to determine by measuring Hb concentration using a measurement kit or the like.

[0065] <Acquisition of RNA Data> The expression levels of multiple small RNAs are measured for the obtained 100 types of hemolyzed samples using NGS, and small RNA data showing the measurement results are obtained. From the obtained small RNA data, for example, small RNAs with zero expression levels are excluded, and small RNA data for the top 100 small RNAs with high expression levels are used.

[0066] <Setting the Degree of Hemolysis for Determination in the Disease-Trained Model> Here, the degree of hemolysis at which a correct determination can be made in the disease determination step 16 was confirmed. Specifically, small RNA data from the hemolyzed samples of the 100 types of healthy individuals was input into a disease-trained model for determining the presence of lung cancer, and the accuracy rate for correctly determining that lung cancer was negative was confirmed. As a result, the accuracy rate was 100% at level 0, 90% at level 1, 50% at level 2, and 10% or less at levels 3 to 9. Therefore, the degree of hemolysis at which a correct determination can be made in the disease-trained model was set to levels 0 and 1.

[0067] In the disease-trained model used to determine lung cancer incidence, the hemolysis levels at which correct determination was possible were levels 0 and 1. However, the hemolysis level at which correct determination was possible is thought to vary depending on the disease-trained model. For example, in the disease-trained model used to determine diabetes incidence, the hemolysis level at which correct determination was possible was only level 0. Here, it was found that the hemolysis threshold at which correct determination was possible in the disease-trained model used to determine lung cancer incidence exists between levels 1 and 2. When comparing hemolysis levels 1 and 2, the difference is very small in terms of color tone and Hb concentration. It was difficult to detect the difference by visual inspection or by measuring Hb concentration using a measurement kit. It was shown that the use of the present invention makes it possible to determine such extremely small differences in hemolysis level.

[0068] <Generation of hemolysis level trained model> Based on the above setting results, hemolysis level levels 0 and 1 are used as training data for the good data group, and hemolysis level levels 2 to 9 are used as training data for the bad data group to generate a hemolysis level trained model. In other words, the hemolysis level trained model is a model that makes a binary judgment of good and bad. [Good data group] Hemolysis level: Level 0 and 1 [Bad data group] Hemolysis level: Level 2 to 9

[0069] <Examples of learning models> Various linear and non-linear algorithms known as machine learning algorithms can be used, or multiple algorithms can be combined. For example, the following algorithms can be used:

[0070] Random forest Gradient boosted trees Extreme gradient boosted trees Light-gradient boosted machine Neural networks Regularized regression Elastic-net K-Nearest neighbors Support vector machine Generalized additive model

[0071] <Performance Verification of the Hemolysis Level Trained Model> Small RNA data from serum samples obtained from 46 healthy volunteers and 46 lung cancer patients were input into the hemolysis level trained model to determine the appropriateness of the hemolysis level. As a result, the accuracy rate was 100% (46 / 46 correct answers) for the 46 healthy volunteers and 100% (46 / 46 correct answers) for the 46 lung cancer patients. Note that spontaneous hemolysis occurred in one lung cancer patient sample, but no hemolysis occurred in the other samples. As described above, it was confirmed that it is possible to set criteria for determining the appropriateness of the hemolysis level according to the disease-trained model and generate a hemolysis level trained model capable of performing this judgment. Furthermore, it was confirmed that even a hemolysis level trained model created using artificially generated hemolytic samples can correctly determine the appropriateness of the hemolysis level for naturally occurring hemolysis.

[0072] The present invention is not limited to the above-described embodiments, and various modifications, alterations, and improvements are possible without departing from the spirit and scope of the present invention. In the testing method according to this embodiment, the disease to be determined is cancer, but this is not limited thereto. The disease to be determined may be, for example, other diseases such as diabetes. Furthermore, although the testing method according to this embodiment includes a disease determination step, this is not limited thereto. The testing method may include, for example, a step of performing determination, analysis, experiment, etc. using blood other than the disease, instead of the disease determination step. Furthermore, the testing method is sufficient as long as it includes at least a testing step, and it is not essential to perform the disease determination step, and the testing method may not include the disease determination step.

[0073] As described above, in testing method 10, the presence or absence of a disease is indicated by inputting the small RNA data obtained in measurement step 12 into a disease-trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood and disease prevalence data showing the presence or absence of a disease in the subject from whom the blood was collected. Therefore, the measured expression levels of small RNAs can be said to indicate the presence or absence of a disease based on the judgment criteria in the disease-trained model. Therefore, test method 10 is a test method for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the subject's blood, wherein the expression levels of the small RNAs indicate the presence or absence of a disease based on judgment criteria in a disease-trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in the blood and disease incidence data showing the presence or absence of a disease in the subject from whom the blood was collected, and the degree of hemolysis in the hemolysis level data machine-trained in the hemolysis level trained model can also be said to be a test method based on whether the degree of hemolysis is appropriate for judgment in the disease-trained model.

[0074] <Additional Notes> (Aspect 1) A testing method for testing the degree of hemolysis of blood collected from a subject using small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the subject's blood, comprising: a testing step of testing the degree of hemolysis by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine a correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data indicating the degree of hemolysis of the blood collected from the multiple subjects. (Aspect 2) The testing method according to Aspect 1 further comprises: a disease determination step of determining the presence or absence of a disease by inputting the small RNA data into a disease-trained model that has been machine-learned to determine a correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood and disease level data indicating the presence or absence of a disease in the subject from whom the blood was collected, wherein the degree of hemolysis in the hemolysis level data that is machine-learned by the hemolysis level trained model is based on whether the degree of hemolysis is appropriate for determination in the disease-trained model. (Aspect 2-2) The test method according to Aspect 1, wherein the expression level of the small RNA indicates the presence or absence of the disease based on a judgment criterion in a disease-trained model that has been machine-learned to learn a correlation between measurement data showing the results of measuring expression levels of a plurality of small RNAs in blood and morbidity data showing the presence or absence of the disease in the subject from whom the blood was collected, and the degree of hemolysis in the hemolysis level data that has been machine-learned in the hemolysis level-trained model is based on whether the degree of hemolysis for judgment in the disease-trained model is appropriate. (Aspect 3) The test method according to Aspect 1 or Aspect 2, wherein the hemolysis level-trained model judges whether the degree of hemolysis is appropriate and outputs the appropriateness. (Aspect 4) The test method according to Aspect 3, wherein the testing step inspects the appropriateness of the degree of hemolysis. (Aspect 5) The test method according to any one of Aspects 1 to 4, wherein the hemolysis level-trained model judges the degree of hemolysis and outputs a judgment value having a predetermined numerical range.(Aspect 6) The testing method according to Aspect 5, wherein the testing step tests the appropriateness of the degree of hemolysis based on a judgment result of a judgment unit that judges the appropriateness of the degree of hemolysis based on a comparison of the judgment value output by the hemolysis level trained model with a threshold. (Aspect 7) The testing method according to any one of Aspects 2 to 6, wherein the testing step tests the appropriateness of the degree of hemolysis, and the disease judgment step judges the presence or absence of the disease if the test result in the testing step indicates that the degree of hemolysis is appropriate. (Aspect 8) A testing device that tests the degree of hemolysis of blood collected from a subject using small RNA data indicating results of measuring expression levels of multiple small RNAs in the subject's blood, comprising a processor, wherein the processor tests the degree of hemolysis by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine a correlation between measurement data indicating results of measuring expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data indicating the degree of hemolysis of the blood collected from the multiple subjects. (Aspect 9) A testing system for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the subject's blood, comprising: a measurement unit that acquires small RNA data showing the results of measuring the expression levels of multiple small RNAs in the subject's blood; and a testing unit that tests the degree of hemolysis by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine the correlation between the measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and the hemolysis level data showing the hemolysis level of the blood collected from the multiple subjects.(Aspect 10) A testing program for causing a computer to execute a testing process for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the blood of the subject, by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data showing the hemolysis level of the blood collected from the multiple subjects. (Aspect 11) A non-transitory recording medium having recorded thereon a testing program for causing a computer to execute a testing process for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the blood of the subject, by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data showing the hemolysis level of the blood collected from the multiple subjects. (Aspect 12) A method for generating a trained model of the degree of hemolysis used in a testing method for testing the degree of hemolysis of blood collected from a subject, the testing method comprising inputting small RNA data showing the results of measuring the expression levels of multiple small RNAs in the subject's blood into a trained model of the degree of hemolysis, the method generating the trained model of the degree of hemolysis by machine learning a correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data showing the degree of hemolysis of the blood collected from the multiple subjects.

[0075] The disclosure of Japanese Patent Application No. 2023-182777, filed on October 24, 2023, is incorporated herein by reference in its entirety. All documents, patent applications, and technical standards mentioned herein are incorporated herein by reference to the same extent as if each individual document, patent application, and technical standard was specifically and individually indicated to be incorporated by reference.

Claims

1. A testing method for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the blood of the subject, comprising: a testing step for testing the degree of hemolysis by inputting the small RNA data into a hemolysis level trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data showing the degree of hemolysis of the blood collected from the multiple subjects.

2. The testing method described in claim 1 further comprises a disease determination process for determining the presence or absence of a disease by inputting the small RNA data into a disease-trained model that has been machine-learned to determine the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood and disease data indicating the presence or absence of a disease in the subject from whom the blood was taken, wherein the degree of hemolysis in the hemolysis degree data machine-learned in the hemolysis degree trained model is based on whether the degree of hemolysis for determination in the disease trained model is appropriate.

3. The inspection method according to claim 1, wherein the hemolysis degree trained model judges whether the hemolysis degree is appropriate or not and outputs the appropriateness or not.

4. The inspection method according to claim 3, wherein the inspection step inspects whether the degree of hemolysis is appropriate.

5. The inspection method according to claim 1, wherein the hemolysis degree trained model judges the hemolysis degree and outputs a judgment value having a predetermined numerical range.

6. The inspection method described in claim 5, wherein in the inspection process, the appropriateness of the degree of hemolysis is inspected based on the judgment result of a judgment unit that judges the appropriateness of the degree of hemolysis based on a comparison of the judgment value output by the hemolysis degree learned model with a threshold value.

7. The testing method according to claim 2, wherein in the testing step, the appropriateness of the degree of hemolysis is tested, and in the disease determination step, the presence or absence of the disease is determined if the test result in the testing step indicates that the degree of hemolysis is appropriate.

8. A testing device that tests the degree of hemolysis of blood taken from a subject using small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the subject's blood, comprising a processor, wherein the processor tests the degree of hemolysis by inputting the small RNA data into a hemolysis degree trained model that has machine-learned the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood taken from multiple subjects and hemolysis degree data indicating the degree of hemolysis in the blood taken from the multiple subjects.

9. A testing system for testing the degree of hemolysis of blood taken from a subject using small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the subject's blood, comprising: a measurement unit that acquires small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the subject's blood; and a testing unit that tests the degree of hemolysis by inputting the small RNA data into a hemolysis degree trained model that has been machine-learned to determine the correlation between the measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood taken from multiple subjects and the hemolysis degree data indicating the degree of hemolysis in the blood taken from the multiple subjects.

10. A testing program for causing a computer to execute a testing process for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the blood of the subject, the testing process testing the degree of hemolysis by inputting the small RNA data into a hemolysis degree trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis degree data showing the degree of hemolysis of the blood collected from the multiple subjects.

11. A non-temporary recording medium having recorded thereon a testing program for causing a computer to execute a testing process for testing the degree of hemolysis of blood collected from a subject using small RNA data showing the results of measuring the expression levels of multiple small RNAs in the blood of the subject, the testing process being performed by inputting the small RNA data into a hemolysis degree trained model that has been machine-learned to determine the correlation between measurement data showing the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis degree data showing the degree of hemolysis of the blood collected from the multiple subjects.

12. A method for generating a trained model of the degree of hemolysis used in a testing method for testing the degree of hemolysis of blood collected from a subject, the method comprising inputting small RNA data indicating the results of measuring the expression levels of multiple small RNAs in the subject's blood into a trained model of the degree of hemolysis, the method generating the trained model of the degree of hemolysis by machine learning the correlation between measurement data indicating the results of measuring the expression levels of multiple small RNAs in blood collected from multiple subjects and hemolysis level data indicating the degree of hemolysis of the blood collected from the multiple subjects.

Images