Disease or health status detection system
By extracting biomarkers from biological samples, ionizing and generating three-dimensional profiles, combined with the learning algorithm of the recognizer, the problem of insufficient disease recognition accuracy in the prior art is solved, and higher disease recognition accuracy and multiple types are achieved Accurate diagnosis of the disease.
Patent Information
- Application Number
- CN202280066428.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2021-11-30
- Filing Date
- 2022-11-28
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-11-28
AI Technical Summary
The prior art uses principal component analysis to identify diseases, the accuracy is insufficient and cannot meet higher identification needs.
Using a disease or health status examination method based on biomarkers in biological samples, the biomarkers are removed from the biological samples, and ionized to flow between the two electrodes to generate a three-dimensional profile and use it with the disease A type-associated recognizer is used to learn to improve recognition accuracy.
Achieve higher disease recognition accuracy, able to identify a variety of diseases and health status, including non-disease, mild and severe cancers, and improve the accuracy of clinical diagnosis.
Smart Images

Figure CN118043659B_ABST
Abstract
Description
Technical Field
[0001] A method and system for examining a disease or health condition. Background Art
[0002] As shown in Non-Patent Document 1, there is known a method for identifying a predetermined disease based on a biomarker in a biological sample. Summary of the invention
[0003] Problems to be solved by the invention
[0004] When principal component analysis is used to identify a disease as described in Non-Patent Document 1, the accuracy of identifying a predetermined disease becomes higher than that of conventional clinical markers, but it is still insufficient and higher accuracy is required.
[0005] Means for solving problems
[0006] The method for inspecting a disease or health state of the first viewpoint is a method for inspecting a disease or health state based on a biomarker in a biological sample. The method for inspecting a disease or health state comprises a first step, a second step, a third step, and a fourth step. The first step is to extract a biomarker from a biological sample. The second step is to ionize the extracted biomarker. The third step is to make the ionized biomarker flow between two electrodes, provide a plurality of electrical signals to the two electrodes, and thereby obtain a three-dimensional profile. The fourth step is to use an identifier obtained by learning by associating the three-dimensional profile with the type of disease to identify a specified disease.
[0007] The disease or health status inspection method of the first aspect can identify a predetermined disease with higher accuracy by using an identifier obtained by learning by associating a three-dimensional profile with the type of disease.
[0008] The disease or health state examination method according to the second aspect is the disease or health state examination method according to the first aspect, wherein the identifier simultaneously identifies a plurality of diseases.
[0009] The method for examining a disease or health condition according to the third aspect is the method for examining a disease or health condition according to any one of the first aspect or the second aspect, wherein the biological sample is at least one of exhaled breath, serum, blood, saliva, plasma, nipple aspirate, synovial fluid, cerebrospinal fluid, sweat, urine, stool, tears, skin emanations, tracheal washing material, material collected with a cotton swab, needle aspirate, semen, vaginal secretions, and pre-ejaculatory material.
[0010] The disease or health condition examination method according to a fourth aspect is the disease or health condition examination method according to any one of the first to third aspects, wherein the biological sample is exhaled breath or urine.
[0011] The method for examining a disease or health state according to the fifth aspect is any one of the methods for examining a disease or health state according to the first to fourth aspects, wherein the biomarker is at least one of a gene biomarker, a cell biomarker, an organic compound biomarker, a metabolic biomarker, a carbohydrate biomarker, a lipid biomarker, a heterocyclic biomarker, an elemental compound biomarker, an image biomarker, an anthropological biomarker, a personal habit biomarker, a disease state biomarker, and an expression biomarker.
[0012] The method for examining a disease or a health state according to a sixth aspect is the method for examining a disease or a health state according to any one of the first to fifth aspects, wherein the biomarker is an organic compound biomarker produced by metabolism in a living body.
[0013] The method for examining a disease or a health state according to a seventh aspect is the method for examining a disease or a health state according to the sixth aspect, wherein the organic compound biomarker is a volatile organic compound.
[0014] The method for examining a disease or a health state according to an eighth aspect is the method for examining a disease or a health state according to the seventh aspect, wherein the volatile organic compound is a volatile substance group.
[0015] The disease or health state examination method of a ninth aspect is the disease or health state examination method of any one of the first to eighth aspects, wherein the identifier distinguishes non-disease, mild cancer, and severe cancer.
[0016] The method for examining a disease or a healthy state according to the tenth aspect is the method for examining a disease or a healthy state according to any one of the first to ninth aspects, wherein the third step utilizes changes in ion mobility in a low electric field and a high electric field to obtain a three-dimensional profile.
[0017] The disease or health state inspection system of the 11th viewpoint is a disease or health state inspection system based on biomarkers in a biological sample. The disease or health state inspection system comprises a first device, a second device, and a third device. The first device extracts a biomarker from the biological sample. The second device ionizes the extracted biomarker. The second device causes the ionized biomarker to flow between two electrodes, provides a plurality of electrical signals to the two electrodes, and thereby obtains a three-dimensional profile. The third device uses an identifier obtained by learning by associating the three-dimensional profile with the type of disease to identify a specified disease.
[0018] The disease or health condition inspection system of the eleventh aspect can identify a predetermined disease with higher accuracy by using an identifier obtained by learning by associating a three-dimensional profile with the type of disease. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 It is a structural diagram of the inspection system.
[0020] Figure 2 This is a diagram showing an example of a FAIMS profile.
[0021] Figure 3 It is a graph used to compare the FAIMS profiles of disease patients and healthy people.
[0022] Figure 4 It is a control block diagram of the identification device.
[0023] Figure 5 is a diagram showing the structure of a recognizer.
[0024] Figure 6 It is a figure which shows the recognition result based on principal component analysis.
[0025] Figure 7 It is a diagram showing the recognition result by the identifier.
[0026] Figure 8 This is a flowchart for explaining an example of processing of the inspection system. DETAILED DESCRIPTION
[0027] (1) Overall structure
[0028] Figure 1 1 is a block diagram of a disease or health state inspection system 1. The inspection system 1 identifies a predetermined disease based on a biomarker 50 in a biological sample 40.
[0029] like Figure 1 As shown, the inspection system 1 includes a sampling device 10 (first device), an analysis device 20 (second device), and an identification device 30 (third device).
[0030] (2) Detailed structure
[0031] (2-1) Sampling device
[0032] The sampling device 10 extracts the biomarker 50 from the biological sample 40 .
[0033] The biological sample 40 is at least one of exhaled air, serum, blood, saliva, plasma, nipple aspirate, synovial fluid, cerebrospinal fluid, sweat, urine, stool, tears, skin emanation, tracheal washings, cotton swab collection, needle aspirate, semen, vaginal secretions, and pre-ejaculatory material. The biological sample 40 in this embodiment is exhaled air or urine.
[0034] Biomarker 50 is at least one of a gene biomarker, a cell biomarker, an organic compound biomarker, a metabolic biomarker, a carbohydrate biomarker, a lipid biomarker, a heterocyclic biomarker, an elemental compound biomarker, an image biomarker, an anthropology biomarker, a personal habit biomarker, a disease state biomarker, and an expression biomarker. The biomarker 50 in this embodiment is an organic compound biomarker produced in the metabolism in the organism. In addition, in this embodiment, VOC (Volatile Organic Compound) is used as an organic compound biomarker. In other words, the VOC in this embodiment is a volatile group (volatile metabolites).
[0035] like Figure 1 As shown, sampling device 10 allows VOC in biological sample 40 to be adsorbed by an adsorbent, thereby extracting VOC from biological sample 40. Furthermore, sampling device 10 allows VOC to be desorbed from the adsorbent, thereby allowing VOC to flow into analysis device 20 uniformly (stably) for a certain period of time.
[0036] (2-2) Analysis device
[0037] The analysis device 20 uses FAIMS (Field Asymmetric Ion Mobility Spectrometry) to analyze the VOCs extracted from the biological sample 40 by the sampling device 10. The analysis device 20 is, for example, Lonestar from Owlstone.
[0038] FAIMS is an analytical technique that utilizes the change in ion mobility between low and high electric fields.
[0039] The analysis device 20 has a processor such as a CPU and a GPU, and a storage device such as a RAM, a ROM, and a HDD. In addition, the analysis device 20 has a keyboard and a mouse for inputting various commands and various information. In addition, the analysis device 20 has a monitor for displaying the FAIMS profile 21 described later. In addition, the analysis device 20 has a network interface device for communicating with the recognition device 30 via a communication line not shown in the figure.
[0040] The analysis device 20 first ionizes the VOCs taken out from the biological sample 40 by the sampling device 10. Then, the analysis device 20 allows the ionized VOCs to flow between two electrodes and provides a plurality of electrical signals to the two electrodes, thereby generating a FAIMS profile 21.
[0041] The FAIMS profile 21 records the current value of cations and the current value of anions flowing between two electrodes for each combination of DF (Dispersion Field) and CV (Compensation Voltage). Figure 2 This is a diagram showing an example of the FAIMS profile 21 . Figure 2 In FIG. 2 , the FAIMS profile 21 is represented by a two-dimensional arrangement in which the first dimension (vertical axis) is a dispersion electric field (Dispersion Field. Hereinafter, sometimes referred to as DF) and the second dimension (horizontal axis) is a compensation voltage (Compensation Voltage. Hereinafter, sometimes referred to as CV). Figure 2 The left figure is a FAIMS profile 21 recording the current value of the cation (hereinafter, sometimes referred to as the FAIMS profile 21 of the cation). Figure 2 The right figure is a FAIMS profile 21 recording the current value of negative ions (hereinafter, sometimes referred to as the FAIMS profile 21 of negative ions). Figure 2 In the example, the DF of "0% to 100%" is divided into 51 stages, and the CV of "-6V to +6V" is divided into 512 stages. In other words, Figure 2 The FAIMS profiles 21 of the cations and anions in are two-dimensional arrays having a size of "51×512", respectively. Figure 2 In FIG. 2 , the magnitude of the current value of each block of the FAIMS profile 21 is expressed by the depth of the color.
[0042] It is known that when biological samples 40 are collected from disease patients and healthy persons, the concentration of specific substances contained in VOCs extracted from the biological samples 40 differs between the disease patients and the healthy persons.
[0043] For example, Non-Patent Document 2 shows that when exhaled air is collected from colorectal cancer patients and healthy subjects, the concentrations of the substances shown in Table 1 below contained in VOCs extracted from the exhaled air are different between the patients and the healthy subjects.
[0044]
Table 1
[0045]
[0046]
[0047] Furthermore, Non-Patent Document 3 shows that when exhaled air is collected from colorectal cancer patients and healthy subjects, the concentrations of the substances shown in Table 2 below contained in VOCs extracted from the exhaled air are different between the patients and the healthy subjects.
[0048]
Table 2
[0049]
[0050] Furthermore, Non-Patent Document 4 shows that when urine is collected from colorectal cancer patients and healthy subjects, the concentrations of the substances shown in Table 3 below contained in VOCs extracted from the urine differ between the patients and the healthy subjects.
[0051]
Table 3
[0052]
[0053] Furthermore, Non-Patent Document 5 shows that when urine is collected from colorectal cancer patients and healthy subjects, the concentrations of the substances shown in Table 4 below contained in VOCs extracted from the urine differ between the patients and the healthy subjects.
[0054]
Table 4
[0055]
[0056] Furthermore, Non-Patent Document 6 shows that when exhaled breath is collected from liver cancer patients and healthy subjects, the concentrations of the substances shown in Table 5 below contained in VOCs extracted from the exhaled breath differ between the patients and the healthy subjects.
[0057]
Table 5
[0058]
[0059] Furthermore, Non-Patent Document 7 shows that when exhaled air is collected from pancreatic cancer patients and healthy subjects, the concentrations of the substances shown in Table 6 below contained in the VOCs extracted from the exhaled air are different between the patients and the healthy subjects.
[0060]
Table 6
[0061]
[0062] Such a difference in the concentration of a specific substance included in VOC affects the FAIMS profile 21. Specifically, the distribution of the current value of the FAIMS profile 21 differs between disease patients and healthy people according to the difference in the concentration of the specific substance included in VOC. Figure 3 This is a diagram for comparing the FAIMS profile 21 of disease patients and healthy persons. Figure 3 In this study, the biological sample was urine, and the FAIMS profiles of patients with kidney disease and healthy subjects were compared. Figure 3 In the area to the left of the arrow in FIG. 1 , the left side is the FAIMS profile 21 of a patient with kidney disease, and the right side is the FAIMS profile 21 of a healthy person. In addition, the upper side is the FAIMS profile 21 of a cation, and the lower side is the FAIMS profile 21 of an anion. Figure 3In the area to the right of the arrow in FIG. 1 , a graph showing the difference between the FAIMS profile 21 of a kidney disease patient and a healthy person is drawn for the FAIMS profile 21 of cations and anions, respectively. In the graph showing the difference, the magnitude of the difference in the current value of each block of the FAIMS profile 21 is expressed by the depth of the color. In the graph showing the difference, the depth of the color can be seen, so it can be seen that the distribution of the current value of the FAIMS profile 21 is different between kidney disease patients and healthy people.
[0063] Next, the analysis device 20 stacks the generated FAIMS profiles 21 of the cations and anions in the third dimension to obtain a three-dimensional profile 22 (three-dimensional arrangement). Figure 2 In the case of , the analysis device 20 obtains a three-dimensional profile 22 having a size of “51×512×2” by stacking the FAIMS profiles 21 of the cations and the anions.
[0064] (2-3) Identification device
[0065] The recognition device 30 is a general computer and recognizes a predetermined disease using a recognizer 342 obtained by learning by associating the three-dimensional profile 22 with the type of disease.
[0066] Figure 4 30 is a control block diagram of the recognition device 30. Figure 4 As shown, the recognition device 30 mainly includes an input unit 31 , a display unit 32 , a storage unit 33 , a communication unit 34 , and a control unit 39 .
[0067] (2-3-1) Input section
[0068] The input unit 31 is a keyboard and a mouse. Various commands and various information to the recognition device 30 can be input using the input unit 31 .
[0069] (2-3-2) Display section
[0070] The display unit 32 is a monitor and can display the learning status and the like on the display unit 32 .
[0071] (2-3-3)Communication Department
[0072] The communication unit 34 is a network interface device for communicating with the analysis device 20 via a communication line (not shown).
[0073] (2-3-4) Storage unit
[0074] The storage unit 33 is a storage device such as a RAM, a ROM, and a HDD. The storage unit 33 stores programs executed by the control unit 39, data necessary for the execution of the programs, and the like.
[0075] The storage unit 33 particularly stores learning data 341 and an identifier 342 .
[0076] The learning data 341 is data for learning the identifier 342. Specifically, the learning data 341 is data that associates the three-dimensional profile 22 of the VOC extracted from the biological sample 40 with the type of disease of the person who collected the biological sample 40. The three-dimensional profile 22 is obtained from the analysis device 20 via the communication unit 34, for example. The type of disease is input by the user from the input unit 31, for example.
[0077] (2-3-5) Control Unit
[0078] The control unit 39 is a processor such as a CPU or a GPU. The control unit 39 reads and executes the program stored in the storage unit 33 to realize various functions of the recognition device 30. In addition, the control unit 39 can write the calculation results to the storage unit 33 or read out the information stored in the storage unit 33 according to the program.
[0079] The control unit 39 includes a learning unit 391 and a recognition unit 392 as functional blocks.
[0080] (2-3-5-1) Learning Department
[0081] The learning unit 391 uses the learning data 341 to learn the identifier 342. In the present embodiment, the size of the three-dimensional profile 22 of the learning data 341 is "51×512×2". In addition, the learning data 341 is data that corresponds the three-dimensional profile 22 to either a "prescribed disease patient" or a "healthy person". In other words, the identifier 342 identifies whether the person is a "prescribed disease patient" or a "healthy person" based on the three-dimensional profile 22. In addition, the identifier 342 uses a model that combines CNN (Convolutional Neural Network) and RNN (Recurrent Neural Network).
[0082] Figure 5 342 is a diagram showing the structure of the identifier 342. Figure 5 As shown, the identifier 342 is composed of an input layer, a first intermediate layer to a fourth intermediate layer, and an output layer.
[0083] Between the input layer and the first intermediate layer, two-dimensional convolution (Conv2D) is performed on the dimensions of DF and CV. As a result, the size of the data flowing through the identifier 342 is changed from "51×512×2" to "25×128×4".
[0084] Between the first and second intermediate layers, the permutations are taken out column by column (25 times in total) in the order of DF from small to large, and input into RNN (Conv1D+RNN). In each step of RNN, a one-dimensional convolution is performed on the dimension of CV, and the dimension is reduced. The FAIMS profile 21 has the continuity of the current value in the dimension of DF, so by inputting into the recursive layer (RNN), the parameters are reduced and the performance of the model is also improved. As a result, the size of the data flowing through the identifier 342 changes from "25×128×4" to "128×12".
[0085] Between the second intermediate layer and the third intermediate layer, a one-dimensional convolution (Conv1D) is performed on the dimension of CV. As a result, the size of the data flowing through the identifier 342 is changed from "128×12" to "32×6".
[0086] Between the third intermediate layer and the fourth intermediate layer, the dimension is reduced by average pooling. As a result, the size of the data flowing through the identifier 342 is changed from "32×6" to "6".
[0087] Between the 4th intermediate layer and the output layer, the probability of being a "patient with a prescribed disease" and the probability of being a "healthy person" are calculated by the fully connected layer. Based on the calculated probabilities, a prescribed threshold is used to identify whether the person is a "patient with a prescribed disease" or a "healthy person". As a result, the size of the data flowing through the identifier 342 changes from "6" to "2".
[0088] (2-3-5-2) Identification Unit
[0089] The identification unit 392 identifies a predetermined disease using the identifier 342 learned by the learning unit 391. In the present embodiment, the identifier 342 identifies whether the person is a "predetermined disease patient" or a "healthy person" based on the three-dimensional profile 22.
[0090] (3) Verification
[0091] In this verification, urine collected from “disease patients” and “healthy people” was used to compare the recognition accuracy of the conventional principal component analysis and the recognition device 342 .
[0092] Any identification method is the same until the three-dimensional profile 22 is obtained by the analysis device 20. In the principal component analysis, the three-dimensional profile 22 having a size of "51×512×2" is processed into a one-dimensional array having 52224 (=51×512×2) elements. In the identifier 342, as the learning data 341, data that associates the three-dimensional profile 22 with either "disease patient" or "healthy person" is used.
[0093] Figure 6 It is a figure which shows the recognition result based on principal component analysis. Figure 6In , "disease patients" and "healthy people" are plotted on a two-dimensional eigenspace composed of the first principal component PC1 and the second principal component PC2. The contribution rate of the first principal component PC1 is 20.4%, and the contribution rate of the second principal component PC2 is 12.3%. Figure 6 In the figure, the lighter colored plotted points represent “healthy people” and the darker colored plotted points represent “disease patients”. Figure 6 As shown in the figure, the analysis results are mixed with "disease patients" and "healthy people". Usually, in principal component analysis, the nonlinearity between feature values (current values) and the mixing of a small number of special sample data will damage the representativeness of the correlation coefficient and distort the principal components extracted thereafter. In addition, Figure 6 As shown in FIG. 1 , when only the first principal component PC1 and the second principal component PC2 that have a great influence on the sample data are plotted and identified, in the principal component analysis, since the influence of the remaining principal components is ignored, all feature quantities (current values) cannot be considered for identification. Therefore, in the principal component analysis, it is difficult to identify "disease patients" and "healthy people".
[0094] Figure 7 3 is a diagram showing the recognition result by the identifier 342 . Figure 7 The ROC curves of each time when learning by stratified ten-fold cross validation and the average ROC curve of these curves are depicted in FIG. Figure 7 As shown, the AUC (Area Under the Curve) of the average ROC curve is 0.86±0.08, so it can be known that the discriminator 342 can distinguish "disease patients" and "healthy people" with high accuracy. In the discriminator 342, all feature quantities (current values) constituting the FAIMS profile 21 are considered for identification, so that the identification accuracy can be improved.
[0095] (4) Processing
[0096] use Figure 8 The flowchart of FIG. 1 illustrates an example of the processing of the inspection system 1 .
[0097] As shown in step S1 , the sampling device 10 extracts VOCs from the biological sample 40 .
[0098] When step S1 is completed, as shown in step S2 , the analyzer 20 ionizes the VOC extracted from the biological sample 40 .
[0099] When step S2 is finished, as shown in step S3 , the analysis device 20 allows the ionized VOC to flow between the two electrodes, and provides a plurality of electrical signals to the two electrodes.
[0100] When step S3 is finished, as shown in step S4 , the analysis device 20 generates FAIMS profiles 21 of cations and anions.
[0101] When step S4 is finished, as shown in step S5 , the analysis device 20 stacks the FAIMS profiles 21 of the cations and the anions to obtain a three-dimensional profile 22 .
[0102] When step S5 is finished, as shown in step S6, the recognition device 30 uses the identifier 342 obtained by learning by associating the three-dimensional profile 22 with "prescribed disease patients" or "healthy people" to recognize whether it is a "prescribed disease patient" or a "healthy person".
[0103] (5) Features
[0104] (5-1)
[0105] As described in Non-Patent Document 1, a method for identifying a predetermined disease based on a biomarker in a biological sample is known.
[0106] However, if principal component analysis is used to identify a disease as in Non-Patent Document 1, the accuracy of identifying a predetermined disease becomes higher than that of conventional clinical markers. However, as shown in verification, this is still insufficient, and higher accuracy is required.
[0107] The inspection system 1 of the present embodiment is an inspection system for diseases or health conditions based on VOCs (biomarkers) in a biological sample 40. The inspection system 1 includes a sampling device 10, an analysis device 20, and an identification device 30. The sampling device 10 extracts VOCs from the biological sample 40. The analysis device 20 ionizes the extracted VOCs. The analysis device 20 causes the ionized VOCs to flow between two electrodes, and provides a plurality of electrical signals to the two electrodes, thereby obtaining a three-dimensional profile 22. In particular, the analysis device 20 obtains the three-dimensional profile 22 using FAIMS. The identification device 30 uses an identifier 342 obtained by learning by associating the three-dimensional profile 22 with the type of disease to identify a specified disease.
[0108] The inspection system 1 can identify a predetermined disease with higher accuracy by using the identifier 342 obtained by learning by associating the three-dimensional profile 22 with the type of disease.
[0109] (6) Modification
[0110] (6-1) Modification 1A
[0111] In the present embodiment, the learning data 341 is data in which the three-dimensional profile 22 is associated with either a “predetermined disease patient” or a “healthy person”.
[0112] However, the learning data 341 may also be data that associates the three-dimensional profile 22 with a plurality of diseases. For example, the learning data 341 is data that associates the three-dimensional profile 22 with any one of "colorectal cancer patients", "kidney cancer patients" and "healthy people". In this case, the identifier 342 can simultaneously identify a plurality of diseases by setting the number of nodes to be identified in the output layer.
[0113] (6-2) Modification 1B
[0114] In the present embodiment, the learning data 341 is data in which the three-dimensional profile 22 is associated with either a “predetermined disease patient” or a “healthy person”.
[0115] However, the learning data 341 may be data that associates the three-dimensional profile 22 with any one of "non-disease", "cancer stage I", "cancer stage II", "cancer stage III", and "cancer stage IV". In this case, the identifier 342 can identify "non-disease", "mild cancer (e.g., cancer stage I and II)", and "severe cancer (e.g., cancer stage III and IV)" by setting the number of nodes to be identified in the output layer.
[0116] (6-3)
[0117] As mentioned above, although embodiment of this disclosure was described, it should be understood that various changes can be made to the aspect and detail without departing from the meaning and scope of this disclosure described in the claims.
[0118] Explanation of symbols
[0119] 1. Check the system
[0120] 10 Device 1
[0121] 20 Device 2
[0122] 22 Three-dimensional profile
[0123] 30 Device 3
[0124] 342 Identifier
[0125] 40 Biological samples
[0126] 50 Biomarkers
[0127] S1Step 1
[0128] S2 Step 2
[0129] S3~5 Step 3
[0130] S6 Step 4
[0131] Prior art literature
[0132] Patent literature
[0133] Non-patent literature 1: Christopher Psutka, Marina Yamada, Akihisa Matsuda, Kazuya Yamahatsu, Satoshi Matsumoto, Toshihiko Kitayama, Nobuo Nakano, Jyunichi Koyano, Tohru Mikoshiba, Masao Miyashita “FAIMS technology in urinary volatile organic compound analysis to detect colorectal cancer” AACR 2017, Abstract No. 5303
[0134] Non-patent literature 2: D.F. Altomare, A. Picciariello, M.T. Rotelli, M. De Fazio, A. Aresta, C.G. Zambonin, L. Vincenti, P. Trerotoli, N. De Vietro “Chemical signature of colorectal cancer: case–control study for profiling the breath print” BJSOpen 2020; 4: 1189-1199
[0135] Non-patent literature 3: D F Altomare, M Di Lena, F Porcelli, L Trizio, E Travaglio, M Tutino, S Dragonieri, V Memeo, G de Gennaro “Exhaled volatile organic compounds identify patients with colorectal cancer” British Journal of Surgery 2013; 100: 144-150
[0136] 4:Ramesh P Arasaradnam,Michael J McFarlane,Courtenay Ryan-Fisher,Erik Westenbrink,Phoebe Hodges,Matthew G Thomas,Samantha Chambers,Nicola O'Connell,Catherine Bailey,Christopher Harmston,Chuka U Monday,KarnaD Bardhan,James A Covington“Detection of colorectal cancer(CRC)by urinaryvolatile organic compound analysis”PLOS ONE 2014:9:Issue 9e108750
[0137] Publication 5:Heena Tyagi,Emma Daulton,Ayman S.Bannaga,RameshP.Arasaradnam and James A.Covington“Non-Invasive Detection and Staging ofColorectal Cancer Using a Portable Electronic Nose”Sensors 2021:21:5440-
[0138] 6:Galen Miller-Atkins,Lou-Anne Acevedo-Moreno,David Grove,Raed A.Dweik,Adriano R.Tonelli,J.Mark Brown,Daniela S.Allende,FedericoAucejo,Daniel M.Rotroff“Breath Metabolomics Provides Accurate andNoninvasive Approach to Screening for Cirrhosis,Primary,and Secondary Liver Tumors”Hepatology Communications,Vol.4,No.7,2
[0139] Non-Patent Document 7: Andrea Princivalle, Lorenzo Monasta, Giovanni Butturini, Claudio Bassi & Luigi Perbellini, "Pancreatic ductal adenocarcinoma can be detected by analysis of volatile organic compounds (VOCs) in alveolar air", BMC Cancer volume 18, Article number: 529 (2018)
Claims
1. A disease or health state inspection system (1), which is a disease or health state inspection system (1) based on a biomarker (50) in a biological sample (40), wherein: The inspection system (1) comprises: a first device (10) for extracting the biomarker from the biological sample; The second device (20) is a high-field asymmetric waveform ion mobility spectrometer, which ionizes the extracted biomarker, allows the ionized biomarker to flow between two electrodes, and provides a plurality of electrical signals to the two electrodes, thereby obtaining a three-dimensional profile (22); and The third device (30) identifies a predetermined disease using an identifier (342) obtained by learning by associating the three-dimensional profile with the type of disease, The recognizer is a model that combines CNN and RNN. The three-dimensional profile records the current value according to each combination of the dispersion electric field (DF), the compensation voltage (CV) and the type of ions. The identifier has a migration between layers in which the CNN is applied to the dimension of the compensation voltage and the RNN is applied to the dimension of the dispersion electric field.
2. The disease or health status detection system (1) according to claim 1, wherein: The identifier identifies multiple diseases simultaneously.
3. The disease or health status detection system (1) according to claim 1 or 2, wherein: The biological sample (40) is at least one of a substance collected by a cotton swab and a substance aspirated by a needle.
4. The disease or health status detection system (1) according to claim 1 or 2, wherein: The biological sample (40) is at least one of exhaled air, serum, blood, saliva, plasma, nipple aspirate, synovial fluid, cerebrospinal fluid, sweat, urine, stool, tears, skin emanations, tracheal washings, semen, vaginal secretions, and pre-ejaculatory material.
5. The disease or health status detection system (1) according to claim 1 or 2, wherein: The biological sample (40) is exhaled breath or urine.
6. The disease or health status detection system (1) according to claim 1 or 2, wherein: The biomarker (50) is at least one of a gene biomarker, a cell biomarker, an organic compound biomarker, a metabolic biomarker, a carbohydrate biomarker, a lipid biomarker, a heterocyclic biomarker, an elemental compound biomarker, an image biomarker, an anthropology biomarker, a personal habit biomarker, a disease state biomarker, and an expression biomarker.
7. The disease or health status detection system (1) according to claim 1 or 2, wherein: The biomarker (50) is an organic compound biomarker produced during metabolism in an organism.
8. The disease or health status detection system (1) according to claim 7, wherein: The organic compound biomarker is a volatile organic compound.
9. The disease or health status detection system (1) according to claim 8, wherein: The volatile organic compounds are a group of volatile compounds.
10. The disease or health status detection system (1) according to claim 1 or 2, wherein: The identifier identifies non-disease, mild cancer, and severe cancer.
11. The disease or health status detection system (1) according to claim 1 or 2, wherein: The high-field asymmetric waveform ion mobility spectrometer obtains the three-dimensional profile by utilizing the change in ion mobility in a low electric field and a high electric field.
