Determination method, determination device, and computer program

Abstract

Provided are a determination method and the like capable of improving the accuracy with which a property of a sample is determined using a response profile detected by a sensor.　The determination method involves: acquiring a response profile, for a target sample, of a first sensor element that exhibits a responsiveness to a specific property of the sample, and a response profile, for the target sample, of a second sensor element that exhibits a lower responsiveness to the specific property of the sample than the first sensor element; correcting the response profile of the first sensor element corresponding to identification information about the target sample on the basis of a comparison between the response profile of the second sensor element corresponding to the identification information about the target sample and a reference profile indicating a reference response of the second sensor element; determining the property of the target sample on the basis of a comparison between the response profile of the first sensor element following correction and a reference profile indicating a reference response of the first sensor element; and storing the response profiles and the determination result.

Description

Determination method, determination device, and computer program

[0001] The present invention relates to a determination method, a determination device, and a computer program.

[0002] Olfactory sensors have been proposed as artificial devices that substitute for the sense of smell, one of the five human senses. For example, Patent Document 1 discloses a low-cost odor sensor that includes a transistor with a gate electrode containing aluminum or aluminum oxide, an insect cell with an olfactory receptor placed on the gate electrode, and a detection device that detects the current generated in the transistor when the insect cell responds to an odor.

[0003] Japanese Patent Application Laid-Open No. 2018-113957

[0004] However, even if the response profile is detected using the same odor sensor, there is a possibility that the response may differ for each detection target. The influence of such response differences reduces the accuracy of judgments using the response profile. This problem can occur not only with odor sensors but also with various other sensors in general.

[0005] An object of the present disclosure is to provide a determination method and the like that can improve the accuracy of determining the properties of a sample using a response profile detected by a sensor.

[0006] A determination method according to one aspect of the present disclosure includes a computer-implemented process of acquiring a response profile of a first sensor element, which exhibits reactivity in accordance with a specific property of the sample, to a target sample, and a response profile of a second sensor element, which exhibits a lower reactivity in accordance with the specific property of the sample than the first sensor element, correcting the response profile of the first sensor element, which corresponds to the identification information of the target sample, based on a comparison of the response profile of the second sensor element, which corresponds to the identification information of the target sample, with a reference profile, which indicates a reference response of the second sensor element; determining the properties of the target sample based on a comparison of the corrected response profile of the first sensor element with the reference profile, which indicates the reference response of the first sensor element; and storing the response profile of the first sensor element, the response profile of the second sensor element, and the determination result in association with the identification information of the target sample.

[0007] A determination device according to one aspect of the present disclosure includes a control unit that executes a process of acquiring a response profile of a first sensor element, which exhibits reactivity according to a specific property of the sample, to a target sample, and a response profile of a second sensor element, which exhibits a lower reactivity according to the specific property of the sample than the first sensor element, correcting the response profile of the first sensor element, which corresponds to the identification information of the target sample, based on a comparison of the response profile of the second sensor element, which corresponds to the identification information of the target sample, with a reference profile, which indicates a reference response of the second sensor element; determining the properties of the target sample based on a comparison of the corrected response profile of the first sensor element with the reference profile, which indicates the reference response of the first sensor element; and storing the response profile of the first sensor element, the response profile of the second sensor element, and the determination result in association with the identification information of the target sample.

[0008] A computer program according to one aspect of the present disclosure causes a computer to execute a process of acquiring a response profile of a first sensor element, which exhibits reactivity in accordance with a specific property of the sample, to a target sample, and a response profile of a second sensor element, which exhibits a lower reactivity in accordance with the specific property of the sample than the first sensor element, correcting the response profile of the first sensor element, which corresponds to identification information of the target sample, based on a comparison of the response profile of the second sensor element, which corresponds to identification information of the target sample, with a reference profile, which indicates a reference response of the second sensor element; determining the properties of the target sample based on a comparison of the corrected response profile of the first sensor element with the reference profile, which indicates the reference response of the first sensor element; and storing the response profile of the first sensor element, the response profile of the second sensor element, and the determination result in association with the identification information of the target sample.

[0009] According to the present disclosure, it is possible to improve the accuracy of determining the properties of a sample using a response profile detected by a sensor.

[0010] 1 is a schematic diagram of a determination system. FIG. 1 is a schematic diagram showing an example of the configuration of an olfactory sensor. FIG. 2 is a block diagram showing the configuration of a determination device. FIG. 3 is a diagram showing an example of the content of information stored in a detection DB. FIG. 4 is a block diagram showing the configuration of a terminal device. FIG. 5 is a diagram showing an example of changes in luminescence intensity of a sensor cell over time. FIG. 6 is a schematic diagram showing an example of a reception screen. FIG. 7 is a diagram showing an example of a reference profile related to cells selected for disease determination. FIG. 8 is a diagram showing an example of a reference profile related to cells not selected for disease determination. FIG. 9 is a schematic diagram showing an example of a result screen showing determination results. FIG. 10 is a schematic diagram showing another example of a result screen showing determination results. FIG. 11 is a flowchart showing an example of a reference profile generation process. FIG. 12 is a flowchart showing an example of a process procedure for determining the possibility of disease. FIG. 13 is a flowchart showing an example of a process procedure executed by a determination device of a second embodiment. FIG. 14 is a diagram showing an example of a response profile depending on whether or not a correction process is performed. FIG. 15 is a diagram showing an example of a response profile depending on whether or not a correction process is performed. FIG. 16 is a block diagram showing an example of the configuration of a determination device of a third embodiment. FIG. 17 is an explanatory diagram showing an overview of a learning model. FIG. 18 is a flowchart showing an example of a process procedure executed by a determination system of a third embodiment.

[0011] The present disclosure will be specifically described with reference to the drawings showing embodiments thereof.

[0012] (First embodiment) Fig. 1 is a schematic diagram of a determination system 100. The determination system 100 includes a determination device 1, a terminal device 2, a detection device 3, and an olfactory sensor 4. The olfactory sensor 4 is an example of a sensor. The determination device 1 is communicably connected to the terminal device 2 and the detection device 3 via a network N such as the Internet. The determination system 100 determines the health condition of a subject based on the detection results of odor molecules in a specimen derived from the subject detected by the detection device 3 and the olfactory sensor 4, and provides a service in which the determination device 1 presents the determination results via the terminal device 2.

[0013] In the following embodiments, an example will be described in which a subject's possible disease is determined based on odor molecules detected in the subject's urine. The types of diseases to be determined are not limited, but include, for example, infectious diseases such as influenza and COVID-19, cancer, type 2 diabetes, lifestyle-related diseases such as heart disease and cerebrovascular disease, and neurological diseases such as Parkinson's disease and Alzheimer's disease. When the disease is cancer, it may be any of various types, such as lung cancer, esophageal cancer, breast cancer, stomach cancer, liver cancer, pancreatic cancer, gallbladder cancer, bile duct cancer, colon cancer, kidney cancer, bladder cancer, ovarian cancer, uterine cancer, prostate cancer, oral cancer, and pharyngeal cancer.

[0014] The determination device 1 is an information processing device capable of various information processing and information transmission and reception, such as a server computer, a personal computer, a quantum computer, etc. The determination device 1 acquires a signal indicating an odor response to urine from a subject, determines the possibility of disease in the subject based on the acquired signal, and provides the determination result to a user via the terminal device 2. The determination device 1 may be a local computer provided in the facility where the detection device 3 is installed.

[0015] The terminal device 2 is an information processing terminal used by the subject, and is, for example, a personal computer, a smartphone, a tablet terminal, etc. The terminal device 2 displays the determination result received from the determination device 1. The subject is an example of a user who receives the determination result. The determination result is not limited to being provided to the subject, but may also be provided to, for example, medical professionals, personnel at the analytical institution that performs the analysis, etc. The number of terminal devices 2 connected to the determination device 1 may be one or three or more.

[0016] The olfactory sensor 4 and the detection device 3 are managed, for example, by an analytical laboratory that performs analysis of the subject. The olfactory sensor 4 includes cells 42 (see FIG. 2 ) that express olfactory receptors and outputs a signal indicating the response of the olfactory receptors to odor molecules. The cells 42 are an example of a sensor element. The detection device 3 detects the response signal from the olfactory sensor 4 over time. In addition to functioning as a detector that detects the response signal described above, the detection device 3 also functions as a computer that processes various data and communicates with external devices, and transmits detection data of the response signal obtained by detection to the determination device 1 via the network N. The detector and the computer may be provided separately and configured to be able to communicate with each other. The number of detection devices 3 connected to the determination device 1 may be one or three or more.

[0017] FIG. 2 is a schematic diagram showing an example configuration of the olfactory sensor 4. FIG. 2 is a view of the olfactory sensor 4 as seen from above. The olfactory sensor 4 includes a substrate 41, wells 43 formed in the substrate 41, and cells 42 arranged in the wells 43. For example, a "384-well plate" having 384 wells 43 formed therein is used as the substrate 41. For simplicity of illustration, FIG. 2 shows a "24-well plate" having 24 wells 43 formed therein. In this embodiment, cells 42 consisting of a plurality of cells of the same type are arranged in each of the multiple wells 43. That is, a large number of cells of the same type are seeded in each well 43 at high density. Note that the cells 42 may consist of a single cell. Each cell has an olfactory receptor. The cells 42 function as sensor cells that output signals indicating the response of the olfactory receptor to odor molecules. The olfactory receptor responds to specific odor molecules. By detecting a response signal based on the binding between the olfactory receptors provided in the olfactory sensor 4 and odor molecules, it is possible to quantitatively detect the odor contained in the subject's urine.

[0018] The substrate 41 is made of a material such as glass, silicon, ceramics, resin, or metal. The surface of the substrate 41 may be subjected to surface treatment such as plasma treatment, corona treatment, or UV-Ozone treatment, or may be coated with a polypeptide or the like. The shape of the substrate 41 may be any shape that allows odor detection using the cells 42 arranged on the substrate 41, such as a rectangular plate. The size of the substrate 41 is not particularly limited and can be set appropriately depending on the number of cells 42 arranged on the substrate 41, etc.

[0019] The olfactory receptor may be derived from an animal. Examples of animals include insects, vertebrates, and mammals, and examples of olfactory receptors that can be used include flies, mosquitoes, mice, rats, rabbits, cows, dogs, and humans. Insect olfactory receptors are preferred. Insect olfactory receptors are ionotropic receptors, and when they bind to odor molecules, an olfactory receptor complex formed by the olfactory receptor and an olfactory receptor co-receptor is activated, resulting in the influx of cations into the cell.

[0020] The amino acid sequences and coding sequences of olfactory receptors and olfactory receptor co-receptors are known or can be easily identified by sequence identity searches based on known sequences. Furthermore, amino acid mutations relative to the known amino acid sequences can be included. Amino acid mutations include, for example, amino acid substitutions, insertions, additions, or deletions.

[0021] The cells 42 may be specific cells that naturally express an olfactory receptor, or genetically modified cells into which an olfactory receptor gene has been incorporated. Genetically modified cells can be produced by transforming cells with a vector into which an olfactory receptor gene has been incorporated. When the olfactory receptor is an insect olfactory receptor, it is preferable to further incorporate a gene for an olfactory receptor co-receptor.

[0022] The cell 42 may further contain a fluorescent protein or a luminescent protein. When an odor molecule binds to an ionotropic olfactory receptor in the cell 42, cations such as calcium ions flow into the cell. By introducing into the cell 42 a gene that expresses a fluorescent protein whose fluorescence intensity changes depending on the ion concentration or a luminescent protein whose luminescence intensity changes, the response of the cell to the odor molecule can be detected by changes in fluorescence intensity or luminescence intensity. In other words, it is possible to detect odor molecules by changes in fluorescence intensity or luminescence intensity. Examples of such proteins include aequorin, Yellow Cameleon, and GCaMP.

[0023] A calcium ion-dependent fluorescent dye may be introduced into the cells 42. By introducing the calcium ion-dependent fluorescent dye into the cells, the influx of calcium ions into the cells due to the binding of odor molecules to olfactory receptors can be detected by changes in fluorescence intensity. Examples of such calcium ion-dependent fluorescent dyes include Fura-2, Fluo-3, and Fluo-4.

[0024] The detection device 3 detects a signal indicating a response of the cells 42 based on the binding between the olfactory receptors and odor molecules. The detection device 3 detects the luminescence of each well 43 in the olfactory sensor 4 as a detection target. In other words, the detection device 3 can detect a response signal for each cell 42 included in each well 43 and for each type of olfactory receptor possessed by the cell 42. The detection device 3 includes, for example, a photomultiplier tube, and detects fluorescence or luminescence based on changes in intracellular ion concentration. The response signal of the cells 42 detected by the detection device 3 is not limited to fluorescence or luminescence intensity, but may also be an electrical signal (electric potential) based on changes in intracellular ion concentration. The response signal may also be a moving image or still image of the luminescence of the cells 42 captured by an imaging device such as a CCD camera. The following describes an example in which the response signal is luminescence intensity.

[0025] As shown in Fig. 2, the olfactory sensor 4 of this embodiment includes a plurality of cells 42 arranged on a substrate 41. In the example shown in Fig. 2, the plurality of cells 42 are arranged in rows and columns at regular intervals on the upper surface of the substrate 41.

[0026] In general, olfactory receptors have selectivity for odor molecules. Therefore, by arranging multiple cells 42 expressing different olfactory receptors on a substrate 41 and detecting the responses of each of these cells 42, it is possible to detect multiple types of odors. A single cell 42 may have one type of olfactory receptor, or multiple types of olfactory receptors. The olfactory sensor 4 may include multiple identical cells 42. The olfactory sensor 4 of this embodiment includes different types of cells 42, each of which has one type of olfactory receptor.

[0027] The number, types, and arrangement of olfactory receptors used in the olfactory sensor 4 can be determined appropriately depending on the odor molecules to be detected and the type of disease to be diagnosed. The olfactory receptors may be a specific combination for the odor molecules to be detected, or a comprehensive combination of multiple olfactory receptors may be used for various odor molecules. For example, the number of cells 42 mounted on one olfactory sensor 4, i.e., the total number of sensor cells mounted on one olfactory sensor 4, can be from 1 to 2000, and the number of types of cells 42 mounted on one olfactory sensor 4 can be from 1 to 2000.

[0028] The method for detecting the response signal is not limited to the above example, and any appropriate method can be used depending on the olfactory receptors in the olfactory sensor 4 and the types of odor molecules sensed by the olfactory receptors.

[0029] 3 is a block diagram showing the configuration of the determination device 1. The determination device 1 includes a control unit 11, a storage unit 12, and a communication unit 13. The determination device 1 may be a single computer, or may be a computer system configured by multiple computers and peripheral devices. The determination device 1 may be a virtual machine whose entity is virtualized, or may be a cloud.

[0030] The control unit 11 includes one or more arithmetic processing devices such as a central processing unit (CPU) or a graphics processing unit (GPU). The control unit 11 controls each component and executes processing using built-in memories such as a read-only memory (ROM) or a random access memory (RAM), a clock, a counter, etc. The functional units of the control unit 11 may be realized by software, or some or all of them may be realized by hardware such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

[0031] The storage unit 12 includes a nonvolatile memory such as a hard disk, flash memory, or SSD (Solid State Drive). The storage unit 12 may be separate from the determination device 1 and may be one or more external storage devices connected externally. The storage unit 12 stores various computer programs and data referenced by the control unit 11. The storage unit 12 stores a program 1P for causing a computer to execute processing related to determining the possibility of disease, and a detection DB (Data Base) 121. The storage unit 12 may further store a reference profile, which will be described later.

[0032] A computer program (program product) including program 1P may be provided by a non-transitory recording medium 1A on which the computer program is readably recorded. The storage unit 12 stores the computer program read from the recording medium 1A by a reading device (not shown). The recording medium 1A may be, for example, a magnetic disk, an optical disk, or a semiconductor memory. Alternatively, the computer program may be downloaded from an external server connected to a communications network and stored in the storage unit 12. Program 1P may be a single computer program or may be composed of multiple computer programs. Program 1P may also be executed on a single computer or may be executed cooperatively by multiple computers.

[0033] The communication unit 13 includes a communication device that realizes communication via the network N. The control unit 11 transmits and receives data between the terminal device 2 and the detection device 3 via the communication unit 13.

[0034] The configuration of the determination device 1 is not limited to the above example, and may include, for example, an operation unit for accepting user operations, a display unit for displaying images, and the like.

[0035] The determination device 1 and the detection device 3 are not limited to transmitting and receiving data via the network N. The determination device 1 may, for example, be provided with an input interface for connecting to the detection device 3 and may receive data output from the detection device 3 via a signal line or the like.

[0036] 4 is a diagram showing an example of the contents of information stored in the detection DB 121. The detection DB 121 is a database that stores detection information related to each of a plurality of subjects, subject information related to subjects, and reference profile information related to reference profiles.

[0037] The detection information table stores records that link information such as subject ID, collection date, detection device ID, detection date, detection data, and determination results, using, for example, subject ID as a key. The subject ID is identification information for uniquely identifying the subject collected from the subject. The subject ID may be an ID attached to a container that contains a specific subject. The subject ID is identification information for identifying the subject. The collection date indicates the date and time the subject was collected. The detection device ID is identification information for identifying the detection device 3 used to detect a response signal for the subject. The detection date indicates the date and time the response signal for the subject was detected by the detection device 3.

[0038] The detection data includes information indicating a response signal to the subject. The detection data is, for example, a profile of the response signal detected over time. In this embodiment, the profile of the response signal as detection data is time-dependent data of luminescence intensity. The detection data is generated for each cell 42 in the olfactory sensor 4. The judgment result indicates the possibility of disease based on the detection data. The judgment result may be indicated, for example, by the presence or absence of a possibility, or may be indicated by the degree of possibility classified into multiple levels. The information in the detection information table is collected, for example, via the detection device 3.

[0039] The subject information table stores records that link information such as terminal device information and subject attribute information, using, for example, the subject ID as a key. The detection information table and the subject information table are associated by the subject ID. The terminal device information is information that identifies the terminal device 2 used by the subject, and includes, for example, an address indicating the output destination of the determination result, a device ID, etc. The attribute information includes, for example, the subject's name, age, gender, health information, etc. The health information is information related to the subject's health condition, and may include information such as current symptoms, medical history, test results, and health check results. The information in the subject information table is collected, for example, through the terminal device 2.

[0040] The reference profile information table stores records linking information such as cell information, first reference profile information, and second reference profile information, using, for example, disease information as a key. The disease information is information for identifying the disease to be determined, and includes, for example, a disease ID and a disease name. The cell information is information for identifying the cells 42 in the olfactory sensor 4 and represents the cell type selected for disease determination. The first reference profile information and the second reference profile information are information related to the first reference profile and the second reference profile corresponding to the cell information. Details of the first reference profile and the second reference profile will be described later. Note that FIG. 4 is an example, and the content of the information stored in the detection DB 121 is not limited. Furthermore, the data storage method shown in FIG. 4 is an example, and other storage formats are possible as long as the data content and the relationships between the data are maintained.

[0041] 5 is a block diagram showing the configuration of the terminal device 2. The terminal device 2 includes a control unit 21, a storage unit 22, a communication unit 23, a display unit 24, and an operation unit 25.

[0042] The control unit 21 includes one or more arithmetic processing units such as a CPU, an MPU, a GPU, etc. The control unit 21 controls each component and executes processing using built-in memories such as a ROM or a RAM, a clock, a counter, etc.

[0043] The storage unit 22 includes a non-volatile memory such as a hard disk, flash memory, SSD, etc. The storage unit 22 stores various computer programs and data referenced by the control unit 21. The storage unit 22 stores a program 2P for causing the computer to execute processing related to obtaining the determination result of the possibility of disease.

[0044] A computer program (computer program product) including the program 2P may be provided by a non-transitory recording medium 2A on which the computer program is readably recorded. The storage unit 22 stores the computer program read from the recording medium 2A by a reading device (not shown). The recording medium 2A may be, for example, a magnetic disk, an optical disk, or a semiconductor memory. Alternatively, the computer program may be downloaded from an external server connected to a communications network and stored in the storage unit 22. The program 2P may be a single computer program or may be composed of multiple computer programs. The program 2P may also be executed on a single computer or may be executed cooperatively by multiple computers.

[0045] The communication unit 23 includes a communication device that realizes communication via the network N. The control unit 21 transmits and receives data to and from the determination device 1 via the communication unit 23.

[0046] The display unit 24 includes a display device such as a liquid crystal display, an organic EL (Electro Luminescence) display, etc. The display unit 24 displays various information including the determination results in accordance with instructions from the control unit 21.

[0047] The operation unit 25 is an interface that accepts user operations. The operation unit 25 includes, for example, a keyboard, a mouse, a touch panel device with a built-in display, a speaker, a microphone, etc. The operation unit 25 accepts operation input from the user and sends a control signal according to the operation content to the control unit 21.

[0048] Figure 6 shows an example of the change in luminescence intensity of a sensor cell over time. The vertical axis of the graph in Figure 6 represents luminescence intensity, and the horizontal axis represents the elapsed time (s) from the start of detection. Figure 6 shows the detection results of luminescence intensity for each concentration of odor molecules when a solution containing a specific odor molecule is added to a cell having an olfactory receptor that binds to the odor molecule. In Figure 6, the concentrations of the odor molecules are referred to in descending order as the first concentration, the second concentration, and the third concentration.

[0049] As shown in Figure 6, the profile showing the change in luminescence intensity over time differs depending on the concentration. In this embodiment, the characteristics of a specific group of odor molecules in urine that can be detected by the sensor cell are estimated based on the response profile of the luminescence intensity detected from the subject's urine sample, thereby determining the possibility of disease in the subject.

[0050] The flow of the determination process in the determination system 100 of this embodiment will be described below using a specific example.

[0051] The determination device 1 accepts an application for the determination service from a subject. A user (subject) who wishes to use the service applies for the service using, for example, the terminal device 2 via a reception screen 50 provided by the determination device 1.

[0052] 7 is a schematic diagram showing an example of a reception screen 50. When the determination device 1 receives a request for login using account information and a reception screen by an operation of the subject via the terminal device 2, the determination device 1 outputs the reception screen 50 as shown in FIG.

[0053] The reception screen 50 includes a subject information reception field 501 for receiving input of subject information related to the subject to be assessed, and a disease information reception field 502 for receiving input of the target disease for which assessment is desired. The subject uses the operation unit 25 to input the name and other information of the subject to be assessed in the subject information reception field 501. In the case of a subject who has been registered as a user in advance, the detection DB 121 may be referenced to read out subject information corresponding to the logged-in account information, and the read out subject information may be displayed in the subject information reception field 501.

[0054] The disease information reception field 502 displays a plurality of selectable diseases that can be determined by the determination system 100. The subject can input the designation of one or more diseases for which the subject wishes to be determined, for example, by selecting a check box corresponding to each disease. In the example shown in Fig. 7, lung cancer is selected as the target disease.

[0055] When an application button 503 specifying an application for a diagnosis is selected with various information entered in the subject information reception field 501 and the disease information reception field 502, the terminal device 2 transmits the received subject information and target disease to the determination device 1. The determination device 1 receives the subject information and target disease and accepts the application for a diagnosis. The determination device 1 stores the received subject information and target disease in the memory unit 12, and, if necessary, transmits the subject information and information corresponding to the target disease to the detection device 3 of the analytical laboratory that will perform the test.

[0056] The reception screen 50 may also include a payment method reception field 504 as shown in Fig. 7. The subject can specify the desired payment method by selecting a specific payment method from the payment method reception field 504. The terminal device 2 may execute processing for online payment or credit card payment with a predetermined payment server or the like, depending on the accepted payment method.

[0057] Note that the information related to the application for the assessment service is not limited to that received through the terminal device 2, and the assessment device 1 may acquire subject information, etc., by receiving input from the user, for example.

[0058] Once the application is completed, a urine collection container is sent to the subject. The urine collection container may be distributed through a predetermined store, analytical institution, etc. A label with, for example, a two-dimensional or three-dimensional code representing the subject's ID is attached to the urine collection container. The subject places the collected urine in the urine collection container and submits it to the analytical institution. The urine collection container may be submitted together with the subject's attribute information, such as the subject's name, the collection date, the application number issued when the test was applied for, etc.

[0059] The urine collection container is received at the analysis facility. The detection device 3 receives input from, for example, a person in charge, and thereby acquires the subject ID, subject ID, collection date, etc. of the subject to be tested.

[0060] Next, the analytical laboratory detects the response signal using the olfactory sensor 4. A predetermined number of cells 42, each having a different olfactory receptor, are arranged in an array in the olfactory sensor 4. The olfactory sensor 4 is produced, for example, by integrating DNA containing a specific insect olfactory receptor coding sequence, a specific insect olfactory receptor co-receptor coding sequence, and a calcium-sensitive photoprotein coding sequence, all of which are placed under the control of a promoter sequence, into the chromosomal genomic DNA of each cell.

[0061] A predetermined amount of a urine sample from a subject is brought into contact with the olfactory sensor 4, and the luminescence intensity of each cell is detected over time by the detection device 3. A response profile that indicates changes in luminescence intensity over time is thereby acquired. The response profile is generated for each cell 42 in the olfactory sensor 4. The detection device 3 associates the acquired response profile with the subject ID, subject ID, collection date, detection device ID, detection date, etc. of the subject to be tested, and transmits the response profile to the determination device 1.

[0062] The determination device 1 may generate a response profile by performing predetermined preprocessing on the detection data received from the detection device 3. Measurement of the luminescence intensity of the olfactory sensor 4 may be performed collectively on multiple arrayed sensor cells. In such cases, the raw detection data output from the detection device 3 is expected to include luminescence intensities at various times (detection times) for multiple sensor cells. Alternatively, the detection data may contain a mixture of detection values ​​from multiple urine samples derived from subjects. The determination device 1 generates a response profile by sorting the raw detection data including multiple detection values ​​for each sensor cell or each urine sample derived from the subject, arranging the luminescence intensities over time, and converting them into a predetermined data format. If the luminescence intensities in the raw detection data vary in time, the determination device 1 may unify the times of each data in the response profile by interpolating the data using a predetermined interpolation method.

[0063] The determination device 1 determines the possibility of a disease in a subject based on the response profile of the obtained urine sample from the subject. In this embodiment, the possibility of a disease is determined by comparing the response profile of the urine sample from the subject to be analyzed with a reference profile generated in advance.

[0064] The reference profile is a response profile that indicates a reference for the response signal of the sensor cell. The reference profile includes a first reference profile based on the luminescence intensity detected from urine from a subject in good health, and a second reference profile based on the luminescence intensity detected from a urine sample from a subject in poor health. As an example, if the target disease is lung cancer, a subject in good health refers to a non-cancer patient who is not suffering from cancer, and a subject in poor health refers to a lung cancer patient who is suffering from lung cancer. The reference profile is, for example, generated in advance and stored in the memory unit 12.

[0065] The first and second reference profiles are obtained by detecting the above-described odor molecules using the olfactory sensor 4 using urine collected from a non-cancer patient (hereinafter also referred to as healthy urine) and urine collected from a lung cancer patient (hereinafter also referred to as lung cancer urine) to generate a response profile. The first and second reference profiles are prepared for each cell 42.

[0066] The first reference profile is preferably generated based on the response profiles of multiple healthy urine samples obtained from multiple non-cancer patients. Similarly, the second reference profile is preferably generated based on the response profiles of multiple lung cancer urine samples obtained from multiple lung cancer patients. For example, statistical values ​​of the luminescence intensity detected from each healthy urine sample are calculated over time, and the first reference profile is generated based on the obtained time-dependent statistical values. As the statistical value, the mean or median is preferred, with the weighted mean, geometric mean, or median being more preferred, and the geometric mean being most preferred. Using a similar method, the second reference profile is generated based on the response profiles of multiple lung cancer urine samples obtained from multiple lung cancer patients.

[0067] Furthermore, one or more cells 42 suitable for diagnosing lung cancer, the disease to be diagnosed, are selected from the multiple types of cells 42 contained in the olfactory sensor 4. Cells suitable for diagnosing lung cancer are cells that exhibit reactivity according to lung cancer urine. Cells suitable for diagnosing lung cancer are preferably cells whose reactivity changes significantly depending on whether or not the subject is afflicted with lung cancer, and whose response profile shows a significant difference. Cells suitable for diagnosing lung cancer may be cells that react to odor molecules that are contained in relatively higher amounts in lung cancer urine than in healthy urine, or may be cells that react to odor molecules that are contained in relatively higher amounts in healthy urine than in lung cancer urine. In this embodiment, cells 42 that react to odor molecules that are contained in higher amounts in lung cancer urine than in healthy urine are considered to be cells 42 suitable for diagnosing lung cancer.

[0068] It is not easy to identify in advance the cells 42 that are suitable for diagnosing the disease to be diagnosed at the stage of generating the olfactory sensor 4. For this reason, in this embodiment, an olfactory sensor 4 equipped with multiple types of cells 42 is prepared, and for each disease to be diagnosed, the cells 42 that are suitable for diagnosing the disease are identified from among the multiple types of cells 42.

[0069] The method for identifying cells 42 suitable for disease diagnosis is not limited, but for example, cells 42 can be identified based on the dissimilarity between the first reference profile and the second reference profile. The determination device 1 may calculate the dissimilarity between the first reference profile and the second reference profile, and identify cells 42 for which the calculated dissimilarity is equal to or greater than a predetermined first threshold as cells 42 for disease diagnosis.

[0070] The dissimilarity between the first and second reference profiles may be determined, for example, by the difference between the maximum luminescence intensity in the first and second reference profiles. The area enclosed by the first and second reference profiles (the integral of the intensity difference between the first and second reference profiles) may also be used as the dissimilarity. The larger the difference between the maximum intensities or the integral of the intensity difference between the reference profiles, the lower the similarity between the first and second reference profiles and the greater the change in reactivity depending on the presence or absence of disease. In other words, the difference in response of cells 42 depending on the presence or absence of disease is greater, and this is considered to be highly useful in disease diagnosis. While the dissimilarity between the first and second reference profiles is used in the above example, similarity may also be used.

[0071] The selection of cells 42 for disease assessment may be performed taking into account the accuracy of disease assessment using the cells 42. For example, response profiles for multiple test samples are obtained for multiple cells 42 initially selected for disease assessment, and the acquired response profiles are used to assess the possibility of disease according to the assessment method described below. The accuracy of assessment when using each of the initially selected cells 42 is calculated based on the assessment results for each cell 42 and the known characteristics of the test sample (healthy urine or lung cancer urine). The accuracy of assessment includes, for example, the AUC, recall, specificity, accuracy, and precision of the ROC curve. Based on the calculated accuracy values, the type and number of cells 42 to be selected for disease assessment are identified so as to optimize each assessment accuracy (e.g., maximize each assessment accuracy). Secondary selection of cells 42 for disease assessment is performed in accordance with the identified type and number of cells 42.

[0072] 8A and 8B show examples of a reference profile for cells selected for disease assessment and a reference profile for cells not selected for disease assessment. FIG. 8A shows a reference profile for cells selected for disease assessment, and FIG. 8B shows an example of a reference profile for cells not selected for disease assessment. In the case of cells for disease assessment, the shapes of the response profiles are significantly different between healthy urine and pseudo-lung cancer urine (urine obtained by adding odor molecules derived from lung cancer to healthy urine). In the case of cells for disease assessment, the pseudo-lung cancer urine contains odor substances derived from lung cancer to which the cells for disease assessment respond, causing a temporary, significant increase in luminescence intensity. In contrast, healthy urine does not contain odor substances derived from lung cancer, causing almost no fluctuation in luminescence intensity. On the other hand, in the case of cells not for disease assessment, no significant difference is observed in the response profiles between healthy urine and pseudo-lung cancer urine.

[0073] If the type of cells 42 suitable for determining a specific disease is known, the step of selecting cells 42 for disease determination may be omitted.

[0074] The determination device 1 determines the likelihood of lung cancer by determining the similarity between the response profile of the subject's urine sample corresponding to the selected disease determination cell 42 and each of the first and second reference profiles corresponding to the disease determination cell 42. If the similarity to the first reference profile is higher than the similarity to the second reference profile, it is determined that there is no possibility of lung cancer. If the similarity to the second reference profile is higher than the similarity to the first reference profile, it is determined that there is a possibility of lung cancer. The determination device 1 may determine the likelihood of lung cancer in multiple stages, as a percentage, or the like, depending on the similarity value.

[0075] The method for determining the degree of similarity is not limited, but for example, the absolute value of the difference between the luminescence intensity in the response profile of the subject's urine sample and the luminescence intensity in the reference profile may be calculated for each elapsed time, and the sum of the absolute values ​​of the calculated differences may be used as the index for determining the degree of similarity. A smaller sum of the absolute values ​​of the luminescence intensity differences indicates a higher degree of similarity.

[0076] When multiple cells 42 for disease assessment are selected, the assessment device 1 individually assesses the possibility of lung cancer based on each of the cells 42 for disease assessment and integrates the individually assessed lung cancer possibilities to make an overall assessment of the possibility of lung cancer. The assessment device 1, for example, determines the overall assessment based on a majority vote of the assessment results based on each cell 42. When making the overall assessment, weighting may be performed so that a greater weight is given to an assessment result based on a specific cell 42. The cells 42 to be weighted may be cells 42 that are more useful in disease assessment, cells 42 that have a higher degree of similarity, cells 42 that have a smaller variation in response signals, etc.

[0077] The possibility of lung cancer may be determined using a machine learning technique. The determination device 1 has prepared in advance a determination model that outputs the possibility of lung cancer when the similarity between the response profile of each cell 42 for disease determination and a reference profile is input. The similarity includes at least one of the similarity between the response profile of the subject's urine sample and a first reference profile and the similarity between the response profile and a second reference profile.

[0078] The determination device 1 inputs the similarity between the response profiles of each cell 42 for disease determination into a determination model, and obtains the possibility of lung cancer output from the determination model. The type of cell 42 may be input to the determination model along with the similarity. The determination result of the possibility of lung cancer output from the determination model corresponds to the comprehensive determination result. In this case, individual determination based on each cell 42 may be omitted.

[0079] If the type of cells 42 suitable for diagnosing a specific disease is known, an olfactory sensor 4 may be individually generated for each type of disease to be diagnosed using only the cells 42 suitable for disease diagnosis. In this case, luminescence intensity is detected using an individual olfactory sensor 4 corresponding to the selected target disease, and the possibility of disease is determined based on the detection results of each cell 42 in the individual olfactory sensor 4. The determination device 1 associates a series of obtained information, such as the subject ID, subject ID, collection date, detection device ID, detection date, response profile, and determination result, and stores them in the detection DB 121. The determination device 1 also outputs the determination result to the subject's terminal device 2.

[0080] 9 is a schematic diagram showing an example of a result screen 51 showing the determination result. The result screen 51 includes a first display section 511 that displays information about the subject to be determined, and a second display section 512 that displays the determination result.

[0081] The determination device 1 displays information about the subject and the specimen to be determined on the first display unit 511 based on the information stored in the detection DB 121. The first display unit 511 displays, for example, the subject's name, the collection date of the specimen, and the specimen ID.

[0082] The determination device 1 also displays the determination result indicating the possibility of the disease on the second display unit 512. When determination results for multiple types of diseases are obtained, the result screen 51 may be configured to include multiple second display units 512 corresponding to the respective target diseases. Each second display unit 512 displays the target disease and the determination result related to the target disease.

[0083] Fig. 10 is a schematic diagram showing another example of the result screen 51 showing the determination result. In the example shown in Fig. 10, the second display section 512 of the result screen 51 further includes a detection result display field 513 that displays the detection result for the analyte detected by the olfactory sensor 4. The detection result display field 513 displays a response profile corresponding to the analyte.

[0084] In addition to the response profile corresponding to the subject, the criteria used in the disease assessment process, i.e., the first and second reference profiles, may be displayed in the detection result display field 513. For ease of explanation, only one type of response profile is shown in Fig. 10, but the detection result display field 513 may also display multiple response profiles indicating the detection results for each cell 42 used for disease assessment. The detection result display field 513 may also display an image showing the luminescence of the cells 42 in the olfactory sensor 4, captured by an imaging device.

[0085] When displaying response profiles for multiple cells 42 in the detection result display field 513, the determination device 1 may determine the display order on the screen based on the usefulness in disease determination or the similarity between each response profile and a reference profile. The determination device 1 may, for example, preferentially display the detection results of cells 42 with high usefulness or similarity in the detection result display field 513. The determination device 1 may also display the detection results for a predetermined number of cells 42 in the detection result display field 513 in descending order of priority. The determination device 1 may determine the display order based on the contribution of input information in the above-described determination model, such that cells 42 with a higher contribution of similarity are displayed preferentially. The contribution can be calculated based on, for example, a SHAP (Shapeley Additive exPlanation) value, a Gini coefficient, a LIME (Local Interpretable Model-Agnostic Explanations), a PFI (Permutation Feature Importance), or the like.

[0086] The detection result is not limited to being presented to the subject via the terminal device 2. The determination device 1 may output the detection result to, for example, another computer or a predetermined printing device.

[0087] 11 is a flowchart showing an example of a reference profile generation process. The process in the flowchart below is executed by the control unit 11 in accordance with a program 1P stored in the storage unit 12 of the determination device 1.

[0088] The control unit 11 of the determination device 1 acquires, via the detection device 3, response profiles detected from healthy urine samples from multiple non-cancer patients and response profiles detected from lung cancer urine samples from multiple lung cancer patients (step S11). The response profiles are, for example, time-course data of luminescence intensity, and are generated for each cell 42 in the olfactory sensor 4. Each response profile may be linked to information indicating the corresponding cell 42.

[0089] The control unit 11 generates a first reference profile based on each response profile for the plurality of normal urine samples (step S12). For example, the control unit 11 generates the first reference profile by calculating the geometric mean of the luminescence intensity detected from each normal urine sample for each elapsed time.

[0090] The control unit 11 generates a second reference profile based on each response profile for the plurality of lung cancer urine samples (step S13). The control unit 11 generates the second reference profile by, for example, calculating the geometric mean of the luminescence intensity detected from each lung cancer urine sample for each elapsed time.

[0091] The control unit 11 selects one or more cells 42 to be used for lung cancer assessment from among the multiple types of cells 42 included in the olfactory sensor 4 (step S14). The control unit 11, for example, calculates the maximum luminescence intensity or the intensity difference integral value in each of the first and second reference profiles for each cell 42 included in the olfactory sensor 4. The control unit 11 selects cells 42 for which the calculated difference in the maximum luminescence intensity or the intensity difference integral value is equal to or greater than a predetermined first threshold as cells 42 for lung cancer assessment. In step S14, the control unit 11 may identify the type and number of cells 42 to be selected for disease assessment based on the accuracy of disease possibility assessment when each cell 42 is used, so as to optimize each assessment accuracy, and select cells 42 for lung cancer assessment according to the identification results.

[0092] The control unit 11 associates the disease information, the cell information of the selected cell 42 for lung cancer diagnosis, and the first and second reference profiles corresponding to the cell 42 and stores them in the detection DB 121 (step S15), and then completes the series of processes.

[0093] The control unit 11 executes the above-described process for all diseases that can be subject to diagnosis, and generates and stores first and second reference profiles for various disease diagnosis.

[0094] 12 is a flowchart showing an example of a procedure for determining the possibility of a disease. The following process is executed by the control unit 11 in accordance with a program 1P stored in the storage unit 12 of the determination device 1, and by the control unit 21 in accordance with a program 2P stored in the storage unit 22 of the terminal device 2.

[0095] The control unit 21 of the terminal device 2 receives the subject information and the target disease of the subject who wishes to be diagnosed based on the subject's operation using the reception screen (step S21), and transmits the received subject information and the target disease to the assessment device 1 (step S22).

[0096] The control unit 11 of the determination device 1 receives the subject information and the target disease (step S23).

[0097] The control unit 11 acquires a response profile generated from a urine sample derived from the subject through the detection device 3 (step S24). The response profile is generated for each cell 42 in the olfactory sensor 4. The response profile is associated with a subject ID, subject ID, collection date, detection device ID, detection date, etc. The control unit 11 may acquire raw detection data obtained by detection from the detection device 3 and generate a response profile by performing various preprocessing operations on the acquired raw detection data.

[0098] The control unit 11 calculates the degree of similarity between the response profile of the urine sample derived from the subject and each of the first and second reference profiles for each cell 42 for disease determination according to the target disease (step S25).

[0099] The control unit 11 compares the calculated similarities and identifies the health state corresponding to the reference profile with the highest similarity as the health state of the subject, thereby individually determining whether or not the subject has a disease (step S26). In step S26, the presence or absence of a disease is determined for each cell 42. The control unit 11 makes an overall determination of the presence or absence of a disease, for example, by a majority vote of the individual determinations for each cell 42 (step S27). The control unit 11 stores the subject ID, subject ID, collection date, detection device ID, detection date, response profile by cell type, and determination results in the detection DB 121 in association with each other (step S28).

[0100] The control unit 11 generates a result screen showing the obtained disease possibility assessment result (step S29). The control unit 11 transmits the generated result screen to the terminal device 2 corresponding to the subject identified by the subject information acquired in step S23 (step S30).

[0101] The control unit 21 of the terminal device 2 receives the result screen from the determination device 1 (step S31). The control unit 21 displays the received result screen on the display unit 24 (step S32), and ends the series of processes.

[0102] In the above-described process, the user may request the display of a result screen using the terminal device 2 and accept the result screen in response to the request, thereby being able to check the judgment result at any time.

[0103] In the above-described process, the control unit 11 of the determination device 1 may update the reference profile based on the response profile and determination result newly stored in the detection DB 121 by the process of step 28. The control unit 11 extracts one or more newly added response profiles from the detection DB 121, for example, at appropriate intervals. The control unit 11 regenerates a first reference profile based on the multiple extracted response profiles to which a response profile determined to be unlikely to be cancer has been newly added. Alternatively, the control unit 11 regenerates a second reference profile based on the multiple extracted response profiles to which a response profile determined to be likely to be cancer has been newly added.

[0104] In the above, the olfactory sensor 4 is used to determine the possibility of disease based on detection data for a urine sample. The specimen to be analyzed is not limited to urine, but may also be, for example, blood, sweat, saliva, tears, exhaled breath, skin gas, tissue fluid, synovial fluid, follicular fluid, cerebrospinal fluid, semen, milk, vaginal fluid, etc. Furthermore, the health condition to be determined is not limited to the possibility of disease, but may also be the possibility of other abnormalities, etc. The health condition determination may be a stress check, a bad breath check, a body odor check, etc. The subject to be determined for health condition is not limited to humans, but may also be an animal.

[0105] According to this embodiment, health status can be determined based on detection data from the olfactory sensor 4, improving the practicality of the olfactory sensor 4. Health status can be determined with high accuracy by using a response profile that indicates the response of olfactory receptors. Health status determination processing is facilitated by determining health status through comparison with a pre-generated reference profile.

[0106] The subject can obtain the assessment result by submitting the subject's sample and registering the necessary information, thereby reducing the burden required for testing and increasing the utilization of the service. The assessment result can be confirmed using the terminal device 2, so the assessment result can be reliably grasped at any time. In addition to the assessment result, the detection result can be displayed in a visually recognizable manner, allowing the results to be confirmed more reliably and in detail. Displaying the reference profile that serves as the assessment standard in addition to the subject's own detection result improves the explainability of the assessment result.

[0107] Second Embodiment In the second embodiment, correction is performed to eliminate individual differences in the response profile. In the following embodiments, differences from the first embodiment will be mainly described, and the same reference numerals will be used to designate components common to the first embodiment, and detailed descriptions thereof will be omitted.

[0108] The response signal detected from a test sample may vary from individual to individual due to various factors. For example, due to the influence of impurities contained in a urine sample, even if two urine samples contain the same concentration of odor compounds, the luminescence intensity detected from the urine sample containing the impurities may be greater or smaller than that of a urine sample without the impurities. In other words, the influence of the impurities causes individual differences in the correlation between the concentration of odor molecules and the luminescence intensity. The response signal may also vary from individual to individual due to the influence of the balance of odor molecules. Individual differences may occur not only in luminescence intensity but also in various response signals. Samples, including urine samples, may contain various individual difference-causing factors (i.e., error factors) that affect the response signal, such as impurities.

[0109] When performing a determination based on a response profile, the occurrence of individual differences as described above leads to a decrease in determination accuracy. In particular, as described in the first embodiment, when determining whether a urine sample resembles a profile of healthy urine or lung cancer urine by comparing a target response profile with a reference profile, the possibility of erroneous determination due to the influence of individual differences increases. In this embodiment, a correction process is performed to eliminate individual differences in the response profile, thereby improving determination accuracy.

[0110] 13 is a flowchart showing an example of a processing procedure executed by the determination device 1 of the second embodiment. The processing of FIG. 13 is executed, for example, between steps S24 and S25 of the first embodiment. By the processing of step S24, a response profile corresponding to each cell 42 included in the olfactory sensor 4 is acquired.

[0111] The control unit 11 of the determination device 1 selects one or more cells 42 to be used for correcting individual differences from the multiple types of cells 42 contained in the olfactory sensor 4 (step S41). Cells 42 for correcting individual differences are preferably cells 42 whose reactivity to a target disease (e.g., lung cancer) is lower than that of cells for disease determination. More preferably, cells for correcting individual differences are cells whose reactivity does not change significantly depending on whether or not a subject has lung cancer, and whose response profile does not show a significant difference. Cells for correcting individual differences may be cells that respond to odor molecules present at approximately the same concentrations in both healthy urine and lung cancer urine, or cells that respond to odor molecules artificially added to urine that are not present in the urine before addition. Cells for correcting individual differences may be cells that do not respond to odor substances derived from the target disease but do respond to odor substances contained in a urine sample. It is preferable that the direction of increase or decrease in the response signal due to the influence of error factors is the same for cells for correcting individual differences and cells for disease determination. In other words, it is preferable that both the cells for correcting individual differences and the cells for disease assessment are cells in which the value of the response signal obtained under the influence of an error factor is uniformly increased or uniformly decreased compared to the value of the response signal obtained without the influence of the error factor.

[0112] The control unit 11 selects cells 42 for individual variability correction based on, for example, the dissimilarity between a first reference profile based on non-cancer urine and a second reference profile based on lung cancer urine. Specifically, the control unit 11 calculates the dissimilarity between the first reference profile and the second reference profile for each cell 42 included in the olfactory sensor 4. The control unit 11 selects cells 42 for which the calculated dissimilarity is less than a predetermined second threshold as cells 42 for individual variability correction. The dissimilarity between the first reference profile and the second reference profile may be, for example, the difference between the maximum luminescence intensity in the first reference profile and the maximum luminescence intensity in the second reference profile, or the integral value of the intensity difference between the first reference profile and the second reference profile, as in the case of identifying cells 42 for disease assessment. Note that although the dissimilarity between the first reference profile and the second reference profile is used in the above, similarity may also be used.

[0113] Preferably, the first and second reference profiles are each generated based on response profiles of multiple urine samples obtained from multiple individuals. The second threshold value may be the same as or smaller than the first threshold value used to select the cells 42 for disease assessment. The cells 42 for correcting individual differences may be selected from among the cells 42 in the olfactory sensor 4, other than the cells 42 for disease assessment. If the type of the cells 42 for correcting individual differences is known, the selection process described above may be omitted.

[0114] The control unit 11 acquires a correction profile corresponding to the selected correction cell 42 (step S42). The correction profile may be a response profile generated based on statistical values ​​of the first and second reference profiles. If the maximum value of the luminescence intensity or the integral value of the intensity difference between the first and second reference profiles is approximately zero, either the first or second reference profile may be used as the correction profile. A correction profile is acquired for each correction cell 42.

[0115] The control unit 11 calculates a correction coefficient (correction value) for correcting the response profile based on the acquired correction profile and the response profile of the subject's urine sample using the correction cells 42 (step S43).

[0116] The correction coefficient can be calculated, for example, by the following method. The area enclosed by the response profile and correction profile of the subject's urine sample from the correction cells 42 and the x-axis (time axis) is divided into sections at predetermined time intervals. For each section, the ratio between the area of ​​the first region enclosed by the response profile and the area of ​​the second region enclosed by the correction profile is calculated. The geometric mean value of the ratios in all sections is used as the correction coefficient. Note that the method for calculating the correction coefficient is not limited to the above example, and any method may be used as long as it is capable of correcting individual differences in the response profile. The correction coefficient may be adjusted depending on the type of the determination cells 42. For example, the correction coefficient for each determination cell 42 may be calculated by multiplying the above correction coefficient calculated based on the correction profile and the response profile by a predetermined coefficient that is set in advance for each determination cell 42.

[0117] When multiple cells 42 are selected as correction cells 42, the control unit 11 may calculate the above-mentioned correction coefficient for each correction cell 42 and determine the final correction coefficient by obtaining the statistical value (e.g., geometric mean, median, etc.) of each calculated correction coefficient.

[0118] The control unit 11 uses the calculated correction coefficient to correct the response profile of each cell 42 selected for disease determination (step S44). Specifically, the control unit 11 multiplies each luminescence intensity in the response profile for disease determination by the correction coefficient to generate a corrected response profile. The control unit 11 uses the corrected response profile to perform the processes from step S25 onwards, thereby determining the possibility of disease based on the corrected response profile.

[0119] According to the above-described process, a correction coefficient is calculated based on the response profile and reference profile of the correction cell 42 selected from the cells 42 in the olfactory sensor 4, and the response profile of the disease assessment cell 42 can be corrected using the calculated correction coefficient. The disease assessment cell 42 is an example of a first sensor element, and the correction cell 42 is an example of a second sensor element. The correction cell 42 does not exhibit different response signal characteristics depending on a specific property (e.g., whether or not there is a possibility of disease), and the response profile is largely independent of the specific property. On the other hand, the disease assessment cell 42 exhibits different response signal characteristics depending on the specific property, and the response profile is highly dependent on the specific property. It is expected that the influence of error factors on the luminescence intensity in the same urine sample will occur in the same way in multiple cells 42. The luminescence intensities of the disease assessment cell 42 and the correction cell 42 for the same urine sample are predicted to uniformly increase or decrease due to the influence of error factors. By using the response profile of the correction cell 42, the influence of factors causing individual differences included in the response profile of the disease assessment cell 42 can be eliminated.

[0120] FIG. 14 shows examples of response profiles with and without correction processing. The vertical axis of the graph shown in FIG. 14 represents luminescence intensity, and the horizontal axis represents the elapsed time (s) from the start of measurement. In FIG. 14, open triangles (△) represent the response profile of the disease determination cells for a urine specimen affected by impurities before correction, and closed triangles (▲) represent the response profile of the disease determination cells for the urine specimen after correction. Also in FIG. 14, open squares (□) represent the first reference profile generated based on the response profiles obtained from multiple healthy urine samples, and open circles (○) represent the second reference profile generated based on the response profiles obtained from multiple lung cancer-suspected urine samples. The corrected response profile was created by correcting the pre-correction response profile using a correction coefficient calculated based on the deviation between the response profile of the correction cells and the reference profile indicating the response standard. The cells selected for disease assessment were those confirmed by known screening methods to respond to lung cancer-derived odor molecules added to lung cancer-mimicking urine samples, and that showed a significant increase in luminescence intensity in both healthy and lung cancer-mimicking urine samples.The cells selected for correction were those confirmed by known screening methods to not respond to lung cancer-derived odor molecules added to lung cancer-mimicking urine samples, and that showed a significant increase in luminescence intensity in both healthy and lung cancer-mimicking urine samples.

[0121] Figure 14A shows an example of a response profile for a lung cancer-mimicking urine specimen containing activity-reducing contaminants. In the example shown in Figure 14A, the response profile before correction shows a decrease in overall luminescence intensity due to the influence of the activity-reducing contaminants. Similarly, a correction coefficient for increasing luminescence intensity was calculated based on the response profile of correction cells, which also showed a decrease in overall luminescence intensity. The response profile after correction using the correction coefficient is corrected to have a higher luminescence intensity. The correction shown in Figure 14A can reduce the possibility of erroneously determining that a subject with a possible disease is not diseased.

[0122] FIG. 14B shows an example of a response profile for a healthy urine sample containing impurities that enhance activity. In the example shown in FIG. 14B, the response profile before correction shows an overall increase in luminescence intensity due to the influence of the impurities that enhance activity. Similarly, a correction coefficient was calculated to reduce the luminescence intensity based on the response profile of the correction cells, which showed an overall increase in luminescence intensity. The response profile after correction using the above correction coefficient has been corrected to reduce the luminescence intensity. The correction shown in FIG. 14B can reduce the possibility of erroneously determining that a subject who is not likely to have a disease is likely to have a disease.

[0123] In this embodiment, an example of correcting a response profile obtained by an olfactory sensor 4 equipped with sensor cells has been described, but the above-described correction method can be widely applied to response profiles for samples detected by various sensors equipped with sensor elements. Examples of sensors include biosensors equipped with biological elements (e.g., enzymes, antibodies, DNA, RNA, peptides, lipid membranes, cells, etc.), physical sensors equipped with sensor elements that detect physical quantities, and chemical sensors equipped with sensor elements that detect chemical substances. Furthermore, the object of determination using the response profile is not limited to the possibility of disease, and is not particularly limited as long as it is the property of the sample.

[0124] In this case, a sensor is prepared that includes a first sensor element that exhibits a reactivity in response to a specific property in a sample and a second sensor element that has a lower reactivity in response to the specific property than the first sensor element. The first sensor element may have a reactivity that changes significantly in response to the specific property, while the second sensor element may have a reactivity that does not change significantly in response to the specific property. Using the above-described correction method, the response profile of the first sensor element is corrected based on the response profile of the second sensor element to the target sample, and the property of the target sample is determined based on the corrected response profile of the first sensor element. Note that the first sensor element and the second sensor element are not limited to being provided in the same sensor, but may also be provided in separate sensors.

[0125] An example application of the correction method of this embodiment is described below. A biosensor containing a first protein that responds to a component in a food sample originating from a specific region and a second protein that does not respond to the component originating from the specific region is used to obtain a response profile for each protein in the food sample. The response profile of the first protein, corrected by the above-described correction method based on the response profile of the second protein, is used to determine whether the target food sample originated from the specific region. The sensor used for the above-described determination is not limited to a protein-based sensor; it may also be a sensor using biological elements such as DNA, RNA, peptides, or lipid membranes, or a chemical sensor such as a molecular functional membrane that changes its physical properties upon adsorption with a component in a food sample and generates an electrical signal. Furthermore, the detection target is not limited to food components; it can be widely applicable to a wide range of biological components, such as nucleic acids, amino acids, low-molecular-weight metabolites, and vesicles, which can be used in the fields of testing and diagnosis, as well as narcotic and explosive components, which can be used in the field of security.

[0126] According to this embodiment, it is possible to correct for individual differences between specimens, thereby suppressing a decrease in determination accuracy due to individual differences. When measuring a sample using a sensor, it is thought that differences in response signals may occur due to various factors. According to this embodiment, such individual differences can be appropriately eliminated.

[0127] According to this embodiment, it is possible to efficiently determine the sensor element for correcting individual differences based on the characteristics of the response profile. By preparing a sensor having a plurality of sensor elements and acquiring the response profile of each sensor element, it is possible to select an appropriate sensor element for correcting individual differences according to the various properties to be determined.

[0128] Third Embodiment In a third embodiment, a learning model is used to determine the possibility of disease.

[0129] 15 is a block diagram showing an example of the configuration of a determination device 1 according to the third embodiment. The determination device 1 according to the third embodiment stores a learning model 122 in the storage unit 12. The learning model 122 is a machine learning model that has learned predetermined training data. The learning model 122 is expected to be used as a program module that constitutes part of artificial intelligence software.

[0130] 16 is an explanatory diagram showing an overview of the learning model 122. The learning model 122 receives a response profile of luminescence intensity detected from a urine sample of a subject, and outputs information indicating whether or not the response profile is likely to be associated with a target disease. The learning model 122 is, for example, a convolutional neural network (CNN), which is a type of neural network.

[0131] The learning model 122 includes an input layer to which a response profile is input, an output layer that outputs whether or not there is a possibility of a disease, and an intermediate layer (hidden layer). The intermediate layer may include a convolutional layer, a pooling layer, a fully connected layer, etc. The intermediate layer has multiple nodes that extract features of the response profile and passes the features extracted using various parameters to the output layer. When a response profile is input to the input layer, calculations are performed in the intermediate layer using the learned parameters, and output information indicating the classification result of whether or not there is a possibility of the target disease is output from the output layer.

[0132] The learning model 122 can be generated by preparing training data in which labels indicating the possibility of a target disease are associated with response profiles, and using the training data to train an untrained neural network. For example, a diagnosis by an experienced doctor is used as the correct label. The training data includes response profiles detected from multiple subjects who may have the disease, and response profiles detected from multiple subjects who may not have the disease. The learning model 122 learns the relationship between these response profiles and the possibility of the disease.

[0133] The determination device 1 inputs multiple response profiles contained in training data into the input layer of a pre-learning neural network model, undergoes arithmetic processing in the intermediate layer, and obtains the presence or absence of a disease possibility output from the output layer. The determination device 1 compares the presence or absence of a disease possibility output from the output layer with the presence or absence of a disease possibility included in the training data, and optimizes parameters such as the weights between neurons using, for example, an error backpropagation method so that the presence or absence of a disease possibility output from the output layer approaches a correct value. Note that the learning model 122 may be constructed by an external device and deployed to the determination device 1.

[0134] The input data input to the learning model 122 is not limited to an image representing a response profile of luminescence intensity, but may be values ​​of luminescence intensity over time. The input data to the learning model 122 may further include the type of cell 42 corresponding to the luminescence intensity. Of course, the input to the learning model 122 may be a response signal other than luminescence intensity.

[0135] The output layer of the learning model 122 is not limited to estimating the possibility of disease as either present or absent, but may output, for example, a probability level classified into multiple categories according to the degree of probability, or may output a numerical value indicating the probability as a percentage.

[0136] The learning model 122 may be constructed such that one model is constructed for each type of cell 42, one model is constructed for each type of disease, or one model is constructed for each combination of cell 42 and disease. The learning model 122 may receive response signals relating to multiple types of cells 42 as input and estimate the possibility of a disease corresponding to the response signals relating to the multiple types of cells 42.

[0137] The learning model 122 may be configured to receive response profiles for samples collected from the subject on multiple collection dates as input and output a disease probability. In this case, the learning model 122 may receive the most recent response profile and several previous response profiles as input and output the current disease probability of the subject, or the future disease probability.

[0138] The configuration of the learning model 122 is not limited to the above example, and may be any model that can identify the possibility of a target disease based on the time-series data of the response signal. The learning model 122 may be a model constructed using other learning algorithms, such as a recurrent neural network (RNN), a graph neural network (GNN), a transformer, a support vector machine (SVM), logistic regression, or eXtreme Gradient Boosting (XGBoost).

[0139] FIG. 17 is a flowchart illustrating an example of a processing procedure executed by the determination system 100 according to the third embodiment.

[0140] The control unit 21 of the terminal device 2 executes the same processes as steps S21 to S22, and receives the subject information and the target disease (step S51) and transmits them (step S52).

[0141] The control unit 11 of the determination device 1 executes the same processes as steps S23 to S24, receives the subject information and the target disease (step S53), and acquires a response profile (step S54).

[0142] The control unit 11 selects a learning model 122 corresponding to the received target disease and the disease determination cells 42 corresponding to the target disease from among the multiple learning models 122 stored in the memory unit 12 (step S55).

[0143] The control unit 11 inputs the response profile of the corresponding cell 42 for disease assessment into each selected learning model 122 (step S56). The control unit 11 acquires the possibility of disease output from the learning model 122 (step S57). The learning model 122 outputs the presence or absence of possibility for, for example, each cell 42 and each type of disease. Thereafter, the control unit 11 executes the same processes as steps S27 to S31.

[0144] In the above-described processing, the control unit 11 may perform a comprehensive determination of the possibility of disease using a determination model, as in the first embodiment. In this case, the determination model may be configured to receive the possibility of disease for each type of cell 42 output from the learning model 122 as an input and output the possibility of disease.

[0145] According to this embodiment, the possibility of disease can be determined easily and accurately using the learning model 122.

[0146] The following supplementary note is further disclosed in relation to the above-described embodiments: (Supplementary Note 1) A determination method in which a computer executes processes of acquiring a response profile of a first sensor element, which exhibits reactivity according to a specific property of the sample, to the target sample, and a response profile of a second sensor element, which exhibits a lower reactivity according to the specific property of the sample than the first sensor element, storing the acquired response profile of the first sensor element and the response profile of the second sensor element in association with identification information of the target sample, correcting the response profile of the first sensor element corresponding to the identification information of the target sample based on a comparison of the response profile of the second sensor element corresponding to the identification information of the target sample with a reference profile indicative of a reference response of the second sensor element, determining the property of the target sample based on a comparison of the corrected response profile of the first sensor element with the reference profile indicative of the reference response of the first sensor element, and storing the response profile of the first sensor element, the response profile of the second sensor element, and a determination result in association with the identification information of the target sample. (Supplementary Note 2) The determination method according to Supplementary Note 1, further comprising: determining a correction value based on a deviation between the response profile of the second sensor element and a reference profile of the second sensor element; and correcting the response profile of the first sensor element using the determined correction value.(Supplementary Note 3) A determination method in which a computer executes a process of acquiring response profiles indicating responses to a target sample under the influence of error factors obtained from a first sensor element having reactivity according to a specific property of the sample and a second sensor element not having reactivity according to the specific property, deriving a correction value based on a deviation between the acquired response profile of the second sensor element to the target sample and a reference profile indicating a reference for the response of the second sensor element, correcting the acquired response profile of the first sensor element to the target sample using the derived correction value, determining the property of the target sample based on a comparison of the corrected response profile of the first sensor element with the reference profile indicating a reference for the response of the first sensor element, and storing the determination result of the property of the target sample, the response profile of the first sensor element, and the response profile of the second sensor element in association with identification information of the target sample. (Supplementary Note 4) The determination method according to Supplementary Note 3, in which the first sensor element and the second sensor element have the same direction of variation in response signals due to the influence of the error factors. (Supplementary Note 5) The determination method according to any one of Supplementary Notes 1 to 4, comprising: acquiring a response profile of the first sensor element and a response profile of the second sensor element for a plurality of samples; generating a reference profile of the first sensor element based on a statistical value of a response signal in the response profile of the first sensor element for each of the acquired samples; and generating a reference profile of the second sensor element based on a statistical value of a response signal in the response profile of the second sensor element for each of the acquired samples. (Supplementary Note 6) The determination method according to any one of Supplementary Notes 1 to 5, comprising: acquiring a response profile for a sample having the specific property and a response profile for a sample not having the specific property detected using a plurality of sensor elements; and selecting the second sensor element from the plurality of sensor elements based on each acquired response profile for each sensor element.(Supplementary Note 7) The determination method according to any one of Supplementary Notes 1 to 6, comprising: acquiring a response profile for a sample having the specific property and a response profile for a sample not having the specific property detected using a plurality of sensor elements; and selecting, from the plurality of sensor elements, a sensor element for which the dissimilarity between the response profile for the sample having the specific property and the response profile for the sample not having the specific property is less than a predetermined value as the second sensor element based on the dissimilarity between the response profile for each sensor element for the sample having the specific property and the response profile for the sample not having the specific property. (Supplementary Note 8) The determination method according to any one of Supplementary Notes 1 to 7, comprising detecting the response profile using a sensor including a plurality of sensor elements including the first sensor element and the second sensor element. (Supplementary Note 9) The determination method according to any one of Supplementary Notes 1 to 8, wherein the first sensor element and the second sensor element are biological elements. (Supplementary Note 10) The determination method according to any one of Supplementary Notes 1 to 9, wherein response profiles obtained from the first sensor element and the second sensor element under the influence of an error factor are acquired; and the first sensor element and the second sensor element have the same direction of variation in response signals due to the influence of the error factor.

[0147] The embodiments disclosed herein are illustrative in all respects and should not be considered limiting. The technical features described in each embodiment can be combined with each other, and the scope of the present invention is intended to include all modifications within the scope of the claims and equivalents thereto. The sequences shown in each embodiment are not limited, and within the scope of no contradiction, each processing step may be executed in a different order, or multiple processes may be executed in parallel. The entity that performs each process is not limited, and within the scope of no contradiction, the process of each device may be executed by another device.

[0148] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. A multiple claim (multi-multi claim) that references at least one other multiple claim may also be used.

[0149] REFERENCE SIGNS LIST 100 Determination system 1 Determination device 11 Control unit 12 Memory unit 13 Communication unit 1A Recording medium 1P Program 121 Detection DB 122 Learning model 2 Terminal device 21 Control unit 22 Memory unit 23 Communication unit 24 Display unit 25 Operation unit 2A Recording medium 2P Program 3 Detection device 4 Olfactory sensor 42 Cell

Claims

1. A determination method in which a computer executes a process of acquiring a response profile of a first sensor element, which exhibits reactivity according to a specific property of the sample, to a target sample, and a response profile of a second sensor element, which exhibits a lower reactivity according to the specific property of the sample than the first sensor element, correcting the response profile of the first sensor element, which corresponds to the identification information of the target sample, based on a comparison of the response profile of the second sensor element, which corresponds to the identification information of the target sample, with a reference profile, which indicates a reference response of the second sensor element; determining the properties of the target sample based on a comparison of the corrected response profile of the first sensor element with the reference profile, which indicates the reference response of the first sensor element; and storing the response profile of the first sensor element, the response profile of the second sensor element, and the determination result in association with the identification information of the target sample.

2. The determination method according to claim 1, wherein a correction value is determined based on the deviation between the response profile of the second sensor element and the reference profile of the second sensor element, and the response profile of the first sensor element is corrected using the determined correction value.

3. A determination method according to claim 1 or claim 2, comprising: acquiring a response profile of the first sensor element and a response profile of the second sensor element for a plurality of samples; generating a reference profile of the first sensor element based on statistical values ​​of response signals in the response profile of the first sensor element for each of the acquired samples; and generating a reference profile of the second sensor element based on statistical values ​​of response signals in the response profile of the second sensor element for each of the acquired samples.

4. A determination method according to claim 1 or claim 2, which obtains a response profile for a sample having the specific property and a response profile for a sample not having the specific property detected using a plurality of sensor elements, and selects the second sensor element from among the plurality of sensor elements based on each response profile obtained for each sensor element.

5. A determination method according to claim 1 or claim 2, which obtains a response profile for a sample having the specific property and a response profile for a sample not having the specific property detected using a plurality of sensor elements, and selects, from among the plurality of sensor elements, a sensor element for which the dissimilarity is less than a predetermined value as the second sensor element based on the dissimilarity between the response profile for the sample having the specific property and the response profile for the sample not having the specific property for each sensor element.

6. The determination method according to claim 1 or 2, wherein the response profile is detected using a sensor having a plurality of sensor elements including the first sensor element and the second sensor element.

7. The determination method according to claim 1 or 2, wherein the first sensor element and the second sensor element are biological elements.

8. A determination device comprising: a control unit that executes a process of acquiring a response profile of a first sensor element, which exhibits reactivity according to a specific property of the sample, to a target sample, and a response profile of a second sensor element, which exhibits a lower reactivity according to the specific property of the sample than the first sensor element, to the target sample; correcting the response profile of the first sensor element, which corresponds to the identification information of the target sample, based on a comparison between the response profile of the second sensor element, which corresponds to the identification information of the target sample, and a reference profile, which indicates a reference response of the second sensor element; determining the properties of the target sample based on a comparison between the corrected response profile of the first sensor element and the reference profile, which indicates the reference response of the first sensor element; and storing the response profile of the first sensor element, the response profile of the second sensor element, and the determination results in association with the identification information of the target sample.

9. A computer program that causes a computer to execute the following process: obtain a response profile of a first sensor element, which exhibits reactivity according to a specific property of the sample, to a target sample; and obtain a response profile of a second sensor element, which exhibits a lower reactivity according to the specific property of the sample than the first sensor element; correct the response profile of the first sensor element, which corresponds to the identification information of the target sample, based on a comparison between the response profile of the second sensor element, which corresponds to the identification information of the target sample, and a reference profile, which indicates a reference response of the second sensor element; determine the properties of the target sample based on a comparison between the corrected response profile of the first sensor element and the reference profile, which indicates the reference response of the first sensor element; and store the response profile of the first sensor element, the response profile of the second sensor element, and the determination result in association with the identification information of the target sample.

Images