Blood glucose level estimation device, learning device, blood glucose level estimation method, method for producing learning information, and program

Abstract

To solve the problem that, conventionally, it has been difficult to easily acquire an appropriate blood sugar value of a person in a noninvasive manner.SOLUTION: A blood sugar value estimation device 7 includes: a learning information housing unit 711 for housing learning information acquired by using two or more pieces of teacher data having one or more pieces of NIRS information acquired by using reflection light of near infrared light emitted to a human body and a blood sugar value; an NIRS acquisition unit 63 for acquiring one or more pieces of NIRS information acquired by emitting near infrared light to a user; an estimation unit 734 for acquiring an estimated blood sugar value by using the one or more NIRS information acquired by the NIRS acquisition unit 63 and the learning information; and a blood sugar value output unit 741 for outputting the blood sugar value acquired by the estimation unit 734. The blood sugar value estimation device enables an appropriate blood sugar value of a person to be acquired easily and in a noninvasive manner.SELECTED DRAWING: Figure 2

Description

Technical Field

[0004] , , , ,

[0005] , , , , ,

[0003] , , ,

[0001] The present invention relates to a blood glucose level estimation device that acquires and outputs an estimated blood glucose level.

Background Art

[0002] Conventionally, there has been a technique for non-invasively quantifying the blood glucose level of a subject (see, for example, Patent Document 1). In such a conventional technique, the concentration of a body component of a subject is determined using a calibration curve and an absorbance spectrum measured from the subject using near-infrared light. The calibration curve is obtained by obtaining a plurality of differential absorbance spectra that are differences between a plurality of near-infrared absorbance spectra of a living body and a reference absorbance spectrum selected therefrom, and synthesizing the reference absorbance spectrum of the subject measured in advance with each of the differential absorbance spectra to obtain a plurality of synthesized absorbance spectra, and creating the calibration curve by performing multivariate analysis using the obtained plurality of synthesized absorbance spectra.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the conventional technique, since learning information is not used, it has been difficult to non-invasively and easily acquire an appropriate blood glucose level of a person.

Means for Solving the Problems

[0005] The blood glucose level estimation device of the present invention comprises: a learning information storage unit that stores learning information acquired using two or more training data sets having one or more NIRS information obtained using reflected near-infrared light emitted towards the human body and blood glucose levels; a NIRS acquisition unit that acquires one or more NIRS information obtained by emitting near-infrared light towards the user; an estimation unit that acquires an estimated blood glucose level using the one or more NIRS information acquired by the NIRS acquisition unit and the learning information; and a blood glucose level output unit that outputs the blood glucose level acquired by the estimation unit.

[0006] This configuration allows for the easy and non-invasive acquisition of a person's appropriate blood glucose level.

[0007] Furthermore, the blood glucose level estimation device of this second invention, compared to the first invention, includes, as training data, one or more NIRS information obtained by emitting near-infrared light of two or more wavelengths to the human body, and as a NIRS acquisition unit, it acquires one or more NIRS information obtained by emitting near-infrared light of two or more wavelengths to the user.

[0008] This configuration allows for the non-invasive and easy acquisition of more accurate blood glucose levels using near-infrared light of two or more wavelengths.

[0009] Furthermore, the wavelength of this third invention is a blood glucose level estimation device in which, compared to the second invention, two or more wavelengths are in the range of approximately 760 nm to approximately 1300 nm.

[0010] This configuration allows for the non-invasive and easy acquisition of more accurate blood glucose levels using near-infrared light of two or more appropriate wavelengths.

[0011] Furthermore, the blood glucose level estimation device of this fourth invention is a blood glucose level estimation device in which, compared to any one of the first to third inventions, the number of near-infrared lights of two or more different wavelengths used to acquire learning information is greater than the number of near-infrared lights of two or more different wavelengths used to acquire one or more NIRS pieces of information by the NIRS acquisition unit.

[0012] This configuration allows for the easy and non-invasive acquisition of more accurate blood glucose levels.

[0013] Furthermore, the blood glucose level estimation device of the fifth invention is a blood glucose level estimation device that, in relation to any one of the first to fourth inventions, has training data consisting of one or more user dynamic attribute values ​​from among skin temperature, acceleration, pulse, pulse wave, pulse rate, blood saturated oxygen concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level of a person having a human body, and further comprises a user dynamic attribute value acquisition unit that acquires one or more user dynamic attribute values ​​from among skin temperature, acceleration, pulse, pulse wave, pulse rate, blood saturated oxygen concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level of a single user, and an estimation unit that acquires an estimated blood glucose level using one or more NIRS information acquired by the NIRS acquisition unit, one or more user dynamic attribute values ​​acquired by the user dynamic attribute value acquisition unit, and learning information.

[0014] This configuration allows for the non-invasive and easy acquisition of more accurate blood glucose levels using one or more user dynamic attribute values.

[0015] Furthermore, the blood glucose level estimation device of the sixth invention is a blood glucose level estimation device that, in addition to any one of the first to fifth inventions, has training data which consists of one or more user static attribute values ​​which are static attribute values ​​of a person having a human body, and further comprises a user static attribute value acquisition unit which acquires one or more user static attribute values ​​of a user, and an estimation unit which acquires an estimated blood glucose level using one or more NIRS information acquired by the NIRS acquisition unit, one or more user static attribute values ​​acquired by the user static attribute value acquisition unit, and learning information.

[0016] This configuration allows for the non-invasive and easy acquisition of more accurate blood glucose levels, using one or more user-static attribute values.

[0017] Furthermore, the blood glucose level estimation device of the seventh invention is a blood glucose level estimation device in which one or more user static attribute values ​​are one or more pieces of information from the user's age, height, weight, and forearm circumference, compared to the sixth invention.

[0018] This configuration allows for the non-invasive acquisition of more accurate blood glucose levels by also using one or more appropriate user static attribute values.

[0019] Furthermore, the blood glucose level estimation device of the eighth invention is a blood glucose level estimation device that, with respect to any one of the first to seventh inventions, has training data which consists of one or more device characteristic values ​​which are characteristic values ​​of a device used to acquire one or more NIRS information, and further comprises a device characteristic value acquisition unit which acquires one or more device characteristic values ​​of a device used to acquire one or more NIRS information of a user, and an estimation unit which acquires an estimated blood glucose level using one or more NIRS information acquired by the NIRS acquisition unit, one or more device characteristic values ​​acquired by the device characteristic value acquisition unit, and learning information.

[0020] This configuration allows for the non-invasive acquisition of more accurate blood glucose levels using one or more device characteristic values.

[0021] Furthermore, the blood glucose level estimation device of the ninth invention is a blood glucose level estimation device in which one or more device characteristic values ​​are one or more pieces of information from among distance information that specifies the distance between two or more light-receiving units that receive reflected light emitted from a near-infrared light-emitting unit, the number of light-receiving units, and the number of light-emitting units.

[0022] This configuration allows for the non-invasive acquisition of more accurate blood glucose levels using one or more appropriate device characteristic values.

[0023] Furthermore, the blood glucose level estimation device of the tenth invention is a blood glucose level estimation device in which, with respect to any one of the first to seventh inventions, the NIRS acquisition unit acquires one or more NIRS information or one or more NIRS information sources from two or more NIRS devices that acquire one or more NIRS information or one or more NIRS information sources that become the basis of one or more NIRS information from the user, and the blood glucose level output unit transmits the blood glucose level or blood glucose level-related information related to the blood glucose level acquired by the estimation unit to a NIRS device or a terminal device corresponding to a NIRS device.

[0024] With such a configuration, the appropriate blood glucose levels of two or more users can be easily obtained non-invasively.

[0025] In addition, the blood glucose level estimation device of the eleventh invention is a blood glucose level estimation device which, with respect to any one of the first to tenth inventions, uses two or more pieces of teacher data to perform a learning process of machine learning, and is a learning device obtained thereby.

[0026] With such a configuration, a more appropriate blood glucose level can be easily obtained non-invasively by an algorithm of machine learning.

[0027] In addition, the learning device of the twelfth invention includes a NIRS acquisition unit that acquires one or more pieces of NIRS information regarding the reflected light, which is the reflected light of near-infrared light emitted from one or more light emitting units and received by one or more light receiving units, a measured blood glucose value acquisition unit that acquires the blood glucose level of a user, a teacher data configuration unit that configures teacher data using one or more pieces of NIRS information and the blood glucose level, a learning unit that acquires learning information using the teacher data, and a storage unit that stores the learning information.

[0028] With such a configuration, learning information for non-invasively acquiring a person's appropriate blood glucose level can be obtained.

[0029] In addition, the learning device of the thirteenth invention, with respect to the twelfth invention, has one or more light emitting units that emit near-infrared light of two or more wavelengths, one or more light receiving units that receive the reflected light of near-infrared light of each of the two or more wavelengths, and the NIRS acquisition unit that acquires one or more pieces of NIRS information regarding the reflected light of near-infrared light of each of the two or more wavelengths.

[0030] With such a configuration, learning information for non-invasively acquiring a person's appropriate blood glucose level can be obtained.

[0031] Furthermore, the learning device of the fourteenth invention is a learning device that, compared to the twelfth or thirteenth invention, further comprises a user dynamic attribute value acquisition unit that acquires one or more user dynamic attribute values ​​from among the user's skin temperature, acceleration, pulse rate, pulse wave, pulse rate, blood saturated oxygen concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level, and a training data structuring unit that constructs training data using one or more NIRS information, one or more user dynamic attribute values, and blood glucose levels.

[0032] This configuration allows for obtaining learning information to non-invasively acquire more appropriate blood glucose levels for a person, using one or more user dynamic attribute values.

[0033] Furthermore, the learning device of the fifteenth invention is a learning device that, in addition to any one of the twelfth to fourteenth inventions, further comprises a user static attribute value acquisition unit that acquires one or more user static attribute values ​​which are static attribute values ​​of the user, and a training data structuring unit that constructs training data using one or more NIRS information, one or more user static attribute values ​​and blood glucose levels.

[0034] This configuration allows for obtaining learning information to non-invasively acquire more appropriate blood glucose levels for a person, using one or more user static attribute values.

[0035] Furthermore, the learning device of the sixteenth invention further comprises a device characteristic value acquisition unit that acquires one or more device characteristic values ​​of a device having one or more light receiving units and one or more light emitting units, and the training data structuring unit constructs training data using one or more NIRS information, one or more device characteristic values, and blood glucose levels, in addition to any one of the twelve to fifteen inventions, and is a learning device.

[0036] With this configuration, learning information for non-invasively acquiring more appropriate blood glucose levels in a person can be obtained using one or more device characteristic values. [Effects of the Invention]

[0037] The blood glucose level estimation device according to the present invention allows for the easy and non-invasive acquisition of an appropriate blood glucose level for a person. [Brief explanation of the drawing]

[0038] [Figure 1] Conceptual diagram of the blood glucose level estimation system C in Embodiment 1 [Figure 2] This diagram shows an example of a block diagram for the blood glucose estimation system C. [Figure 3] Flowchart explaining an example of operation of the blood glucose level estimation device 7. [Figure 4] This figure shows an example of the NIRS device 6. [Figure 5] This figure shows an example of the NIRS device 6. [Figure 6] This figure shows an example of the NIRS device 6. [Figure 7] This diagram shows an example of a block diagram for the blood glucose estimation system C. [Figure 8] This figure shows a conceptual diagram of the blood glucose level estimation system D in Embodiment 2. [Figure 9] Block diagram of the blood glucose estimation system D. [Figure 10] Flowchart explaining an example of operation of the blood glucose level estimation device 7. [Figure 11] Block diagram of the learning system E in Embodiment 3 [Figure 12] Block diagram of the learning system E [Figure 13] A flowchart illustrating an example of the operation of the learning device 9. [Figure 14] A diagram illustrating a specific example of the operation of the learning system E. [Figure 15] Overview of the computer system in the above embodiment [Figure 16] Block diagram of the computer system [Modes for carrying out the invention]

[0039] The embodiments of the blood glucose level estimation device and the like will be described below with reference to the drawings. In the embodiments, components that are denoted by the same reference numerals perform the same operation, and therefore, further explanation may be omitted.

[0040] (Embodiment 1) In this embodiment, a blood glucose estimation system including a blood glucose estimation device is described, which applies NIRS information obtained by emitting near-infrared light (hereinafter referred to as "NIRS" as appropriate) towards a user to training information (e.g., a learner) composed of two or more training data sets having NIRS information obtained by emitting near-infrared light (hereinafter referred to as "NIRS") towards the human body and blood glucose levels, to acquire and output blood glucose levels. Note that the training data may include information other than NIRS information. Information other than NIRS information may include, for example, one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values. The blood glucose estimation device may also receive NIRS information from a device and transmit the estimated blood glucose level to an external device. Furthermore, the training information may be, for example, a learner obtained by machine learning.

[0041] Furthermore, in this embodiment, a blood glucose estimation system including a blood glucose estimation device that estimates blood glucose levels using learning information acquired using two or more wavelengths in order to obtain training data will be described. Note that the number of wavelengths used during learning may be greater than the number of wavelengths used during prediction.

[0042] Figure 1 is an example of a conceptual diagram of the blood glucose level estimation system C in this embodiment. The blood glucose level estimation system C comprises a NIRS device 6 and a blood glucose level estimation device 7. The blood glucose level estimation system C may also be a system comprising two or more NIRS devices 6 and one blood glucose level estimation device 7. The NIRS device 6 and the blood glucose level estimation device 7 may also be a single integrated device.

[0043] The NIRS device 6 comprises a near-infrared light emitter and a light receiver, acquires NIRS information or a source of NIRS information described later, and passes it to the blood glucose level estimation device 7. The NIRS device 6 is preferably watch-type, but its shape is not limited. The NIRS device 6 can be any device that emits NIRS light towards any part of the human body and receives reflected light.

[0044] The blood glucose level estimation device 7 acquires NIRS information based on information from the NIRS device 6 and estimates blood glucose levels using this NIRS information and learned information. The blood glucose level estimation device 7 can be any type of device, such as a personal computer, smartphone, tablet, or server. The server can be any type, such as a cloud server or ASP server.

[0045] Figure 2 is a block diagram of the blood glucose level estimation system C in this embodiment. The NIRS device 6 that constitutes the blood glucose level estimation system C includes a light-emitting unit 61, a light-receiving unit 62, a NIRS acquisition unit 63, and a NIRS output unit 64.

[0046] The blood glucose level estimation device 7 comprises a storage unit 71, a reception unit 72, a processing unit 73, and an output unit 74. The storage unit 71 comprises a learning information storage unit 711, a user information storage unit 712, and a device characteristic value storage unit 713. The reception unit 72 comprises a NIRS information reception unit 721. The processing unit 73 comprises a user dynamic attribute value acquisition unit 731, a user static attribute value acquisition unit 732, a device characteristic value acquisition unit 733, and an estimation unit 734. The output unit 74 comprises a blood glucose level output unit 741.

[0047] The light-emitting unit 61, which constitutes the NIRS apparatus 6, emits near-infrared light. The light-emitting unit 61 normally emits near-infrared light towards the human body. For example, the light-emitting unit 61 emits near-infrared light towards parts of the human body such as the wrist, fingertips, and earlobes. However, the location where the light-emitting unit 61 emits near-infrared light is not limited.

[0048] The NIRS apparatus 6 may have two or more light-emitting units 61. It is preferable that the light-emitting units 61 emit near-infrared light with wavelengths in the range of approximately 760 nm to approximately 1300 nm. For example, each of the two light-emitting units 61 constituting the NIRS apparatus 6 preferably emits near-infrared light with wavelengths of approximately 1200 nm and approximately 1300 nm, respectively. One light-emitting unit 61 may emit two or more near-infrared lights with different wavelengths. Alternatively, one light-emitting unit 61 may emit two or more near-infrared lights with different wavelengths at different times. It is preferable that the wavelengths of each of the two or more near-infrared lights are in the range of approximately 760 nm to approximately 1300 nm. For example, the wavelengths of each of the two or more near-infrared lights include wavelengths of approximately 1200 nm and approximately 1300 nm. For example, the wavelengths of each of the two or more near-infrared lights include wavelengths of approximately 760 nm and approximately 850 nm. The wavelengths of the two or more near-infrared light sources are, for example, approximately 760 nm, approximately 850 nm, approximately 1200 nm, and approximately 1300 nm. The light-emitting unit 61 can be realized by, for example, a light-emitting diode, an LED, etc., but the means of realization are not limited.

[0049] The light-receiving unit 62 receives the reflected light when the light-emitting unit 61 emits near-infrared light towards the human body. The light-receiving unit 62 may also receive the transmitted light when the light-emitting unit 61 emits near-infrared light towards the human body. The light-receiving unit 62 can be implemented using a known device.

[0050] The NIRS acquisition unit 63 acquires one or more NIRS pieces of information obtained by emitting near-infrared light to the user. The NIRS acquisition unit 63 acquires one or more NIRS pieces of information which are information about the light received by the light receiving unit 62. The NIRS acquisition unit 63 usually acquires one or more NIRS pieces of information which are the reflected light of the near-infrared light emitted by the light emitting unit 61 and the reflected light received by the light receiving unit 62.

[0051] The NIRS acquisition unit 63 preferably acquires one or more pieces of NIRS information obtained by emitting near-infrared light of two or more wavelengths to the user.

[0052] NIRS information is, for example, an optical signal. NIRS information is, for example, a collection of light intensities for each wavelength. NIRS information is, for example, the intensity of reflected near-infrared light at wavelengths of approximately 1200 nm and approximately 1300 nm.

[0053] The NIRS acquisition unit 63 may acquire one or more user dynamic attribute values. One or more user dynamic attribute values ​​are, for example, one or more pieces of information from among a user's skin temperature, acceleration, pulse rate, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level.

[0054] The NIRS acquisition unit 63 may, for example, have one or more sensors and use each of these sensors to acquire one or more user dynamic attribute values. The NIRS acquisition unit 63 may, for example, include a temperature sensor to acquire the user's skin temperature. The NIRS acquisition unit 63 may, for example, include a gyroscope or accelerometer to acquire acceleration. The NIRS acquisition unit 63 may, for example, include a pulse sensor to acquire pulse rate. The NIRS acquisition unit 63 may, for example, include a pulse wave meter to acquire pulse wave. The NIRS acquisition unit 63 may, for example, include a heart rate monitor to acquire pulse rate. The NIRS acquisition unit 63 may, for example, include a blood oxygen saturation meter to acquire blood saturated oxygen concentration. The NIRS acquisition unit 63 may, for example, include a blood pressure monitor to acquire blood pressure. The NIRS acquisition unit 63 may, for example, include an image sensor to acquire skin color. The NIRS acquisition unit 63 may, for example, include a body composition analyzer to acquire muscle mass. The NIRS acquisition unit 63 includes, for example, a hemoglobin measuring device to acquire the amount of hemoglobin in the blood.

[0055] The NIRS acquisition unit 63 may acquire one or more user static attribute values ​​from a storage unit (not shown) of the NIRS device 6.

[0056] The NIRS acquisition unit 63 may acquire one or more device characteristic values ​​from a storage unit (not shown) of the NIRS device 6.

[0057] The NIRS output unit 64 outputs one or more NIRS information acquired by the NIRS acquisition unit 63. The NIRS output unit 64 may also output one or more user dynamic attribute values ​​acquired by the NIRS acquisition unit 63. The NIRS output unit 64 may also output one or more user static attribute values ​​acquired by the NIRS acquisition unit 63. The NIRS output unit 64 may also output one or more device characteristic values ​​acquired by the NIRS acquisition unit 63.

[0058] In this context, output typically refers to transmission to the blood glucose estimator 7. When the NIRS device 6 and the blood glucose estimator 7 are an integrated device, output refers to handover to the blood glucose estimator 7. The NIRS output unit 64 may also store one or more pieces of NIRS information on a recording medium, transmit them to other devices, or hand over processing results to other processing devices or other programs.

[0059] The storage unit 71, which constitutes the blood glucose level estimation device 7, stores various types of information. These types of information include, for example, learning information (described later), user information (described later), and device characteristic values ​​(described later).

[0060] The learning information storage unit 711 stores learning information. Learning information is information used when acquiring blood glucose levels using one or more NIRS information sets. Learning information is information acquired using two or more training data sets. Training data sets consist of one or more NIRS information sets and blood glucose levels. NIRS information is information acquired using reflected near-infrared light emitted onto the human body. NIRS information may also be information acquired using transmitted light.

[0061] In the training data, for example, 1 or more NIRS data points are explanatory variables, and blood glucose levels are the dependent variable.

[0062] The training data preferably has one or more user dynamic attribute values. User dynamic attribute values ​​are dynamic attribute values ​​of the user. User dynamic attribute values ​​are information that can change moment by moment or frequently. Preferably, the one or more user dynamic attribute values ​​are one or more pieces of information from among the following for a person with a human body: skin temperature, acceleration, pulse, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level.

[0063] The training data preferably has one or more user static attribute values. User static attribute values ​​are static attribute values ​​of a person. One or more user static attribute values ​​include, for example, gender, age, age group, height, weight, and forearm circumference.

[0064] The training data preferably has one or more device characteristic values. The device characteristic values ​​are characteristic values ​​of the device (in this case, the NIRS apparatus 6) used to acquire one or more pieces of NIRS information. The one or more device characteristic values ​​are, for example, distance information that specifies the distance between two or more light-receiving units 62, the number of light-receiving units 62, and the number of light-emitting units 61.

[0065] Furthermore, the learning information includes, for example, a learning device, a correspondence table, and an arithmetic formula. The details of the learning information will be explained below. It is preferable that the learning information is information acquired by the learning device 9, which will be described later. (1) When the learning information is a learning device

[0066] The learner is information obtained by performing machine learning training using two or more training data points. The machine learning algorithm is not restricted. Examples of machine learning algorithms that can be used for training include deep learning, decision trees, random forests, and SVR, but the type of machine learning is not restricted. Furthermore, the learner takes information containing one or more NIRS data points as input and outputs blood glucose levels. The learner can also be called a classifier, predictor, or model. The module used for machine learning training is also not restricted. Examples of such modules include various functions within the TensorFlow library, the R language's random forest module, and tinySVM. (2) When the learning information is a correspondence table

[0067] The correspondence table is information that has two or more correspondence pieces showing the correspondence between information containing one or more NIRS information and blood glucose levels. The correspondence pieces may also be information containing one or more NIRS information and blood glucose levels. The correspondence pieces may also be information containing a pointer to information containing one or more NIRS information and a pointer to blood glucose levels. The correspondence pieces only need to be information that associates information containing one or more NIRS information with blood glucose levels. The correspondence table is preferably information acquired by the learning device 9 described later. Note that the information containing one or more NIRS information may be just one or more NIRS information, or it may also be information that includes one or more of the following: one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values. (3) When the learning information is an arithmetic expression

[0068] The calculation formula is a formula that takes one or more NIRS information values ​​as parameters to obtain blood glucose levels. The calculation formula may also take one or more of the following information values ​​as parameters: one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0069] The user information storage unit 712 stores one or more user information. The user information has one or more static user attribute values. One or more static user attribute values ​​are, for example, the user's age, age group, height, weight, and forearm circumference. The user information may also have one or more dynamic user attribute values. The user information may also be associated with a user identifier that identifies the user. The user identifier is, for example, a user ID, email address, mobile phone number, etc.

[0070] The device characteristic value storage unit 713 stores one or more sets of device characteristic values. The set of one or more device characteristic values ​​may be associated with a user identifier.

[0071] The reception unit 72 receives various instructions and information. These instructions and information include, for example, user information, one or more static user attribute values, one or more dynamic user attribute values, and one or more device characteristic values.

[0072] The reception unit 72 accepts, for example, one or more user static attribute values ​​from the user. The reception unit 72 accepts, for example, one or more device characteristic values ​​from the user.

[0073] The means of inputting various instructions and information can be anything, such as a touch panel, keyboard, mouse, or menu screen.

[0074] The reception unit 72 receives various instructions and information from other devices, for example. Other devices include, for example, the NIRS device 6 and a user terminal device (not shown).

[0075] The reception unit 72 receives, for example, one or more types of information from the NIRS device 6, including one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0076] The NIRS information receiving unit 721 receives one or more pieces of NIRS information. For example, the NIRS information receiving unit 721 receives one or more pieces of NIRS information from the NIRS device 6.

[0077] The processing unit 73 performs various processes. These various processes include, for example, the processes performed by the user dynamic attribute value acquisition unit 731, the user static attribute value acquisition unit 732, the device characteristic value acquisition unit 733, and the estimation unit 734.

[0078] The user dynamic attribute value acquisition unit 731 acquires one or more user dynamic attribute values. For example, the user dynamic attribute value acquisition unit 731 acquires one or more user dynamic attribute values ​​received by the reception unit 72. Alternatively, the blood glucose level estimation device 7 may have one or more sensors that acquire one or more user dynamic attribute values, and may acquire one or more user dynamic attribute values ​​using each of these sensors. One or more user dynamic attribute values ​​are, for example, one or more pieces of information from among a user's skin temperature, acceleration, pulse rate, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level.

[0079] The user static attribute value acquisition unit 732 acquires one or more user static attribute values ​​of the user. The user static attribute value acquisition unit 732 acquires one or more user static attribute values ​​from, for example, the user information storage unit 712. The user static attribute value acquisition unit 732 acquires one or more user static attribute values ​​corresponding to the received user identifier from the user information storage unit 712. The user static attribute value acquisition unit 732 acquires one or more received user static attribute values. The one or more user static attribute values ​​are, for example, one or more pieces of information from the user's age, age group, height, weight, and forearm circumference.

[0080] The device characteristic value acquisition unit 733 acquires one or more device characteristic values ​​of a device used to acquire one or more NIRS information of a user. The device characteristic value acquisition unit 733 acquires one or more device characteristic values ​​from, for example, the device characteristic value storage unit 713. The device characteristic value acquisition unit 733 acquires one or more device characteristic values ​​corresponding to the received user identifier from the device characteristic value acquisition unit 733. The device characteristic value acquisition unit 733 acquires one or more received device characteristic values.

[0081] The device characteristic value acquisition unit 733 may inspect the NIRS device 6 and automatically acquire one or more device characteristic values. For example, the device characteristic value acquisition unit 733 may perform image recognition on a camera image taken of the NIRS device 6, extract the contour of the light-emitting unit 61, calculate the coordinates of the centroid of the figure enclosed by the contour, and acquire distance information which is the distance between the centroids of two or more contours.

[0082] The one or more device characteristic values ​​are, for example, one or more pieces of information from among distance information that specifies the distance between two or more light-receiving units that receive reflected light emitted from the near-infrared light-emitting unit 61, the number of light-receiving units 62, and the number of light-emitting units 61.

[0083] The estimation unit 734 obtains an estimated blood glucose level using one or more NIRS data points acquired by the NIRS acquisition unit 63 and the learning information. The estimation unit 734 performs prediction processing using one or more NIRS data points acquired by the NIRS acquisition unit 63 and the learning information to obtain the blood glucose level.

[0084] The estimation unit 734 obtains an estimated blood glucose level using, for example, one or more NIRS information obtained by the NIRS acquisition unit 63, one or more user dynamic attribute values ​​obtained by the user dynamic attribute value acquisition unit 731, and learning information.

[0085] The estimation unit 734 obtains an estimated blood glucose level using, for example, one or more NIRS information obtained by the NIRS acquisition unit 63, one or more user static attribute values ​​obtained by the user static attribute value acquisition unit 732, and learning information.

[0086] The estimation unit 734 obtains an estimated blood glucose level using, for example, one or more NIRS information acquired by the NIRS acquisition unit 63, one or more device characteristic values ​​acquired by the device characteristic value acquisition unit 733, and learning information.

[0087] The estimation unit 734 obtains an estimated blood glucose level using, for example, one or more NIRS information acquired by the NIRS acquisition unit 63, two or more types of information from one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values, and learning information.

[0088] The following describes examples of the prediction process performed by the estimation unit 734 for each case: when the learning information is a learning device, when it is a correspondence table, and when it is a calculation formula. (1) When the learning information is a learning device

[0089] The estimation unit 734 provides the one or more NIRS data points acquired by the NIRS acquisition unit 63 and the training data to a machine learning prediction processing module, executes the module, performs prediction processing, and obtains the estimated blood glucose level. The machine learning algorithm is not limited. For example, deep learning, decision trees, random forests, SVR, etc., can be used as the machine learning prediction processing algorithm, but the type of machine learning is not limited. The module that performs the machine learning prediction processing is also not limited. For example, the module can be various functions in the TensorFlow library, the R language random forest module, tinySVM, etc.

[0090] The estimation unit 734 provides, for example, one or more types of information from among one or more NIRS information acquired by the NIRS acquisition unit 63, one or more user dynamic attribute values ​​acquired by the user dynamic attribute value acquisition unit 731, one or more user static attribute values ​​acquired by the user static attribute value acquisition unit 732, and one or more device characteristic values ​​acquired by the device characteristic value acquisition unit 733, along with learning information, to a machine learning prediction processing module, executes the module, performs prediction processing, and obtains the estimated blood glucose level. (2) When the learning information is a correspondence table

[0091] The estimation unit 734 determines one or more corresponding pieces of information from the corresponding information in the correspondence table that match the approximation conditions for the vector containing one or more NIRS information acquired by the NIRS acquisition unit 63, and obtains the estimated blood glucose value using the blood glucose value associated with each of the one or more corresponding pieces of information. The approximation conditions are, for example, having the greatest degree of approximation, having the degree of approximation within a threshold, or having the degree of approximation greater than a threshold. The degree of approximation is information that becomes smaller as the distance between the two vectors increases, and is calculated, for example, by a decreasing function with the distance between the two vectors as a parameter. The method for obtaining the degree of approximation of the two vectors is not specified. A vector containing one or more NIRS information is, for example, a vector whose elements are one or more individual NIRS information. Alternatively, a vector containing one or more NIRS information is, for example, a vector whose elements are one or more individual NIRS information and one or more of the following: one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0092] The estimation unit 734 obtains, for example, the blood glucose value from a correspondence table that corresponds to the vector that most closely approximates the vector containing one or more NIRS information obtained by the NIRS acquisition unit 63. This blood glucose value is the estimated blood glucose value.

[0093] The estimation unit 734 obtains, for example, representative values ​​of two or more blood glucose levels from corresponding information whose degree of similarity to a vector containing one or more NIRS information obtained by the NIRS acquisition unit 63 matches the approximation condition. These representative values ​​are the estimated blood glucose levels. The representative values ​​are, for example, the mean, median, and weighted mean. The weighted mean is a value obtained by giving greater weight to vectors that are closer in distance from each other. (3) When the learning information is an arithmetic expression

[0094] The estimation unit 734 provides one or more NIRS information obtained by the NIRS acquisition unit 63 to a calculation formula, executes the calculation formula, and obtains the estimated blood glucose level.

[0095] The estimation unit 734 provides, for example, one or more types of information from one or more NIRS information acquired by the NIRS acquisition unit 63, one or more user dynamic attribute values ​​acquired by the user dynamic attribute value acquisition unit 731, one or more user static attribute values ​​acquired by the user static attribute value acquisition unit 732, and one or more device characteristic values ​​acquired by the device characteristic value acquisition unit 733 to a calculation formula, executes the calculation formula, and obtains the estimated blood glucose level.

[0096] The output unit 74 outputs various types of information. These types of information include, for example, blood glucose levels. Here, output is a concept that includes displaying on a screen, projecting using a projector, printing with a printer, outputting sound, transmitting to an external device, storing on a recording medium, and transferring processing results to other processing devices or other programs.

[0097] The blood glucose output unit 741 outputs, for example, the blood glucose value acquired by the estimation unit 734. The blood glucose output unit 741 may also transmit the blood glucose value or blood glucose-related information acquired by the estimation unit 734 to the NIRS device 6 or a terminal device corresponding to the NIRS device 6. The terminal device may be, for example, a smartphone or a tablet device.

[0098] The NIRS acquisition unit 63, processing unit 73, user dynamic attribute value acquisition unit 731, user static attribute value acquisition unit 732, device characteristic value acquisition unit 733, and estimation unit 734 can typically be implemented using a processor, memory, etc. The processing procedures of the processing unit 73, etc., are typically implemented in software, and this software is recorded on a recording medium such as ROM. However, it may also be implemented in hardware (dedicated circuitry). The processor can be, for example, a CPU, MPU, GPU, etc., and its type is not limited.

[0099] The NIRS output unit 64 can be implemented, for example, by wireless or wired communication means.

[0100] The storage unit 71, the learning information storage unit 711, the user information storage unit 712, and the device characteristic value storage unit 713 are preferably made of non-volatile recording media, but can also be made of volatile recording media.

[0101] The process by which information is stored in the storage unit 71, etc. is not relevant. For example, information may be stored in the storage unit 71, etc. via a recording medium, information transmitted via a communication line, etc. may be stored in the storage unit 71, etc., or information input via an input device may be stored in the storage unit 71, etc.

[0102] The reception unit 72 and the NIRS information reception unit 721 can be implemented, for example, by wireless or wired communication means. The reception unit 72 can also be implemented, for example, by a device driver for an input means such as a touch panel or keyboard, or by control software for a menu screen.

[0103] The output unit 74 and the blood glucose output unit 741 may be implemented, for example, by wireless or wired communication means. The output unit 74 and the blood glucose output unit 741 may also be implemented by output devices such as a display or a speaker. The output unit 74 and the blood glucose output unit 741 may be implemented by driver software for the output device, or by driver software for the output device and the output device, etc.

[0104] Next, we will explain an example of the operation of the blood glucose estimation system C. First, we will explain an example of the operation of the NIRS device 6.

[0105] Each of the one or more light-emitting units 61 of the NIRS device 6 emits near-infrared light. Each of the one or more light-receiving units 62 receives the reflected near-infrared light. Next, the NIRS acquisition unit 63 acquires one or more pieces of NIRS information from the received signal. The NIRS acquisition unit 63 also acquires, for example, one or more user dynamic attribute values. The NIRS acquisition unit 63 also acquires, for example, one or more user static attribute values. The NIRS acquisition unit 63 also acquires, for example, one or more device characteristic values. Next, the NIRS output unit 64 passes the one or more pieces of NIRS information acquired by the NIRS acquisition unit 63 to the blood glucose level estimation device 7. The one or more pieces of NIRS information may be just one or more pieces of NIRS information, or it may be one or more pieces of NIRS information and one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values. The NIRS output unit 64 transmits, for example, one or more pieces of NIRS information to the blood glucose level estimation device 7.

[0106] Next, an example of the operation of the blood glucose level estimation device 7 will be explained using the flowchart in Figure 3.

[0107] (Step S301) The NIRS information receiving unit 721 determines whether it has received one or more pieces of NIRS information. If it has received one or more pieces of NIRS information, it proceeds to step S302; otherwise, it returns to step S301. At this point, the NIRS information receiving unit 721 may also receive one or more pieces of information from among one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0108] (Step S302) The processing unit 73 determines whether or not to use user dynamic attribute values ​​when acquiring blood glucose levels. If user dynamic attribute values ​​are to be used, the process proceeds to step S303; otherwise, the process proceeds to step S304. The decision of whether or not to use user dynamic attribute values ​​may be predetermined, based on user instructions, or determined based on whether or not user dynamic attribute values ​​can be acquired.

[0109] (Step S303) The user dynamic attribute value acquisition unit 731 acquires one or more user dynamic attribute values.

[0110] (Step S304) The processing unit 73 determines whether or not to use user static attribute values ​​when acquiring blood glucose levels. If user static attribute values ​​are to be used, the process proceeds to step S305; otherwise, the process proceeds to step S306. The decision of whether or not to use user static attribute values ​​may be predetermined, based on user instructions, or determined by whether or not user static attribute values ​​are stored.

[0111] (Step S305) The user static attribute value acquisition unit 732 acquires one or more user static attribute values.

[0112] (Step S306) The processing unit 73 determines whether or not to use device characteristic values ​​when acquiring blood glucose levels. If device characteristic values ​​are to be used, the process proceeds to step S307; otherwise, the process proceeds to step S308. The decision of whether or not to use device characteristic values ​​may be predetermined, based on user instructions, or determined based on whether or not device characteristic values ​​are stored.

[0113] (Step S307) The device characteristic value acquisition unit 733 acquires one or more device characteristic values.

[0114] (Step S308) The estimation unit 734 prepares data for prediction processing. The estimation unit 734 obtains, for example, a vector to be given to machine learning along with the learner. The data for prediction processing may consist of only one or more pieces of NIRS information. The data for prediction processing may also consist of one or more pieces of NIRS information and information that includes one or more types of information from one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0115] (Step S309) The estimation unit 734 uses the data prepared in step S308 and the learning information to perform prediction processing and obtain the blood glucose level. An example of the prediction processing is described above. The estimation unit 734

[0116] (Step S310) The blood glucose output unit 741 outputs the blood glucose value obtained in step S309. Return to step S301. The blood glucose output unit 741 transmits the blood glucose value to an external device, such as the NIRS device 6.

[0117] In the flowchart in Figure 3, processing is terminated by power-off or processing termination interrupts.

[0118] The following describes a specific example of the operation of the blood glucose level estimation system C in this embodiment. Figure 1 shows a conceptual diagram of the blood glucose level estimation system C.

[0119] Let's assume that the learning information storage unit 711 of the smartphone-based blood glucose estimation device 7 stores a learner acquired by a machine learning algorithm (for example, a random forest). Let's assume that this learner was acquired by a learning device 9, described later, using a large amount of training data having explanatory variables containing one or more NIRS information and an objective variable being blood glucose level. Let's also assume that the one or more NIRS information here refers to the light intensity of reflected near-infrared light at each wavelength. Here, let's assume that the NIRS information is a 4-dimensional vector (light intensity of reflected near-infrared light at a wavelength of 760 nm, light intensity of reflected near-infrared light at a wavelength of 850 nm, light intensity of reflected near-infrared light at a wavelength of 1200 nm, light intensity of reflected near-infrared light at a wavelength of 1300 nm). Let's also assume that the training data has one or more user static attribute values, one or more user dynamic attribute values, and one or more device characteristic values. Here, let's assume that the one or more user static attribute values ​​are gender, age, height, weight, and forearm circumference. Furthermore, it is assumed that the user dynamic attribute values ​​of one or more are the person's skin temperature, acceleration, pulse, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level. In addition, it is assumed that the device characteristic value of one or more is distance information that identifies the distance between the two light-receiving units 62. In other words, the structure of the training data is assumed to be "<explanatory variables>(light intensity of reflected near-infrared light with a wavelength of 760 nm, light intensity of reflected near-infrared light with a wavelength of 850 nm, light intensity of reflected near-infrared light with a wavelength of 1200 nm, light intensity of reflected near-infrared light with a wavelength of 1300 nm, gender, age, height, weight, forearm circumference, skin temperature, acceleration, pulse, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, blood hemoglobin level, distance information of the light-receiving unit 62),<dependent variable> blood glucose level".

[0120] Furthermore, as shown in Figure 4, the NIRS apparatus 6 is assumed to have a light-emitting unit 61(1) and a light-emitting unit 61(2), and two light-receiving units 62 corresponding to each light-emitting unit 61. The light-emitting unit 61(1) emits near-infrared light with wavelengths of 760 nm and 850 nm. The light-emitting unit 61(2) emits near-infrared light with wavelengths of 1200 nm and 1300 nm. The light-receiving unit 62(1) receives the reflected light of the near-infrared light emitted from the light-emitting unit 61(1). The light-receiving unit 62(2) receives the reflected light of the near-infrared light emitted from the light-emitting unit 61(2). In such cases, the NIRS information used by the blood glucose estimator 7 to estimate blood glucose levels is a four-dimensional vector (light intensity of reflected near-infrared light with a wavelength of 760 nm, light intensity of reflected near-infrared light with a wavelength of 850 nm, light intensity of reflected near-infrared light with a wavelength of 1200 nm, and light intensity of reflected near-infrared light with a wavelength of 1300 nm). Note that 401 in Figure 4 represents the internal tissue of the human body. The upper part of 401 is closer to the skin surface, and the lower part is deeper.

[0121] The light-emitting unit 61 and light-receiving unit 62 of the NIRS device 6 may be configured as shown in Figure 5. In Figure 5, multiple reflected light beams of near-infrared light emitted from a single light-emitting unit 61, each with a different optical path length, are received by different light-receiving units 62(2). Here, there are two reflected light beams with different optical path lengths, but there may be three or more. Also, the light-emitting unit 61 in Figure 5 emits near-infrared light with a wavelength of 760 nm and near-infrared light with a wavelength of 850 nm. In this case, the NIRS information used by the blood glucose level estimation device 7 for blood glucose level estimation is a four-dimensional vector (light intensity of reflected light of short-path-length near-infrared light with a wavelength of 760 nm, light intensity of reflected light of short-path-length near-infrared light with a wavelength of 850 nm, light intensity of reflected light of long-path-length near-infrared light with a wavelength of 1200 nm, and light intensity of reflected light of long-path-length near-infrared light with a wavelength of 1300 nm). Note that 501 in Figure 5 represents the internal tissue of the human body. The upper part of 501 is closer to the skin surface, and the lower part is deeper.

[0122] Furthermore, the light-emitting unit 61 and light-receiving unit 62 of the NIRS device 6 may be configured as shown in Figure 6. In Figure 6, there is one light-emitting unit 61 and one light-receiving unit 62. The light-emitting unit 61 emits near-infrared light of several different wavelengths. Here, the light-emitting unit 61 emits near-infrared light with a wavelength of 760 nm, near-infrared light with a wavelength of 850 nm, near-infrared light with a wavelength of 1200 nm, and near-infrared light with a wavelength of 1300 nm. In this case, the NIRS information used by the blood glucose level estimation device 7 to estimate blood glucose levels is a four-dimensional vector (light intensity of reflected near-infrared light with a wavelength of 760 nm, light intensity of reflected near-infrared light with a wavelength of 850 nm, light intensity of reflected near-infrared light with a wavelength of 1200 nm, and light intensity of reflected near-infrared light with a wavelength of 1300 nm). Note that 601 in Figure 6 represents the internal tissue of the human body. The upper part of 501 is closer to the skin surface, while the lower part is deeper.

[0123] Furthermore, as shown in Figures 4 to 6, the light-emitting unit 61 is preferably constantly emitting near-infrared light of two or more wavelengths, but it may also emit light in a time-division manner.

[0124] Furthermore, the NIRS device 6 in this context is said to have various sensors such as a temperature sensor and an acceleration sensor, and is capable of acquiring one or more user dynamic attribute values ​​as described above.

[0125] Furthermore, it is assumed that the user information storage unit 712 of the blood glucose level estimation device 7 stores one or more of the above-mentioned static user attribute values. Also, it is assumed that the device characteristic value storage unit 713 of the blood glucose level estimation device 7 stores one or more of the above-mentioned device characteristic values.

[0126] In the above situation, for example, one or more light-emitting units 61 of a wristwatch-type NIRS device 6 emit near-infrared light of approximately 760 nm, approximately 850 nm, approximately 1200 nm, and approximately 1300 nm onto the wrist area of ​​the human body. Then, one or more light-receiving units 62 of the NIRS device 6 receive the reflected light of each wavelength. Next, a NIRS acquisition unit 63 acquires one or more pieces of NIRS information. Here, one or more pieces of NIRS information are a four-dimensional vector (light intensity of reflected near-infrared light of 760 nm wavelength, light intensity of reflected near-infrared light of 850 nm wavelength, light intensity of reflected near-infrared light of 1200 nm wavelength, light intensity of reflected near-infrared light of 1300 nm wavelength).

[0127] Next, the NIRS output unit 64 transmits four NIRS information values ​​and one or more user dynamic attribute values ​​to the blood glucose level estimation device 7.

[0128] Next, the reception unit 72 of the blood glucose estimation device 7 receives four NIRS information items and one or more user dynamic attribute values ​​from the NIRS device 6.

[0129] Next, the user static attribute value acquisition unit 732 acquires one or more user static attribute values ​​from the user information storage unit 712.

[0130] Furthermore, the device characteristic value acquisition unit 733 acquires one or more device characteristic values ​​from the device characteristic value storage unit 713.

[0131] Next, the estimation unit 734 uses the four received NIRS information values, one or more user dynamic attribute values, one or more acquired user static attribute values, and one or more acquired device characteristic values ​​to construct data to be provided to the machine learning prediction module. The structure of this data is as follows: (light intensity of reflected near-infrared light with a wavelength of 760 nm, light intensity of reflected near-infrared light with a wavelength of 850 nm, light intensity of reflected near-infrared light with a wavelength of 1200 nm, light intensity of reflected near-infrared light with a wavelength of 1300 nm, gender, age, height, weight, forearm circumference, skin temperature, acceleration, pulse, pulse wave, pulse rate, blood saturated oxygen concentration, blood blood pressure, skin color, muscle mass, blood hemoglobin level, distance information of the light receiving unit 62).

[0132] Next, the estimation unit 734 provides the configured data and the learner in the learning information storage unit 711 to a machine learning prediction module (for example, a random forest prediction module), executes the module, and obtains the estimated blood glucose level.

[0133] Next, the blood glucose output unit 741 outputs the acquired blood glucose value.

[0134] In summary, according to this embodiment, it is possible to obtain an appropriate blood glucose level for a person in a non-invasive manner.

[0135] In this embodiment, the NIRS device 6 and the blood glucose level estimation device 7 may be integrated into a single device. In this case, an example of a block diagram of the blood glucose level estimation system C in which the NIRS device 6 and the blood glucose level estimation device 7 are integrated is shown in Figure 7. In Figure 7, there may be multiple light-emitting units 61 and light-receiving units 62.

[0136] Furthermore, in this embodiment, the number of near-infrared light beams of two or more different wavelengths used to acquire learning information may be greater than the number of near-infrared light beams of two or more different wavelengths used for the process of estimating blood glucose levels.

[0137] Furthermore, the processing in this embodiment may be implemented in software. This software may be distributed by software download or the like. Alternatively, this software may be recorded on a recording medium such as a CD-ROM and distributed. This also applies to other embodiments in this specification. The software that implements the blood glucose level estimation device 7 in this embodiment is the following program. In other words, this program causes a computer that can access a learning information storage unit, which stores learning information acquired using two or more training data sets having one or more NIRS information obtained using reflected near-infrared light emitted from the human body and blood glucose levels, to function as a NIRS acquisition unit that acquires one or more NIRS information obtained by emitting near-infrared light to the user, an estimation unit that acquires an estimated blood glucose level using the one or more NIRS information acquired by the NIRS acquisition unit and the learning information, and a blood glucose level output unit that outputs the blood glucose level acquired by the estimation unit.

[0138] (Embodiment 2) In this embodiment, the difference from Embodiment 1 is that the NIRS device and the blood glucose estimation device have a many-to-one relationship. In other words, in this embodiment, each user has their own NIRS device. The blood glucose estimation device receives NIRS information from one or more users from the NIRS device, estimates the blood glucose level, and functions as a server to transmit the blood glucose level. The destination of the blood glucose level can be the NIRS device or another device.

[0139] Figure 8 is a conceptual diagram of the blood glucose estimation system D in this embodiment. The blood glucose estimation system D comprises one or more NIRS devices 8 and a blood glucose estimation device 7. In addition to the configuration of the NIRS device 6, the NIRS device 8 has the function of receiving blood glucose values ​​from the NIRS device 8 and outputting those blood glucose values. The NIRS device 8 that constitutes the blood glucose estimation system D may have the same configuration as the NIRS device 6. The blood glucose estimation device 7 is a so-called server. The server can be of any type, such as a cloud server or an ASP server. Each of the one or more NIRS devices 8 and the blood glucose estimation device 7 can communicate with each other via a network such as the Internet or a LAN.

[0140] Figure 9 is a block diagram of the blood glucose level estimation system D in this embodiment. The NIRS device 8 that constitutes the blood glucose level estimation system D includes a light-emitting unit 61, a light-receiving unit 62, a NIRS acquisition unit 63, a NIRS output unit 64, a blood glucose level receiving unit 81, and a blood glucose level output unit 741.

[0141] The NIRS output unit 64 of the NIRS device 8 transmits one or more NIRS information acquired by the NIRS acquisition unit 63 to the blood glucose estimator 7. The NIRS output unit 64 may also transmit one or more types of information from one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values ​​to the blood glucose estimator 7. The NIRS output unit 64 may also transmit a user identifier to the blood glucose estimator 7 to identify the user. The NIRS output unit 64 can be implemented by wireless or wired communication means.

[0142] The blood glucose receiving unit 81 receives the estimated blood glucose level from the blood glucose estimator 7. The blood glucose receiving unit 81 can be implemented using wireless or wired communication means.

[0143] The blood glucose output unit 741 of the NIRS device 8 outputs the blood glucose values ​​received by the blood glucose receiving unit 81. Here, output usually refers to display on a screen, but it may also be a concept that includes projection using a projector, printing with a printer, sound output, transmission to an external device, storage on a recording medium, and transfer of processing results to other processing devices or other programs. The blood glucose output unit 741 may or may not be considered to include output devices such as a display or speaker. The blood glucose output unit 741 can be implemented by driver software for an output device, or by driver software for an output device and an output device.

[0144] The learning information storage unit 711 of the blood glucose level estimation device 7 may store learning information for each set of explanatory variables used. In other words, the learning information storage unit 711 may contain learning information when the explanatory variable is only 1 or more NIRS information, and learning information when the explanatory variable is 1 or more NIRS information and one or more types of information from 1 or more user dynamic attribute values, 1 or more user static attribute values, and 1 or more device characteristic values. In addition, the learning information storage unit 711 may also store learning information associated with a user identifier.

[0145] It is preferable that the user information storage unit 712 stores user information associated with a user identifier. The user information includes one or more types of information from one or more static user attribute values ​​and one or more dynamic user attribute values.

[0146] It is preferable that the device characteristic value storage unit 713 stores one or more device characteristic values ​​associated with a user identifier.

[0147] The NIRS information receiving unit 721 of the blood glucose level estimation device 7 receives one or more NIRS information transmitted by the NIRS device 8. The NIRS information receiving unit 721 may also receive one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values ​​transmitted by the NIRS device 8. The NIRS information receiving unit 721 may also receive a user identifier. The NIRS information receiving unit 721 can be implemented using wireless or wired communication means, etc.

[0148] The blood glucose output unit 741 of the blood glucose estimation device 7 transmits the blood glucose value acquired by the estimation unit 734. The destination of the blood glucose value transmission is preferably the NIRS device 8, but other devices may also be used. Other devices include, for example, a user's terminal device (not shown). The terminal device may be, for example, a smartphone, tablet device, or so-called personal computer, but the type is not limited.

[0149] Next, we will explain an example of the operation of the blood glucose estimation system D. First, we will explain an example of the operation of the NIRS device 8.

[0150] Each of the one or more light-emitting units 61 of the NIRS device 8 emits near-infrared light. Each of the one or more light-receiving units 62 receives the reflected near-infrared light. Next, the NIRS acquisition unit 63 acquires one or more pieces of NIRS information from the received signal. The NIRS acquisition unit 63 also acquires, for example, one or more user dynamic attribute values. The NIRS output unit 64 transmits the one or more pieces of NIRS information acquired by the NIRS acquisition unit 63 to the blood glucose level estimation device 7. The NIRS output unit 64 may also transmit one or more user dynamic attribute values ​​or a user identifier to the blood glucose level estimation device 7. The user identifier is stored in a storage unit (not shown) of the NIRS device 6.

[0151] An example of the operation of the blood glucose level estimation device 7 will be explained using the flowchart in Figure 10.

[0152] (Step S1001) The NIRS information receiving unit 721 determines whether or not it has received NIRS information, etc. If NIRS information, etc. has been received, the unit proceeds to step S1002; otherwise, it returns to step S1001. Note that the NIRS information, etc. includes one or more pieces of NIRS information. The NIRS information, etc. may include, for example, one or more user dynamic identifiers, one or more user static identifiers, and one or more device characteristic values.

[0153] (Step S1002) The processing unit 73 determines whether or not to use user dynamic attribute values ​​when acquiring blood glucose levels. If user dynamic attribute values ​​are to be used, the process proceeds to step S1003; otherwise, the process proceeds to step S1004. The decision of whether or not to use user dynamic attribute values ​​may be predetermined, based on user instructions, or determined based on whether or not user dynamic attribute values ​​can be acquired.

[0154] (Step S1003) The user dynamic attribute value acquisition unit 731 acquires one or more user dynamic attribute values. The user dynamic attribute value acquisition unit 731 acquires, for example, one or more received user dynamic attribute values.

[0155] (Step S1004) The processing unit 73 determines whether or not to use user static attribute values ​​when acquiring blood glucose levels. If user static attribute values ​​are to be used, the process proceeds to step S1005; otherwise, the process proceeds to step S1006. The decision of whether or not to use user static attribute values ​​may be predetermined, based on user instructions, or determined based on whether or not user static attribute values ​​are stored.

[0156] (Step S1005) The user static attribute value acquisition unit 732 acquires one or more user static attribute values. For example, the user static attribute value acquisition unit 732 acquires one or more user static attribute values ​​corresponding to the received user identifier from the user information storage unit 712.

[0157] (Step S1006) The processing unit 73 determines whether or not to use device characteristic values ​​when acquiring blood glucose levels. If device characteristic values ​​are to be used, the process proceeds to step S1007; otherwise, the process proceeds to step S1008. The decision of whether or not to use device characteristic values ​​may be predetermined, based on user instructions, or determined based on whether or not device characteristic values ​​are stored.

[0158] (Step S1007) The device characteristic value acquisition unit 733 acquires one or more device characteristic values. For example, the device characteristic value acquisition unit 733 acquires one or more device characteristic values ​​corresponding to the received user identifier from the device characteristic value storage unit 713.

[0159] (Step S1008) The estimation unit 734 prepares data for prediction processing. The estimation unit 734 obtains, for example, vectors to be given to machine learning along with the learner.

[0160] (Step S1009) The estimation unit 734 uses the data prepared in step S1008 and the learning information stored in the learning information storage unit 711 to perform prediction processing and obtain the blood glucose level. An example of the prediction processing is described above.

[0161] (Step S1010) The blood glucose output unit 741 transmits the blood glucose value obtained in step S1009. Return to step S1001. The blood glucose output unit 741 transmits the blood glucose value to, for example, the NIRS device 8.

[0162] In the flowchart shown in Figure 10, processing is terminated by power-off or processing termination interrupts.

[0163] In summary, according to this embodiment, a server can be provided that can non-invasively acquire appropriate blood glucose levels for two or more users.

[0164] The software that implements the NIRS device 8 in this embodiment is as follows. In other words, this program causes the computer to function as a NIRS acquisition unit that acquires one or more pieces of NIRS information obtained by emitting near-infrared light to the user, a NIRS output unit that transmits the one or more pieces of NIRS information to a blood glucose level estimation device, a blood glucose level receiving unit that receives blood glucose levels in response to the transmission of the one or more pieces of NIRS information, and a blood glucose level output unit that outputs the blood glucose levels.

[0165] Furthermore, the software that realizes the blood glucose level estimation device 7 in this embodiment is the following program. In other words, this program causes the computer to function as a NIRS information receiving unit that receives one or more NIRS information obtained by emitting near-infrared light to the user, an estimation unit that obtains an estimated blood glucose level using the one or more NIRS information obtained by the NIRS acquisition unit and the learning information, and a blood glucose level output unit that transmits the blood glucose level obtained by the estimation unit.

[0166] (Embodiment 3) In this embodiment, a learning system comprising a learning device for acquiring learning information will be described.

[0167] Figure 11 is a block diagram of the learning system E in this embodiment. The learning system E comprises one or more NIRS devices 6 and a learning device 9. As shown in NIRS device 6(1) and NIRS device 6(2) in Figure 11, the functions of the NIRS device 6 may be shared among two or more devices. Preferably, the NIRS device 6 is a device used by two or more users.

[0168] Figure 12 is a block diagram of the learning system E in this embodiment. The NIRS device 6 that constitutes the learning system E includes a light-emitting unit 61, a light-receiving unit 62, a NIRS acquisition unit 63, and a NIRS output unit 64.

[0169] The learning device 9 comprises a learning storage unit 91, a learning reception unit 92, and a learning processing unit 93. The learning storage unit 91 comprises a user information storage unit 712 and a device characteristic value storage unit 713.

[0170] The learning reception unit 92 includes a NIRS information reception unit 721. The learning processing unit 93 includes a user dynamic attribute value acquisition unit 731, a user static attribute value acquisition unit 732, a device characteristic value acquisition unit 733, a measured blood glucose value acquisition unit 931, a training data structuring unit 932, a learning unit 933, and a storage unit 934.

[0171] The NIRS acquisition unit 63 of the NIRS device 6 acquires one or more pieces of NIRS information. The NIRS acquisition unit 63 may also acquire one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0172] The NIRS acquisition unit 63 may also acquire the measured blood glucose level (hereinafter referred to as "measured blood glucose level" as appropriate). The measured blood glucose level is, for example, information acquired by a known blood glucose meter. However, the method of acquiring the measured blood glucose level is not specified. The measured blood glucose level is usually an accurate blood glucose level.

[0173] The NIRS output unit 64 transmits one or more NIRS information acquired by the NIRS acquisition unit 63 to the learning device 9. The NIRS output unit 64 may also transmit one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values ​​acquired by the NIRS acquisition unit 63 to the learning device 9. The NIRS output unit 64 may also transmit a user identifier to the learning device 9. The NIRS output unit 64 may also transmit the measured blood glucose level to the learning device 9.

[0174] Various types of information are stored in the learning storage unit 91, which constitutes the learning device 9. These types of information include, for example, user information and device characteristic values. It is preferable that the user information and device characteristic values ​​are associated with a user identifier.

[0175] The learning reception unit 92 receives various types of information. The learning reception unit 92 receives various types of information from the NIRS device 6. These various types of information include, for example, one or more NIRS information items. These various types of information include, for example, one or more user dynamic attribute values, one or more user static attribute values, one or more device characteristic values, and measured blood glucose values.

[0176] The learning processing unit 93 performs various processes. These processes are carried out by the user dynamic attribute value acquisition unit 731, the user static attribute value acquisition unit 732, the device characteristic value acquisition unit 733, the measured blood glucose value acquisition unit 931, the training data structuring unit 932, the learning unit 933, and the storage unit 934.

[0177] The user dynamic attribute value acquisition unit 731 acquires, for example, one or more user dynamic attribute values ​​received by the learning reception unit 92. The user dynamic attribute value acquisition unit 731 acquires, for example, one or more user dynamic attribute values ​​from the user information storage unit 712 that are paired with the user identifier received by the learning reception unit 92.

[0178] The user static attribute value acquisition unit 732 acquires, for example, one or more user static attribute values ​​received by the learning reception unit 92. The user static attribute value acquisition unit 732 acquires, for example, one or more user static attribute values ​​from the user information storage unit 712 that are paired with the user identifier received by the learning reception unit 92.

[0179] The device characteristic value acquisition unit 733 acquires, for example, one or more device characteristic values ​​received by the learning reception unit 92. The device characteristic value acquisition unit 733 acquires, for example, one or more device characteristic values ​​from the device characteristic value storage unit 713 that are paired with the user identifier received by the learning reception unit 92.

[0180] The actual blood glucose level acquisition unit 931 acquires the user's actual blood glucose level. The actual blood glucose level acquisition unit 931 also acquires the actual blood glucose level received by, for example, the learning reception unit 92.

[0181] The training data structuring unit 932 constructs training data using one or more NIRS information values ​​and measured blood glucose values.

[0182] The training data constructor 932 preferably constructs training data using one or more types of information from among one or more NIRS information and measured blood glucose values, and one or more user dynamic attribute values, one or more user static attribute values, and one or more device characteristic values.

[0183] The training data structuring unit 932 typically acquires two or more training data points.

[0184] The learning unit 933 acquires learning information using the training data acquired by the training data structuring unit 932. Typically, the learning unit 933 acquires learning information using two or more training data acquired by the training data structuring unit 932.

[0185] The learning unit 933 acquires learning information using one of the following methods (1) to (3). (1) When the learning information is a learning device

[0186] The learning unit 933 provides two or more training data points acquired by the training data structuring unit 932 to a module that performs machine learning training, executes the module, and obtains a learner. The machine learning training algorithm can be any type, such as deep learning, decision trees, random forests, or SVR. The learner is designed to output blood glucose levels using one or more NIRS data points as input. The machine learning training module is also not limited; examples include various functions from the TensorFlow library, the R language's random forest module, tinySVM, etc. (2) When the learning information is a correspondence table

[0187] The learning unit 933 acquires correspondence information for each training data set, showing the correspondence between the set of explanatory variables contained in each training data set (2 or more) and the measured blood glucose value contained in each training data set. The learning unit 933 then constructs a correspondence table containing 2 or more correspondence information sets. The set of explanatory variables includes 1 or more NIRS information. Preferably, the set of explanatory variables includes one or more types of information from 1 or more user dynamic attribute values, 1 or more user static attribute values, and 1 or more device characteristic values. (3) When the learning information is an arithmetic expression

[0188] The learning unit 933 takes a set of explanatory variables from two or more training data as input and obtains a calculation formula that outputs the measured blood glucose value from each training data. The learning unit 933 obtains the calculation formula, for example, by regression analysis. The learning unit 933 obtains the calculation formula, for example, by multiple regression analysis.

[0189] The storage unit 934 stores the learning information acquired by the learning unit 933. The storage location for the learning information is, for example, the learning storage unit 91, but it may be any other device.

[0190] The learning storage unit 91 is preferably made of a non-volatile recording medium, but it can also be made of a volatile recording medium.

[0191] The process by which information is stored in the learning storage unit 91 is not relevant. For example, information may be stored in the learning storage unit 91 via a recording medium, information transmitted via a communication line or the like may be stored in the learning storage unit 91, or information input via an input device may be stored in the learning storage unit 91.

[0192] The learning reception unit 92 can be implemented by wireless or wired communication means.

[0193] The learning processing unit 93, the actual blood glucose level acquisition unit 931, the training data structuring unit 932, the learning unit 933, and the storage unit 934 can typically be implemented using a processor, memory, etc. The processing procedures of the learning processing unit 93, etc., are usually implemented in software, and this software is recorded on a recording medium such as ROM. However, it may also be implemented in hardware (dedicated circuitry). The processor can be, for example, a CPU, MPU, GPU, etc., and the type is not limited.

[0194] Next, we will describe an example of the operation of the learning system E. First, we will describe an example of the operation of the NIRS device 6 here.

[0195] Each of the one or more light-emitting units 61 of the NIRS device 6 emits near-infrared light. Each of the one or more light-receiving units 62 receives the reflected near-infrared light. Next, the NIRS acquisition unit 63 acquires one or more pieces of NIRS information from the received signal. The NIRS acquisition unit 63 also acquires, for example, one or more user dynamic attribute values. The NIRS acquisition unit 63 also acquires the measured blood glucose value. The method of acquiring the measured blood glucose value is not specified. The NIRS output unit 64 passes the one or more pieces of NIRS information acquired by the NIRS acquisition unit 63 and the measured blood glucose value to the learning device 9. Here, the NIRS output unit 64 may also pass one or more user dynamic attribute values, etc., to the learning device 9. The NIRS output unit 64 transmits, for example, one or more pieces of NIRS information, etc., to the learning device 9.

[0196] Next, an example of the operation of the learning device 9 will be explained using the flowchart in Figure 13.

[0197] (Step S1301) The NIRS information receiving unit 721 determines whether or not NIRS information has been received. If NIRS information has been received, the unit proceeds to step S1302; otherwise, it proceeds to step S1310. The NIRS information includes one or more NIRS information values ​​and the measured blood glucose value. The NIRS information may also include, for example, one or more user dynamic identifiers, one or more user static identifiers, and one or more device characteristic values.

[0198] (Step S1302) The processing unit 73 determines whether or not to use user dynamic attribute values ​​when acquiring blood glucose levels. If user dynamic attribute values ​​are to be used, the process proceeds to step S1303; otherwise, the process proceeds to step S1304. The decision of whether or not to use user dynamic attribute values ​​may be predetermined, based on user instructions, or determined based on whether or not user dynamic attribute values ​​can be acquired.

[0199] (Step S1303) The user dynamic attribute value acquisition unit 731 acquires one or more user dynamic attribute values. For example, the user dynamic attribute value acquisition unit 731 acquires one or more received user dynamic attribute values.

[0200] (Step S1304) The processing unit 73 determines whether or not to use user static attribute values ​​when acquiring blood glucose levels. If user static attribute values ​​are to be used, the process proceeds to step S1305; otherwise, the process proceeds to step S1306. The decision of whether or not to use user static attribute values ​​may be predetermined, based on user instructions, or determined by whether or not user static attribute values ​​are stored.

[0201] (Step S1305) The user static attribute value acquisition unit 732 acquires one or more user static attribute values. For example, the user static attribute value acquisition unit 732 acquires one or more user static attribute values ​​corresponding to the received user identifier from the user information storage unit 712.

[0202] (Step S1306) The processing unit 73 determines whether or not to use device characteristic values ​​when acquiring blood glucose levels. If device characteristic values ​​are to be used, the process proceeds to step S1307; otherwise, the process proceeds to step S1308. The decision of whether or not to use device characteristic values ​​may be predetermined, based on user instructions, or determined based on whether or not device characteristic values ​​are stored.

[0203] (Step S1307) The device characteristic value acquisition unit 733 acquires one or more device characteristic values. For example, the device characteristic value acquisition unit 733 acquires one or more device characteristic values ​​corresponding to the received user identifier from the device characteristic value storage unit 713.

[0204] (Step S1308) The measured blood glucose acquisition unit 931 acquires the received measured blood glucose value. The training data structuring unit 932 acquires one or more NIRS information received. The training data structuring unit 932 also constructs training data using information including one or more NIRS information and the measured blood glucose value.

[0205] (Step S1309) The teacher data structuring unit 932 stores the teacher data constructed in step S1308 in the learning storage unit 91. Return to step S1301.

[0206] (Step S1310) The learning unit 933 determines whether or not to start the learning process. If it decides to start the learning process, it proceeds to step S1311; if it decides not to start the learning process, it returns to step S1301. The learning unit 933 may decide to start the learning process, for example, upon receiving instructions from the user. The learning unit 933 may also decide to start the learning process, for example, when training data exceeding a threshold has been accumulated in the learning storage unit 91. The conditions for starting the learning process are not specified.

[0207] (Step S1311) The learning unit 933 acquires two or more training data stored in the learning storage unit 91. The learning unit 933 uses the two or more training data to perform a learning process and acquire learning information.

[0208] (Step S1312) The storage unit 934 stores the learning information acquired in step S1311. Return to step S1301.

[0209] In the flowchart shown in Figure 13, processing is terminated by power-off or processing termination interrupts.

[0210] The specific operation of the learning system E in this embodiment will be described below with reference to Figure 14. A conceptual diagram of the learning system E is shown in Figure 12.

[0211] The NIRS device 6, placed on the subject's wrist, emits near-infrared light at wavelengths of 850 nm, 1200 nm, and 1300 nm, and receives the reflected light. The NIRS device 6 then acquires the intensity of the reflected light corresponding to each wavelength in a time series (Figure 14 (1)). The NIRS device 6 also acquires time series information of skin temperature, an example of a user dynamic attribute value, using a temperature sensor included in the NIRS device 6 (Figure 14 (1)). The NIRS device 6 also acquires time series information of the measured blood glucose level. The NIRS device 6 then acquires a large amount of training data, which associates the input from the sensor at each time point in the time series (intensity of reflected light corresponding to each wavelength, skin temperature) with the measured blood glucose level at the same time (Figure 14 (2)). The NIRS device 6 then transmits this large amount of training data to the learning device 9.

[0212] Next, the learning device 9 receives and stores a large amount of training data. When the amount of training data exceeds a threshold, it provides this large amount of training data to a machine learning learning module, executes the learning module, acquires a learner, and stores it.

[0213] In summary, according to this embodiment, learning information for non-invasively obtaining appropriate blood glucose levels in a person can be obtained.

[0214] In this embodiment, the NIRS device 6 and the learning device 9 may be integrated. In this case, the learning device 9 is a learning device comprising: one or more light-emitting units that emit near-infrared light onto the user's body; one or more light-receiving units corresponding to each of the one or more light-emitting units that receive reflected near-infrared light emitted by each of the one or more light-emitting units; a blood glucose value acquisition unit that acquires the user's blood glucose value; a NIRS information acquisition unit that acquires one or more pieces of NIRS information related to the reflected light using the reflected light received by each of the one or more light-receiving units; a training data configuration unit that constructs training data using the one or more pieces of NIRS information and the blood glucose value; a learning unit that acquires learning information using the training data; and a storage unit that stores the learning information.

[0215] Furthermore, the software that realizes the learning device 9 in this embodiment is the following program. In other words, this program is a program that causes the computer to function as a NIRS acquisition unit that acquires one or more NIRS pieces of information related to near-infrared light, which is reflected light emitted from one or more light-emitting units and received by one or more light-receiving units; an actual blood glucose value acquisition unit that acquires the user's blood glucose value; a training data configuration unit that constructs training data using the one or more NIRS pieces of information and the blood glucose value; a learning unit that acquires learning information using the training data; and a storage unit that stores the learning information.

[0216] Figure 15 also shows the appearance of a computer that runs the program described herein to realize the blood glucose level estimation device 7 and learning device 9 of the various embodiments described above. The embodiments described above can be realized with computer hardware and computer programs that run on it. Figure 15 is an overview of this computer system 300, and Figure 16 is a block diagram of the system 300.

[0217] In Figure 15, the computer system 300 includes a computer 301 with a CD-ROM drive, a keyboard 302, a mouse 303, a monitor 304, and a microphone 305.

[0218] In Figure 16, the computer 301 includes, in addition to the CD-ROM drive 3012, an MPU 3013, a bus 3014 connected to the CD-ROM drive 3012, a ROM 3015 for storing programs such as boot-up programs, a RAM 3016 connected to the MPU 3013 for temporarily storing application program instructions and providing temporary storage space, and a hard disk 3017 for storing application programs, system programs, and data. Although not shown here, the computer 301 may further include a network card for providing connectivity to a LAN.

[0219] The program that causes the computer system 300 to execute the functions of the blood glucose level estimation device 7, learning device 9, etc., as described above, may be stored on CD-ROM 3101, inserted into CD-ROM drive 3012, and then transferred to hard disk 3017. Alternatively, the program may be transmitted to computer 301 via a network (not shown) and stored on hard disk 3017. The program is loaded into RAM 3016 during execution. The program may also be loaded directly from CD-ROM 3101 or the network.

[0220] The program does not necessarily have to include an operating system (OS) or third-party program that causes the computer 301 to execute functions such as the blood glucose level estimation device 7 of the above-described embodiment. The program only needs to include the instruction portion that calls the appropriate function (module) in a controlled manner and obtains the desired result. How the computer system 300 operates is well known, so a detailed explanation is omitted.

[0221] In the above program, steps such as sending information and receiving information do not include hardware-based processing, such as processing performed by a modem or interface card in the transmission step (processing that can only be performed by hardware).

[0222] Furthermore, the computer running the above program may be a single unit or multiple units. That is, it may perform centralized processing or distributed processing. In other words, the blood glucose estimation device 7 and the learning device 9 may be standalone devices or consist of two or more devices.

[0223] Furthermore, it goes without saying that in each of the above embodiments, two or more communication means present in a single device may be physically implemented in a single medium.

[0224] Furthermore, in each of the above embodiments, each process may be implemented by centralized processing by a single device, or by distributed processing by multiple devices.

[0225] It goes without saying that the present invention is not limited to the embodiments described above, and various modifications are possible, all of which are also included within the scope of the present invention. [Industrial applicability]

[0226] As described above, the blood glucose level estimation device according to the present invention has the effect of easily and non-invasively obtaining an appropriate blood glucose level for a person, and is useful as a blood glucose level estimation device, etc. [Explanation of Symbols]

[0227] C, D Blood glucose level estimation system E Learning System 6, 8 NIRS device 7. Blood glucose level estimation device 9. Learning device 61 Light-emitting part 62 Light receiving part 63 NIRS Acquisition Department 64 NIRS output section 71 Storage section 72 Reception Department 73 Processing Unit 74 Output section 81 Blood glucose receiving unit 91 Learning Storage Unit 92 Learning Reception Department 93 Learning Processing Unit 711 Learning Information Storage Unit 712 User Information Storage Unit 713 Device characteristic value storage unit 721 NIRS Information Reception Department 731 User Dynamic Attribute Value Acquisition Unit 732 User Static Attribute Value Acquisition Unit 733 Device characteristic value acquisition unit 734 Estimation Department 741 Blood glucose output unit 760n, 850n, 1200n, 1300n abbreviation 931 Actual blood glucose level acquisition unit 932 Training Data Organizer 933 Learning Department 934 Storage Unit

Claims

1. A learning information storage unit stores learning information acquired using two or more training data sets, each containing one or more NIRS information obtained using reflected near-infrared light emitted from the human body and blood glucose levels. A NIRS acquisition unit that acquires one or more NIRS pieces of information by emitting near-infrared light towards the user, An estimation unit obtains an estimated blood glucose level using one or more NIRS information obtained by the NIRS acquisition unit and the learning information. The system comprises a blood glucose output unit that outputs the blood glucose value acquired by the estimation unit, The aforementioned training data includes one or more NIRS pieces of information obtained by emitting near-infrared light of two or more wavelengths onto the human body. The NIRS acquisition unit is, By emitting near-infrared light of two or more wavelengths toward the user, one or more pieces of NIRS information are obtained. A blood glucose level estimation device in which the number of two or more different wavelength types used to acquire learning information is greater than the number of two or more different wavelength types used to acquire one or more NIRS information by the NIRS acquisition unit.

2. The blood glucose level estimation device according to claim 1, wherein the two or more wavelengths are two or more wavelengths in the range of 760 nm to 1300 nm.

3. The aforementioned training data includes one or more user dynamic attribute values ​​from among the following of a person having a human body: skin temperature, acceleration, pulse rate, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level. The system further comprises a user dynamic attribute value acquisition unit that acquires one or more user dynamic attribute values ​​from among the user's skin temperature, acceleration, pulse rate, pulse wave, pulse rate, blood oxygen saturation concentration, blood pressure, skin color, muscle mass, and blood hemoglobin level. The estimation unit, A blood glucose level estimation device according to claim 1 or claim 2, which obtains an estimated blood glucose level using one or more NIRS information obtained by the NIRS acquisition unit, one or more user dynamic attribute values ​​obtained by the user dynamic attribute value acquisition unit, and the learning information.

4. The aforementioned training data has one or more user static attribute values, which are static attribute values ​​of a person having a human body. The system further comprises a user static attribute value acquisition unit that acquires one or more user static attribute values ​​for the aforementioned user, The estimation unit, A blood glucose level estimation device according to any one of claims 1 to 3, wherein an estimated blood glucose level is obtained using one or more NIRS information obtained by the NIRS acquisition unit, one or more user static attribute values ​​obtained by the user static attribute value acquisition unit, and the learning information.

5. The blood glucose level estimation device according to claim 4, wherein the one or more user static attribute values ​​are one or more pieces of information from the user's age, height, weight, and forearm circumference.

6. The aforementioned training data has one or more device characteristic values, which are characteristic values ​​of the device used to acquire the one or more NIRS pieces of information. The system further comprises a device characteristic value acquisition unit that acquires one or more device characteristic values ​​of a device used to acquire one or more NIRS information of the user, The estimation unit, A blood glucose level estimation device according to any one of claims 1 to 5, wherein an estimated blood glucose level is obtained using one or more NIRS information obtained by the NIRS acquisition unit, one or more device characteristic values ​​obtained by the device characteristic value acquisition unit, and the learning information.

7. The blood glucose level estimation device according to claim 6, wherein the one or more device characteristic values ​​are one or more pieces of information from distance information that specifies the distance between two or more light-receiving units that receive reflected light emitted from a near-infrared light-emitting unit, the number of light-receiving units, and the number of light-emitting units.

8. The NIRS acquisition unit is, From two or more NIRS devices that acquire one or more NIRS information or one or more NIRS information sources that form the basis of one or more NIRS information from a user, the system acquires the one or more NIRS information or the one or more NIRS information sources. The aforementioned blood glucose output unit is A blood glucose estimator according to any one of claims 1 to 7, wherein the estimation unit transmits the blood glucose level or blood glucose level-related information relating to said blood glucose level to the NIRS device or a terminal device corresponding to the NIRS device.

9. The blood glucose level estimation device according to any one of claims 1 to 8, wherein the learning information is a learner obtained by performing a machine learning learning process using the two or more training data sets.

10. A method for estimating blood glucose levels, wherein all processing performed by the blood glucose level estimator described in any one of claims 1 to 9 is performed by a computer.

11. Computers, A program for causing a blood glucose level estimation device to function as described in any one of claims 1 to 9.

Images