Method for estimating cause of extension of blood coagulation time, and information processing apparatus

Abstract

To estimate a cause of extension of coagulation time of the blood of a specimen to be inspected.SOLUTION: A method for estimating a cause of extension of blood coagulation time includes: acquiring a coagulation waveform indicating a change with time of a light amount due to a coagulation reaction of reaction liquid obtained by mixing a specimen to be inspected and an agent (S301b); acquiring a first waveform through before-and-after difference processing on the coagulation waveform (S302b); acquiring first and second fitting waveforms through fitting processing on the coagulation waveform and the first waveform (S303b, S304b); acquiring a third waveform through the before-and-after difference processing on the second fitting waveform (S305b); acquiring first to third normalized waveforms by normalizing light amount axes and time axes of the first and second fitting waveforms and the third waveform (S306b); extracting a feature quantity from each of the first to third normalized waveforms (S307b); and estimating the cause of extension of the coagulation time of the blood of the specimen to be inspected on the basis of a known feature quantity and the extracted feature quantity.SELECTED DRAWING: Figure 3B

Description

[Technical field]

[0001] The present disclosure relates to a method for estimating a cause of prolongation of blood coagulation time, and an information processing device. [Background technology]

[0002] Blood coagulation tests are performed for the purpose of grasping the pathology of the blood coagulation and fibrinolysis system, diagnosing DIC (disseminated intravascular coagulation), confirming the effectiveness of thrombus treatment, diagnosing hemophilia, etc. In particular, blood coagulation time measurement involves mixing a sample with a reagent and measuring the time it takes for a fibrin clot to form (hereinafter referred to as blood coagulation time). If there is a congenital or acquired abnormality in blood coagulation ability, the blood coagulation time will be prolonged. Causes of blood coagulation time prolongation include a deficiency of blood coagulation factors (deficiency type) and inhibition of blood coagulation reaction by antibodies against blood coagulation factors or components (e.g., phospholipids) in the blood coagulation time measurement reagent (inhibitor type). Since the treatment policy differs depending on the cause of prolongation of blood coagulation time, it is necessary to identify the cause of prolongation, but the cause of prolongation cannot be identified by measuring the blood coagulation time alone.

[0003] A typical method for identifying the cause of prolonged blood clotting time is the cross-mixing test. In the cross-mixing test, the blood clotting time of multiple plasma samples is measured by mixing a test specimen and a normal specimen in different ratios (e.g., 10:0, 9:1, 8:2, 5:5, 2:8, 1:9, 0:10). The measurements are performed immediately after sample preparation (immediate type) and after 2 hours of incubation (delayed type). The blood clotting time obtained is plotted on the vertical axis against the test specimen mixing ratio on the horizontal axis, and the results are connected by a line to create a graph, and the cause of prolongation is determined from the shape of the graph for the immediate and delayed types. This determination is qualitative, and requires the person making the determination to have a high level of skill and experience. In addition, the complicated process of preparing the specimen and incubating it during the test places a burden on the examiner. In addition, the need for additional blood sampling for this test places a burden on the patient.

[0004] As an evaluation method that reduces the burden on the examiner and the patient and provides more quantitative results, a method based on clot waveform analysis has been proposed. The target clot waveform is a waveform obtained by optical measurement, and is a record of the change in turbidity over time that accompanies the formation of a fibrin clot.

[0005] Patent Document 1 discloses a method for distinguishing and estimating whether the cause of the prolongation of the blood clotting time in a test specimen is a deficiency type or an inhibitor type, using features extracted from a clot waveform obtained from a plasma sample obtained by mixing a test specimen and a normal specimen and its differential waveform. Patent Document 2 discloses a method for estimating the presence or absence of blood clotting abnormality in a test specimen and the activity value (concentration) of blood clotting factors by template matching using 50 parameters related to the center of gravity of a waveform obtained by first-order differentiation of a clot waveform obtained from the test specimen. In particular, a method for estimating the concentration of blood clotting factor VIII (FVIII) and the concentration of blood clotting factor IX (FIX) and distinguishing and estimating which blood clotting factor is deficient in the test specimen is disclosed. Qualitative and quantitative abnormality of FVIII is called hemophilia A, and qualitative and quantitative abnormality of FIX is called hemophilia B, and which blood clotting factor is deficient is important because it is related to the selection of a treatment policy. Patent Document 3 discloses an APTT prolongation factor estimation system equipped with the template matching algorithm. In addition, Non-Patent Document 1 discloses a method for determining whether a test specimen is a deficiency type, an inhibitor type, or an anticoagulant-added type, by using a flow chart that uses the slope and area of ​​a part of the waveform obtained by first-order differentiation of the clot waveform obtained from the test specimen, the ratio of the time width of the waveform, etc. In particular, in the inhibitor type, the test specimen is determined to be a lupus anticoagulant (LA) positive group or an inhibitor expression group against FVIII. LA is one of the inhibitors against phospholipids, and its expression leads to antiphospholipid syndrome. In addition, Patent Document 4 describes extracting multiple parameters from the first-order differential waveform and the second-order differential waveform of the clot waveform, determining the predicted concentration of each blood coagulation factor from the multiple parameters by multivariate correlation, and using a neural network trained for the prediction. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Patent No. 6994528 [Patent Document 2] International Publication No. 2020 / 158948 [Patent Document 3] Patent No. 6811975 [Patent Document 4] U.S. Patent No. 6,524,861 [Non-patent literature]

[0007] [Non-Patent Document 1] D.Shimomura et al., The First-Derivative Curve of the Coagulation Waveform Reveals the Cause of aPTT Prolongation, Clinical and Applied Thrombosis / Hemostasis, Volume 26(2020)P1-8 Summary of the Invention [Problem to be solved by the invention]

[0008] (1) The objective of this project is to use features extracted from the analysis of the clot waveform obtained by measuring the clotting reaction in the reaction solution produced by mixing the test sample with a reagent to distinguish and estimate whether the cause of the prolonged blood clotting time of the test sample is FVIII deficiency, FIX deficiency, or LA positive sample, which are important for diagnosing hemophilia A, hemophilia B, and antiphospholipid syndrome.

[0009] (2) The more severe the hemophilia patient's blood clotting time is, the longer it becomes. In order to improve throughput, it is common knowledge among engineers in the field that blood clotting time measuring devices stop measuring each sample when the blood clotting reaction has ended and move on to measuring the next sample. As a result, the data length of the clotting waveform differs for each sample.

[0010] Patent Document 4 describes the use of a neural network to estimate the predicted concentration of each blood coagulation factor from multiple parameters using multivariate backward correlation, and the use of a neural network to estimate cases, and the clustering method including the neural network uses fixed data length feature values ​​as input data. Currently known clustering methods such as neural networks, mixed Gaussian distribution, and K-means method are technologies that assume that the number of input data is constant. When using existing clustering methods including these for clot waveform analysis, the clustering method functions only after appropriate preprocessing is performed on the clot waveform data, which is of variable data length, to make it a fixed data length. As an example of preprocessing for such variable data length, data compression technology and zero padding technology are widely known as excellent preprocessing for image and voice recognition. If it becomes possible to estimate the cause of prolongation of blood coagulation time using a neural network directly using a series of clot waveform groups obtained by a cross-mixing test using a clot waveform of only a test specimen or a multiple mixed specimen in which a test specimen and a normal specimen are mixed at a predetermined ratio, the cause of prolongation of blood coagulation time can be estimated including hidden information contained in the clot waveform in addition to experimentally determined feature values. To achieve this, it is necessary to demonstrate that suitable preprocessing can be used to discriminate clot waveforms, which are of variable data length, using a neural network, and that practical detection accuracy can be achieved using a neural network using clot waveforms or groups of clot waveforms, as well as to disclose the neural network configuration and specific numerical values ​​such as hyper-parameters to achieve this.

[0011] Therefore, the present disclosure aims to estimate the cause of the prolonged blood clotting time of a test sample by using features extracted from a clotting waveform that indicates the change in light intensity over time due to the clotting reaction of a reaction solution generated by mixing a test sample with a reagent.

[0012] In addition, the present disclosure aims to estimate the cause of prolonged blood clotting time with practical accuracy using a neural network by performing suitable preprocessing on clot waveforms, which have variable data lengths. [Means for solving the problem]

[0013] The method of estimating the cause of prolongation of blood clotting time of the present disclosure includes obtaining a clot waveform showing a change in light intensity over time due to a clot reaction of a reaction solution produced by mixing a test specimen made of plasma obtained by separating blood obtained from a subject with a reagent, obtaining a first waveform by performing a front-to-back differential process on the clot waveform, obtaining a first fitted waveform by performing a fitting process on the first waveform, obtaining a second fitted waveform by performing a fitting process on the first waveform, and obtaining a third waveform by performing a front-to-back differential process on the second fitted waveform; The method includes obtaining a first normalized waveform, a second normalized waveform, and a third normalized waveform by normalizing the light intensity axis and the time axis of each of the first fitted waveform, the second fitted waveform, and the third waveform, respectively; extracting feature amounts from each of the first normalized waveform, the second normalized waveform, and the third normalized waveform; and estimating a cause of the prolongation of the blood coagulation time of the test sample based on known feature amounts extracted from a group of samples in which the cause of the prolongation of the blood coagulation time is known and the feature amounts extracted from each of the first normalized waveform, the second normalized waveform, and the third normalized waveform.

[0014] In addition, the method for estimating the cause of prolongation of blood clotting time disclosed herein is a method for estimating the cause of prolongation of blood clotting time of a test specimen from a clot waveform that indicates the change in light intensity over time due to the clotting reaction of a reaction liquid generated by mixing a test specimen consisting of plasma obtained by separating blood obtained from a subject with a reagent, and includes aligning the data length of the clot waveform data, estimating multiple causes of prolongation of blood clotting time using a neural network from the clot waveform data with aligned data lengths, and presenting the causes of prolongation of blood clotting time of the test specimen based on the probability of belonging to the estimated causes of prolongation of blood clotting time. Effect of the Invention

[0015] According to the present disclosure, it is possible to estimate the cause of the prolonged blood clotting time of a test sample by using features extracted from a clotting waveform that indicates the change in light intensity over time due to the clotting reaction of a reaction solution produced by mixing a test sample with a reagent.

[0016] In addition, according to the present disclosure, by performing suitable preprocessing on the clot waveform, which has a variable data length, it is possible to estimate the cause of prolonged blood clotting time with practical accuracy using a neural network.

[0017] Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0018] [Figure 1] FIG. 2 shows the clot waveforms of two FVIII-deficient samples in Example 1 and the waveforms corresponding to their first derivatives. [Diagram 2] This figure shows the clot waveforms of the FVIII-deficient sample group, the FIX-deficient sample group, and the LA-positive sample group of Example 1, the waveforms equivalent to the first derivative of the clot waveform, and the waveforms equivalent to the second derivative of the clot waveform normalized. [Figure 3A] 11 is a flowchart showing a method for estimating the cause of prolongation of blood coagulation time in a comparative example. [Figure 3B]1 is a flowchart showing a method for estimating the cause of prolongation of blood coagulation time in Example 1. [Figure 4] 1 is a diagram showing an example of the overall configuration of an automatic analyzer according to a first embodiment. [Diagram 5] FIG. 13 is a diagram showing WaveNor0, 1, and 2, which are waveforms obtained by normalizing Waves 0, 1, and 2 of the first embodiment, and feature vectors. [Figure 6] FIG. 13 is a diagram showing a two-dimensional map made up of feature amounts of WaveNor0, 1, and 2 in the first embodiment. [Figure 7] FIG. 13 is a diagram showing an example of plotting feature amounts of a test sample in a two-dimensional map created from feature amounts of a sample for which the cause of prolongation of blood coagulation time in Example 1 is known. [Figure 8] This figure shows normalized Waves 0, 1, and 2 of the test samples in Example 1, indicators showing the probability of belonging to the FVIII deficient sample group, FIX deficient sample group, LA positive sample group, normal sample group, and unknown, and an example of the identification estimation result. [Figure 9] FIG. 2 is a diagram showing the relationship between the analysis unit, PC, and communication interface in Example 1, and an example of displaying the results of a test sample on a display unit. [Figure 10] FIG. 11 is a diagram showing an example of WaveSrc1 and Wave1 in the first embodiment. [Figure 11A] FIG. 11 is a diagram illustrating a preprocessing method for a neural network according to the second embodiment. [Figure 11B] FIG. 11 is a diagram illustrating a preprocessing method for a neural network according to the second embodiment. [Figure 11C] FIG. 11 is a diagram illustrating a preprocessing method for a neural network according to the second embodiment. [Figure 12] FIG. 11 is a diagram showing the structure of a detector using a neural network according to a second embodiment. [Figure 13A] 13 is an experimental result showing the relationship between dropout and detection accuracy of a detector using a preprocessing method and a neural network according to the second embodiment. [Figure 13B] 13 is an experimental result showing the relationship between dropout and detection accuracy of a detector using a preprocessing method and a neural network according to the second embodiment. [Figure 13C] 13 is an experimental result showing the relationship between dropout and detection accuracy of a detector using a preprocessing method and a neural network according to the second embodiment. [Figure 14A] These are experimental results showing the relationship between the preprocessing method of Example 2 and the hyper-parameters of the neural network-based detector and the detection accuracy. [Figure 14B] These are experimental results showing the relationship between the preprocessing method of Example 2 and the hyper-parameters of the neural network-based detector and the detection accuracy. [Figure 15] 13 shows experimental results showing that the preprocessing method of the second embodiment and the classifier using a neural network suppress overlearning. [Figure 16] 13 shows the experimental results of Example 4 showing the difference in clot waveforms between FVIII-deficient samples and FIX-deficient samples. [Figure 17A] 1 shows experimental results showing the relationship between the maximum value of the first derivative, which is a known index, and the log clotting time. [Figure 17B] 13 shows the experimental results showing the relationship between vec1_y and log clotting time in Example 4. [Figure 18A] 13 shows the experimental results of representative waveforms of each sample group for WaveNor0, in which the clot waveforms of commercially available samples in Example 4 were standardized based on the NRC method. [Figure 18B] 13 shows the experimental results of representative waveforms for each sample group for WaveNor1, in which the clot waveforms of commercially available samples in Example 4 were standardized based on the NRC method. [Figure 18C] 13 shows the experimental results of representative waveforms for each sample group for WaveNor2, in which the clot waveforms of commercially available samples in Example 4 were standardized based on the NRC method. [Figure 19] 13 is an experimental result showing a frequency spectrum of WaveNor2 in Example 4. [Figure 20] 13 shows an example of a feature set according to the fourth embodiment. [Figure 21] 13 shows the experimental results of the commercially available sample group of Example 4 displayed in feature space #6. [Figure 22A]13 is a display example showing a summary in which the commercially available sample group of Example 4 is plotted in a feature space. [Figure 22B] 1 is a display example showing the discrimination results of one FVIII-deficient sample in Example 4. [Figure 23] FIG. 13 is a schematic diagram showing a method for dimensionally compressing a series of clot waveform data obtained for one test specimen in the cross-mixing test of Example 5 into two area indices F1 and F2. [Figure 24] FIG. 13 shows the results of discrimination of samples in a cross-mixing test using the multidimensional mixed Gaussian distribution of Example 5. [Figure 25A] FIG. 13 is a diagram showing cluster regions of indexes (F1, F2) in Example 5. [Figure 25B] FIG. 13 shows the experimental results of adding data points to the cluster regions of the indicators (F1, F2) in Example 5. [Figure 26] FIG. 13 is a diagram explaining the indexes (G1, G2) of Example 5. [Figure 27A] This figure shows the experimental results of plotting immediate (F1, G1) and delayed (F2, G2) values ​​obtained by combining index F calculated from a series of clot waveforms obtained from one sample in Example 5 and index G calculated from feature space #1 using the same sample. [Figure 27B] FIG. 1 shows experimental results in which the indices (F1, G1) and (F2, G2) obtained from the 34 test samples in Example 5 are plotted. [Figure 28A] 13 is an experimental result showing the average shape of trajectory data for each cause of prolongation of blood coagulation time obtained by the cross-mixing test of Example 5. [Figure 28B] FIG. 13 shows the experimental results of discrimination performed by separating the trajectory data of each specimen in the cross-mixing test of Example 5 into a 2-dimensional x14 space. [Figure 29] FIG. 13 is a diagram summarizing the conditions for determining the total number of specimens required for the discrimination method of Example 5. [Diagram 30]FIG. 13 is a schematic diagram showing the relationship between the number of clot waveforms used in the discrimination in Example 5 and the worst probability (the minimum value of Pti across all samples, when Pti is the probability that the i-th sample belongs to the correct sample group cluster). [Diagram 31] 13 is a display example showing the discrimination results of the cross-mixing test of Example 5. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Hereinafter, the present embodiment will be described with reference to the drawings. In the drawings, functionally identical elements may be indicated by the same numbers. Note that the drawings show embodiments and implementation examples according to the principles of the present disclosure, but these are for understanding the present disclosure and are not to be used to interpret the present disclosure in a limited manner. The description in this specification is merely a typical example and does not limit the scope or application of the present disclosure in any sense.

[0020] In the present embodiment, the description is given in sufficient detail for a person skilled in the art to implement the present disclosure, but it should be understood that other implementations and forms are possible, and that changes in configuration and structure and substitutions of various elements are possible without departing from the scope and spirit of the technical ideas of the present disclosure. Therefore, the following description should not be interpreted as being limited thereto.

[0021] [Example 1] In the first embodiment, the cause of the prolonged blood coagulation time is estimated using features extracted from waveforms (WaveNor0, WaveNor1, WaveNor2) obtained by waveform fitting, which are normalized based on the NRC (Normalization based on the Reaction Constants) method described below.

[0022] The samples used to obtain the clot waveform contain multiple blood coagulation factors, and if the concentration of each factor differs, the speed of the coagulation reaction differs, leading to a variety (variation) of blood coagulation time. In this field, the first derivative of the clot waveform is treated as the coagulation velocity, and the second derivative is treated as the coagulation acceleration. Even in the case of the same sample group, the clot waveform and the waveform related to the coagulation velocity do not match, making it difficult to visually recognize the characteristics of the sample group. As an example, Figure 1 shows the clot waveforms of two FVIII-deficient samples (hereinafter, the clot waveforms are appropriately referred to as "WaveSrc0") and the first waveform equivalent to the first derivative of the two clot waveforms (WaveSrc0) (hereinafter, the waveform equivalent to the first derivative of the clot waveform is appropriately referred to as "WaveSrc1"). In WaveSrc0, the rise time of the waveform (position of the arrowhead a1, a2), the total change in the amount of light (arrow width b1, b2), and the blood coagulation time (for example, the time c1, c2 at which the change in the amount of light (scattered light intensity (count) in Figure 1) becomes 50%) differ for each sample. The total change in light intensity depends on the concentration of fibrinogen, which is one of the blood coagulation factors. Blood clotting time is affected by the speed of the clotting reaction, and in the case of hemophilia samples, it is self-evident that the blood clotting time will be longer as the symptoms become more severe. Even in WaveSrc1, there are differences in the height and width of the waveform for each sample. It is also an experimental fact that waveform features such as the clotting speed (change in the amount of light (scattered light intensity (counts)) per unit time) become smaller in inverse proportion to the blood clotting time as the blood clotting time becomes longer, and they differ for each sample due to the difference in blood clotting time. However, these two samples are both FVIII-deficient samples, and it is necessary to find common features to classify both samples into the FVIII-deficient sample group.

[0023] The inventors considered that suppressing the influence of variations in blood clotting time within each sample group is effective in revealing common characteristics of the same sample group. Based on this idea, in the present disclosure, by performing normalization based on the NRC method that simultaneously normalizes not only the light intensity axis but also the time axis, it has been found that it is possible to converge WaveSrc0 of the same sample group, WaveSrc1 obtained by differential processing before and after WaveSrc0, and WaveSrc2 obtained by differential processing before and after WaveSrc1 to reveal waveform characteristics within the same sample group, and that it is possible to reveal differences in waveform shapes for each sample group when comparing waveforms for each sample group. A specific method of normalization will be described in detail later (sample identification program). WaveSrc2 is a waveform equivalent to the first differential of WaveSrc1 (equivalent to the second differential of the clotting waveform).

[0024] FIG. 2 shows WaveSrcNor0, WaveSrcNor1, and WaveSrcNor2, which are waveforms obtained by normalizing WaveSrc0, WaveSrc1, and WaveSrc2 of the FVIII-deficient sample group, the FIX-deficient sample group, and the lupus anticoagulant (LA)-positive sample group, respectively. In each graph, it can be seen that the waveforms generally converge within the same sample group. In addition, when the same type of waveform (for example, WaveSrcNor0) is compared for each sample group, differences in shape were observed. For example, in WaveSrcNor0, differences were observed in the rising position of the waveform for each sample group. The auxiliary line L1 was placed so as to be located at the center of the rising position of the waveform for the LA-positive group. The rising position of the waveform for the FVIII-deficient sample group was located to the right of this auxiliary line L1, and the rising position of the waveform for the FIX-deficient sample group was located to the left of this auxiliary line L1, and a difference was observed. In WaveSrcNor1 and WaveSrcNor2, differences were observed in the width between the apexes of the two peaks and the height of the apexes. In both the waveforms of WaveSrcNor1 and WaveSrcNor2, the auxiliary lines L2 and L4 were placed at the center of the peak of the left peak of the LA positive group, and the auxiliary lines L3 and L5 were placed so as to be located at the center of the peak of the right peak of the LA positive group. When the peak position of the right peak was confirmed using the auxiliary lines L3 and L5 as a guide, no significant difference was observed between the sample groups. On the other hand, when the peak position of the left peak was confirmed using the auxiliary lines L2 and L4 as a guide, the peak of the FVIII deficient sample group was located to the right of the auxiliary lines L2 and L4, and the peak of the FIX deficient sample group was located to the left of the auxiliary lines L2 and L4, and differences were observed. Since the differences between these sample groups are larger than the variations within the same sample group, the inventors came up with the idea of ​​quantifying these differences as features and using them for sample identification.

[0025] The samples used to obtain WaveSrc0, which is the original data for generating WaveSrc0 shown in FIG. 1 and WaveSrcNor0 shown in FIG. 2, are all commercially available products. Specifically, they are as follows. FVIII-deficient samples are Factor VIII Deficient Plasma (manufactured by Precision BioLogic, Inc.), FVIII Deficient Plasma (manufactured by George King Bio-Medical, Inc.), and Factor VIII Deficient Plasma (manufactured by Affinity Biologicals, Inc.). FIX-deficient samples are Factor IX Deficient Plasma (manufactured by Precision BioLogic, Inc.), FIX Deficient Plasma (manufactured by George King Bio-Medical, Inc.), and Factor IX Deficient Plasma (manufactured by Affinity Biologicals, Inc.). The LA-positive samples were Lupus Positive Control and Weak Lupus Positive Control (Precision BioLogic, Inc.), and Positive LA Plasma (George King Bio-Medical, Inc.).

[0026] The specimen to be disclosed herein is not particularly limited as long as it is a specimen derived from a subject that causes a coagulation reaction, but a specimen made of plasma obtained by separating blood obtained from a subject is preferable. The specimen made of plasma may contain a small amount of impurities within a range that does not cause problems in the analysis by the automatic analyzer of the disclosure, such as the measurement of the clot waveform. The reagent is not particularly limited as long as it is a reagent for measuring the thromboplastin time (PT) item, the activated partial thromboplastin time (APTT) item, and the fibrinogen (Fbg) item. The device may be configured to measure the change in the light amount (light amount is an index including turbidity) of the reaction solution over time due to the coagulation reaction. The reagent used in the measurement of the clot waveform shown in this embodiment is Coagpia APTT-N (manufactured by Sekisui Medical Co., Ltd.), and the device is Hitachi Automated Analyzer 3500 (manufactured by Hitachi High-Tech Corporation).

[0027] The present disclosure is characterized not only in that it performs standardization based on the NRC method to make the differences in waveform shapes apparent, but also in that it performs waveform fitting to quantify the made-apparent differences in waveform shapes while suppressing the effects of noise.

[0028] Next, an automatic analyzer that acquires a clot waveform, a method for acquiring a clot waveform, a sample screening program, and the like will be specifically described.

[0029] (Automatic analyzer 100) An automatic analyzer 100 used to acquire a clot waveform will be described. Fig. 4 is a diagram showing an example of the overall configuration of the automatic analyzer 100 of Example 1. Here, a basic operation of the device will be described with reference to Fig. 4, but is not limited to the following example.

[0030] 4, the automatic analyzer 100 is generally composed of an analysis unit 130, an operation computer 118, a storage unit 119, and a control computer 120. The analysis unit 130 is generally composed of a specimen dispensing mechanism 101, a specimen disk 102, a reagent dispensing mechanism 106, a reagent disk 107, a reaction vessel stock unit 111, a reaction vessel transport mechanism 112, a detection unit 113, and a reaction vessel disposal unit 117. The control computer 120 is connected to the analysis unit 130 so as to be able to communicate with it.

[0031] The specimen dispensing mechanism 101 aspirates specimens 103a contained in specimen containers 103 arranged on a specimen disk 102 that rotates clockwise and counterclockwise, and dispenses the specimens into reaction containers 104. The specimen dispensing mechanism 101 executes the aspirating and dispensing operations of the specimen by the operation of a specimen syringe pump 105 controlled by a control computer 120.

[0032] The reagent dispensing mechanism 106 aspirates the reagent 108a contained in the reagent container 108 arranged on the reagent disk 107, and dispenses the reagent into the reaction container 104. The reagent dispensing mechanism 106 executes the aspirating operation and dispensing operation of the reagent by the operation of a reagent syringe pump 110 controlled by a control computer 120.

[0033] In addition, a reagent heating mechanism 109 is built into the reagent dispensing mechanism 106, and the temperature of the reagent 108a aspirated by the reagent dispensing mechanism 106 is raised to an appropriate temperature (predetermined temperature) by the reagent heating mechanism 109 controlled by the control computer 120.

[0034] The reaction vessel transport mechanism 112 transports and sets the reaction vessel 104. The reaction vessel transport mechanism 112 holds the reaction vessel 104 and rotates it in the horizontal direction, thereby transporting and setting the reaction vessel 104 from the reaction vessel stock section 111 to the reaction vessel setting section 114 of the detection unit 113.

[0035] The detection unit 113 has one or more reaction vessel installation parts 114 (in this embodiment, one case is shown as an example) for placing the reaction vessel 104, and measures the light intensity of the reaction liquid (mixture of sample 103a and reagent 108a) in the reaction vessel 104 inserted in the reaction vessel installation part 114. The detection unit 113 controls the temperature so that the reaction liquid in the reaction vessel 104 inserted in the reaction vessel installation part 114 is, for example, 37°C. Note that, although the present embodiment shows a case where one detection unit 113 is arranged, a configuration having multiple detection units 113 may be used. An example of the detection principle in the detection unit 113 will be described below. Light irradiated from the light source 115 is scattered by the reaction liquid in the reaction vessel 104. The detection part (optical sensor) 116 receives the scattered light scattered by the reaction liquid in the reaction vessel 104. For the light source 115, for example, a halogen lamp or an LED is used. The detection section (optical sensor) 116 is composed of a photodiode and the like. The signal received by the detection section (optical sensor) 116 is converted into a digital signal by an A / D converter 121 to become a light quantity value, which is input to a control computer 120 as reaction process data (clotting waveform data showing the time change of the detected light quantity) and is taken into a storage section 119. The operation of the detection unit 113 is controlled by the control computer 120. Here, a detector using scattering of light is used, but there are also other types, such as those using transmitted light.

[0036] The reaction vessel transport mechanism 112 holds the reaction vessel 104 after measurement has been completed, and discards it into the reaction vessel discard unit 117 .

[0037] For the purpose of improving the processing capacity, an incubator 122 without a detector may be provided for warming the sample before the addition of the measurement start reagent.

[0038] The control computer 120 is an information processing device having a computer system having a processor, a memory, and the like, and not only controls the operation of the automatic analyzer such as dispensing the specimen 103a and the reagent 108a, transferring the reaction vessel 104, and discarding the reaction vessel 104, but also calculates the blood coagulation time from the measured value (clotting waveform data) of the light intensity that changes with time according to the coagulation reaction of the reaction liquid, and performs identification and estimation of the cause of the prolongation of the blood coagulation time. The calculated blood coagulation time and the identification and estimation result of the cause of the prolongation of the blood coagulation time are output to the display unit 118c and stored in the memory unit 119. These results may be printed out by the printer 123 via the operation computer 118. The memory unit 119 connected to the control computer 120 stores measured waveform data, quantification results, specimen identification results, and the like in addition to programs such as a control program, a measurement program, a calibration curve generation program, a quantification program, and a specimen identification program. Various programs are read and executed according to a request input to the operation computer 118 or a request sent from the communication interface 124. Input to the operation computer 118 may be made by touching the display unit 118c or via the connected keyboard 118b. Also, input may be made by selecting what is displayed on the display unit 118c with the mouse 118a. Waveform data, quantitative results, specimen examination results, and the like stored in the memory unit 119 are output to the display unit 118c, or sent to the communication interface 124, or printed by the printer 123 as necessary. The communication interface 124 is connected to, for example, a network within a hospital, and communicates with a Hospital Information System (HIS) or a Laboratory Information System (LIS).

[0039] As described above, the sample identification program and the waveform data of the sample stored in the storage unit 119 are called up by the control computer 120 and executed to estimate the cause of the prolongation of the blood coagulation time of the test sample. Alternatively, the waveform data of the test sample may be transferred to the analysis computer 125 that executes the sample identification program, and the analysis computer 125 may estimate the cause of the prolongation of the blood coagulation time of the test sample and display the result on the screen of the analysis computer 125. Alternatively, the waveform data stored in the storage unit 119 may be written to another external storage medium or the like via the control computer 120 and the operation computer 118, and an independent analysis computer having the sample identification program may read the written waveform data and estimate the cause of the prolongation of the blood coagulation time of the test sample.

[0040] (How to obtain the clot waveform) Next, the analysis operation of the blood clotting time item and the acquisition of the clotting waveform (WaveSrc0) will be described. There are three main blood clotting time items: thromboplastin time (PT) item, activated partial thromboplastin time (APTT) item, and fibrinogen (Fbg) item. In each item, the parameters required for the analysis are set first. In the parameter setting, the item to be analyzed, the sample amount, the reagent amount, the output unit, and the like are input. In the PT item and the Fbg item, the calibration method is also input, and calibration is performed as necessary. After the parameters are set, the sample container 103 and the reagent container 108 containing the sample 103a and the reagent 108a, respectively, are placed on the sample disk 102 and the reagent disk 107, respectively, and analysis is performed. The reagent 108a can be, for example, a commercially available PT measurement reagent, APTT measurement reagent, or Fbg measurement reagent.

[0041] (PT item analysis operation) The analysis operation of the PT item will be described. The specimen dispensing mechanism 101 dispenses a specimen 103a into an empty reaction vessel 104 housed in the reaction vessel stock section 111. The reaction vessel 104 containing the specimen 103a is moved to the reaction vessel installation section 114 of the detection unit 113 by the reaction vessel transport mechanism 112. The reagent 108a is aspirated from the reagent vessel 108 by the reagent dispensing mechanism 106 and heated to an appropriate temperature by the reagent heating mechanism 109. The temperature at this time is preferably 37°C. The heated reagent 108a is discharged into the reaction vessel 104 containing the specimen 103a already installed in the reaction vessel installation section 114. At the same time as the discharge starts, the turbidity change over time of the reaction liquid, which is a mixture of the specimen 103a and the reagent 108a, is optically measured and becomes WaveSrc0.

[0042] (APTT item analysis operation) The analysis operation of the APTT item will be described. The specimen dispensing mechanism 101 dispenses the specimen 103a into an empty reaction vessel 104 housed in the reaction vessel stock section 111. The reaction vessel 104 containing the specimen 103a is moved to the reaction vessel installation section 114 of the detection unit 113 by the reaction vessel transport mechanism 112. The first reagent contains an activator such as ellagic acid or kaolin, and is aspirated from the reagent vessel 108 by the reagent dispensing mechanism 106 and heated to an appropriate temperature by the reagent heating mechanism 109. The temperature at this time is preferably 37°C. The heated first reagent is discharged into the reaction vessel 104 containing the specimen 103a already installed in the reaction vessel installation section 114. The mixture of the specimen 103a and the first reagent is temperature-regulated in the reaction vessel 104. The temperature at this time is preferably 37°C. At this time, the mixture of the specimen 103a and the first reagent may be stirred by a stirring mechanism not shown. The second reagent is a calcium chloride solution, which is aspirated from a reagent container 108 by a reagent dispensing mechanism 106 and heated to an appropriate temperature by a reagent heating mechanism 109. The temperature at this time is preferably 37°C. The heated second reagent is discharged into a reaction container 104 containing a mixture of a specimen 103a and the first reagent already installed in a reaction container installation section 114. Before the second reagent is discharged, the mixture of the specimen 103a and the first reagent is temperature-controlled for a certain period of time, which is preferably 180 seconds. Simultaneously with the start of discharge of the second reagent, the change in turbidity over time of the reaction liquid, which is a mixture of the specimen 103a and the first and second reagents, is optically measured, resulting in WaveSrc0.

[0043] (Analysis of Fbg items) The analysis operation of the Fbg item will be described. The specimen 103a is dispensed by the specimen dispensing mechanism 101 into an empty reaction vessel 104 housed in the reaction vessel stock section 111. The reaction vessel 104 containing the specimen 103a is moved to the reaction vessel installation section 114 of the detection unit 113 by the reaction vessel transport mechanism 112. The specimen dilution liquid is aspirated from the reagent vessel 108 by the reagent dispensing mechanism 106 and heated to an appropriate temperature by the reagent heating mechanism 109. The temperature at this time is preferably 37°C. The heated specimen dilution liquid is discharged into the reaction vessel 104 containing the specimen 103a already installed in the reaction vessel installation section 114. The mixture of the specimen 103a and the specimen dilution liquid is temperature-regulated in the reaction vessel 104. The temperature at this time is preferably 37°C. At this time, the mixture of the specimen 103a and the specimen dilution liquid may be stirred by a stirring mechanism (not shown). The reagent is aspirated from the reagent container 108 by the reagent dispensing mechanism 106 and heated to an appropriate temperature by the reagent heating mechanism 109. The temperature at this time is preferably 37°C. The heated reagent is discharged into the reaction container 104 containing a mixture of the specimen 103a and specimen dilution solution already installed in the reaction container installation section 114. At the same time as the discharge of the reagent starts, the change in turbidity over time of the reaction solution, which is the mixture of the specimen 103a, specimen dilution solution, and reagent, is optically measured and becomes WaveSrc0.

[0044] In either case, the measurement ends when the coagulation reaction is completed (when no change in turbidity is observed). Alternatively, the measurement ends after a certain period of time has elapsed. The certain period of time is, for example, 5 minutes. The optically measured change in turbidity over time is input to the memory unit 119 as a coagulation waveform via the control computer 120.

[0045] An example of the clot waveform is shown in Figure 1, where the horizontal axis represents the measurement time and the vertical axis represents the light intensity. Figure 1 shows an example of the clot waveform of an FVIII-deficient sample (Precision BioLogic Inc.) in the APTT item, and the vertical axis represents the scattered light intensity.

[0046] (Sample Screening Program) FIG. 3B is a flowchart showing a method for estimating the cause of prolongation of blood coagulation time, which is executed by executing a sample identification program. The computer system of the control computer 120 estimates the cause of prolongation of blood coagulation time in Example 1 by executing the sample identification program. A specific description will be given with reference to FIG. 3B. Each step of the flowchart in FIG. 3B may be executed by the control computer 120, the operation computer 118, the analysis computer 125, or a computer on another network. Here, an example in which the computer system of the control computer 120 executes each step will be described.

[0047] The control computer 120 reads the clot waveform (WaveSrc0) data acquired by the clot reaction measurement (FIG. 3B-S301b). Specific analysis operations related to acquisition of WaveSrc0 are as described above in (Method of acquiring clot waveform).

[0048] Next, the control computer 120 generates WaveSrc1 by performing differential processing before and after WaveSrc0 (FIG. 3B-S302b). WaveSrc1 is calculated, for example, by Equation 1. Here, Δt in Equation 1 below is the measurement interval of the coagulation reaction, and is, for example, 0.1 seconds.

[0049]

number

[0050] Next, the control computer 120 generates a first fitted waveform Wave0 (hereinafter, the waveform that has been subjected to fitting processing on the clot waveform is referred to as the first fitted waveform and is appropriately referred to as "Wave0") by waveform fitting of WaveSrc0 (FIG. 3B-S303b), and generates Wave1 (hereinafter, the waveform that has been subjected to fitting processing on the first waveform is referred to as the second fitted waveform and is appropriately referred to as "Wave1") by waveform fitting of WaveSrc1 (FIG. 3B-S304b). Considering that the clot waveform is a sigmoid curve (S-shaped curve), it is desirable for the fitting equation for WaveSrc0 to be an equation based on a sigmoid curve. For example, it is Equation 2. In Equation 2, when the amount of light after the reagent is discharged is Y dc This is a formula that follows the clot waveform:

[0051]

number

[0052] Y in the above formula 2 dc is the base light intensity [counts], Y0 is the base scattered light intensity [counts] of the reaction solution itself, Y1 is the saturated scattered light intensity [counts], t is the measurement time [seconds], T0 is the time (Y0 + Y1) ÷ 2 [seconds], and T1 is the reaction time constant [seconds].

[0053] Equation 2 assumes fitting of the clot waveform measured by scattered light, but in the case of the clot waveform measured by absorbance, the waveform measured by scattered light is inverted upside down, so it is sufficient to improve it based on Equation 2 so that it fits the clot waveform measured by absorbance.

[0054] The fitting equation for WaveSrc1 should be one that can reproduce the bimodal distribution more accurately. For example, the left peak of the bimodal distribution should be expressed as a Gaussian distribution, and the right peak as a sigmoidal derivative, and the two should be connected with a cubic spline function. The shape of WaveSrc1 often becomes bimodal, and this tendency is particularly noticeable in samples with prolonged blood clotting times.

[0055] Next, the control computer 120 generates Wave2 (hereinafter, a waveform equivalent to the first differential of the first fitting waveform is regarded as a third waveform and is appropriately referred to as "Wave2") by performing differential processing before and after Wave1 (FIG. 3B-S305b). Wave2 is calculated, for example, by Equation 3.

[0056]

number

[0057] Next, the control computer 120 normalizes both the time axis and the light amount axis of Wave0, 1, and 2 based on the NRC method (FIG. 3B-S306b). The time axis is normalized by Equation 4. The light amount axes of Wave0, 1, and 2 are normalized by Equation 5, Equation 6, and Equation 7, respectively. The waveform obtained by normalizing the time axis and the light amount axis of Wave0 is the first normalized waveform, which is hereinafter appropriately called "WaveNor0". The waveform obtained by normalizing the time axis and the light amount axis of Wave1 is the second normalized waveform, which is hereinafter appropriately called "WaveNor1". The waveform obtained by normalizing the time axis and the light amount axis of Wave2 is the third normalized waveform, which is hereinafter appropriately called "WaveNor2". The normalization is based on the reaction constant, and is called the NRC (Normalization based on the Reaction Constants) method. The reaction constant used in the NRC method can be a numerical value obtained by fitting WaveSrc0. Waveform fitting is performed in order to convert the difference into a numerical value while suppressing the influence of noise, and to extract it as a unique feature.

[0058]

number

[0059]

number

[0060]

number

[0061]

number

[0062] Here, n is the number of data points (n≧1), and the relationship between t(n) and n is t(n)=n / 10 [seconds]. In addition, T0, T1, Wave0(n), Y dc , Y1 is obtained from Equation 2.

[0063] Finally, the control computer 120 extracts features from the normalized waveform (FIG. 3B-S307b). The features must be quantified portions where differences in waveform shape are recognized for each sample group. Here, reference is made again to FIG. 2. FIG. 2 shows a waveform (WaveSrcNor0, 1, 2) normalized from the original waveform (WaveSrc0, 1, 2) before waveform fitting. The normalization is in accordance with Equations 4 to 7, but in FIG. 2, the original waveform is normalized, and WaveSrc0, WaveSrc1, and WaveSrc2 are substituted for Wave0(n) in Equation 5, Wave1(n) in Equation 6, and Wave2(n) in Equation 7, respectively.

[0064] As mentioned above, for example, in WaveSrcNor0, differences were observed in the rising position of the waveform. The order of the rising was FIX-deficient specimen group, LA-positive specimen group, and FVIII-deficient specimen group, in order of the shortest normalized time. According to the time normalization formula, Equation 4, the smaller the reaction time constant T1 (i.e., the faster the reaction), the farther the distance from the normalized time 0. In other words, the difference in reaction speed of each specimen group was expressed as a difference in the rising timing. FVIII becomes activated FVIII, and its presence enhances a part of the reaction in the coagulation reaction mechanism, so it is considered that the reaction time constant became large and the rising timing became delayed in the FVIII-deficient specimen group. LA has the property of binding to phospholipids, which are the scaffolding of the coagulation reaction, and inhibiting the binding of coagulation factors, and therefore inhibits the coagulation reaction uniformly regardless of the activity of blood coagulation factors. Since the degree of inhibition varies depending on the quantity and quality (potency) of LA, the speed of the coagulation reaction becomes diverse, which leads to a diversity of reaction time constants T1, and it is considered that the sample group is located between the FIX-deficient sample group and the FVIII-deficient sample group. In this way, the rise timing is considered to reflect the difference originating from the cause of the extension of blood coagulation time, and it was determined that a feature that quantifies the difference in the rise timing is useful. Here, the feature was a vector with the point where the normalized time of each waveform is 0 as the start point and the point where the waveform rises as the end point.

[0065] In addition, in WaveSrcNor1, differences were observed in the width between the apexes of the two peaks and the height of the apexes. In the FVIII-deficient sample group, a small amount of FVIII was used in the early stage of the reaction, and then it became insufficient, which is thought to have caused the first peak to become larger (higher). On the other hand, in the FIX-deficient sample group and the LA-positive sample group, FVIII was present in sufficient amounts, so the reaction rate was faster in the amplification reaction called positive feedback that occurred later than in the early stage of the reaction, and the second peak became larger (higher). In addition, focusing on the normalized time, there was a difference in the timing of the rise of the waveform for the same reason as in WaveSrcNor0, which led to a difference in the normalized time at which the first peak occurred. The difference in the occurrence time of the first peak and the difference in the height of the two peaks are thought to reflect differences derived from the causes of the extension of the blood coagulation time, and it was determined that a feature that quantifies these differences is useful. Here, the feature was a vector with the normalized time of 0 as the starting point and the peak point of the first peak as the ending point.

[0066] In WaveSrcNor2, differences were observed between the first peak height and its occurrence time, influenced by waveform differences equivalent to the first derivative. Here, the peak height and normalized time information of the first peak were combined, and the vector information with the normalized time of 0 as the start point and the peak point of the first peak as the end point was used as the feature.

[0067] In order to suppress the influence of noise and to quantify the vector feature, it is extracted from the waveform (WaveNor0,1,2) obtained by normalizing the fitting waveform (Wave0,1,2). Figure 5 shows WaveNor0,1,2, which are the waveforms obtained by normalizing Wave0,1,2, and the feature vector. The horizontal axis represents the normalized time, and the vertical axis represents the normalized level. The feature vectors of WaveNor0,1,2 are vectors 0,1,2, respectively. The x and y components of each vector are the feature (X,Y) of each waveform. Here, the x component of WaveNor0 is designed to subtract T0 / T1 so that the difference in the waveform shape of the sample group is emphasized. The feature (X,Y) of WaveNor0,1,2 is expressed as (vec0_x,vec0_y), (vec1_x,vec1_y), and (vec2_x,vec2_y), respectively.

[0068] Finally, the extracted feature values ​​of the test specimen are compared with the feature values ​​of specimens whose sample background (cause of prolonged blood clotting time) is known, and the cause of prolonged blood clotting time of the test specimen is identified and estimated. The feature values ​​(X, Y) extracted from specimens whose cause of prolonged blood clotting time is known are used independently, and in a two-dimensional map in which these feature values ​​are plotted, a probability density distribution is given to the feature group (cluster) for each specimen group. For example, a probability density distribution based on a Gaussian mixture distribution model is used. Figure 6 shows an example of a two-dimensional map consisting of feature values ​​extracted from normalized waveforms WaveNor0, 1, and 2. In the map, they are represented as Wave0, 1, and 2. The feature values ​​are extracted by analyzing the waveforms obtained from specimens whose cause of prolonged blood clotting time is known. The ellipses shown for the feature group (cluster) of each specimen group (FVIII deficiency, FIX deficiency, LA positive, normal) represent the spread of the feature data. In order to prevent the ellipses of normal and abnormal samples from overlapping, the size of each ellipse is set to 4σ (σ: standard deviation in each direction, called Mahalanobis distance in two or more dimensions) from the center. These ellipses are expressed as the probability density of a Gaussian distribution. In the figure, the area that does not belong to any ellipse indicates that it does not belong to any sample group within the range of the given data, and indicates that it is a new clot waveform depending on the artificial mixing of two or more types of test samples or the drug administration status of a patient under treatment. This area is defined as the "unknown" group. The ellipses of abnormal samples overlap in all detectors, and the boundary can be set to any value of the probability of belonging calculated according to the probability density distribution. Figure 6 shows the case where the boundary is a probability of belonging of 50%. The abnormal samples used here are the commercially available products described above. The normal samples were also substituted with commercially available products. Specifically, these are Normal Reference Plasma and Pooled Normal Plasma (manufactured by Precision BioLogic, Inc.), Factor Assay Control Plasma and Borderline Factor Assay Control Plasma (manufactured by George King Bio-Medical, Inc.).

[0069] The feature quantities extracted from WaveNor0, 1, and 2 of the test sample are applied to these three 2D maps to calculate the probability of the test sample belonging to each map (probability of which sample group it belongs to). The test sample is identified and estimated according to the calculated probability of belonging. For example, the test sample may be identified and estimated as the sample group with the highest probability of belonging, or the probability of belonging of each map may be averaged and the sample group with the highest probability of belonging may be identified and estimated, or the probability of belonging of each map may be weighted and the sample group with the highest probability of belonging may be estimated.

[0070] 7 and 8 show examples of display of the test specimen identification and estimation results. FIG. 7 shows an example of where the feature values ​​of the test specimen are plotted in a two-dimensional map created from the feature values ​​of known specimens. FIG. 8 shows WaveNor0, 1, and 2 (indicated as Wave0, Wave1×2, and Wave2 in the drawing) of the test specimen, indicators showing the probability of belonging to the FVIII-deficient specimen group, the FIX-deficient specimen group, the LA-positive specimen group, the normal specimen group, and the unknown, and the identification and estimation results. FIG. 9 shows the relationship between the analysis unit 130, the control computer 120, and the communication interface 124, and an example of display of the test specimen results on the display unit 118c. The display unit 118c displays a column for selecting the specimen number (S_No.) of the test specimen whose results are to be displayed and the two-dimensional map. The display unit 118c may display the blood coagulation time of the selected specimen number, the result of identification and estimation, and the estimation result of the blood coagulation factor activity. Although it is not necessary to display the estimated results of blood coagulation factor activity, it is better to do so. For example, blood coagulation factor VIII deficiency is a state in which the activity value of blood coagulation factor VIII is less than 1%, and blood coagulation factor IX deficiency is a state in which the activity value of blood coagulation factor IX is less than 1%. In FIG. 9, the lower left side of the screen shows the selected map and the feature plot of the test sample in that map. By displaying in this way, it is possible to intuitively express with high visibility what is the cause of the extension of the blood coagulation time of the test sample. On the other hand, it is also important to quantitatively show the cause of the extension, and in FIG. 9, "Hemophilia A (probability: 95%)" is displayed as the "discrimination estimation result". In addition, as shown in the lower right side of the screen in FIG. 9, it is better to display the waveform and details of the belonging probability (for example, the contents of FIG. 8) of the sample of the selected sample number, and it is better to display an indicator that visualizes the waveform and displays the details of the belonging probability.

[0071] In the first embodiment, pre-processing such as smoothing may be performed after reading WaveSrc0 (FIG. 3B-S301b) and / or generating WaveSrc1 (FIG. 3B-S302b). Pre-processing (first filter processing to remove noise from WaveSrc0 and second filter processing to remove noise from WaveSrc1) refers to filter processing for reducing noise on each waveform. This processing step is not necessarily required. However, in order to perform waveform fitting with higher accuracy, it is desirable to perform it before waveform fitting. Specific examples of filter processing include N-point moving average processing (N is a positive integer and odd number, 3≦N≦number of measurement points) and smoothing processing by selecting the minimum value in a certain section. In addition, it is also possible to use a T1-point moving average using the reaction time constant of the fitting parameter according to Equation 2, in order not to lose the original waveform shape with respect to the variation of the value of the blood coagulation time. In this case, for example, the value of T1 can be rounded off to one decimal place to make it an integer, and if it is an even number, 1 can be added to it or subtracted from it to make it an odd number.

[0072] (Example of the results of running the sample screening program) The sample screening program identified and estimated the test samples, and the FVIII deficiency, FIX deficiency, and LA positive sample groups were identified at 85.7% (=12 / 14*100%), 100% (=9 / 9*100%), and 66.7% (=4 / 6*100%) respectively. Overall, the probability was 86% (=25 / 29*100%). Normal samples were identified at 100% (=10 / 10*100%), and it was confirmed that normal samples were not misclassified as FVIII deficiency, FIX deficiency, or LA positive. The test samples used were different lots of the commercially available product mentioned above.

[0073] (Effects of Example 1) Figure 10 shows examples of WaveSrc1 and Wave1. It can be seen that the waveform of WaveSrc1 contains noise. As a method for removing noise, pre-processing such as smoothing processing has been published, but there are issues such as excessive processing leading to loss of waveform shape, and insufficient processing leading to issues such as multiple values ​​being present and not uniquely determined when extracting a certain feature as a numerical value. When extracting a feature inherent in a clot waveform as a numerical value, multiple values ​​being present and not uniquely determined means, for example, when trying to extract the maximum value of the left peak of the two peaks in Figure 10, multiple maximum values ​​exist at different times in the raw data containing noise, and it is not determined to be a single point. The method of extracting features by performing waveform fitting solves such issues by determining waveform model parameters from the observed waveform containing noise so that, for example, the RMS error is minimized, and then quantifying the features inherent in the clot waveform from the fitted waveform without noise.

[0074] The cause of the prolonged blood clotting time is estimated using the extracted feature values. Specifically, feature values ​​are extracted using samples for which the cause of the prolonged blood clotting time is known, and a probability density distribution is given to the feature groups (clusters) for each sample group in a two-dimensional map on which these feature values ​​are plotted. The feature values ​​of the test sample are applied to this two-dimensional map to calculate the probability of the test sample belonging to the sample group (probability of which sample group it belongs to), and the cause of the prolonged blood clotting time of the test sample can be intuitively estimated with high visibility according to the probability of belonging.

[0075] This disclosure presents a novel method to solve the problems of the conventional technology described above by extracting the inherent features of the clot waveform as numerical values ​​using a fitting model, and by performing standardization processing using the NRC method based on the reaction constant to reveal the characteristics of each cause of prolonged blood coagulation time.

[0076] Here, FIG. 3A shows an example of a sample identification flow according to a comparative example. In the example of a sample identification flow according to a comparative example (FIG. 3A), first, clot waveform data acquired by coagulation reaction measurement is read (FIG. 3A-S301a). Next, preprocessing such as smoothing is performed (FIG. 3A-S302a). Next, normalization processing (sometimes expressed as correction processing) is performed (FIG. 3A-S303a). The normalization processing is performed only on the light quantity axis. Next, the time derivative (sometimes expressed simply as differentiation) of the waveform data or further time derivative is calculated to calculate a waveform related to the coagulation velocity or coagulation acceleration (FIG. 3A-S304a). Finally, the waveform parameters are extracted as features (FIG. 3A-S305a), and, for example, the identification and estimation of the test sample is performed based on the correlation between the features of a sample whose cause of prolongation of blood coagulation time is known and the features of the test sample. At this time, preprocessing and normalization processing of the light quantity axis are not necessarily required, and the flow may be skipped. Also, although not shown in the flow, there are cases where the axis representing the rate of change in light intensity of the waveform relating to coagulation speed or coagulation acceleration is normalized, and it is disclosed that normalization is performed by setting the maximum value of each waveform to 100%.

[0077] In contrast, the present disclosure is significantly different from the comparative example in that waveform fitting is performed, both the time axis and the light amount axis are normalized, and highly visible and intuitive identification estimation is possible from the extracted feature amount. In an example of a sample identification flow according to Example 1 (FIG. 3B), first, coagulation waveform (WaveSrc0) data acquired by coagulation reaction measurement is read (FIG. 3B-S301b). Next, WaveSrc1 is generated by a difference processing before and after WaveSrc0 (FIG. 3B-S302b). Next, Wave0 is generated by waveform fitting of WaveSrc0 (FIG. 3B-S303b), and Wave1 is generated by waveform fitting of WaveSrc1 (FIG. 3B-S304b). Next, Wave2 is generated by a difference processing before and after Wave1 (FIG. 3B-S305b). WaveSrc1 or Wave1 in the present disclosure corresponds to a waveform related to the coagulation velocity in the comparative example, and Wave2 corresponds to a waveform related to the coagulation acceleration in the comparative example. Next, Wave0, 1, and 2 are normalized based on the NRC method (FIG. 3B-S306b). The normalization based on the NRC method in the present disclosure is normalization based on a reaction constant, and means normalization of both the time axis and the light amount axis. As disclosed in the comparative example, when normalizing the light amount axis, the waveforms related to the clot waveform, the clot velocity, or the clot acceleration are not normalized by the maximum value of each waveform, but in the present disclosure, normalization is performed by making Wave0, 1, and 2 related to each other. Finally, feature values ​​are extracted from the normalized waveform (FIG. 3B-S307b), and the feature values ​​of the test specimen are applied to a two-dimensional map having a probability density distribution created using feature values ​​extracted from a specimen whose cause of prolongation of blood coagulation time is known, and the test specimen is identified and estimated. In the present disclosure, preprocessing such as smoothing processing may be performed appropriately, for example, after reading WaveSrc0 (FIG. 3B-S301b) and / or after generating WaveSrc1 (FIG. 3B-S302b).

[0078] The main inventive step of the present disclosure is that by normalizing the time axis direction using the reaction constant measured using the fitting model of WaveSrc0, and by introducing a difference using the time constant contained in the reaction constant (hereinafter referred to as the reaction time constant difference) instead of the time derivative of the clot waveform (simply described as derivative) disclosed in Patent Documents 1 to 4 and Non-Patent Document 1, it becomes possible to estimate the cause of prolonged blood clotting time while reducing the influence of the severity of symptoms expressed by the length of the blood clotting time.

[0079] [Example 2] In Example 2, identification of clot waveforms using a neural network (limited to an example using WaveSrc0 from the viewpoint of the uniformity of the invention) will be described.

[0080] Here, an embodiment is shown regarding a method for estimating the cause of prolongation of blood clotting time by a neural network using clot waveform data as input, based on the time axis normalization method disclosed in the present invention. As mentioned above, estimation of the cause of prolongation of blood clotting time by a neural network directly using clot waveform data has not been disclosed as a set of techniques that can be easily reproduced by a technician in the field, so the following three points are the gist of this embodiment. (1) Since the measured coagulation waveform is variable-length data, a preprocessing technique compatible with neural networks is disclosed. (2) To prove that neural network-based clot waveform clustering is practical, we aimed to demonstrate a detection accuracy of >90%. (3) In order to easily reproduce the above neural network and to facilitate its expansion to measurement data from various devices and samples from patients undergoing treatment, we present configuration parameters such as the number of nodes and the number of hidden layers, as well as hyper-parameters related to the learning conditions, such as the number of epochs and the number of batches.

[0081] (1) Disclosure of pre-processing First, we describe pre-processing techniques suitable for neural networks.

[0082] Figures 11A-C show experimental results showing the clot waveform (WaveSrc0) of a commercially available sample and the clot waveform preprocessed to fit the neural network. Figure 11A shows WaveSrc0 of a commercially available sample. Here, 103 waveforms were used, consisting of FVIII-deficient sample group (46 waveforms), FIX-deficient sample group (22 waveforms), LA-positive sample group (6 waveforms), and normal sample group (29 waveforms). The measuring device used for the measurements was a Hitachi Automated Analyzer 3500 (manufactured by Hitachi High-Tech Corporation). The vertical axis shows the count value of quantized data by an A / D converter corresponding to the scattered light intensity.

[0083] FIG. 11B shows a waveform obtained by processing WaveSrc0 by a zero padding method well known in the field of image and voice recognition as a preprocessing of a neural network. Here, the time axis was normalized according to Equation 4 of the time axis normalization using the reaction time constant and the response time, which is the main inventive step of the present disclosure, and the clot waveform was resampled by linear interpolation with the range of -5τ to +10τ being 501 points of equally spaced data. The vertical axis is normalized so that the value at the start of measurement is zero and the change in scattered light intensity of 10,000 counts is 1. This normalization is for adapting to a general sigmoid or hyperpublisher tangent as an activation function of a neural network. The normalization used here is for using the change in scattered light intensity as a feature amount, and is Equation 8, which is different from Equation 5 described above.

[0084]

number

[0085] In addition, for the time range of -5τ to +10τ, the measurement was stopped and there is no data, so the data length is made uniform by filling in the data "0".

[0086] FIG. 11C shows waveform data when a fitting model is used to complement missing data. By using the parameters τ, T0, Y0, and Y1 obtained by the fitting model of Equation 2, it is easy to obtain continuous numerical data even for the time after the measurement is completed. In this sense, the introduction of a fitting model can be said to have a high affinity for the detection of clot waveforms using a neural network. As a similar simple method, the missing data can be complemented by copying the value of the last measured data point. FIG. 11C differs from FIG. 11B in that the data indicating the scattered light intensity after the measurement end point is smooth. As preprocessing for use with a neural network, FIG. 11B and FIG. 11C, in which the data length is fixed, are suitable.

[0087] (2) Construction of neural networks and proof of their practical performance Next, the configuration of the neural network will be described.

[0088] FIG. 12 shows the structure of the neural network of the present disclosure. FIG. 12 is a summary by Keras included in TensorFlow, which is a general tool for studying neural networks. As can be seen in the figure, the neural network of the present disclosure can provide the practical performance shown above with a configuration of 4 hidden layers, 512 nodes, and tightly connected in all layers. The number of epochs for learning is 10,000, the batch size is 32, and the hyperpublic tangent is used as the activation function. As for the activation function, it should be easy for a technician in the field to understand that the same classification performance can be obtained even if a sigmoid is used instead of the hyperpublic tangent.

[0089] In the following explanation, the hyper-parameters are fixed at 10,000 epochs and 32 batch size. The clot waveform data of commercially available samples is divided into 60% for learning, 20% for validation, and 20% for testing, and cross-entropy is used as the loss function. The reason for dividing the data into three parts is explained below. The weights and other parameters that make up the neural network are automatically updated for the learning data, so the hyper-parameters such as the number of hidden layers, number of nodes, dropout, number of epochs, and batch size are appropriately selected so that the cross-entropy of the validation data is minimized. In order to objectively evaluate the performance of the neural network determined by these, it is necessary to evaluate the cross-entropy and the classification accuracy using test data that was not used in the above procedure.

[0090] Figures 13A-C show the experimental results showing the relationship between learning results and dropout. Here, the results are shown for the number of nodes = 256 and the number of hidden layers = 3. Figure 13A shows the results when using a clot waveform whose data length was made uniform by the zero padding method shown in Figure 11B. It can be seen that even if the amount of dropout, which is effective in suppressing overlearning, is changed, the classification accuracy of the test data never exceeds 90%.

[0091] Figure 13B shows the results when using a clot waveform preprocessed by data extrapolation using the fitting model shown in Figure 11C. As can be seen from the figure, an accuracy of 90% or more, which is a practical guideline for detection accuracy, can be obtained under conditions where the dropout amount is 0.4 or less. This proves that the preprocessing method disclosed herein and the sample detection method based on clot waveform analysis using a neural network configuration are detection methods that can be easily reproduced and expanded by engineers in the relevant field.

[0092] Figure 13C shows the results when using a clot waveform in which the data length was made uniform by complementation through copying the final data point of measurement. As above, it can be seen that the detection accuracy is 90% or more when the dropout amount is 0.4 or less. This method has the advantage of requiring less calculation compared to the complementation method shown in Figure 13B. When the amount of sample data is further increased or noise is mixed in during measurement for some reason, it can be said that data extrapolation complementation using a fitting model is superior in terms of the stability of detection performance. In the following, unless otherwise specified, preprocessing is assumed to be data extrapolation complementation using a fitting model.

[0093] (3) Present configuration parameters such as the number of nodes and the number of hidden layers, as well as hyper-parameters related to the learning conditions, such as the number of epochs and the number of batches. Next, parameter ranges suitable for implementing the present disclosure are described.

[0094] The number of epochs, dropout, and other factors suitable for neural network training are as described above. Here, the number of nodes and examples will be described as values ​​indicating the configuration of the neural network.

[0095] Figures 14A and 14B show experimental results showing the range of hyper-parameters where a classifier that combines the preprocessing method of the present disclosure with a neural network shows practical performance. Figure 14A shows the relationship between the number of hidden layers and the classification accuracy. Here, the number of nodes is set to 1024, and the dropout is set to 0.25. As can be seen from the figure, the classification method of the present disclosure can achieve a practical classification accuracy of 90% or more when the number of hidden layers is 3 or more.

[0096] FIG. 14B shows the experimental results showing the relationship between the number of nodes and the detection accuracy. Here, the number of hidden layers is 3. As can be seen from the figure, the detection method of the present disclosure can obtain a practical detection accuracy of 90% or more when the number of nodes is 16 or more. These results show that the detection method of the present disclosure can obtain a practical detection accuracy by configuring a neural network with 3 or more hidden layers and 16 or more nodes.

[0097] FIG. 15 shows experimental results indicating that the preprocessing method and the neural network classification method disclosed herein realize sufficient suppression of overlearning. Here, the relationship between the number of nodes and the cross entropy for learning, validation, and test samples is shown. In the fields of voice and face recognition, where neural networks are widely used, the cross entropy is generally about 0.1 to 0.3. In contrast, in the classification method disclosed herein, when the number of nodes is 16 or more, the cross entropy is about 0.1 or less, and it can be seen that the cross entropy values ​​are almost the same for each group of learning, validation, and test samples. This result indicates that the neural network configuration disclosed herein can sufficiently suppress overlearning.

[0098] As described above, according to the present disclosure, the clot waveform obtained from a sample is preprocessed to a predetermined data length using the reaction time constant and response time quantified by the fitting model, and then a neural network with 3 or more hidden layers and 16 or more nodes is used to estimate the cause of prolongation of blood clotting time with a practical detection accuracy. Needless to say, the neural network detector outputs the probability of belonging to each cluster of causes of prolongation of blood clotting time as an output, so it is self-evident that the neural network detector shown here can also display the probability of belonging as exemplified in FIG.

[0099] In addition, by applying the NRC method to the waveform data used for learning, it is possible to further improve the detection accuracy.

[0100] (Effects of Example 2) By performing suitable pre-processing on the clot waveform, which has a variable data length, it is possible to estimate the cause of prolonged blood clotting time with a practical degree of accuracy using a neural network.

[0101] [Example 3] In Example 3, application to a cross-mixing test will be described.

[0102] The above describes the identification of causes of blood clotting time extension using the clot waveform obtained by measuring a sample collected from a subject. On the other hand, a cross-mixing test, which measures the blood clotting time of a plurality of mixed plasma samples obtained by mixing the sample and a normal sample at a predetermined ratio when the blood clotting time of the sample collected from the subject is extended, and screens the causes of blood clotting time extension from the graph shape plotted against the mixing ratio of the sample, is recognized as a representative method for identifying causes of blood clotting time extension. According to the present disclosure, not only the mixing ratio dependency of the blood clotting time but also the plurality of feature amounts and the probability of belonging to each cause of blood clotting time included in the series of clot waveforms can be obtained from a series of clot waveforms obtained from a plurality of mixed plasma samples, so that the sample identification method of the present disclosure can be easily applied to clot waveform data obtained by the cross-mixing test, and is expected to improve the identification accuracy more than ever before. When emphasis is placed on compatibility with conventional methods, for example, a preferred method is to quantify the mixing ratio dependency of blood clotting time, and at the same time, quantify blood clotting factor VIII deficiency and blood clotting factor IX deficiency from the clot waveform of a sample collected from a subject, and map the quantification in two-dimensional space.

[0103] [Example 4] In the fourth embodiment, the feature quantities of the clot waveform obtained by normalization processing using the NRC method based on the reaction constant are expanded.

[0104] That is, in Example 4, in addition to the feature values ​​extracted from the first normalized waveform (WaveNor0), the second normalized waveform (WaveNor1), and the third normalized waveform (WaveNor2), feature values ​​are extracted from waveform data obtained by performing nonlinear operations on these waveforms, and the cause of the prolongation of the blood coagulation time of the test sample is inferred based on the extracted feature values ​​and distribution information of the feature values ​​for each cause of the prolongation of the blood coagulation time that has been prepared in advance.

[0105] The method for estimating the cause of prolongation of the blood coagulation time of the test sample in Example 4 is as follows: (A) acquiring a clot waveform showing a change in light intensity over time due to a clot reaction of a reaction solution produced by mixing a test sample, which is made of plasma obtained by separating blood obtained from a subject, with a reagent; (B) Obtaining a first waveform by subtraction processing of the clot waveform; (C) Obtaining a first fitted waveform by fitting the clot waveform; (D) obtaining a second fitted waveform by a fitting process on the first waveform; (E) Obtaining a third waveform by performing a front-back subtraction process on the second fitted waveform; (F) normalizing the light intensity axis and the time axis of each of the first fitted waveform, the second fitted waveform, and the third waveform to obtain a first normalized waveform, a second normalized waveform, and a third normalized waveform; (G) extracting a feature amount from at least one of the first normalized waveform, the second normalized waveform, the third normalized waveform, and waveform data obtained by performing a nonlinear operation on the first normalized waveform, the second normalized waveform, the third normalized waveform, and the third normalized waveform; (H) estimating a cause of prolongation of the blood clotting time of the test sample based on distribution information of the feature values ​​for each cause of prolongation of the blood clotting time prepared in advance and the feature values ​​extracted from the clot waveform of the test sample; has.

[0106] In the following, in order to simplify the drawings and aid in understanding the technology, the FVIII-deficient sample group will be referred to as "FVIII-deficient", the FIX-deficient sample group as "FIX-deficient", the LA-positive sample group as "LA", and the normal sample group as "Normal", with the following symbols. Here, the clot waveform measured by Hitachi Autoanalyzer 3500 (manufactured by Hitachi High-Tech Corporation) using commercially available samples is used. The measurement reagent is Coagpia APTT-N (manufactured by Sekisui Medical Co., Ltd.). In addition to the commercially available products described in Example 1, Coagpia Control PN I (manufactured by Sekisui Medical Co., Ltd.) and Coagtrol N (manufactured by Sysmex Corporation) were used as normal samples.

[0107] Figure 16 compares WaveSrc0 of one representative sample each of the FVIII-deficient sample and the FIX-deficient sample. As can be seen in Figure 16, the blood coagulation times of both are approximately the same at about 100 seconds. On the other hand, it can be seen that the FVIII-deficient sample has a slower convergence of the reaction (a longer time for the scattered light intensity to reach the saturation value from the initial value) compared to the FIX-deficient sample. This represents the difference in the influence of the deficient factors of both on the reaction process. This characteristic means that, for example, in Equation 2, T0 of both is approximately equal, and T1 is such that FIX deficiency < FVIII deficiency. In the present disclosure, (T0 / T1) was subtracted from vec0_x.

[0108] Note that the coagulation time is calculated as the time until the level reaches 50% when the initial value of the scattered light intensity is 0% and the saturation value is 100%, but the method for calculating the blood coagulation time is not limited to this method. For example, the time until reaching any arbitrary level other than 50% may be calculated as the blood coagulation time.

[0109] Figure 17A shows the experimental results of the relationship between the maximum value of the first derivative of the coagulation waveform and the log coagulation time. This maximum value of the first derivative corresponds to the index |min1| described in Patent Document 1. As shown in Figure 17A, the R2 value obtained by least-squares fitting with a quadratic function was 0.9323. That is, it can be seen that the maximum value of the first derivative has a very high correlation with the blood coagulation time. It can be seen that the maximum values of the first derivatives of the FVIII-deficient sample group, FIX-deficient sample group, and LA-positive sample group (LA) in the figure are distributed in a very small region compared to the overall change amount of the index including the normal sample group.

[0110] FIG. 17B shows the experimental results showing the relationship between vec1_y and log clotting time in Example 4. The R2 value obtained by least squares fitting with a quadratic function was 0.12. In other words, it can be seen that vec1_y in Example 4 has almost no correlation with blood clotting time. The distribution of vec1_y in the FVIII-deficient sample group, FIX-deficient sample group, and LA-positive sample group (LA) in the figure shows a large change compared to the overall change in the index including the normal sample group. Moreover, it can be seen that each sample group is separated into an independent region where the vec1_y values ​​have the relationship FVIII deficiency>LA>FIX deficiency.

[0111] 17A , the results were similar to those in Fig. 17A for the maximum value of the second derivative of the clot waveform obtained by scattered light measurement, which corresponds to another index |min2| described in Patent Document 1. As is clear from the above experimental facts, the normalization based on the NRC method disclosed herein generates a waveform independent of the blood clotting time and extracts feature amounts, thereby improving the performance of identifying the cause of prolonged blood clotting time compared to conventional indexes.

[0112] 18A shows the experimental results showing representative waveforms of each sample group for WaveNor0, which is obtained by standardizing the clot waveforms of the commercially available samples in Example 4 based on the NRC method. It can be seen that there is a large difference between each sample group in the rising portion of the waveform.

[0113] 18B shows the experimental results showing representative waveforms of each sample group for WaveNor1, which is obtained by standardizing the clot waveforms of the commercially available samples in Example 4 based on the NRC method. Similarly, it can be seen that the differences in the waveforms of each sample group are clearly evident.

[0114] 18C shows the experimental results showing representative waveforms of each sample group for WaveNor2, which is obtained by standardizing the clot waveforms of the commercially available samples in Example 4 based on the NRC method. Similarly, it can be seen that the differences in the waveforms of each sample group are clearly evident.

[0115] 19 shows the experimental results showing the frequency spectrum of WaveNor2 in Example 4. Here, for the representative specimen of each specimen group, a WaveNor2 waveform array of i=0 to 2047 was generated in the normalized time interval -50T1 to +50T1, and a Fourier transform was performed. For example, the amplitude spectrum values ​​of frequencies FREQ1 (HB minimum, i=40) and FREQ2 (HA maximum, i=115) can be added as the feature of each specimen group.

[0116] FIG. 20 summarizes an example of a feature set in the fourth embodiment. In general, the more features there are, the more distinguishable the causes of prolongation of blood clotting time are from different perspectives, which is preferable. However, there are limitations to this. For example, when features α and β are present, it is not preferable to add a new feature γ=aα+bβ (a and b are constants) to the features. γ is a linear combination of α and β, and adding γ to the features does not increase the amount of information required for distinguishing, and distinguishing the causes of prolongation of blood clotting time from different perspectives is not possible. Furthermore, the addition of γ increases the influence of measurement noise and numerical calculation errors, which reduces the reliability of distinguishing the causes of prolongation of blood clotting time. On the other hand, the waveform data obtained by mapping the normalized waveform (WaveNor0, WaveNor1, WaveNor2) to another space (converting it into new information) by nonlinear operations such as power operations, differential operations, Fourier transform operations, Galois transform operations, and convolution operations is easy to add effective features because it is guaranteed to be mathematically orthogonal to the original data. Of course, it is also possible to select features by a method of listing multiple feature candidates from the same waveform data, calculating the cross-correlation coefficients of all combinations of those, and eliminating those with large cross-correlation coefficients (which add less information for discrimination). Below, an example is shown in which values ​​obtained from waveform data mapped into separate independent spaces are added. As shown in the figure, here, two quantities from each of the seven feature spaces (#0 to #6) can be used as features. (1) Feature Space #0 In the first embodiment, these are vec0_x-(T0 / T1) and vec0_y. (2) Feature Space #1 These are vec1_x and vec1_y in the first embodiment. (3) Feature Space #2 These are vec2_x and vec2_y in the first embodiment. (4) Feature Space #3 This is the RMS error between WaveNor2 and the representative waveform of the FIX-deficient sample group, and the RMS error between WaveNor2 and the representative waveform of the normal sample group. (5) Feature Space #4 These are the x and y coordinates of the intersection point between WaveNor1 and WaveNor2. (6) Feature Space #5 These are the amplitude spectrum values ​​of FREQ1 and FREQ2 in the frequency spectrum of WaveNor2 shown in FIG. (7) Feature Space #6 When the clotting time is represented by t50, the Severity_Index is a value defined by the empirical formula (9) below which indicates the severity, and is the RMS error between WaveNor1 and the representative waveform of the normal sample group.

[0117] The representative waveform may be an average waveform, which is obtained by normalizing waveforms obtained from a group of samples having similar blood coagulation factor activity levels based on the NRC method, and then averaging the normalized levels for each normalized time.

[0118]

number

[0119] In addition, in Fig. 20, for convenience in implementing numerical analysis, the x and y ranges recommended for illustrating each feature space are added. In the following, in Example 4, the diagrams will be disclosed according to these ranges.

[0120] In the fourth embodiment, as a feature discrimination process, discrimination according to a probability density distribution based on a two-dimensional mixed Gaussian distribution is performed in each feature space. Since engineers in this field are familiar with the basic mixed Gaussian distribution, it will not be described here. In the fourth embodiment, in order to improve the reliability of the discrimination result, a probability density distribution of each cluster is used with an upper limit of the Mahalanobis distance applied, based on a mixed Gaussian distribution in which the influence of measurement error can be physically contained according to the central limit theorem. The Mahalanobis distance is an amount equivalent to the standard deviation of a one-dimensional Gaussian distribution, and if σ is used as a unit according to convention, the upper limit is preferably about 3 to 6σ. By applying the upper limit of the Mahalanobis distance, the basic shape of each cluster area in the feature space becomes an ellipse. When two clusters are adjacent to each other, the boundary is a condition where the probability of belonging to each is 50%. Data of one sample is displayed as one point in each feature space, and at the same time, the probability of belonging to each cluster of each sample group is calculated. The discrimination process of the fourth embodiment is performed by averaging the probability of belonging to the clusters in each feature space, and the cluster and the probability of belonging of the sample group to which the sample belongs with the highest probability are obtained as the discrimination result. In this embodiment, an example was shown in which the probability of belonging to each sample group cluster of the sample was obtained by averaging the probability of belonging to the sample group clusters in each feature space, but for example, the probability of belonging to the clusters in each feature space may be weighted and the sample may be classified as the sample group with the highest probability of belonging, or it may be classified from the probability of belonging to the sample group cluster in any of the feature spaces.

[0121] The imposition of an upper limit on the Mahalanobis distance introduced in Example 4 means that, in light of the standard deviation of the measurement data, for example, a data point that is more than 3σ away from the center of a specimen group cluster is not allowed to belong to that cluster. In discrimination using general statistical distributions, SVM, K-means, deep neural networks, etc., the space is divided into a specified number of clusters, so new or abnormal data is classified into one of the known clusters.

[0122] Data for each cluster is generated by extracting two feature data from each of seven feature spaces using clot waveform data from FVIII-deficient samples (FVIII deficiency), FIX-deficient samples (FIX deficiency), LA-positive samples (LA), and normal samples (Normal), and calculating distribution data consisting of the center position, height, and variance-covariance matrix to form a cluster for each sample group.

[0123] A note on feature space #6 will be added. FIG. 21 shows the experimental results of displaying the commercially available sample group of Example 4 on feature space #6. As shown in FIG. 21, this uses the Severity_Index shown in Equation 9 as the x-axis feature, and is specialized in distinguishing between normal samples and others, whereas other feature spaces distinguish between clusters of four sample groups. Among the commercially available samples prepared, it was confirmed that Borderline Factor Assay Control Plasma (manufactured by George King Bio-Medical, Inc.) with a clotting time of 42 seconds and an LA-positive sample with a clotting time of 46 seconds could be separated at a Mahalanobis distance of 3σ or more, and almost perfect distinction was possible. The FVIII and FIX activity values ​​of the Borderline Factor Assay Control Plasma were 41% and 55%, respectively, according to the attached assay values, and since this sample could not be classified into any of the LA positive sample group, the coagulation factor inhibitor positive sample group, the FVIII deficient sample group, and the FIX deficient sample group, for convenience in this example, it was treated as being in the range of normal samples. The normal range can be adjusted by changing the upper limit of the Mahalanobis distance.

[0124] FIG. 22A is a display example showing a summary of the commercially available specimen group of Example 4 plotted on a feature space. In FIG. 22A, the commercially available specimen group used in the experiment of Example 4 is plotted on seven feature spaces, and WaveNor1 obtained by normalization and a summary of the accuracy rate of the discrimination results are displayed. Here, the upper limit of the Mahalanobis distance is set to 4σ. The accuracy rate was 100% (106 / 106*100%).

[0125] Figure 22B is a display example of the discrimination result of one FVIII-deficient sample in Example 4. In each of the feature spaces in Example 4, the sample is represented as a single point, and the probability of belonging to each cluster of the sample group is displayed. In this case, it is FVIII-deficient (83%), LA (17%), FIX-deficient (0%), and Normal (0%).

[0126] [Example 5] In Example 5, a specific method for applying the technology of the present invention to a cross-mixing test will be described along the outline shown in Example 3. A series of sample specimens obtained by mixing plasma and normal plasma obtained by separating blood obtained from a subject at a plurality of mixing ratios are obtained by extracting feature values ​​from the first normalized waveform (WaveNor0), the second normalized waveform (WaveNor1), and the third normalized waveform (WaveNor2) for each sample specimen, and feature values ​​are extracted from the waveform data obtained by performing nonlinear calculation on these feature values, and the cause of the extension of the blood coagulation time of the test specimen is estimated using at least one of these feature values.

[0127] Regarding the cross-mixing test, as described in paragraph 0121 of Patent Document 1, "The changes in the graph patterns of LA-positive samples and coagulation factor inhibitor-positive samples were similar. Therefore, it is clear that the discrimination between the two based on the changes in the graph patterns must be a qualitative evaluation, and it is difficult for non-experts to make a judgment." It is a well-known fact that it is difficult to distinguish between LA-positive samples and coagulation factor inhibitor-positive samples based on the relationship between the mixing ratio of normal plasma and test plasma and the coagulation time. In this embodiment, the objective is to provide a method that applies the solving technology of problem (1) to the cross-mixing test, and enables not only the discrimination between LA-positive samples and coagulation factor inhibitor-positive samples, but also the quantitative discrimination of the cause of prolonged blood coagulation time, including FVIII-deficient samples and FIX-deficient samples, even if the person is not an expert.

[0128] The method for estimating the cause of prolongation of the blood coagulation time of the test sample in Example 5 is as follows: (A) preparing a series of sample specimens by mixing plasma obtained by separating blood obtained from a subject and normal plasma at multiple mixing ratios; (B) Acquiring a series of clot waveforms showing the change in light intensity over time due to the clot reaction of a reaction solution produced by mixing a series of sample specimens and a reagent; (C) Obtaining a first series of waveforms by pre-post subtraction of each of the series of clot waveforms; (D) Obtaining a series of first fitted waveforms by performing a fitting process on each of the series of clot waveforms; (E) obtaining a series of second fitted waveforms by a fitting process for each of the series of first waveforms; (F) Obtaining a series of third waveforms by performing pre- and post-subtraction on each of the series of second fitted waveforms; (G) normalizing the light intensity axis and the time axis of each of the series of first fitted waveforms, the series of second fitted waveforms, and the series of third waveforms to obtain a series of first normalized waveforms, a series of second normalized waveforms, and a series of third normalized waveforms; (H) extracting a feature amount from at least one of a series of first normalized waveforms, a series of second normalized waveforms, a series of third normalized waveforms, and a series of waveform data obtained by performing a nonlinear operation on the series of first normalized waveforms, the series of second normalized waveforms, the series of third normalized waveforms, and the series of waveform data obtained by performing a nonlinear operation on the series of first normalized waveforms, the series of second normalized waveforms, the series of third normalized waveforms, and the series of waveform data obtained by performing a nonlinear operation on the series of first normalized waveforms, the series of third normalized waveforms, and the series of waveform data obtained by performing a nonlinear operation on the series of first normalized waveforms, the series of third normalized waveforms, and the series of waveform data obtained by performing a nonlinear operation on the series of third normalized waveforms, ... third normal (I) estimating a cause of prolongation of blood clotting time of a sample specimen using at least one of a series of features extracted from a series of clot waveforms; has. Moreover, the computer system such as the control computer 120 described in the first embodiment executes each of the above-mentioned processes (A) to (I).

[0129] In the following, in this embodiment, to simplify the diagrams and aid in understanding the technology, the FVIII deficient sample group will be referred to as "FVIII deficient", the FIX deficient sample group as "FIX deficient", the LA positive sample group as "LA", and the coagulation factor inhibitor positive sample group (here, a commercially available product in which a coagulation factor inhibitor has been added to the FVIII deficient sample group) as "FVIII deficient inh" respectively. Commercially available samples were used in the experiment. Specifically, they are as follows. The FVIII deficient sample was Factor VIII Deficient Plasma (manufactured by Precision BioLogic, Inc.). The FIX deficient sample was Factor IX Deficient Plasma (manufactured by Precision BioLogic, Inc.). The LA positive samples were Lupus Positive Control and Weak Lupus Positive Control (manufactured by Precision BioLogic, Inc.). The coagulation factor inhibitor positive specimens were Mild Factor VIII Inhibitor Plasma and Strong Factor VIII Inhibitor Plasma (manufactured by Affinity Biologicals, Inc.). The normal specimen was Pooled Normal Plasma (manufactured by Precision BioLogic, Inc.). Here, FVIII deficient specimens, FIX deficient specimens, LA positive specimens, and coagulation factor inhibitor positive specimens were used as test specimens. In this example, the data from repeated measurements were treated as one specimen, and a total of 34 test specimens (FVIII deficient: 7, FIX deficient: 7, LA: 8, FVIII deficient inh: 12) were considered. The measuring device was Hitachi Automated Analyzer 3500 (manufactured by Hitachi High-Tech Corporation), and the reagent was Coagpia APTT-N (manufactured by Sekisui Medical Co., Ltd.).

[0130] In this example, a series of plasma samples were prepared by mixing the test specimen and normal specimen at different ratios (10:0, 9:1, 8:2, 5:5, 2:8, 1:9, 0:10), and the clot waveform was measured immediately after sample preparation (immediate type) and after the sample was heated at 37°C for 2 hours (delayed type). For one test specimen, a series of 14 clot waveforms (seven series of immediate and delayed types) were measured, and the feature values ​​extracted from the waveforms normalized based on the NRC method were calculated. In the following, the seven mixing ratios (0.0, 0.1, 0.2, 0.5, 0.8, 0.9, 1.0) of the test specimen and the measurement conditions (immediate or delayed) are used to represent the mixed plasma samples.

[0131] FIG. 23 is a schematic diagram showing a method for dimensional compression of a series of clotting time data obtained for one test specimen in the cross-mixing test of Example 5 into two area indices F1 and F2. In the cross-mixing test, there are conditions under which the change in blood clotting time is small relative to the change in the mixing ratio of the test specimen. In such cases, the blood clotting time data obtained varies due to being strongly influenced by the mixing conditions and measurement errors. Therefore, the average value of the two clotting times (instant and delayed) when the mixing ratio is 0.0 is calculated as t s The average of the two clotting times (immediate and delayed) when the mixing ratio is 1.0 is t e The difference from the linear approximate coagulation time is quantified as area indices F1 and F2 based on the following formula, and the variance of the measurement results is reduced by dimensionality reduction and averaging to obtain an index. In the following formula, n is an integer (n=1, 2), N is the number of mixture ratios (here, N=7), i is an index integer representing the mixture ratio, t i is the clotting time, t^ i is s and e The linear clotting time, Δx i is the i-th mixture ratio x i is the integration interval corresponding to x i-1 , x i+1 The width connecting the midpoints of i-1 =0.0, rightmost case x i+1 = 1.0). In the experimental conditions used in this example, the mixing ratio x iFor Δx = (0.0, 0.1, 0.2, 0.5, 0.8, 0.9, 1.0), i =(0.05, 0.1, 0.2, 0.3, 0.2, 0.1, 0.05).

number

[0132] When the instantaneous measurement index F1 and the delayed measurement index F2 are calculated according to the above formula, for example, even if the mixing ratios used in the instantaneous measurement and the delayed measurement are different, indices of the same dimension can be obtained, with -1≦F1≦+1 and -1≦F2≦+1.

[0133] FIG. 24 shows the results of discrimination of samples in a cross-mixing test using the multidimensional mixed Gaussian distribution of Example 5. Here, a case where a series of 14 test samples are prepared and tested by changing the mixing ratio of one original sample and a normal sample, and the time conditions of immediate and delayed are described. The method disclosed in Example 1 is a case where one clot waveform is linked to one original test sample. If this is expanded to a case where 14 clot waveforms are linked to one original test sample, 28 features are quantified per feature space. In each feature space, the two-dimensional mixed Gaussian distribution shown in Example 1 is expanded to a 28-dimensional mixed Gaussian distribution to quantify the discrimination accuracy. In addition, as a conventional example, raw data of blood clotting time is discriminated with a 14-dimensional mixed Gaussian distribution, and the dimensionally compressed indexes (F1, F2) shown above are discriminated with a two-dimensional mixed Gaussian distribution, and the results are summarized. As shown in Fig. 24, the discrimination accuracy was in the range of 32.4% to 88.2%, with the lowest discrimination accuracy when using the raw data of blood clotting time and the highest discrimination accuracy when using the indicators (F1, F2). Although the two are essentially the same, as described above, (F1, F2) with reduced influence of variation can obtain higher discrimination accuracy, so that in the cross-mixing test, the preparation conditions of the series of mixed plasma samples, etc. are a highly difficult measurement that cannot be handled by simply expanding the method disclosed in Example 1.

[0134] When the indices of each feature space (28 dimensions) obtained by normalization based on the NRC method of this embodiment were used, the maximum discrimination accuracy of 76.5% was obtained in feature space #3. It is believed that the influence of measurement variability is also strongly involved in this.

[0135] FIG. 25A is a diagram showing the cluster regions of the indicators (F1, F2) of Example 5. As can be seen from the figure, the FVIII deficiency cluster and the FIX deficiency cluster are formed close to each other so as to almost overlap, and although they are independent clusters as blood coagulation factor deficiency groups, it is difficult to distinguish with high accuracy which of the FVIII factor and the FIX factor is deficient. Although it is difficult to specify which blood coagulation factor is deficient, it is possible to determine that it is a factor deficiency type, as is the characteristic of the cross-mixing test, which is a conventional method. In addition, the LA positive cluster (LA) is formed overlapping the center of the coagulation factor inhibitor positive cluster (FVIII deficiency inh), and the experimental results reproduce the description in paragraph 0121 of Patent Document 1 that "the changes in the graph patterns of the LA positive specimen and the coagulation factor inhibitor positive specimen were similar. Therefore, it is found that the distinction between the two based on the changes in the graph pattern is inevitably a qualitative evaluation, and it is difficult to judge unless the person is an expert."

[0136] FIG. 25B shows the experimental results when data points were added to the cluster regions of the indicators (F1, F2) of Example 5. As can be seen from the figure, for the coagulation factor inhibitor positive cluster (FVIII deficiency inh), the prepared commercial samples have two types of inhibitor titers, "Mild" and "Strong," and are distributed on the lower left and upper right sides of the LA positive cluster (LA). Therefore, assuming a patient sample between "Mild" and "Strong," FVIII deficiency inh is approximately equal to LA by this method, and differentiation is considered impossible.

[0137] FIG. 26 is a diagram for explaining the indices (G1, G2) of Example 5. As described above, it is difficult to separate FVIII deficiency clusters from FIX deficiency clusters by using only blood clotting time. Therefore, the index is determined to be the length of a perpendicular line drawn from a position vector to data P of 100% of the test specimens, with the midpoint of the line segment connecting the center of the FVIII deficiency cluster and the center of the FIX deficiency cluster in each feature space as the origin. The indices (G1, G2) corresponding to immediate measurement and delayed measurement are defined by the following equation.

[0138]

number

[0139] In the above formula, the center of the FVIII deficiency cluster is +0.5, and the center of the FIX deficiency cluster is -0.5, and the larger the value, the more likely the sample is to be FVIII deficient.

[0140] FIG. 27A is a diagram (display example) showing the experimental results of plotting immediate (F1, G1) and delayed (F2, G2) indexes F calculated from a series of clot waveforms obtained from one sample in Example 5 and index G calculated from the data in feature space #1 using the same sample according to Equation 11. Here, one sample is represented as two points (F1, G1) and (F2, G2) and a straight line connecting them. The appearance is close to a schematic diagram of a dipole. Hereinafter, this will be referred to as a dipole. As can be seen in the figure, the clusters of the four sample groups (FVIII deficiency, FIX deficiency, LA, FVIII deficiency inh) are formed in almost independent regions. The upper limit of the Mahalanobis distance was set to 6σ. The reason why the region of FVIII deficiency inh is relatively wide is because, as mentioned above, there are two types of inhibitor titers, "Mild" and "Strong".

[0141] FIG. 27B is a diagram (display example) showing the experimental results in which the indices (F1, G1) and (F2, G2) obtained from 34 samples in Example 5 are plotted. Here, the index G is calculated according to Equation 11 from the data for feature space #1. It can be seen that the dipoles representing each sample are plotted separately for each cause of prolongation of blood clotting time. The discrimination accuracy was 97%. It can be seen that the "Mild" and "Strong" of the coagulation factor inhibitor positive group (FVIII deficiency inh), the LA positive group (LA), the FVIII deficiency group, and the FIX deficiency group are plotted spatially separated. It was confirmed that this method significantly improves the discrimination accuracy compared to the conventional discrimination using only blood clotting time.

[0142] Next, as a specific example of a suitable method for applying the technology of the present invention to a cross-mixing test, a method for discriminating the results of the cross-mixing test using feature spaces #0 to #5 will be shown.

[0143] In the feature space, seven-point sequence obtained by immediate measurement and delayed measurement of one sample are connected in ascending or descending order of the mixture ratio of the test sample, and are referred to as trajectory data here. FIG. 28A is an experimental result (display example) showing the average shape of trajectory data for each cause of blood clotting time extension obtained by the cross-mixing test of Example 5. As can be seen from the figure, the shape of the trajectory data is characteristic for each cause of blood clotting time extension. For example, the FVIII deficiency group and the coagulation factor inhibitor positive group (FVIII deficiency inh) are close in terms of morphology (shape similarity), but the coagulation factor inhibitor positive group (FVIII deficiency inh) has a large deviation between immediate (Wait=0h) and delayed (Wait=2h), which can be judged to be a correct result in light of the essence of the cross-mixing test.

[0144] One effective method for identifying causes of prolonged blood clotting time from trajectory data is to apply handwritten character recognition technology such as CNN (Convolutional Neural Network). In this method, each point that constitutes the trajectory data is connected using straight lines or spline interpolation to generate image data, which is then discriminated using the same method as handwritten character recognition. This method is capable of identifying many causes of prolonged blood clotting time, not only in terms of the clusters of the four sample groups handled in this study, but also in terms of the condition of patients during treatment (including medication, etc.).

[0145] As another method for identifying causes of prolonged blood clotting time from trajectory data, a multidimensional mixture Gaussian distribution can be used. As shown in Figure 24, it has been found that this method does not provide high accuracy. The main reason is that, according to the above conditions, the clusters of each sample group are defined as inside a 28-dimensional solid, but the mathematical restriction that this solid must be a Gaussian distribution no matter which surface it is sliced ​​through is deemed to be insufficient in terms of degrees of freedom to be applied to the identification of trajectory data from cross-mixing tests. As is well known, a multidimensional mixture Gaussian distribution must be non-degenerate, but the restriction is relaxed in a 2-dimensional mixture Gaussian distribution. Therefore, by separating the 28-dimensional trajectory data into a 2-dimensional x14 space and using the product of the probability of belonging to the cluster of the sample group obtained in each space to identify causes of prolonged blood clotting time, it is possible to identify causes of prolonged blood clotting time without being bound by the restriction of the continuity of the trajectory data with respect to changes in the mixture ratio of the sample.

[0146] FIG. 28B shows the experimental results (display example) in which the trajectory data of each specimen in the cross-mixing test was separated into a 2-dimensional x 14 space and discrimination was performed. Here, discrimination was performed from the probability of belonging to a cluster of normalized specimen groups based on the product of the belonging probabilities obtained in individual spaces, and the trajectory data was plotted divided into four regions. In the figure, the result of feature space #4 is shown as an example. A significant improvement was demonstrated compared to FIG. 24 (discrimination accuracy = 32.4% to 88.2%), which is the result of simply extending the method disclosed in Example 1, and the discrimination accuracy was 100%.

[0147] The complicated process of sample preparation and incubation required for the cross-mixing test places a burden on the examiner. In addition, additional blood sampling for the test places an additional burden on the patient. Therefore, the fewer the sample conditions (total number of samples prepared due to differences in mixing ratio and incubation) used in the cross-mixing test, the better.

[0148] FIG. 29 is a diagram summarizing the conditions for determining the total number of specimens required for the differentiation method of Example 5. As mentioned above, this time, 14 clot waveforms (seven series of immediate and delayed types) were obtained for one specimen. Each row indicates the conditions for the clot waveform, and hatched conditions indicate that they were not used in the clot waveform analysis. In the figure, "Condition A" indicates the case where all 14 clot waveforms are used, and "Condition B" indicates the conditions for differentiation using the seven clot waveforms excluding the hatched waveforms.

[0149] FIG. 30 is a schematic diagram showing the relationship between the number of clot waveforms used in the discrimination in Example 5 and the worst probability (the minimum value of Pti across all samples when the probability of the i-th sample belonging to the correct sample group cluster is Pti). If the worst probability on the vertical axis is 50% or more, discrimination errors will not occur, and if it is 50% or less, discrimination errors will occur. Here, the performances of three methods were compared: (1) discrimination using indices F1 and F2 (F Factor Method), (2) discrimination using a combination of indices F1, G1, F2, and G2 (FG Factor Method), and (3) discrimination dividing orbit data into two-dimensional sets (Orbit Discrimination Method). For (1) the F Factor Method, the worst probability was 50% or less under all conditions, and discrimination errors always occurred. For (2) the FG Factor Method, discrimination performance was improved, and if five or more clot waveforms were used, the worst probability was 50% or more. (3) The Orbit Discrimination Method had the best performance, and the worst probability was 77% or more when two or more clot waveforms were used.

[0150] The F Factor Method uses an index that represents the relationship between the mixing conditions of the test sample and normal sample and the blood coagulation time according to the above definition. Therefore, its discrimination performance can be considered to be equivalent to the discrimination results of the conventional cross-mixing test.

[0151] In the FG Factor Method, in addition to the indicators F1 and F2, the shape characteristics of the clot waveform quantified by the NRC method of the present invention are taken into account as indicators G1 and G2, thereby improving the discrimination performance as shown in the figure. In addition, the amount of data to be processed is small, at four per sample, and information processing calculations can be performed in a shorter time than with the Orbit Discrimination Method. Therefore, the FG Factor Method is suitable for use as auxiliary information when judging the discrimination results, in conjunction with the conventional method of discriminating the cause of prolonged blood coagulation time from the relationship between the mixing ratio of the sample and the blood coagulation time.

[0152] The Orbit Discrimination Method distinguishes the results of the cross-mixing test only by the index of the feature space normalized and quantified by the NRC method of the present invention, and can obtain high discrimination performance. At the same time, since there is no calculation of the index F by the formula 10, Δx i This enables information calculation independent of the relationship with adjacent data points via the orbit discrimination method, and allows for flexible selection of any mixing ratio. Due to these advantages, the orbit discrimination method is suitable as a means of realizing fully automatic discrimination in cross-mixing tests using machine learning technology.

[0153] 31 is a display example showing the discrimination results of the cross-mixing test of Example 5. The relationship between the analysis unit 130, the control computer 120, and the communication interface 124, and the display of the test sample results on the display unit 118c follow the results of FIG.

[0154] According to Example 5, a series of clot waveforms obtained by a cross-mixing test enables quantitative discrimination based on probability into LA-positive sample group, coagulation factor inhibitor-positive sample group, FVIII-deficient sample group, and FIX-deficient sample group.

[0155] [Variations] The present disclosure is not limited to the above-mentioned embodiments, and includes various modified examples. The above-mentioned embodiments have been described in detail to clearly explain the present disclosure, and are not necessarily limited to those having all of the configurations described. It is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. In addition, it is possible to add, delete, or replace the same configuration or other configurations with respect to a part of the configuration of each embodiment.

[0156] For example, in the first embodiment, the cause of the prolongation of the blood clotting time is estimated from the feature amounts of WaveNor0, 1, and 2, but in the present disclosure, the cause of the prolongation of the blood clotting time may be estimated from at least two feature amounts of WaveNor0, 1, and 2. For example, the cause of the prolongation may be estimated from the feature amounts of the x component and the y component of WaveNor0, or the cause of the prolongation may be estimated from the feature amounts of the x component of WaveNor0 and the y component of WaveNor2. [Explanation of symbols]

[0157] 100: automatic analyzer, 101: specimen dispensing mechanism, 102: specimen disk, 103: specimen container, 103a: specimen, 104: reaction container, 105: specimen syringe pump, 106: reagent dispensing mechanism, 107: reagent disk, 108: reagent container, 108a: reagent, 109: reagent heating mechanism, 110: reagent syringe pump, 111: reaction container stock section, 112: reaction container transport mechanism, 113: detection unit, 114: reaction container installation section, 115: light source, 116: detection section (optical sensor), 117: reaction container disposal section, 118: operation computer, 118a: mouse, 118b: keyboard, 118c: display section, 119: memory section, 120: control computer, 121: A / D converter, 122: incubator, 123: printer, 124: communication interface, 125: analysis computer, 130: analysis unit

Claims

1. Acquiring a clot waveform showing a change in light intensity over time due to a clot reaction of a reaction solution produced by mixing a test specimen, which is made of plasma obtained by separating blood obtained from a subject, with a reagent; Obtaining a first waveform by performing pre-post differential processing on the clot waveform; Obtaining a first fitted waveform by fitting the clot waveform; Obtaining a second fitted waveform by performing a fitting process on the first waveform; obtaining a third waveform by performing a front-back subtraction process on the second fitting waveform; obtaining a first normalized waveform, a second normalized waveform, and a third normalized waveform by normalizing a light amount axis and a time axis of each of the first fitted waveform, the second fitted waveform, and the third waveform; extracting a feature amount from each of the first normalized waveform, the second normalized waveform, and the third normalized waveform; and and estimating a cause of the prolongation of the blood coagulation time of the test sample based on known feature amounts extracted from a group of samples for which the cause of the prolongation of the blood coagulation time is known and the feature amounts extracted from each of the first normalized waveform, the second normalized waveform, and the third normalized waveform. A method for estimating the cause of prolonged blood coagulation time.

2. The normalization is based on a numerical value obtained by fitting the clot waveform. The method for estimating the cause of prolongation of blood coagulation time according to claim 1.

3. The step of predicting the cause of the prolongation of the blood clotting time includes predicting that the cause of the prolongation of the blood clotting time of the test sample is blood clotting factor VIII deficiency, blood clotting factor IX deficiency, or lupus anticoagulant (LA) positivity. The method for estimating the cause of prolongation of blood coagulation time according to claim 1.

4. The blood coagulation factor VIII deficiency refers to a state in which the activity of the blood coagulation factor VIII is less than 1%, and the blood coagulation factor IX deficiency refers to a state in which the activity of the blood coagulation factor IX is less than 1%. The method for estimating the cause of prolongation of blood coagulation time according to claim 3.

5. performing a first filtering process to remove noise from the clot waveform before the fitting process for the clot waveform; and and performing a second filtering process to remove noise from the first waveform before the fitting process for the first waveform. The method for estimating the cause of prolongation of blood coagulation time according to claim 1.

6. In the second filtering process, a moving average process is performed using a reaction time constant, which is one of the parameters obtained by the fitting process for the clot waveform. The method for estimating the cause of prolongation of blood coagulation time according to claim 5.

7. and displaying the estimated cause of the extension on a display unit. The method for estimating the cause of prolongation of blood coagulation time according to claim 1.

8. creating a map showing the relationship between known feature amounts extracted from the sample group for which the cause of prolongation of blood clotting time is known and the cause of prolongation of blood clotting time; and The method further comprises displaying the map and information obtained by plotting the feature amount of the test sample on the map on a display unit. The method for estimating the cause of prolongation of blood coagulation time according to claim 1.

9. and displaying on a display unit a result of the diagnosis based on the estimated cause of the prolongation of the blood coagulation time of the test sample. The method for estimating the cause of prolongation of blood coagulation time according to claim 1.

10. An information processing device capable of communicating with an automatic analyzer that acquires a clot waveform that indicates a change in light quantity over time due to a clot reaction of a reaction solution generated by mixing a test specimen, which is made of plasma obtained by separating blood obtained from a subject, with a reagent, the information processing device comprising: The information processing device includes a computer system having a processor and a memory; The computer system includes: A process of acquiring the clot waveform from the automated analyzer; A process of obtaining a first waveform by performing pre-post differential processing on the clot waveform; A process of obtaining a first fitted waveform by performing a fitting process on the clot waveform, and obtaining a second fitted waveform by performing a fitting process on the first waveform; A process of obtaining a third waveform by performing a front-back difference process on the second fitting waveform; a process of obtaining a first normalized waveform, a second normalized waveform, and a third normalized waveform by normalizing a light quantity axis and a time axis of each of the first fitted waveform, the second fitted waveform, and the third waveform; A process of extracting a feature amount from each of the first normalized waveform, the second normalized waveform, and the third normalized waveform; and and estimating a cause of prolongation of the blood coagulation time of the test sample based on known feature amounts extracted from a group of samples for which the cause of prolongation of the blood coagulation time is known and the feature amounts extracted from each of the first normalized waveform, the second normalized waveform, and the third normalized waveform.

23. An information processing apparatus comprising:

11. The computer system includes: Normalization is performed based on the numerical value obtained by fitting the clot waveform.

11. The information processing apparatus according to claim 10,

12. The process of estimating the cause of the prolongation of the blood coagulation time includes a process of estimating that the cause of the prolongation of the blood coagulation time of the test sample is blood coagulation factor VIII deficiency, blood coagulation factor IX deficiency, or lupus anticoagulant (LA) positivity.

11. The information processing apparatus according to claim 10,

13. The blood coagulation factor VIII deficiency refers to a state in which the activity of the blood coagulation factor VIII is less than 1%, and the blood coagulation factor IX deficiency refers to a state in which the activity of the blood coagulation factor IX is less than 1%.

13. The information processing apparatus according to claim 12.

14. The computer system includes: a first filtering process for removing noise from the clot waveform before the fitting process for the clot waveform; and and performing a second filtering process to remove noise from the first waveform before the fitting process for the first waveform.

11. The information processing apparatus according to claim 10,

15. The computer system includes: In the second filtering process, a moving average process is performed using a reaction time constant, which is one of the parameters obtained by the fitting process for the clot waveform.

15. The information processing apparatus according to claim 14,

16. The device further includes a display unit that displays the estimated cause of the extension.

11. The information processing apparatus according to claim 10,

17. The computer system includes: A process of creating a map showing the relationship between the known feature amount extracted from the sample group, the cause of which is known, and the cause of the prolongation of the blood coagulation time; and and displaying the map and information obtained by plotting the feature amount of the test sample on the map on a display unit.

11. The information processing apparatus according to claim 10,

18. The method further includes a display unit that displays the result of the diagnosis based on the estimated cause of the prolongation of the blood coagulation time of the test sample.

11. The information processing apparatus according to claim 10,

19. A method for estimating a cause of prolongation of a blood coagulation time of a test specimen from a coagulation waveform showing a change in light intensity over time due to a coagulation reaction of a reaction solution produced by mixing a test specimen, which is made of plasma obtained by separating blood obtained from a test subject, with a reagent, comprising: Aligning the data length of the clot waveform data with respect to the time axis of the clot waveform; applying the clot waveform data having the same data length to a neural network to estimate a plurality of causes of prolongation of blood clotting time; and presenting a cause of prolongation of the blood coagulation time of the test sample based on the estimated probability of the cause of the prolongation of the blood coagulation time; A method for estimating the cause of prolonged blood coagulation time.

20. The neural network in the estimation method includes: A neural network with 16 or more nodes and 3 or more hidden layers. The method for estimating the cause of prolongation of blood coagulation time according to claim 19.

21. Acquiring a clot waveform showing a change in light intensity over time due to a clot reaction of a reaction solution produced by mixing a test specimen, which is made of plasma obtained by separating blood obtained from a subject, with a reagent; Obtaining a first waveform by performing pre-post differential processing on the clot waveform; Obtaining a first fitted waveform by fitting the clot waveform; Obtaining a second fitted waveform by performing a fitting process on the first waveform; obtaining a third waveform by performing a front-back subtraction process on the second fitting waveform; obtaining a first normalized waveform, a second normalized waveform, and a third normalized waveform by normalizing a light amount axis and a time axis of each of the first fitted waveform, the second fitted waveform, and the third waveform; extracting a feature amount from at least one of the first normalized waveform, the second normalized waveform, the third normalized waveform, and waveform data obtained by performing a nonlinear operation on the first normalized waveform, the second normalized waveform, the third normalized waveform, and the third normalized waveform; and and estimating a cause of the prolongation of the blood coagulation time of the test specimen based on distribution information of the feature amount for each cause of the prolongation of the blood coagulation time prepared in advance and the feature amount extracted from the coagulation waveform of the test specimen. A method for estimating the cause of prolonged blood coagulation time.

22. Preparing a series of sample specimens by mixing plasma obtained by separating blood obtained from a subject and normal plasma at multiple mixing ratios; acquiring a series of clot waveforms showing a change in light quantity over time due to a clot reaction of a reaction solution produced by mixing the series of sample specimens and a reagent; Obtaining a series of first waveforms by performing pre-post subtraction processing on each of the series of clot waveforms; Obtaining a series of first fitted waveforms by performing a fitting process on each of the series of clot waveforms; Obtaining a series of second fitted waveforms by performing a fitting process on each of the series of first waveforms; obtaining a series of third waveforms by performing pre-post differential processing on each of the series of second fitted waveforms; normalizing a light quantity axis and a time axis of each of the series of first fitted waveforms, the series of second fitted waveforms, and the series of third waveforms to obtain a series of first normalized waveforms, a series of second normalized waveforms, and a series of third normalized waveforms; extracting a feature amount from at least one of the series of first normalized waveforms, the series of second normalized waveforms, the series of third normalized waveforms, and a series of waveform data obtained by performing a nonlinear operation on the series of first normalized waveforms, the series of second normalized waveforms, the series of third normalized waveforms, and the ... third normalized waveforms, and and estimating a cause of prolongation of the blood clotting time of the sample specimen using at least one of the series of feature amounts extracted from the series of clot waveforms. A method for estimating the cause of prolonged blood coagulation time.

23. The feature quantity extracted from the series of waveform data is An index G calculated from a position where a feature amount including a plurality of components obtained from a clot waveform of a sample specimen having a 100% plasma mixing ratio obtained by separating blood obtained from a subject among the series of sample specimens is plotted in a feature amount space showing the plurality of components. １ and an index F calculated from a series of coagulation times of the real-time measurements of the series of sample specimens. １ Combination with (F １ , G １ ), and An index G calculated from a position where the feature amount including the plurality of components obtained from the clot waveform of the delayed measurement of the sample specimen having a plasma mixing ratio of 100% is plotted in the feature amount space. ２ and an index F calculated from a series of clotting times of the delay measurements of the series of sample specimens. ２ Combination with (F ２ , G ２ ) The combination (F １ , G １ ) and the combination (F ２ , G ２ ) to estimate the cause of the prolongation of the blood coagulation time of the sample specimen. The method for estimating the cause of prolongation of blood coagulation time according to claim 22.

24. The feature quantity extracted from the series of waveform data is a first trajectory data obtained by plotting a series of feature quantities including a plurality of components obtained from a series of clot waveforms of real-time measurements of the series of sample specimens in a feature quantity space representing the plurality of components, and connecting the series of feature quantities in the feature quantity space in ascending or descending order of the plasma mixing ratio; and A series of feature quantities including the plurality of components obtained from a series of clot waveforms of delay measurements of the series of sample specimens are plotted in the feature quantity space, and the series of feature quantities in the feature quantity space are linked in ascending or descending order of the plasma mixing ratio, A cause of the prolongation of the blood coagulation time of the sample specimen is estimated using the first trajectory data and the second trajectory data. The method for estimating the cause of prolongation of blood coagulation time according to claim 22.

25. An information processing device capable of communicating with an automatic analyzer that acquires a clot waveform that indicates a change in light quantity over time due to a clot reaction of a reaction solution generated by mixing a test specimen, which is made of plasma obtained by separating blood obtained from a subject, with a reagent, the information processing device comprising: The information processing device includes a computer system having a processor and a memory; The computer system includes: A process of acquiring the clot waveform from the automated analyzer; A process of obtaining a first waveform by performing pre-post differential processing on the clot waveform; A process of obtaining a first fitted waveform by fitting the clot waveform; A process of obtaining a second fitted waveform by fitting the first waveform; A process of obtaining a third waveform by performing a front-back difference process on the second fitting waveform; a process of obtaining a first normalized waveform, a second normalized waveform, and a third normalized waveform by normalizing a light quantity axis and a time axis of each of the first fitted waveform, the second fitted waveform, and the third waveform; A process of extracting a feature amount from at least one of the first normalized waveform, the second normalized waveform, the third normalized waveform, and waveform data obtained by performing a nonlinear operation on the first normalized waveform, the second normalized waveform, the third normalized waveform, and the third normalized waveform; and and estimating a cause of the prolongation of the blood coagulation time of the test specimen based on distribution information of the feature amounts for each cause of the prolongation of the blood coagulation time prepared in advance and the feature amounts extracted from the coagulation waveform of the test specimen.

23. An information processing apparatus comprising:

26. A process of creating a graph showing a relationship between distribution information of the feature amount for each cause of prolongation of the blood coagulation time prepared in advance and the feature amount extracted from the clot waveform of the test sample; and and displaying the graph and the estimated cause of the prolongation of the blood coagulation time of the test sample on a display unit.

26. The information processing apparatus according to claim 25,

27. An information processing device capable of communicating with an automatic analyzer that prepares a series of sample specimens by mixing plasma and normal plasma obtained by separating blood obtained from a subject at a plurality of mixing ratios, and obtains a series of coagulation waveforms that show a change in light quantity over time due to a coagulation reaction of a reaction solution produced by mixing the series of sample specimens with a reagent, The information processing device includes a computer system having a processor and a memory; The computer system includes: acquiring the series of clot waveforms from the automated analyzer; A process of obtaining a series of first waveforms by performing pre-post subtraction processing on each of the series of clot waveforms; A process of obtaining a series of first fitted waveforms by performing a fitting process on each of the series of clot waveforms; obtaining a series of second fitted waveforms by performing a fitting process on each of the series of first waveforms; obtaining a series of third waveforms by performing a front-back subtraction process on each of the series of second fitted waveforms; a process of obtaining a series of first normalized waveforms, a series of second normalized waveforms, and a series of third normalized waveforms by normalizing a light quantity axis and a time axis of each of the series of first fitted waveforms, the series of second fitted waveforms, and the series of third waveforms; A process of extracting a feature amount from at least one of the series of first normalized waveforms, the series of second normalized waveforms, the series of third normalized waveforms, and a series of waveform data obtained by performing a nonlinear operation on the series of first normalized waveforms, the series of second normalized waveforms, the series of third normalized waveforms, and the ... third normalized waveforms, and the series of waveform data obtained by performing a nonlinear operation on the series of third normalized waveforms, and and performing a process of estimating a cause of prolongation of the blood clotting time of the sample specimen using at least one of the series of feature amounts extracted from the series of clot waveforms.

23. An information processing apparatus comprising:

28. The feature quantity extracted from the series of waveform data is An index G calculated from a position where a feature amount including a plurality of components obtained from a clot waveform of a sample specimen having a 100% plasma mixing ratio obtained by separating blood obtained from a subject among the series of sample specimens is plotted in a feature amount space showing the plurality of components. １ and an index F calculated from a series of coagulation times of the real-time measurements of the series of sample specimens. １ Combination with (F １ , G １ ), and An index G calculated from a position where the feature amount including the plurality of components obtained from the clot waveform of the delayed measurement of the sample specimen having a plasma mixing ratio of 100% is plotted in the feature amount space. ２ and an index F calculated from a series of clotting times of the delay measurements of the series of sample specimens. ２ Combination with (F ２ , G ２ ) The process of estimating the cause of the extension is performed by using the combination (F １ , G １ ) and the combination (F ２ , G ２ ) to estimate the cause of the prolongation of the blood coagulation time of the sample specimen, The index G １ and the index F １ and the index G ２ and the index F ２ and creating a graph showing the relationship between and displaying the graph and the estimated cause of the prolongation of the blood coagulation time of the sample specimen on a display unit.

28. The information processing apparatus according to claim 27,

29. The feature quantity extracted from the series of waveform data is a first trajectory data obtained by plotting a series of feature quantities including a plurality of components obtained from a series of clot waveforms of real-time measurements of the series of sample specimens in a feature quantity space representing the plurality of components, and connecting the series of feature quantities in the feature quantity space in ascending or descending order of the plasma mixing ratio; and A series of feature quantities including the plurality of components obtained from a series of clot waveforms of delay measurements of the series of sample specimens are plotted in the feature quantity space, and the series of feature quantities in the feature quantity space are linked in ascending or descending order of the plasma mixing ratio, the process of estimating a cause of the prolongation is a process of estimating a cause of the prolongation of the blood coagulation time of the sample specimen by using the first trajectory data and the second trajectory data, creating a graph showing the first trajectory data and the second trajectory data; and and displaying the graph and the estimated cause of the prolongation of the blood coagulation time of the sample specimen on a display unit.

28. The information processing apparatus according to claim 27,

Images