Method for detecting occurrence of non-specific reaction, analysis method, analysis device, and program for detecting occurrence of non-specific reaction

Abstract

To provide the occurrence of a nonspecific reaction in an analytical system using an antigen-antibody reaction.SOLUTION: The present invention relates to a method for detecting the occurrence of a nonspecific reaction in analysis of an antigen or an antibody included in a biological sample by using a measurement reagent including an antibody or an antigen causing an antigen-antibody reaction with the antigen or the antibody. The method for detecting the occurrence of the nonspecific reaction includes: generating a data group related to the progress of the antigen-antibody reaction of the antigen or the antibody included in the biological sample with the antibody or the antigen included in the measurement reagent; inputting the data group in a deep learning algorithm; and based on a result output from the deep learning algorithm, generating information on the occurrence of the nonspecific reaction.SELECTED DRAWING: Figure 12

Description

[Technical field]

[0001] The present invention relates to a method for detecting the occurrence of a non-specific reaction, an analysis method, an analysis apparatus, and a program for detecting the occurrence of a non-specific reaction. [Background technology]

[0002] Immunoassays utilizing antigen-antibody reactions are commonly used as methods for measuring antigens and antibodies in tests of biological samples, including blood and urine. It is generally known that in tests using immunoassays, non-specific reactions occur due to rheumatoid factors, complement components C1q, and the like present in biological samples such as blood and urine, resulting in false positive test results. Patent Document 1 discloses a method for suppressing the effects of such non-specific reactions.

[0003] In Patent Document 1, non-specific reactions are suppressed by using an antibody in which the constant region (Fc portion), which is thought to be the cause of such non-specific reactions and whose amino acid sequence shows a high similarity, is blocked with a blocking agent. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] European Patent Publication No. 0566205 Summary of the Invention [Problem to be solved by the invention]

[0005] However, it is difficult to completely suppress the occurrence of non-specific reactions in immunological assays, even when the method described in Patent Document 1 is used. If a non-specific reaction occurs, the test result will be a false positive, so there is a demand for detecting the occurrence of non-specific reactions and preventing false positive test results from being reported.

[0006] An object of the present invention is to provide a method for detecting the occurrence of a non-specific reaction, an analytical method, an analytical apparatus, and a program for detecting the occurrence of a non-specific reaction, which are capable of detecting the occurrence of a non-specific reaction in an analytical system utilizing an antigen-antibody reaction. [Means for solving the problem]

[0007] The present invention relates to a method for detecting the occurrence of a non-specific reaction in an analysis of an antigen or antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antigen or the antibody, the method for detecting the occurrence of a non-specific reaction includes generating a data group related to the progress of an antigen-antibody reaction between the antigen or the antibody contained in the biological sample and the antibody or the antigen contained in the measurement reagent, inputting the data group into a deep learning algorithm, and generating information related to the occurrence of the non-specific reaction based on a result output from the deep learning algorithm.

[0008] The present invention relates to an analytical method for analyzing an antigen or an antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antigen or the antibody, the analytical method comprising the steps of: preparing a measurement sample containing the biological sample and the measurement reagent; analyzing the antigen or the antibody contained in the biological sample; and performing a method for detecting the occurrence of a non-specific reaction.

[0009] The present invention relates to an analytical device (1) that analyzes an antigen or antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antigen. The analytical device (1) includes a sample preparation unit (20) that prepares a measurement sample containing the biological sample and the measurement reagent, a light irradiation unit (10) that irradiates the measurement sample with light, a detection unit (230) that detects the light irradiated from the light irradiation unit through the measurement sample and outputs detection information corresponding to the detected light, and a control device (70). The control device (70) analyzes the antigen or antibody contained in the biological sample and detects the occurrence of a non-specific reaction in the measurement sample, and detects the occurrence of the non-specific reaction by generating a data group related to the progress of an antigen-antibody reaction between the antigen or antibody contained in the biological sample and the antibody or antigen contained in the measurement reagent based on the detection information, inputting the data group into a deep learning algorithm, and generating information related to the occurrence of the non-specific reaction based on the result output from the deep learning algorithm.

[0010] The present invention relates to a detection program for the occurrence of non-specific reactions, which, when executed by a computer, causes the computer to execute the steps of generating a group of data regarding the progress of an antigen-antibody reaction between the antigen or antibody contained in the biological sample and the antibody or antigen contained in the measurement reagent, inputting the group of data into a deep learning algorithm, and generating information regarding the occurrence of non-specific reactions based on the results output from the deep learning algorithm.

[0011] According to the method for detecting the occurrence of a non-specific reaction, the analysis method, the analysis apparatus, and the program for detecting the occurrence of a non-specific reaction, the occurrence of a non-specific reaction can be detected in an analysis system utilizing an antigen-antibody reaction. Effect of the Invention

[0012] It is possible to detect the occurrence of non-specific reactions in an analysis system utilizing an antigen-antibody reaction. [Brief description of the drawings]

[0013] [Figure 1] 1 shows an example of the appearance of an analysis device 1. [Diagram 2] 1 shows an example of the hardware configuration of an analysis device 1. [Diagram 3] 2 shows an example of the configuration of the storage device 202. [Figure 4] 2 shows a configuration example of the light irradiation unit 10. [Diagram 5] 2 shows an example of the configuration of the detection unit 230. [Figure 6] 1 shows a flow of an analysis process executed by the control device 70 of the analysis device 1. [Figure 7] 1 shows the flow of a measurement process executed by the control device 70 based on the measurement program 202a. [Figure 8] 1 shows a flow of a test substance analysis / non-specific reaction detection process executed by the control device 70 based on a test substance analysis / non-specific reaction detection program 202b. [Figure 9A] 1 shows an example of a set of data relating to the progress of an antigen-antibody reaction. [Figure 9B] The vertical axis (Y-axis) indicates absorbance, and the horizontal axis (X-axis) indicates the elapsed time (seconds: sec) from the start of output of detection information. This figure shows a plot of data group D11 corresponding to the wavelength (first wavelength) emitted from the first light source 321, from the data groups illustrated in Figure 9A. [Figure 10] An example of the configuration of the message box MB1 is shown below. [Figure 11] Provides an overview of training deep learning algorithms. [Figure 12] An example of analysis using a deep algorithm is shown below. [Figure 13] An example of the configuration of a message box MB1' is shown below. [Figure 14] 13 shows a flow of a reanalysis process executed by the control device 70. [Figure 15] An example of the output of the dialog box MB3 for prompting the operator to enter reanalysis choices is shown. [Figure 16] An example of the reanalysis item database DB3 is shown. [Figure 17] An example of the configuration of a message box MB1'' is shown below. [Figure 18] An example of the reanalysis result database DB4 is shown. [Figure 19A] 2 shows an example of the appearance of the training device 5. [Figure 19B] 2 shows an example of the hardware configuration of the training device 5. [Figure 20] 13 is a flowchart showing another method for calculating the occurrence probability of a non-specific reaction. [Figure 21] An example of the S value distribution of a group of biological samples in which non-specific reactions do not occur is shown. [Figure 22] The horizontal axis represents the probability (0.0 to 1.0) of the occurrence of a non-specific reaction output from the deep learning algorithm, and the vertical axis represents the frequency of biological samples belonging to each divided range, as shown in the histogram. [Figure 23] The ROC curve obtained as a result of performing ROC analysis on the biological samples shown in FIG. 22 is shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] 1. Analyzer 1 The analysis device of this embodiment (hereinafter simply referred to as "analysis device 1") will be described with reference to Figs. 1 to 12.

[0015] The analytical device 1 measures a biological sample by immunoturbidimetry. Immunoturbidimetry is a measurement method that measures the process of an antigen-antibody reaction between an antigen or antibody in a biological sample and an antibody or antigen in a measurement reagent that specifically binds to the antigen or antibody. The analysis targets of the analytical device 1 include D-dimer, which is a fibrin degradation product, and FDP, which is a fibrin and / or fibrinogen degradation product.

[0016] When measuring D-dimer, a measurement reagent containing carrier particles to which an antibody that recognizes D-dimer is immobilized can be used. Examples of measurement reagents for D-dimer that can be used include RiasAuto (registered trademark) D-Dimer Neo manufactured by Sysmex Corporation, Nanopia (registered trademark) D-Dimer manufactured by Sekisui Medical Co., Ltd., and LPIA Genesis (registered trademark) D-Dimer manufactured by LSI Medience Corporation.

[0017] When measuring FDP, a measurement reagent containing carrier particles to which an antibody that recognizes FDP is fixed can be used. For example, as a measurement reagent for FDP, Rias Auto (registered trademark) P-FDP manufactured by Sysmex Corporation, Nanopia (registered trademark) P-FDP manufactured by Sekisui Medical Co., Ltd., and Lpia (registered trademark) FDP-P manufactured by LSI Medience Corporation can be used.

[0018] The analysis device 1 is not limited to measurement by immunoturbidimetry, and may analyze a biological sample by, for example, immunonephelometry. The analysis target of the analysis device 1 is not limited to D-dimer or FDP, and may be, for example, soluble fibrin monomer complex (hereinafter, sometimes abbreviated as "FMC"); von Willebrand factor antigen amount (hereinafter, sometimes abbreviated as "VWF:Ag"); immunoglobulin IgG, IgA, IgM; complement markers C3, C4; antistreptolysin-O; vancomycin; μ-albumin; prealbumin (P-Alb); lipoprotein (a); adenosine 5'-diphosphate (ADP); collagen; epinephrine; or CRP.

[0019] The biological sample measured by the analysis device 1 is plasma. The biological sample measured by the analysis device 1 is not limited to plasma, and may be a blood sample such as whole blood or serum, or may be urine, pleural fluid, ascites, lymphatic fluid, interstitial fluid, cerebrospinal fluid, etc.

[0020] 1-1. Hardware configuration of analytical device 1 The analytical device 1 is a device capable of preparing a measurement sample by adding to a biological sample a measurement reagent containing an antibody or antigen that specifically binds to an antigen or antibody in the biological sample (i.e., undergoes an antigen-antibody reaction), irradiating the prepared measurement sample with light, detecting the light transmitted through the measurement sample, analyzing the antigen or antibody contained in the biological sample based on the detected light, and detecting the occurrence of a non-specific reaction in the measurement sample. An example of the appearance of the analytical device 1 is shown in Fig. 1. The analytical device 1 includes a measurement device 2 that irradiates the measurement sample with light for a predetermined period of time and obtains detection information indicating the intensity of the light transmitted through the measurement sample, and an input / output device 4 that is capable of inputting and outputting data.

[0021] An example of the hardware configuration of the analysis device 1 is shown in Fig. 2. The measurement device 2 of the analysis device 1 includes a control device 70 and a detection device 80. The control device 70 includes an arithmetic processing device 201 that performs data processing, a storage device 203 used as a work area for data processing, a storage device 202 that stores information transferred to the storage device 203, a bus 209 that transmits data between each device, and interfaces (I / F) 206a, 206b that input and output data to and from external devices. The control device 70 is connected to an electronic medical record system 1000 via a network 99.

[0022] The arithmetic processing device 201 is a CPU (Central Processing Unit). The input / output device 4 is connected to an interface (I / F) 206a, and the network 99 is connected to an interface (I / F) 206b. The interface (I / F) 206a is a USB, and the interface (I / F) 206b is an Ethernet. The storage device 203 is a DRAM and an SRAM. The storage device 202 is a solid state drive. The input / output device 4 is a touch panel display, and receives input from an operator by touching the display, and displays information on the display. Signals within the measuring device are transmitted via a bus 209. The configuration of the control device 70 is not limited to the above configuration, and for example, IEEE1394 may be used as the interface (I / F) 206a or the interface (I / F) 206b, and a hard disk may be used as the storage device 202. The input / output device 4 may be a keyboard and / or a mouse as an input device, and a liquid crystal display or an organic EL display as an output device.

[0023] 3 is a diagram showing information stored in the storage device 202. The storage device 202 stores a measurement program 202a for the measurement device 200 to perform measurement processing and output detection information, a test substance analysis / non-specific reaction detection program 202b for generating and analyzing a data group from the detection information, a deep learning algorithm database DB1 for storing a deep learning algorithm used for the analysis, a calibration curve / threshold database DB2 for storing a calibration curve corresponding to each measurement item, a high concentration determination threshold corresponding to each measurement item, and a non-specific reaction determination threshold corresponding to each measurement item, a reanalysis item database DB3 for storing information on reanalysis items to be executed when reanalysis is required because the occurrence of a non-specific reaction is suspected, and a reanalysis result database DB4 in which a data group relating to the progress of the antigen-antibody reaction of each biological sample identified as having or not having a non-specific reaction by reanalysis is linked to the identification information of the biological sample (sample No., etc.) and the presence or absence of the occurrence of a non-specific reaction.

[0024] Returning to FIG. 2, the detection device 80 includes a sample preparation section 20, a light irradiation section 10, and a detection section 230.

[0025] FIG. 4 shows the configuration of the light irradiation unit 10. The light irradiation unit 10 includes five light sources 321, 322, 323, 324, and 325, five optical fiber units 330a, 330b, 330c, 330d, and 330e provided corresponding to the five light sources 321 to 325, and one holding member 340 for holding the light sources 321 to 325 and the incident ends 331 of the optical fiber units 330a to 330e. The light sources 321 to 325, the optical fiber units 330a to 330e, and the holding member 340 are housed in a metal housing 310. The light sources 321 to 325 are all composed of LEDs, and the first light source 321 generates light of 660 nm (first wavelength). The second light source 322 generates light of 405 nm (second wavelength). The third light source 323 generates light of 800 nm (third wavelength). The fourth light source 324 emits light of 340 nm (fourth wavelength). The fifth light source 325 emits light of 575 nm (fifth wavelength). As described later, the first light source 321 to the fifth light source 325 are controlled by the control device 70 so that one of the first light source 321 to the fifth light source 325 emits light in sequence for a predetermined period (for example, T seconds (T is less than 0.1)) until a predetermined time (for example, 200 seconds) has elapsed from the start of light irradiation.

[0026] Each of the optical fiber parts 330a to 330e is configured as a cable in which optical fiber strands each having one core are bundled. The optical fiber parts 330a to 330e are bundled together at the intermediate part 333, and the bundled optical fiber parts 330a to 330e are divided into two bundles, and the emission ends 332 of the bundles are held by the outlet 311 provided in the housing 310. The outlet 311 also holds the incident end 352 of the optical fiber 21 that connects the light irradiation part 10 and the detection part 230. Between the emission end 332 of the optical fiber parts 330a to 330e and the incident end 352 of the optical fiber 21, a homogenizing member 350 is provided for homogenizing the intensity distribution of the light emitted from the emission end 332. The homogenizing member 350 is a member that causes multiple reflections of the light incident from the incident surface 351 inside, and is, for example, a polygonal rod homogenizer.

[0027] 5(A) and (B) show the configuration of the detection unit 230. The detection unit 230 is provided in each optical fiber 21 connected to the light irradiation unit 10, but the connection configuration between the optical fiber 21 and the detection unit 230 and the hardware configuration of the detection unit 230 are the same for all the detection units 230, so one detection unit 230 will be described here. FIG. 5(A) is a diagram showing the detection unit 230 in a state where it does not hold a transparent container 15 for mixing a measurement reagent and a biological sample. The detection unit 230 has a hole 22b into which the other end 353 of the optical fiber 21, whose incident end 352 (FIG. 4) is held at the outlet 311 of the light irradiation unit 10, is inserted. Furthermore, the detection unit 230 has a communication hole 22c that communicates the hole 22b with the holding unit 22a for holding the container 15.

[0028] The diameter of the hole 22b is larger than the diameter of the communication hole 22c. A lens 22d that collects light from the optical fiber 21 is disposed at the end of the hole 22b. Furthermore, a hole 22f is formed in the inner wall surface of the holding part 22a at a position facing the communication hole 22c. A light receiving part 22g is disposed at the back of this hole 22f. The light receiving part 22g is, for example, a photodiode, and outputs an electrical signal according to the amount of light received. The light transmitted through the lens 22d is incident on the light receiving surface of the light receiving part 22g via the communication hole 22c, the holding part 22a, and the hole 22f. The optical fiber 21 is fixed by the leaf spring 22e with the end 353 inserted into the hole 22b.

[0029] FIG. 5(B) is a diagram showing the detection unit 230 in a state where the container 15 is held. When the container 15 is held by the holder 22a, the light collected by the lens 22d passes through the container 15 and the measurement sample contained in the container 15 and enters the light receiving unit 22g. When the antigen-antibody reaction in the measurement sample progresses, the turbidity of the measurement sample increases. Accordingly, the amount of light passing through the measurement sample (amount of transmitted light) decreases, and the level of the electrical signal output from the light receiving unit 22g decreases. The electrical signal output from the light receiving unit 22g is converted into a digital signal by the A / D converter 22h and transmitted to the control device 70 via the I / F 206 (see FIG. 2). The signal output from the light receiving unit 22g and converted by the A / D converter 22h is digital data reflecting the amount of transmitted light. The detection unit 230 may be configured such that the light receiving unit 22g is disposed on a line intersecting with a line connecting the lens 22d and the light receiving unit 22g, and detects light scattered by the measurement sample.

[0030] The sample preparation section 20 dispenses the biological sample and the measurement reagent into the container 15, and transfers the container 15 containing the measurement sample in which the biological sample and the measurement reagent are mixed to the holder 22a of the detection section 230.

[0031] The light irradiation section 10, the sample preparation section 20 and the detection section 230 may be, for example, devices described in U.S. Patent No. 10,048,249, which is incorporated herein by reference.

[0032] 1-2.Analysis (measurement, test substance analysis, non-specific reaction detection) 6 shows the flow of analysis processing (measurement, test substance analysis, and non-specific reaction detection processing) by the control device 70. In step S1, the control device 70 executes measurement processing based on the measurement program 202a. Next, in step S2, the control device 70 executes test substance analysis and non-specific reaction detection processing based on the test substance analysis and non-specific reaction detection program 202b. The measurement processing is started when the control device 70 accepts a processing start command input by the operator from the input / output device 4.

[0033] 1-3. Processing of measurement programs 7 shows the flow of the measurement process executed by the control device 70 based on the measurement program 202a. In step S11, the control device 70 acquires information on the measurement items ordered for each biological sample based on the information input by the operator from the input / output device 4. The control device 70 may acquire the information on the measurement items via the network 99, for example, from an electronic medical record system 1000 of a medical facility. The biological sample and the information on the measurement items can be linked together by an identifier issued when the test is requested, for example.

[0034] In step S12, the control device 70 controls the sample preparation unit 20 to dispense a biological sample into the container 15. In step S13, the control device 70 controls the sample preparation unit 20 to dispense a measurement reagent corresponding to the measurement item acquired in step S11 into the container 15 and prepare a measurement sample.

[0035] In step S14, the control device 70 controls the light irradiation unit 10 to start irradiating the container 15 in which the measurement sample was prepared in step S13 with light. Specifically, one of the first light source 321 to the fifth light source 325 sequentially emits light for a predetermined period (for example, T seconds (T is less than 0.1)), and this is repeated until a predetermined time (for example, 200 seconds) has elapsed since the start of light irradiation. The detection unit 230 outputs an electric signal (digital data) corresponding to the intensity of the light received through the container 15 (i.e., the transmitted light intensity) one by one during the predetermined period (for example, T seconds). The output digital data indicates the intensity of the light transmitted through the measurement sample. A set of the output digital data is transmitted to the control device 70 as detection information. As a result, each data of each detection information reflects the transmitted light intensity at each measurement time point from the start of light irradiation until the predetermined time (for example, 200 seconds) has elapsed.

[0036] 1-4. Processing of test substance analysis and non-specific reaction detection program 202b i. Analysis of test substances and detection of non-specific reactions FIG. 8 shows the flow of the test substance analysis / non-specific reaction detection process executed by the control device 70 based on the test substance analysis / non-specific reaction detection program 202b.

[0037] In step S21, the control device 70 receives the detection information from the detection unit 230, and generates a data group related to the progress of the antigen-antibody reaction from the received detection information. Specifically, in step S21, the control device 70 receives the detection information from the detection unit 230, classifies the received detection information into the first light source 321 to the fifth light source 325 (i.e., into the wavelengths of the light emitted from the light sources), and stores it in the storage device 202. The control device 70 also converts each digital data of the stored detection information into absorbance, arranges the converted absorbances in chronological order, and stores them in the storage device 202 as a data group related to the progress of the antigen-antibody reaction. Since each digital data of the detection information indicates the transmitted light intensity, for example, the detection information can be converted into absorbance by expressing the reciprocal of each digital data of the detection information using common logarithms. FIG. 9A shows an example of a data group related to the progress of the antigen-antibody reaction. 9A, data groups D11 to D15 include absorbance values ​​generated for each wavelength of light emitted from the light source and arranged in the order of output from the detection unit 230. Each data item in data group D11 reflects the absorbance value at each measurement time point from the start of light irradiation until a predetermined time (e.g., 200 seconds) has elapsed. Similarly, each data item in data groups D12 to D15 reflects the absorbance value at each measurement time point from the start of light irradiation until a predetermined time (e.g., 200 seconds) has elapsed.

[0038] 9B is a graph in which the vertical axis (Y axis) indicates absorbance and the horizontal axis (X axis) indicates the elapsed time (seconds: sec) from the start of output of detection information, and is a diagram in which a data group D11 corresponding to the wavelength (first wavelength) emitted from the first light source 321 is plotted from among the data groups illustrated in FIG. 9A. When an antigen or antibody that is a test substance (hereinafter, the antigen or antibody to be analyzed is referred to as a "test substance") is present in a biological sample, an antigen-antibody reaction between the antigen or antibody and the antibody or antigen fixed to the carrier particle proceeds, causing the carrier particle to aggregate, and the turbidity of the measurement sample in the container 15 increases. Therefore, when a test substance is present in a biological sample, the absorbance increases according to its concentration.

[0039] The control device 70 analyzes the test substance in step S22 shown in FIG. 8. That is, the control device 70 acquires the concentration of the test substance in the biological sample based on a data group (for example, data group D11) corresponding to a wavelength according to the measurement item and a calibration curve stored in the calibration curve / threshold database DB2. Specifically, the control device 70 calculates the amount of change or rate of change in absorbance from the data group. The amount of change or rate of change in absorbance can be calculated from the data group by the Rate method or the VLin method. The Rate method is a method in which a start point and an end point are set in advance, and the amount of change in absorbance per minute is calculated by linear regression for the data group included between the two time points. The VLin method is a method in which a start point and an end point at which the amount of change in absorbance is maximum and the linear approximation is optimal are set for the data group, and the amount of change in absorbance per minute is calculated by linear regression for the data group included between the two time points. The control device 70 determines whether the amount of change or rate of change is calculated by the Rate method or the VLin method depending on the measurement item. Before calculating the amount or rate of change in absorbance, the data group may be subjected to pre-processing such as smoothing, sharpening, or caving.

[0040] The calibration curve is obtained by diluting a standard substance with a known concentration of the test substance at different dilution ratios, measuring each diluted standard substance to obtain the amount or rate of change in absorbance, plotting the amount or rate of change in absorbance on the vertical axis and the concentration of the test substance on the horizontal axis, and performing linear regression. The control device 70 applies the amount or rate of change in absorbance obtained from the data group to the calibration curve to obtain the concentration of the test substance in the biological sample.

[0041] In step S23 shown in Fig. 8, the control device 70 judges whether or not the concentration acquired in step S22 exceeds the threshold value for determining high concentration stored in the calibration curve / threshold value database DB2. If the concentration of the test substance acquired in step S22 is equal to or lower than the threshold value (if step S23 is "No"), the process proceeds to step S27, the concentration of the test substance acquired in step S22 is output to the input / output device 4, and the process ends. If the concentration acquired in step S22 exceeds the threshold value (if step S23 is "Yes"), the control device 70 advances the process to step S24. The control device 70 detects the occurrence of a non-specific reaction in the measurement sample by the processes of steps S24 and S25.

[0042] Here, a method for generating the deep learning algorithm 60 shown in FIG. 12 will be described. The deep learning algorithm 60 is not limited as long as it is an algorithm having a neural network structure. The deep learning algorithm 60 may include a convolutional neural network, a fully connected neural network, and a combination thereof. The deep learning algorithm 60 is generated by training the pre-trained deep learning algorithm 50 (FIG. 11). FIG. 11 shows an overview of the training of the deep learning algorithm. The training of the deep learning algorithm is performed by a training device 5 described later.

[0043] For training data for training the deep learning algorithm 50, a data group related to the progress of an antigen-antibody reaction obtained from a biological sample in which it is known whether or not a non-specific reaction has occurred is used as the first training data. The first training data can be generated according to the method described in step S21 shown in FIG. 8. As shown in FIG. 11, the first training data includes a first data group D1 corresponding to a first wavelength, a second data group D2 corresponding to a second wavelength, a third data group D3 corresponding to a third wavelength, a fourth data group D4 corresponding to a fourth wavelength, and a fifth data group D5 corresponding to a fifth wavelength. The data group D1 includes a plurality of data d11, d12, d13, etc. indicating absorbance. Similarly, the data groups D2 to D5 include a plurality of data indicating absorbance. The first training data is input to the input layer 50a of the neural network 50 shown in FIG. 11. In addition, a label indicating the occurrence or non-occurrence of a non-specific reaction corresponding to the input first training data (in the example of FIG. 11, "occurrence or non-occurrence of a non-specific reaction") is input as second training data to the output layer 50b. With these inputs, the first training data and the second training data are linked for each biological sample, and the weights of each layer of the intermediate layer 50c of the neural network 50 are calculated. By executing the input of the first training data and the second training data and the calculation of the weights of each layer of the intermediate layer 50c for a large number of biological samples, a trained deep learning algorithm 60 shown in FIG. 12 is generated. A softmax function is applied to the output layer 60b of the deep learning algorithm 60. As a result, a numerical value indicating the probability that a non-specific reaction has occurred is output from the deep learning algorithm 60.

[0044] 12, in step S24 shown in FIG. 8, the control device 70 inputs the data group D11 corresponding to the wavelength (first wavelength) emitted from the first light source 321, the data group D12 corresponding to the wavelength (second wavelength) emitted from the second light source 322, the data group D13 corresponding to the wavelength (third wavelength) emitted from the third light source 323, the data group D14 corresponding to the wavelength (fourth wavelength) emitted from the fourth light source 324, and the data group D15 corresponding to the wavelength (fifth wavelength) emitted from the fifth light source 325, which were acquired in step S21, to the input layer 60a of the deep learning algorithm 60 stored in the deep learning algorithm database DB1. The deep learning algorithm 60 processes the input data groups D11 to D15 in the intermediate layer 60c, and outputs the probability that a non-specific reaction has occurred (98.4% in the example of FIG. 12) from the output layer 60b. In addition, the numerical value indicating the probability does not necessarily have to be expressed as a percentage, but may be expressed as a relative value. In this specification, information reflecting a numerical value indicating the probability, such as a percentage or a relative value, is referred to as information indicating the probability.

[0045] "Probability of occurrence of a non-specific reaction" refers to the probability that a non-specific reaction actually occurred in the measurement sample to be analyzed. For example, if the probability of occurrence of a non-specific reaction is 80%, this indicates that the probability that a non-specific reaction actually occurred in the measurement sample is 80%. In other words, a probability of occurrence of a non-specific reaction of 80% means that, if there are 100 measurement samples in which the probability of occurrence of a non-specific reaction is 80%, 80 of the measurement samples actually experienced a non-specific reaction, and the remaining 20 samples did not experience a non-specific reaction.

[0046] As shown in Fig. 11 and Fig. 12, it is preferable that the configuration of the data group in the first training data is the same as the configuration of the data group acquired in step S21, but the present invention is not limited to this. For example, the first wavelength to the fifth wavelength in the first training data are the same as the first wavelength to the fifth wavelength in the data group acquired in step S21, but only some of the wavelengths may be used in the data group in the first training data and / or the data group acquired in step S21. Also, the number of data in each data group in the first training data is the same as the number of data in each data group in the data group acquired in step S21, but the number of data in each data group in the first training data may be different from the number of data in each data group in the data group acquired in step S21.

[0047] Furthermore, the composition of the data group in the first training data and the data group acquired in step S21 do not necessarily have to be acquired for multiple wavelengths; they may be acquired for one wavelength, and the acquired data group may be input to deep learning algorithm 50 or deep learning algorithm 60.

[0048] The deep learning algorithm 60 may be generated so as to be commonly used for a plurality of measurement items, or may be generated for each measurement item. In the case where the deep learning algorithm 60 is generated for each measurement item, in step S24 shown in Fig. 8, the control device 70 selects the deep learning algorithm 60 to be input from among a plurality of deep learning algorithms according to the measurement item acquired in step S11.

[0049] In step S25 shown in FIG. 8, the control device 70 judges whether the probability of occurrence of a non-specific reaction output from the deep learning algorithm 60 in step S24 exceeds the threshold for determining a non-specific reaction stored in the calibration curve / threshold database DB2. If the probability of occurrence of a non-specific reaction output in step S24 is equal to or lower than the threshold (if step S25 is "No"), the process proceeds to step S27, the concentration of the test substance acquired in step S22 is output to the input / output device 4, and the process ends. If the probability of occurrence of a non-specific reaction output in step S24 exceeds the threshold (if step S25 is "Yes"), the control device 70 advances the process to step S26. In step S26, the control device 70 outputs the concentration acquired in step S22 to the input / output device 4. In addition, in step S26, the control device 70 generates information regarding the occurrence of a non-specific reaction and outputs it to the input / output device 4.

[0050] The deep learning algorithm 60 may be generated so as to output the probability that a non-specific reaction did not occur from the output layer 60b. In this case, the control device 70 determines whether the probability that a non-specific reaction did not occur output from the deep learning algorithm 60 in step S24 is smaller than a threshold value for determining a non-specific reaction stored in the calibration curve / threshold database DB2. If the probability that a non-specific reaction did not occur output in step S24 is smaller than the threshold value, the control device 70 proceeds to step S27, and if the probability is equal to or greater than the threshold value, the control device 70 proceeds to step S26.

[0051] FIG. 10 shows an example of a message box MB1 that outputs information about the occurrence of a non-specific reaction, which is generated in step S26 and output to the input / output device 4. The message box MB1 includes a measurement item information area MB11 that indicates measurement item information including the measurement date, measurement time, biological sample identifier (biological sample ID), and measurement result, and a message area MB12 that indicates information about the occurrence of a non-specific reaction related to the measurement result. The measurement result in the measurement item information area MB11 displays the concentration of the test substance of the measurement target obtained in step S22 of FIG. 8. In the example of FIG. 14, the D-dimer concentration (DD C) and the FDP concentration (FDP C) are displayed. The message area MB12 displays the measurement item (DD C), an output label (+++) that qualitatively indicates a suspected non-specific reaction, and information indicating the probability (98.4%). The information indicating the probability is the probability that a non-specific reaction occurred output from the deep learning algorithm 60 in step S24. As output labels, "+", "++", "+++", etc. are used depending on the magnitude of the difference between the probability and the threshold for determining non-specific reactions. Instead of "+", "++", and "+++", "low", "medium", "high", etc. may be used.

[0052] FIG. 13 shows a message box MB1' as a modified example of the message box MB1. In the message box MB1', the measurement item information area MB11' displays the previous measurement results of the same subject for the measurement item in which the probability of non-specific reaction occurrence exceeds the threshold. The previous measurement results of the same subject can be acquired from the electronic medical record system 1000 using the biological sample ID as a key. In addition, in the measurement item information area MB11', the history of past measurement results for the measurement item in which the probability of non-specific reaction occurrence exceeds the threshold may be displayed, or the previous measurement results of a measurement item different from the measurement item may be displayed. In addition, in the message box MB1', if there is a discrepancy between the concentration of the test substance acquired in step S22 and the concentration of another test substance or the clinical information of the subject, a message is displayed in the message area MB12' indicating that the concentration displayed in the measurement item information area MB11' is suspected to be a false high value due to a non-specific reaction. This allows the operator to more clearly grasp the need for reanalysis. In addition, in the message box MB1', the clinical information area MB13 displays the clinical information of the same subject. The clinical information can be obtained from the electronic medical record system 1000 using the biological sample ID as a key. The clinical information is information obtained by examining the subject, and includes information such as CT examination results, MRI examination results, ultrasound examination results, angiography examination results, and medication history. The treatment history includes information on medication history, such as whether anticoagulants are used.

[0053] In addition, in the measurement item information region MB11 and / or the measurement item information region MB11', the measurement results for the measurement items for which the probability of non-specific reactions occurring exceeds a threshold may be hidden. In addition, in the measurement item information region MB11 and / or the measurement item information region MB11', the measurement results for the measurement items for which the probability of non-specific reactions occurring exceeds a threshold may be displayed together with a label indicating that the value is a reference value or a value that requires confirmation. The label may be a text word such as "reference value" or "check required", or may be a symbol such as "*" or "!".

[0054] 8 may be executed even if the concentration of the test substance acquired in step S22 does not exceed the threshold value ("No" in step S23). By not executing the process of step S24 when the concentration of the test substance does not exceed the threshold value, it is possible to prevent a decrease in the processing speed of the control device 70 due to an increase in the amount of calculations. On the other hand, by executing the process of step S24 even when the concentration of the test substance does not exceed the threshold value, it is possible to more reliably detect the occurrence of a non-specific reaction.

[0055] ii. Reanalysis processing The analysis device 1 may be configured to be able to execute the reanalysis process shown below. When the analysis device 1 is configured to be able to execute the reanalysis process, the control device 70 executes the reanalysis process shown in Fig. 14 after step S26 shown in Fig. 8. In step S28, the control device 70 outputs a dialog box MB3 for prompting the operator to input a reanalysis selection to the input / output device 4. As shown in Fig. 15, the dialog box MB3 includes a message area MB31 including a message asking the operator whether or not to execute reanalysis, and a selection area MB33 for prompting the operator to select a reanalysis item. Selection area MB33 includes selection area MB331 for selecting reanalyzing the same measurement item in which a non-specific reaction may have occurred (D-dimer in the example of Figure 15), selection area MB332 for selecting analysis of a measurement item other than the measurement item in which a non-specific reaction may have occurred (D-dimer in the example of Figure 15) (FDP in the example of Figure 15), selection area MB333 for selecting diluting and reanalyzing the biological sample for the same measurement item in which a non-specific reaction may have occurred (D-dimer in the example of Figure 15), and selection area MB334 for selecting not to perform a reanalysis.

[0056] The measurement items displayed in the selection area MB332 are stored in the reanalysis item database DB3. As shown in FIG. 16, the reanalysis item database DB3 stores the measurement items of the initial measurement in association with the measurement items of the reanalysis. It is preferable that the measurement items of the reanalysis are items related to the measurement items of the initial measurement. For example, when the measurement items of the initial measurement are the measurement items of the thrombus and fibrinolysis system, it is preferable that the measurement items of the reanalysis are the measurement items of the thrombus and fibrinolysis system different from the measurement items of the initial measurement. In FIG. 16, when the measurement item of the initial measurement is D-dimer, the measurement item of the reanalysis is FDP. When the measurement item of the initial measurement is FDP, the measurement item of the reanalysis is D-dimer. When the measurement item of the initial measurement is FMC, the measurement items of the reanalysis are soluble fibrin (SF), thrombin-antithrombin III complex (TAT), and prothrombin fragment 1+2 (F1+2). When the measurement item of the initial measurement is VWF:Ag, the measurement items of the reanalysis are von Willebrand factor activity ristocetin cofactor (VWF:RCo), VWF binding ability to recombinant GPIb in the presence of ristocetin (VWF:GPIbR), VWF binding ability to recombinant GPIb mutant (gain-of-function) (VWF:GPIbM), and monoclonal antibody binding ability to the A1 domain of VWF (VWF:Ab). When there are multiple measurement items of the reanalysis corresponding to the measurement item of the initial measurement, the control device 70 displays one selection area MB332 for each measurement item on the input / output device 4.

[0057] The control device 70 judges whether or not the selection region MB334 has been selected by the operator in step S29 shown in FIG. 14. If the selection region MB334 has been selected (if "Yes" in step S29), the control device 70 returns the process to the test substance analysis and non-specific reaction detection process shown in FIG. 8. If the selection region MB334 has not been selected (if "No" in step S29), the control device 70 advances the process to step S30. The control device 70 judges whether or not the selection region MB331 has been selected in step S30. If the selection region MB331 has been selected (if "Yes" in step S30), the control device 70 advances the process to step S12 shown in FIG. 7, and executes the measurement process for the measurement item acquired in step S11. If the selection region MB331 has not been selected (if "No" in step S30), the control device 70 advances the process to step S31. The control device 70 judges whether or not the selection region MB332 has been selected in step S31. If the selection area MB332 is selected (if "Yes" in step S31), the control device 70 advances the process to step S11 shown in FIG. 7, acquires the measurement item designated in the selection area MB332 (FDP in the example of FIG. 15), and executes the measurement process for the measurement item. If the selection area MB332 is not selected (if "No" in step S31), the control device 70 advances the process to step S32. In S32, the control device 70 determines whether or not the selection area MB333 is selected. If the selection area MB333 is selected (if "Yes" in step S32), the control device 70 advances the process to step S12', and controls the sample preparation unit 20 to dispense and dilute the biological sample into the container 15. Next, the control device 70 advances the process to step S13 shown in FIG. 7, and executes the measurement process for the measurement item acquired in step S11. If the selection region MB333 is not selected ("No" in step S32), the control device 70 returns the process to the test substance analysis / non-specific reaction detection process shown in Fig. 8. Note that, in the test substance analysis / non-specific reaction detection process shown in Fig. 8, if the process proceeds to step S27 ("No" in step S23 or step S25), the control device 70 does not execute reanalysis processing.

[0058] The selection regions MB331 to MB334 do not need to be selected by an operator, and may be selected by the control device 70 based on initial settings or settings made by the operator before the start of measurement. The control device 70 may select the selection regions MB331 to MB334 based on the magnitude of the probability that a non-specific reaction has occurred, which is output in step 24 shown in FIG.

[0059] For reanalysis of the same measurement items as those in the initial measurement, the measurement reagent in the same vial as that used in the initial measurement may be used, or a measurement reagent in a different vial from that used in the initial measurement may be used. When using a measurement reagent in a different vial from that used in the initial measurement, the measurement reagent in a different vial from that used in the initial measurement may be used, or a measurement reagent in a different vial from that used in the initial measurement may be used. When using a measurement reagent in a different vial from that used in the initial measurement, the measurement reagent for the same measurement item developed and manufactured by a company different from that used in the initial measurement may be used.

[0060] FIG. 17 shows an example of a message box MB1″ displayed on the input / output device 4 in step S27 shown in FIG. 8 during reanalysis processing. In the message box MB1″, the concentration obtained in the reanalysis, the concentration obtained in the initial measurement, and a message indicating that the concentration obtained in the initial measurement is a reference value are displayed in the measurement item information area MB11″. A message indicating that the displayed information relates to the result of the reanalysis is displayed in the message area MB12″.

[0061] When the operator determines through the reanalysis that a non-specific reaction has occurred or has not occurred, the operator inputs information on the presence or absence of the non-specific reaction from the input / output device 4. The control device 70 may store the input information on the presence or absence of the non-specific reaction in the reanalysis result database DB4 of the storage device 202 in association with the identification information of the biological sample and a data group on the progress of the antigen-antibody reaction acquired for the biological sample in step S21 shown in FIG. 8. FIG. 18 shows an example of the reanalysis result database DB4. The information stored in the reanalysis result database DB4 can be used as the original data of the first training data and the second training data for further training the deep learning algorithm 60, and the analysis accuracy of the deep learning algorithm 60 can be further improved.

[0062] 2.Training equipment FIG. 19(A) shows an example of the appearance of a training device 5 (hereinafter simply referred to as "training device 5") that trains a deep learning algorithm 50. The training device 5 is communicatively connected to an input device 511 and an output device 512. A general-purpose computer can be used as the training device 5.

[0063] FIG. 19(B) shows an example of the hardware configuration of the training device 5. The training device 5 includes a control device 500. The control device 500 includes an arithmetic processing device 501, a storage device 502, a storage device 503, an interface (I / F) 506, and a bus 510. The arithmetic processing device 501 is, for example, a CPU (Central Processing Unit). The input device 511, the output device 512, and the network 99 are connected to the interface (I / F) 506. The interface (I / F) 506 is, for example, a USB, IEEE1394, or Ethernet. The storage device 502 is, for example, a solid state drive or a hard disk drive. The storage device 502 stores a training program 502b that executes training of the deep learning algorithm 50 described above. The storage device 503 is, for example, a DRAM or an SRAM. The input device 511 is, for example, a keyboard or a mouse, and the output device 512 is, for example, a liquid crystal display or an organic EL display. The transmission of signals within the training device 5 is carried out via a bus 510 .

[0064] The training device 5 is connected to the analysis device 1 via a network 99. This makes it possible to transmit the trained deep learning algorithm 60 to the analysis device 1 via the network 99, and to receive information stored in the reanalysis result database DB4 of the analysis device 1 from the analysis device 1 to further train the deep learning algorithm 60.

[0065] 3. Variations Information on the occurrence of non-specific reactions in antigen-antibody reactions can be obtained by the following method in addition to or instead of the method of obtaining information using the deep learning algorithm 60 as described above. That is, in the method of this modification, as shown in FIG. 20, the control device 70 generates a distribution of calculated values ​​(S values) obtained from a predetermined function for a data group related to the progress of antigen-antibody reactions obtained from a plurality of biological samples in which non-specific reactions do not occur, in step S81. In step S82, the control device 70 obtains a calculated value (S value) obtained from the function based on a predetermined function for a data group related to the progress of antigen-antibody reactions obtained from a biological sample to be analyzed. In step S83, the control device 70 obtains information on the occurrence of non-specific reactions based on the distribution generated in step S81 and the calculated value obtained for the biological sample to be analyzed in step S82.

[0066] Specifically, in step S81, the control device 70 executes the processes of step S1 shown in Fig. 6 and step S21 shown in Fig. 8 for a biological sample in which no antigen-antibody reaction occurs, and acquires a data group related to the progress of the antigen-antibody reaction. Based on the acquired data group, the control device 70 calculates an S value shown in the following formula 1.

[0067]

number

[0068]

number

[0069] α1 is determined by fitting equation (2) to the acquired data set regarding the progress of the antigen-antibody reaction. The α1 in equation (2) determined in this manner is substituted into equation (1). The maximum and minimum absorbance values ​​included in equation (1) are determined from the acquired data set regarding the progress of the antigen-antibody reaction. The S value is calculated from equation (1) based on these values.

[0070] The control device 70 similarly calculates the S values ​​for many other biological samples in which no antigen-antibody reaction occurs, and creates a histogram in which the vertical axis represents the number of biological samples and the horizontal axis represents the S value, as shown in Fig. 21. Then, the most frequent value in the distribution of S values ​​(S value = 2.25 in the example of Fig. 21) is taken as a 0% possibility that a nonspecific reaction has occurred, and the maximum value of the S value (S value = 3.9 in the example of Fig. 21) is taken as a 100% possibility that a nonspecific reaction has occurred.

[0071] In step S82, the control device 70 calculates an S value for a group of data relating to the progress of an antigen-antibody reaction obtained from the biological sample to be analyzed, based on the function of Equation 1. In step S83, the control device 70 converts the S value acquired in step S82 into the probability that a non-specific reaction has occurred in the biological sample to be analyzed. The S value can be converted into the probability that a non-specific reaction has occurred, for example, by applying the S value to a linear regression equation generated from the mode value / probability (=0%) in the distribution of S values ​​and the maximum value / probability (=100%) in the distribution of S values.

[0072] 4.Verification of effectiveness In the measurement of D-dimer, the occurrence of non-specific reactions was confirmed by an immunoglobulin absorption test. The prediction accuracy of this analysis method was evaluated using 39 biological samples in which the occurrence of non-specific reactions was confirmed by the immunoglobulin absorption test and 77 biological samples in which the occurrence of non-specific reactions was not confirmed. A set of data groups consisting of a data group D11 corresponding to the first wavelength, a data group D12 corresponding to the second wavelength, a data group D13 corresponding to the third wavelength, and a data group D15 corresponding to the fifth wavelength was input to the deep learning algorithm. The results are shown in Figures 22 and 23. Figure 22 shows a histogram in which the horizontal axis represents the probability of occurrence of a non-specific reaction (0.0 to 1.0) output from the deep learning algorithm, the horizontal axis is divided into 0.05 intervals, and the vertical axis represents the frequency of biological samples belonging to each divided range. The height of the gray bar indicates the frequency of biological samples in which the occurrence of non-specific reactions was confirmed by the immunoglobulin absorption test, and the height of the white bar indicates the frequency of biological samples in which the occurrence of non-specific reactions was not confirmed by the immunoglobulin absorption test. FIG. 23 shows the ROC curve obtained as a result of performing the ROC analysis on the biological samples shown in FIG.

[0073] The result of ROC analysis was AUC=0.899. For example, when the threshold for determining non-specific reactions was set to 0.65, the occurrence of non-specific reactions could be predicted with a sensitivity of 62% and a specificity of 96%. These results demonstrated that the occurrence of non-specific reactions can be detected by the present invention. [Explanation of symbols]

[0074] 1 Analyzer 70 Control device 230 Detection Unit

Claims

1. 1. A method for detecting the occurrence of a non-specific reaction in an analysis of an antigen or an antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antibody, the method comprising the steps of: the biological sample is a blood sample, A measurement sample, which is a mixture of the blood sample and the measurement reagent contained in a transparent container, is irradiated with irradiation light during an antigen-antibody reaction between the antigen or the antibody contained in the blood sample and the antibody or the antigen contained in the measurement reagent; receiving detection light through the measurement sample irradiated with the irradiation light; generating a data set indicative of a progress of the antigen-antibody reaction based on detection information reflecting an intensity of the received detection light; Inputting the data set into a deep learning algorithm; Generate information regarding the occurrence of non-specific reactions based on the results output from the deep learning algorithm; the detection information reflects the intensity of the detection light detected through the measurement sample at a plurality of time points within a predetermined period from the start of irradiation of the irradiation light until a predetermined time has elapsed; The data group is a data group in which digital data indicating the intensities of the detected light at the multiple time points are arranged in a time series. A method for detecting the occurrence of said non-specific reaction.

2. Each data included in the data group indicates an intensity of light transmitted through the measurement sample, an intensity of light absorbed by the measurement sample, or an intensity of light scattered by the measurement sample. The detection method according to claim 1 .

3. the detection information indicates an intensity of the detected light detected through the measurement sample that is sequentially irradiated with a plurality of types of light having different wavelengths; In generating the data group, a plurality of the data groups corresponding to the respective lights are generated from the detection information; In inputting the data groups into the deep learning algorithm, a plurality of the data groups are input into the deep learning algorithm; The detection method according to claim 1 or 2.

4. 1. An analytical method for analyzing an antigen or an antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antigen or the antibody, comprising: preparing a measurement sample containing the biological sample and the measurement reagent; Analyzing the antigen or the antibody contained in the biological sample; Executing the method for detecting the occurrence of a non-specific reaction according to claim 1 . The analytical method.

5. The method further includes irradiating the measurement sample with light for a predetermined period of time and outputting detection information indicating an intensity of the detected light detected through the measurement sample, In the analysis of the antigen or the antibody contained in the biological sample, Analyzing the antigen or the antibody based on the detection information. The analytical method according to claim 4.

6. In generating the data group, generating the data set from the detection information; In the analysis of the antigen or the antibody contained in the biological sample, Analyzing the antigen or the antibody based on the data group generated from the detection information. The analytical method according to claim 5.

7. executing the method for detecting the occurrence of a non-specific reaction when a first condition is satisfied, that is, a concentration of the antigen or the antibody as a result of analyzing the antigen or the antibody contained in the biological sample exceeds a first threshold value; The method of analysis according to any one of claims 4 to 6.

8. When a second condition is satisfied in which the probability of occurrence of the non-specific reaction, which is the result output from the deep learning algorithm, exceeds a second threshold, information regarding the occurrence of the non-specific reaction is output. The method of analysis according to any one of claims 4 to 7.

9. In the output of information regarding the occurrence of the non-specific reaction, Further outputting information regarding the results of analysis of a past biological sample of the subject from whom the biological sample was collected, and / or clinical information obtained by examining the subject. The analytical method according to claim 8.

10. and outputting a message suggesting reanalysis of the biological sample when the result output from the deep learning algorithm satisfies the second condition. The analytical method according to claim 8 or 9.

11. When the result output from the deep learning algorithm satisfies the second condition, an instruction to perform a reanalysis of the biological sample is accepted; and, upon receiving the execution instruction, executing a reanalysis of the biological sample. The analytical method according to claim 10.

12. In the reception of the execution instruction, an instruction to perform an analysis of an antigen or antibody different from the antigen or antibody contained in the biological sample is received. The analytical method according to claim 11.

13. In the receiving of the execution instruction, an instruction is received to dilute the biological sample at a dilution ratio higher than a dilution ratio in the measurement sample and analyze the same type of antigen or antibody as the antigen or antibody contained in the biological sample. The analytical method according to claim 11 or 12.

14. An analyzer for analyzing an antigen or an antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antigen, a sample preparation unit that prepares a measurement sample containing the biological sample and the measurement reagent; a light irradiation unit that irradiates the measurement sample with light; a detection unit that detects the light irradiated from the light irradiation unit through the measurement sample and outputs detection information corresponding to the detected detection light; A control device, the biological sample is a blood sample, The light irradiation unit irradiates the measurement sample, which is a mixture of the blood sample and the measurement reagent contained in a transparent container, with irradiation light during an antigen-antibody reaction between the antigen or the antibody contained in the blood sample and the antibody or the antigen contained in the measurement reagent, The control device includes: Analyzing the antigen or the antibody contained in the biological sample; Detecting the occurrence of a non-specific reaction in the measurement sample; The detection of the occurrence of the non-specific reaction includes: generating a data set indicative of the progress of the antigen-antibody reaction based on the detection information; Inputting the data set into a deep learning algorithm; generating information regarding the occurrence of non-specific reactions based on the results output from the deep learning algorithm; the detection information reflects the intensity of the detection light detected through the measurement sample at a plurality of time points within a predetermined period from the start of irradiation of the irradiation light until a predetermined time has elapsed; The data group is a data group in which digital data indicating the intensities of the detected light at the multiple time points are arranged in a time series. The analytical device.

15. A program for detecting the occurrence of a non-specific reaction in an analysis of an antigen or an antibody contained in a biological sample using a measurement reagent containing the antigen or an antibody that undergoes an antigen-antibody reaction with the antibody, the program being executed by a computer, the biological sample is a blood sample, A measurement sample, which is a mixture of the blood sample and the measurement reagent contained in a transparent container, is irradiated with irradiation light during an antigen-antibody reaction between the antigen or the antibody contained in the blood sample and the antibody or the antigen contained in the measurement reagent, A detection light is received through the measurement sample irradiated with the irradiation light, generating a data group indicative of a progress of the antigen-antibody reaction based on detection information reflecting an intensity of the received detection light; inputting the data set into a deep learning algorithm; generating information regarding the occurrence of non-specific reactions based on the results output from the deep learning algorithm; Run the following on your computer: the detection information reflects the intensity of the detection light detected through the measurement sample at a plurality of time points within a predetermined period from the start of irradiation of the irradiation light until a predetermined time has elapsed; The data group is a data group in which digital data indicating the intensities of the detected light at the multiple time points are arranged in a time series. A program for detecting the occurrence of said non-specific reactions.

Images