Non-invasive blood glucose prediction system and blood glucose prediction method

Abstract

The present invention relates to a non-invasive blood glucose prediction system and blood glucose prediction method. A blood glucose prediction system and prediction method may be provided, wherein the prediction method comprises: receiving an input of body information about a user; measuring information about glucose content in body fluids secreted by the user; storing the input body information about the user and the measured information about the glucose content in the body fluids; and calculating a blood glucose prediction value by using the body information about the user and the measured information about the glucose content in the body fluids. Thus, blood glucose levels can be predicted with high accuracy.

Description

Noninvasive blood glucose prediction system and blood glucose prediction method

[0001] The present invention relates to a blood sugar prediction system and a blood sugar prediction method for predicting blood sugar levels using a non-invasive method.

[0002] Diet, exercise, and medication are used to manage blood sugar levels. Accurately understanding your blood sugar level is essential for these treatments. Furthermore, those with diabetes require regular blood sugar checks.

[0003] Conventional devices for analyzing blood sugar levels utilize invasive methods, collecting blood samples. These samples are then introduced into a chemically treated sensor, which is then inserted into a portable device to measure blood sugar levels. However, invasive methods for measuring blood sugar are painful, increasing the burden of blood collection. Furthermore, requiring periodic measurements can lead to health risks, such as infection from repeated blood draws.

[0004] To solve the problems of the above invasive blood sugar measurement devices, various non-invasive blood sugar measurement devices have been developed, but there is a need to solve the fundamental inaccuracy of blood sugar prediction that occurs because blood sugar levels are not measured directly from the blood.

[0005] In this regard, Korean Patent Publication No. 10-2017-0021216 discloses a contact lens-type glucose detection sensor comprising a complex of cerium oxide (CeO2) nanoparticles and glucose oxidase, and a method for predicting blood sugar levels by detecting glucose in tears using the sensor. However, there is a problem in that cross-validation between actual blood sugar levels and predicted blood sugar levels is not performed.

[0006] Therefore, there is a need to develop a blood sugar prediction system and method that can predict blood sugar levels with high accuracy using a non-invasive method.

[0007]

[0008] The purpose of the present invention is to provide a blood sugar prediction system and a blood sugar prediction method capable of predicting blood sugar with high accuracy by calculating a blood sugar prediction value using the user's body information along with information on the sugar content in the user's body fluid measured by a non-invasive method.

[0009] In addition, the present invention aims to provide a blood sugar prediction system and a blood sugar prediction method that can further increase the accuracy of blood sugar prediction by constructing a model for blood sugar prediction using a previously stored data set, and calculating a blood sugar prediction value by inputting the user's body information and information on sugar content in body fluids into the constructed blood sugar prediction model.

[0010] In addition, the present invention aims to provide a blood sugar prediction system and a blood sugar prediction method that can further increase the accuracy of blood sugar prediction for each individual by constructing a personalized blood sugar prediction model using the user's own data set and calculating a blood sugar prediction value by inputting the user's body information and sugar content information in body fluids into the constructed personalized blood sugar prediction model.

[0011] In addition, the present invention aims to provide a blood sugar prediction system and a blood sugar prediction method that can easily determine a user's blood sugar status periodically during daily life or exercise.

[0012]

[0013] In order to solve the above problem, the present invention provides a blood sugar prediction system including a body information input unit for receiving body information of a user; a body fluid sugar information measurement unit for measuring information on the sugar content in body fluid secreted by the user; an information storage unit for storing the input body information of the user and the measured body fluid sugar content information; and a blood sugar analysis unit for calculating a blood sugar prediction value using the body information of the user and the measured body fluid sugar content information.

[0014] The above physical information may include at least one selected from the group consisting of age, height, weight, and body fat percentage.

[0015] The above body fluid may be any one selected from the group consisting of sweat, tears, saliva, and urine.

[0016] The information regarding the above sugar content may be spectroscopic information measured using a photometric method or electrochemical information measured using an electrochemical method.

[0017] The above blood sugar analysis unit may construct a blood sugar prediction model through machine learning based on a plurality of data sets, and the data sets may include body information, blood sugar information in body fluid, and an actual blood sugar level corresponding to the blood sugar information in body fluid.

[0018] The blood sugar analysis unit may calculate a blood sugar prediction value by inputting the user's physical information and information about the sugar content in the measured body fluid into the constructed blood sugar prediction model.

[0019] The above machine learning may utilize one or more selected from the group consisting of linear regression analysis, nonlinear regression analysis, partial least squares (PLS), Bayesian network, hidden markov model, decision tree, boosting, bagging, support vector machine, convolutional neural network, deep neural network, and recursive neural network (RNN).

[0020] The above-mentioned plurality of data sets may be 30 or more data sets.

[0021] The above data set may be someone else's data set.

[0022] The above-mentioned data set of another person may be previously stored in the above-mentioned information storage unit.

[0023] The above data set may be the user's own data set.

[0024] Information about the sugar content in body fluids secreted by the user may be measured at a specific point after the start of exercise.

[0025] The specific time point after the start of the above exercise may be two different specific time points.

[0026] The specific time after starting the above exercise may be 20 minutes and 30 minutes after starting the exercise.

[0027] The above blood sugar prediction system may further include a display unit that displays the blood sugar prediction value calculated by the blood sugar analysis unit.

[0028] The above blood sugar prediction system may further include a communication unit for transmitting and receiving information with an external device.

[0029] In addition, the present invention provides a blood sugar prediction method including a body information input step of receiving body information of a user; a body fluid sugar information measurement step of measuring information on the sugar content in body fluid secreted from the user; an information storage step of storing the input body information of the user and the information on the measured body fluid sugar content; and a blood sugar analysis step of calculating a blood sugar prediction value using the body information of the user and the information on the measured body fluid sugar content.

[0030] The above physical information may include at least one selected from the group consisting of age, height, weight, and body fat percentage.

[0031] The above body fluid may be any one selected from the group consisting of sweat, tears, saliva, and urine.

[0032] The information regarding the above sugar content may be spectroscopic information measured using a photometric method or electrochemical information measured using an electrochemical method.

[0033] The above blood sugar analysis step includes a step of constructing a blood sugar prediction model through machine learning based on a plurality of data sets, and the data sets may include body information, blood sugar information in body fluid, and an actual blood sugar level corresponding to the blood sugar information in body fluid.

[0034] The above blood sugar analysis step may include a step of calculating a blood sugar prediction value by inputting the user's body information and information about the sugar content in the measured body fluid into the constructed blood sugar prediction model.

[0035] The above machine learning may utilize one or more selected from the group consisting of linear regression analysis, nonlinear regression analysis, partial least squares (PLS), Bayesian network, hidden markov model, decision tree, boosting, bagging, support vector machine, convolutional neural network, deep neural network, and recursive neural network (RNN).

[0036] The above-mentioned plurality of data sets may be 30 or more data sets.

[0037] The data set used in the step of building the above blood sugar prediction model may be a previously stored data set of another person.

[0038] The data set used in the step of building the above blood sugar prediction model may be the user's own data set.

[0039] Information about the sugar content in body fluids secreted by the user may be measured at a specific point after the start of exercise.

[0040] The specific time point after the start of the above exercise may be two different specific time points.

[0041] The specific time after starting the above exercise may be 20 minutes and 30 minutes after starting the exercise.

[0042] The above blood sugar prediction method may further include a step of transmitting and receiving information with an external device.

[0043]

[0044] The present invention can provide a blood sugar prediction system and a blood sugar prediction method capable of predicting blood sugar with high accuracy by calculating a blood sugar prediction value using the user's body information along with information on the sugar content in the user's body fluid measured by a non-invasive method.

[0045] In addition, the present invention can provide a blood sugar prediction system and a blood sugar prediction method that can further increase the accuracy of blood sugar prediction by constructing a model for blood sugar prediction using a previously stored data set, and calculating a blood sugar prediction value by inputting the user's body information and information on sugar content in body fluids into the constructed blood sugar prediction model.

[0046] In addition, the present invention can provide a blood sugar prediction system and a blood sugar prediction method that can further increase the accuracy of blood sugar prediction for each individual by constructing a personalized blood sugar prediction model using the user's own data set and calculating a blood sugar prediction value by inputting the user's body information and sugar content information in body fluids into the constructed personalized blood sugar prediction model.

[0047] In addition, the present invention can provide a blood sugar prediction system and a blood sugar prediction method that can easily determine a user's blood sugar level periodically during daily life or exercise.

[0048]

[0049] Figure 1 is a block diagram of a blood sugar prediction system according to a first embodiment of the present invention.

[0050] Figure 2 is a block diagram of a blood sugar prediction system according to a second embodiment of the present invention.

[0051] Figure 3 is a block diagram of a blood sugar prediction system according to a third embodiment of the present invention.

[0052] Figure 4 is a plan view of a glucose sensor according to one embodiment of the present invention.

[0053] Fig. 5 is a cross-sectional view showing a surface cut along line A-A' of Fig. 4.

[0054] Figure 6 is a flowchart of a blood sugar prediction method according to the first embodiment of the present invention.

[0055] Figure 7 is a flowchart of a blood sugar prediction method according to a second embodiment of the present invention.

[0056] Figure 8 is a flowchart of a blood sugar prediction method according to a third embodiment of the present invention.

[0057] FIG. 9a, FIG. 9b, FIG. 10a, and FIG. 10b are diagrams showing the results of evaluating the accuracy of blood sugar prediction values ​​calculated using the blood sugar prediction system / blood sugar prediction method of the present invention.

[0058]

[0059]

[0060] The present invention relates to a non-invasive blood sugar prediction system and a blood sugar prediction method, and provides a blood sugar prediction system and a blood sugar prediction method capable of predicting blood sugar with high accuracy by calculating a blood sugar prediction value using the user's body information as well as information on the sugar content in the user's body fluid.

[0061] More specifically, the present invention provides a blood sugar prediction system including a body information input unit for receiving body information of a user; a body fluid sugar information measurement unit for measuring information on the sugar content in body fluid secreted from the user; an information storage unit for storing the input body information of the user and the measured body fluid sugar content information; and a blood sugar analysis unit for calculating a blood sugar prediction value using the body information of the user and the measured body fluid sugar content information.

[0062] In addition, the present invention provides a blood sugar prediction method including a body information input step of receiving body information of a user; a body fluid sugar information measurement step of measuring information on the sugar content in body fluid secreted from the user; an information storage step of storing the input body information of the user and the information on the measured body fluid sugar content; and a blood sugar analysis step of calculating a blood sugar prediction value using the body information of the user and the information on the measured body fluid sugar content.

[0063]

[0064] Hereinafter, embodiments of the present invention will be described in more detail with reference to the drawings. However, the following drawings attached to this specification illustrate preferred embodiments of the present invention and, together with the contents of the invention described above, serve to further understand the technical concept of the present invention. Therefore, the present invention should not be interpreted as being limited to the matters described in such drawings.

[0065] In this specification, singular forms also include plural forms, unless specifically stated otherwise. Like reference numerals refer to like elements throughout the specification.

[0066] As used herein, the terms “comprises” and / or “comprising” are used to mean that they do not exclude the presence or addition of one or more other components, steps, operations and / or devices other than the components, steps, operations and / or devices mentioned.

[0067] As used herein, the term "connection" is used to mean both indirectly connecting and directly connecting multiple components, and to mean both physically connecting and electrically connecting.

[0068]

[0069] Blood Sugar Prediction System

[0070] Figures 1 to 3 are block diagrams schematically showing a blood sugar prediction system according to the first to third embodiments of the present invention.

[0071]

[0072] Referring to FIG. 1, a blood sugar prediction system (10) according to a first embodiment of the present invention includes a body information input unit (100) for receiving a user's body information, a body fluid sugar information measurement unit (200) for measuring information on sugar content in body fluid secreted from the user, an information storage unit (300) for storing the input user's body information and the measured body fluid sugar content, and a blood sugar analysis unit (400) for calculating a blood sugar prediction value using the user's body information and the measured body fluid sugar content information.

[0073]

[0074] The body information input unit (100) can receive body information from a user. The body information input unit (100) can include at least one of a mechanical button, a touch panel, and a digitizer.

[0075] The above mechanical buttons may include various types of buttons, such as mechanical key buttons, wheel buttons, etc.

[0076] The above touch panel detects the user's touch input and calculates touch coordinates corresponding to the detected touch signal. When the touch panel is combined with the display unit (500) described below to form a touch screen, the touch screen can be implemented as various types of touch screens, such as electrostatic, pressure-sensitive, or piezoelectric. The electrostatic type uses a dielectric coated on the surface of the touch screen to detect the micro-electricity generated by the user's body when a part of the user's body touches the touch screen surface, thereby calculating touch coordinates. The pressure-sensitive type includes two electrode plates built into the touch screen, and when the user touches the screen, the upper and lower plates of the touched point come into contact, causing current to flow, thereby calculating touch coordinates. The touch signal generated on the touch screen can be mainly generated by a human finger, but can also be generated by an object made of a conductive material that can apply a change in electrostatic capacity.

[0077] The digitizer detects proximity input or touch input of a user's touch pen and calculates a position corresponding to the detected proximity input or touch input signal. The digitizer can detect touch or proximity input based on changes in the intensity of an electromagnetic field caused by proximity or touch of the pen, for example, using electromagnetic resonance (ERM). The digitizer may be provided to have a certain area below the display panel of the display unit (500) described below.

[0078]

[0079] The user's body information may be directly input by the user by operating a mechanical button, touch panel, or digitizer provided in the body information input unit (100), or may be transmitted from an external device through a communication unit (600) described below.

[0080]

[0081] The above user's physical information may include at least one selected from the group consisting of age, height, weight, and body fat percentage, and preferably includes at least two selected from the group consisting of age, height, weight, and body fat percentage, and more preferably includes all of age, height, weight, and body fat percentage.

[0082] The above body fat percentage refers to the ratio (%) of body fat weight to the user's body weight.

[0083] The bodily fluid secreted from the user may be any one selected from the group consisting of sweat, tears, saliva, and urine.

[0084]

[0085] The above body fluid sugar information measuring unit (200) can measure information on the sugar content in the body fluid secreted from the user, and the information on the sugar content can be spectroscopic information measured using a photometric method or electrochemical information measured using an electrochemical method.

[0086] Both the photometric and electrochemical methods basically utilize an oxidase that can react with glucose and oxidize it. The photometric method uses a photometer to obtain spectroscopic information such as light reflectance or transmittance when glucose in body fluid is oxidized by the oxidase, and quantifies this to measure the sugar content. The electrochemical method obtains electrochemical information such as the current generated when oxygen or an oxidized mediator formed when glucose in body fluid is oxidized by the oxidase is converted to hydrogen peroxide or a reduced mediator and then oxidized again to return to its original oxidized form, and quantifies this to measure the sugar content.

[0087]

[0088] The above information storage unit (300) can store the user's body information input from the body information input unit (100) and information on the sugar content in the user's body fluid measured from the body fluid sugar information measuring unit (200).

[0089] The above information storage unit (300) includes both volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data sets, program modules or other data.

[0090]

[0091] The blood sugar analysis unit (400) can calculate a predicted blood sugar level of the user by using the user's physical information and information on the sugar content in the measured body fluid secreted from the user.

[0092] When predicting a user's blood sugar level based on information about the sugar content in body fluids secreted by the user, even with the same blood sugar level, the sugar content in the body fluid may vary depending on each user's individual physical characteristics. Therefore, by applying the user's physical characteristics in addition to the sugar content in body fluids secreted by the user to calculate the predicted blood sugar value and correcting for errors, a user's blood sugar level can be predicted with greater accuracy.

[0093]

[0094] The above blood sugar analysis unit (400) may build a blood sugar prediction model through machine learning based on multiple data sets.

[0095] The above data set may include body information, blood sugar information in body fluid, and actual blood sugar levels corresponding to the blood sugar information in body fluid.

[0096] The actual blood sugar level corresponding to the blood sugar level information in the body fluid refers to the blood sugar level actually measured at substantially the same time as the time at which the blood sugar level information in the body fluid was acquired. The substantially same time may be within 10 minutes from the time at which the blood sugar level information in the body fluid was acquired, preferably within 5 minutes, and more preferably within 3 minutes. The actual measured blood sugar level may be measured through blood sampling and may be input through a known method. The actual measured blood sugar level may be input through the body information input unit (100).

[0097] The blood sugar prediction model constructed through the above machine learning may include at least one selected from a group consisting of a coefficient representing a correlation between the user's physical information and information on sugar content in body fluid secreted from the user, a coefficient representing a correlation between the user's physical information and an actual blood sugar level, and a coefficient representing a correlation between the information on sugar content in body fluid secreted from the user and an actual blood sugar level.

[0098]

[0099] The blood sugar analysis unit (400) may input information about the user's physical information and the sugar content in the measured body fluid into the blood sugar prediction model constructed through machine learning to calculate a blood sugar prediction value. By applying coefficients related to the correlation between each variable obtained through machine learning, the accuracy of blood sugar prediction based on information about the sugar content in the user's body fluid can be further improved.

[0100]

[0101] The above machine learning may use at least one selected from the group consisting of linear regression analysis, nonlinear regression analysis, partial least squares (PLS), Bayesian network, hidden markov model, decision tree, boosting, bagging, support vector machine, convolutional neural network, deep neural network, and recursive neural network (RNN), and it is preferable to use at least one selected from the group consisting of boosting, bagging, support vector machine, and deep neural network.

[0102]

[0103] Multiple data sets can be used to build the blood sugar prediction model using machine learning. To improve the accuracy of the constructed blood sugar prediction model, it is preferable to use at least 30 data sets. For example, the blood sugar prediction model can be built using machine learning based on 30 to 50 data sets.

[0104]

[0105] The data set used to build the blood sugar prediction model through the above machine learning may be another person's data set, and the other person's data set may be pre-stored in the information storage unit (300).

[0106] In addition, the data set used to build the blood sugar prediction model through machine learning may be the user's own data set. Although at least 30 data sets are required to build a blood sugar prediction model through machine learning, a sufficient number of data sets may not be collected when each user initially uses the blood sugar prediction system according to the present invention. Therefore, before the user's own data set is sufficiently collected, a blood sugar prediction model may be built through machine learning based on multiple data sets of others previously stored in the information storage unit (300), and this model may be used to calculate blood sugar prediction values. Thereafter, when the user's own data set is sufficiently stored through repeated use by the user, a new personalized blood sugar prediction model may be built through machine learning based on the user's multiple data sets, and the newly built personalized blood sugar prediction model may be used to calculate blood sugar prediction values.

[0107] In this way, by building a personalized blood sugar prediction model optimized for each user based on the user's own data set and inputting information about the user's physical information and sugar content in body fluids into the model to calculate a blood sugar prediction value, the accuracy of blood sugar prediction for each individual user can be further improved.

[0108]

[0109] The blood sugar information measured by the blood sugar information measuring unit (200) in the body fluid may be measured at a specific point in time after the start of exercise. The specific point in time after the start of exercise may refer to two different specific points in time, and may refer to 20 minutes and 30 minutes after the start of exercise. By repeatedly measuring the blood sugar information in the body fluid at two different specific points in time after the start of exercise, if a large amount of blood sugar information at specific points in time after the start of exercise is collected, the blood sugar trend from the start of exercise to the specific point in time after exercise can be predicted in advance and provided to the user.

[0110] Conventional invasive blood glucose measurement devices require blood sampling, limiting their use in daily life and even during exercise. However, the blood glucose prediction system of the present invention noninvasively measures sugar content in body fluids secreted by the user and uses this information to provide highly accurate blood glucose prediction values, enabling the user to easily monitor their blood glucose status periodically during daily life or exercise.

[0111]

[0112] The blood sugar prediction system according to the present invention can be implemented flexibly, transparently, or wearably.

[0113]

[0114] Referring to FIG. 2, the blood sugar prediction system (10) according to the second embodiment of the present invention further includes a display unit (500) that displays a blood sugar prediction value calculated by the blood sugar analysis unit (400). The description given for the blood sugar prediction system according to FIG. 1 can be equally applied to the blood sugar prediction system illustrated in FIG. 2, and detailed descriptions of substantially identical or similar components are omitted.

[0115]

[0116] The display unit (500) may include a display panel and a controller that controls the operation of the display panel. The display panel may be implemented in various forms, such as a liquid crystal display (LCD), an organic light emitting diode (OLED), an active-matrix organic light-emitting diode (AM-OLED), and a plasma display panel (PDP). The display unit (500) may be implemented in a flexible manner or in a wearable manner, and may be provided as an integrated touch screen combined with a touch panel of the body information input unit (100) in a laminated structure.

[0117]

[0118] Referring to FIG. 3, the blood sugar prediction system (10) according to the third embodiment of the present invention further includes a communication unit (600) for transmitting and receiving information with an external device. The description given for the blood sugar prediction system according to FIGS. 1 and 2 can be equally applied to the blood sugar prediction system illustrated in FIG. 3, and detailed descriptions of substantially identical or similar components are omitted.

[0119]

[0120] The above communication unit (600) can transmit and receive information with external devices according to various wired and wireless communication methods. The communication unit (600) may perform communication using at least one wireless communication method among Wi-Fi, Bluetooth, wireless communication, and NFC.

[0121] The external device may be a user's personal mobile terminal, and may, for example, transmit the user's body information, information on the sugar content in body fluids, and / or blood sugar prediction values ​​to the user's personal mobile terminal through the communication unit (600), or receive the user's own or another person's data set stored in the user's personal mobile terminal through the communication unit (600).

[0122] The external device may be a cloud server, and may transmit, for example, the user's body information, information on the sugar content in body fluids, and / or blood sugar prediction values ​​to the cloud server through the communication unit (600), or may receive the user's own or another person's data set stored in the cloud server through the communication unit (600).

[0123]

[0124] In one embodiment of the present invention, the body fluid sugar information measuring unit (200) may include a glucose sensor that measures the sugar content in the body fluid using an electrochemical method.

[0125]

[0126] FIG. 4 is a plan view of a glucose sensor according to one embodiment of the present invention, and FIG. 5 is a cross-sectional view showing a surface cut along line A-A' of FIG. 4.

[0127]

[0128] Referring to FIG. 4, the glucose sensor may include a substrate (210), a sensor unit (220), and a wiring unit (250), and the sensor unit (220) may include a working electrode (230) and a reference electrode (240) provided spaced apart from the working electrode (230).

[0129]

[0130] The above substrate (210) serves to provide a structural base for the components that constitute the glucose sensor. For example, the substrate (210) may be implemented in the form of a rigid material such as glass or a substrate film having flexible properties.

[0131] Examples of specific materials that can be applied to the substrate film when the substrate (210) is implemented flexibly include polyester resins such as polyethylene terephthalate, polyethylene isophthalate, polyethylene naphthalate, and polybutylene terephthalate; cellulose resins such as diacetyl cellulose and triacetyl cellulose; polycarbonate resins; acrylic resins such as polymethyl (meth)acrylate and polyethyl (meth)acrylate; styrene resins such as polystyrene and acrylonitrile-styrene copolymers; polyolefin resins such as polyethylene, polypropylene, polyolefins having a cyclo- or norbornene structure, and ethylene-propylene copolymers; vinyl chloride resins; amide resins such as nylon and aromatic polyamides; imide resins; polyethersulfone resins; sulfone resins; polyetheretherketone resins; sulfated polyphenylene resins; vinyl alcohol resins; Examples of the transparent optical film include films composed of thermoplastic resins such as vinylidene chloride resins; vinyl butyral resins; allylate resins; polyoxymethylene resins; and epoxy resins. Films composed of blends of the above thermoplastic resins can also be used. In addition, films composed of thermosetting resins such as (meth)acrylic, urethane, acrylic urethane, epoxy, and silicone resins or ultraviolet-curable resins can also be used. The thickness of such a transparent optical film can be appropriately determined, but is generally determined to be 1 to 500 μm in consideration of workability such as strength and handleability, thin layer properties, etc., and is particularly preferably 1 to 300 μm, and more preferably 5 to 200 μm.

[0132] Such a base film may contain one or more suitable additives. Examples of the additives include ultraviolet absorbers, antioxidants, lubricants, plasticizers, release agents, anti-coloring agents, flame retardants, nucleating agents, antistatic agents, pigments, and colorants. The base film may have a structure including various functional layers, such as a hard coating layer, an anti-reflection layer, and a gas barrier layer, on one or both sides of the film. The functional layers are not limited to those described above, and may include various functional layers depending on the intended use.

[0133] Additionally, the substrate film may be surface-treated, if necessary. Such surface treatments include dry treatments such as plasma treatment, corona treatment, and primer treatment, and chemical treatments such as alkaline treatment including saponification treatment.

[0134]

[0135] Referring to FIG. 5, the working electrode (230) may include an electrode portion (231), an electron transfer portion (232), a glucose reaction portion (233), and a filter layer (234) formed on a substrate (210).

[0136] The above electrode unit (231) is composed of a plurality of electrodes formed on a substrate (210). This electrode unit (231) detects an electrical signal generated by a reaction between a substance constituting a glucose reaction unit (233) described below and glucose contained in the user's body fluid.

[0137] The electrode portion (231) is not particularly limited as long as it is a conductive material. The electrode portion (231) may include, for example, one or more selected from the group consisting of gold (Au), silver (Ag), copper (Cu), platinum (Pt), titanium (Ti), nickel (Ni), tin (Ni), molybdenum (Mo), cobalt (Co), and APC, or may be an alloy thereof. APC is an Ag-Pd-Cu alloy.

[0138] The electrode portion (231) can be formed on the substrate (210) by screen printing, physical vapor deposition, etching, or attachment of a conductive tape.

[0139] The sensor unit (220) may be composed of a plurality of electrodes. The plurality of electrodes may include at least one working electrode (230) and at least one reference electrode (240). In addition, the working electrodes and the reference electrodes may be arranged to correspond to each other.

[0140] The above plurality of electrodes may be connected to the information storage unit (300) and / or the blood sugar analysis unit (400) through the wiring unit (250).

[0141]

[0142] The above electron transfer unit (232) is formed on the electrode unit (231). The electron transfer unit (232) is reduced through a redox reaction with an enzyme that has been reduced by reacting with glucose, and the electron transfer medium in a reduced state thus formed generates a current at the electrode surface to which an oxidation potential is applied.

[0143] The electron transfer unit (232) may include an electron transfer mediator, for example, the electron transfer mediator may be hexaammineruthenium(Ⅲ) chloride, potassium ferricyanide, potassium ferrocyanide, dimethylferrocene (DMF), ferricinium, ferrocene monocarboxylic acid (FCOOH), 7,7,8,8-tetracyanoquino-dimethane (TCNQ), tetrathia fulvalene (TTF), nickelocene (Nc), N-methyl acidinium (NMA+), Tetrathiatetracene (TTT), N-methylphenazinium (NMP+), hydroquinone, 3-dimethylaminobenzoic acid (MBTHDMAB), 3-methyl2-benzothiozolinone hydrazone, 2-methoxy-4-allylphenol, 4-aminoantipyrin (AAP), dimethylaniline, 4-aminoantipyrene, 4-methoxynaphthol, 3,3',5,5'-tetramethyl benzidine (TMB), 2,2-azino-di-[3-ethyl-benzthiazoline sulfonate], o-dianisidine, o-toluidine, 2,4-dichlorophenol (2,4-dichlorophenol), 4-aminophenazone, benzidine, Prussian blue, bipyridine-osmium complex compounds, etc. can be used.

[0144] The above electron transfer unit (232) may also play a role in protecting the electrode unit (231). For example, the electron transfer unit (232) may have a structure in which an electrode protector that protects the electrode unit (231) and an electron transfer medium that performs the function of electron transfer are mixed. Here, in order to protect the electrode, the electron transfer unit (232) may further include an electrode protection material.

[0145] For example, Prussian blue included in the electron transfer unit (232) is a component that performs the function of electron transport, and is a blue pigment whose main component is potassium hexacyanoferrate (II) iron (III), and has high oxidation properties. When the electron transfer unit (232) including Prussian blue is formed between the electrode unit (231) and the glucose reaction unit (233), the sensitivity of the electrode can be improved, but the metallic electrode unit (231) located below the Prussian blue can be oxidized and corroded. Therefore, in an additional embodiment of the present invention, in order to prevent oxidation of the electrode unit (231) and protect the electrode, the electron transfer unit (232) may further include an electrode protection material together with the electron transfer medium. In a specific example, the electrode protection material may be carbon.

[0146]

[0147] The above glucose reaction unit (233) is formed on the electron transfer unit (232) and may include an oxidoreductase that reacts with and reduces glucose contained in the user's body fluid. Here, the reduced enzyme reacts with the electron transfer mediator of the electron transfer unit (232) to quantify glucose.

[0148] For example, the glucose reaction unit (233) may include flavin adeninedinucleotide-glucose dehydrogenase (FAD-GDH), nicotinamide adenine dinucleotide-glucose dehydrogenase (NAD-GDH), pyrroloquinoline quinone-glucose dehydrogenase (PQQ-GDH), glucose oxidase (GOx), etc.

[0149]

[0150] The filter layer (234) is formed on the glucose reaction unit (233). For example, it can directly cover the upper surface of the glucose reaction unit (233). The filter layer (234) can protect the glucose reaction unit (233) from external physical force. In addition, it can prevent the oxidizing enzyme or dehydrogenase of the glucose reaction unit (233) from being exposed to the external environment.

[0151] The filter layer (234) can only allow the substance to be detected to pass through the sample. Therefore, the glucose reaction unit (233) can be prevented from being denatured or damaged by substances other than the substance to be detected.

[0152] If the filter layer (234) allows the substance to pass through, an ion exchange membrane commonly used in the art may be used. The ion exchange membrane may include a cation exchange resin such as a perfluorosulfonic acid resin. For example, the ion exchange membrane may include Nafion.

[0153]

[0154] The above reference electrode (240) may be placed on the substrate (210). For example, the reference electrode (240) may be placed on the same surface as the surface on which the working electrode (230) of the substrate (210) is placed. For example, the reference electrode (240) may be placed spaced apart from the working electrode (230), and the reference electrode (240) and the working electrode (230) may be electrically disconnected.

[0155] The reference electrode (240) can provide a reference value for the voltage or potential value measured at the working electrode (230) during sample measurement. Using the potential value of the reference electrode (240) as the reference value, the concentration of the target substance selectively permeated into the working electrode (230) can be specified.

[0156] For example, by comparing the potential value at the reference electrode (240) and the potential value measured at the working electrode (230), the voltage generated purely by the concentration difference of the target substance (e.g., glucose) inside / outside the working electrode (240) can be calculated, and the concentration of the target substance can be derived from the voltage.

[0157]

[0158] The wiring section (250) is formed by a plurality of electrical wires extending from the electrodes constituting the electrode section (231). This wiring section (250) can be connected to a detection and analysis means that performs a current analysis function via an electrical connection medium such as an FPCB (Flexible Printed Circuit Board). A pad area can be provided at the end of the wiring section (250), and the FPCB can be electrically connected by being adhered to this pad area.

[0159]

[0160] <Blood sugar prediction method>

[0161]

[0162] Figures 6 to 8 are flowcharts schematically illustrating blood sugar prediction methods according to the first to third embodiments of the present invention.

[0163]

[0164] Referring to FIG. 6, the blood sugar prediction method of the present invention includes a body information input step (S100) of receiving body information of a user; a body fluid sugar information measurement step (S200) of measuring information on the sugar content in body fluid secreted from the user; an information storage step (S300) of storing the input body information of the user and the measured body fluid sugar content information; and a blood sugar analysis step (S400) of calculating a blood sugar prediction value using the body information of the user and the measured body fluid sugar content information.

[0165]

[0166] The blood sugar prediction method of the present invention may be a method of calculating a blood sugar prediction value using the blood sugar prediction system described above, and the contents of the <blood sugar prediction system> described above may be applied without limitation.

[0167]

[0168] The blood sugar prediction method of the present invention includes a body information input step (S100) for receiving a user's body information. The user's body information may be input from a body information input unit (100), and the user's body information may include one or more items selected from the group consisting of age, height, weight, and body fat percentage.

[0169]

[0170] The blood sugar prediction method of the present invention includes a body fluid sugar information measurement step (S200) for measuring information on the sugar content in body fluid secreted from a user. The information on the sugar content in the body fluid secreted from the user can be measured through a body fluid sugar information measurement unit (200). The information on the sugar content may be spectroscopic information measured using a photometric method or electrochemical information measured using an electrochemical method. The body fluid secreted from the user may be any one selected from the group consisting of sweat, tears, saliva, and urine.

[0171]

[0172] The blood sugar prediction method of the present invention includes an information storage step (S300) for storing information regarding the user's physical information and the measured sugar content in the body fluid. The information regarding the user's physical information and the sugar content in the body fluid may be stored in an information storage unit (300).

[0173]

[0174] The blood sugar prediction method of the present invention includes a blood sugar analysis step (S400) for calculating a blood sugar prediction value using the user's physical information and information regarding the sugar content in the measured body fluid. When predicting a user's blood sugar level based on information regarding the sugar content in the body fluid secreted from the user, the sugar content in the body fluid may differ depending on the physical characteristics of each individual user even if the blood sugar level is the same. Therefore, by applying the user's physical information along with the information regarding the sugar content in the body fluid secreted from the user to the calculation of the blood sugar prediction value to correct for errors, the user's blood sugar level can be predicted with higher accuracy.

[0175]

[0176] The above blood sugar analysis step (S400) may include a step of building a blood sugar prediction model through machine learning based on multiple data sets. The data sets may include body information, blood sugar information in bodily fluids, and actual blood sugar levels corresponding to the blood sugar information in the bodily fluids.

[0177]

[0178] In addition, the blood sugar analysis step (S400) may include a step of calculating a blood sugar prediction value by inputting the user's body information and information about the sugar content in the measured body fluid into the constructed blood sugar prediction model. The blood sugar prediction model constructed through machine learning may include at least one selected from the group consisting of a coefficient indicating a correlation between the user's body information and information about the sugar content in the body fluid secreted from the user, a coefficient indicating a correlation between the user's body information and an actual blood sugar level, and a coefficient indicating a correlation between information about the sugar content in the body fluid secreted from the user and an actual blood sugar level, and by applying the coefficient regarding the correlation of each variable obtained through machine learning, the accuracy of blood sugar prediction based on the information about the sugar content in the user's body fluid can be further increased.

[0179]

[0180] The above machine learning may use at least one selected from the group consisting of linear regression analysis, nonlinear regression analysis, partial least squares (PLS), Bayesian network, hidden markov model, decision tree, boosting, bagging, support vector machine, convolutional neural network, deep neural network, and recursive neural network (RNN), and it is preferable to use at least one selected from the group consisting of boosting, bagging, support vector machine, and deep neural network.

[0181]

[0182] Multiple data sets can be used to build the blood sugar prediction model using machine learning. To improve the accuracy of the constructed blood sugar prediction model, it is preferable to use at least 30 data sets. For example, the blood sugar prediction model can be built using machine learning based on 30 to 50 data sets.

[0183]

[0184] The body fluid sugar information measured in the above body fluid sugar information measurement step (S200) may be measured at a specific point in time after the start of exercise. The specific point in time after the start of exercise may refer to two different specific points in time, such as 20 minutes and 30 minutes after the start of exercise.

[0185] The blood sugar prediction method according to the present invention measures information on the sugar content in body fluid secreted from a user by a non-invasive method, and uses the information to provide a highly accurate blood sugar prediction value, thereby enabling the user to easily check their blood sugar status periodically during daily life or exercise.

[0186]

[0187] The blood sugar prediction method of the present invention may further include a step of transmitting and receiving information with an external device. The transmission and reception of the information may be via at least one of Wi-Fi, Bluetooth, wireless communication, and NFC.

[0188] The external device may be a user's personal mobile terminal, and may transmit, for example, the user's physical information, information on the sugar content in body fluids, and / or blood sugar prediction values ​​to the user's personal mobile terminal in real time, or may receive a data set of the user himself or another person stored in the user's personal mobile terminal.

[0189] The external device may be a cloud server, and may transmit, for example, the user's body information, information on the sugar content in body fluids, and / or blood sugar prediction values ​​to the cloud server in real time, or may receive the user's own or another person's data set stored on the cloud server.

[0190]

[0191] Referring to FIG. 7, a blood sugar prediction method according to a second embodiment of the present invention includes a step (S410a) of constructing a blood sugar prediction model through machine learning based on a plurality of previously stored data sets of others, and a step (S420) of calculating a blood sugar prediction value by inputting information about the user's body information and the sugar content in the measured body fluid into the constructed blood sugar prediction model. The description described for the blood sugar prediction method according to FIG. 6 can be equally applied to the blood sugar prediction method illustrated in FIG. 7, and detailed descriptions of steps, methods, and configurations that are substantially the same or similar are omitted.

[0192]

[0193] Building a blood sugar prediction model through machine learning requires at least 30 data sets. However, when each user initially uses the blood sugar prediction system according to the present invention, a sufficient number of data sets may not be collected. Therefore, before a user's own data set is sufficiently collected, a blood sugar prediction model can be built through machine learning based on multiple data sets of others previously stored in the information storage unit (300), and this model can be used to calculate blood sugar prediction values, thereby further improving the accuracy of blood sugar prediction.

[0194]

[0195] Referring to FIG. 8, a blood sugar prediction method according to a third embodiment of the present invention includes a step (S410b) of constructing a blood sugar prediction model through machine learning based on a plurality of data sets of the user, and a step (S420) of calculating a blood sugar prediction value by inputting the user's body information and information on the sugar content in the measured body fluid into the constructed blood sugar prediction model. The description described for the blood sugar prediction method according to FIGS. 6 and 7 can be equally applied to the blood sugar prediction method illustrated in FIG. 8, and detailed descriptions of steps, methods, and configurations that are substantially the same or similar are omitted.

[0196]

[0197] When a user repeatedly uses the blood sugar prediction method according to the present invention and sufficiently stores his / her own data set, a new personalized blood sugar prediction model can be built through machine learning based on the user's multiple data sets, and the newly built personalized blood sugar prediction model can be used to calculate a blood sugar prediction value.

[0198] In this way, by building a personalized blood sugar prediction model optimized for each user based on the user's own data set and inputting information about the user's physical information and sugar content in body fluids into the model to calculate a blood sugar prediction value, the accuracy of blood sugar prediction for each individual user can be further improved.

[0199]

[0200] FIGS. 9a, 9b, 10a, and 10b are diagrams showing the results of evaluating the accuracy of blood sugar prediction values ​​calculated using the blood sugar prediction system and blood sugar prediction method according to the present invention, expressed as a consensus error grid (CEG) area.

[0201] The definition of the above coincidence error grid (CEG) area is as shown in Table 1 below (refer to “ISO 15197:2013” ​​6.3).

[0202] Risk Level (CEG Zone)Risk for Diabetic PatientsANo effect on clinical measures.BChange in clinical measures - Little or no effect on clinical outcomes.CChange in clinical measures - Likely to affect clinical outcomes.DChange in clinical measures - May result in serious medical risks.EChange in clinical measures - May result in dangerous outcomes.

[0203]

[0204] More specifically, Fig. 9a shows a blood sugar prediction model constructed through linear regression analysis based on multiple data sets of others, and a blood sugar prediction value calculated by inputting user-entered body information and information on sugar content in body fluids into the constructed blood sugar prediction model (blood sugar prediction value) and calculating the predicted blood sugar value together with the actual blood sugar value. As a result, as shown in Fig. 9b, it can be confirmed that the predicted blood sugar value exhibits excellent accuracy, with the occupancy rate of Area A being 79.98% and the combined occupancy rate of Areas A and B being 99.51%.

[0205]

[0206] In addition, Fig. 10a shows a blood sugar prediction model constructed through boosting based on multiple data sets of others, and the blood sugar prediction value calculated by inputting information about the user's body information and the sugar content in body fluids into the constructed blood sugar prediction model (blood sugar prediction value) together with the actual blood sugar value. As a result, as shown in Fig. 10b, it can be confirmed that the occupancy rate of Area A is improved to 98.5%, and the combined occupancy rate of Area A and Area B is improved to 100%, further enhancing the accuracy of blood sugar prediction.

[0207]

[0208] The present invention can provide a blood sugar prediction system and a blood sugar prediction method capable of predicting blood sugar with high accuracy by calculating a blood sugar prediction value using the user's body information along with information on the sugar content in the user's body fluid measured by a non-invasive method.

Claims

1. A body information input section for receiving the user’s body information; A body fluid sugar information measuring unit for measuring information on the sugar content in body fluid secreted from a user; An information storage unit that stores the user's physical information entered above and information on the sugar content in the measured body fluid; and A blood sugar prediction system, comprising a blood sugar analysis unit that calculates a blood sugar prediction value using the user's body information and information on the sugar content in the measured body fluid.

2. In claim 1, A blood sugar prediction system, wherein the above physical information includes at least one selected from the group consisting of age, height, weight, and body fat percentage.

3. In claim 1, A blood sugar prediction system, wherein the above body fluid is any one selected from the group consisting of sweat, tears, saliva and urine.

4. In claim 1, A blood sugar prediction system, wherein the information on the sugar content is spectroscopic information measured using a photometric method or electrochemical information measured using an electrochemical method.

5. In claim 1, The above blood sugar analysis unit builds a blood sugar prediction model through machine learning based on multiple data sets. A blood sugar prediction system, wherein the above data set includes body information, blood sugar information in body fluid, and actual blood sugar levels corresponding to the blood sugar information in the body fluid.

6. In claim 5, The blood sugar analysis unit is a blood sugar prediction system that calculates a blood sugar prediction value by inputting the user's body information and information about the sugar content in the measured body fluid into the constructed blood sugar prediction model.

7. In claim 5, A blood sugar prediction system, wherein the machine learning comprises at least one selected from the group consisting of linear regression analysis, nonlinear regression analysis, partial least squares (PLS), Bayesian network, hidden markov model, decision tree, boosting, bagging, support vector machine, convolutional neural network, deep neural network, and recursive neural network (RNN).

8. In claim 5, A blood sugar prediction system, wherein the above multiple data sets are 30 or more data sets.

9. In claim 5, The above data set is a blood sugar prediction system, which is a data set of others.

10. In claim 9, A blood sugar prediction system, wherein the above-mentioned third party's data set is pre-stored in the information storage unit.

11. In claim 5, The above data set is a blood sugar prediction system, which is the user's own data set.

12. In claim 1, A blood sugar prediction system, wherein information regarding the sugar content in body fluid secreted by a user is measured at a specific point in time after the start of exercise.

13. In claim 12, A blood sugar prediction system in which two different specific points in time are identified after the start of the above exercise.

14. In claim 12, A blood sugar prediction system that is set at specific points after the start of exercise, 20 minutes and 30 minutes after the start of exercise.

15. In claim 1, A blood sugar prediction system further comprising a display unit that displays the blood sugar prediction value calculated by the blood sugar analysis unit.

16. In claim 1, A blood sugar prediction system further comprising a communication unit for transmitting and receiving information with an external device.

17. Body information input step for receiving the user’s body information; A body fluid sugar information measurement step for measuring information on the sugar content in body fluid secreted from a user; An information storage step for storing the user's physical information entered above and information on the sugar content in the measured body fluid; and A blood sugar prediction method, comprising a blood sugar analysis step of calculating a blood sugar prediction value using the user's body information and information on the sugar content in the measured body fluid.

18. In claim 17, A method for predicting blood sugar, wherein the above physical information includes at least one selected from the group consisting of age, height, weight, and body fat percentage.

19. In claim 17, A method for predicting blood sugar, wherein the body fluid is any one selected from the group consisting of sweat, tears, saliva, and urine.

20. In claim 17, A method for predicting blood sugar, wherein the information on the sugar content is spectroscopic information measured using a photometric method or electrochemical information measured using an electrochemical method.

21. In claim 17, The above blood sugar analysis step includes a step of building a blood sugar prediction model through machine learning based on multiple data sets, A blood sugar prediction method, wherein the above data set includes body information, blood sugar information in body fluid, and actual blood sugar levels corresponding to the blood sugar information in the body fluid.

22. In claim 21, A blood sugar prediction method, wherein the blood sugar analysis step includes a step of calculating a blood sugar prediction value by inputting the user's body information and information about the sugar content in the measured body fluid into the constructed blood sugar prediction model.

23. In claim 21, A method for predicting blood sugar, wherein the machine learning comprises at least one selected from the group consisting of linear regression analysis, nonlinear regression analysis, partial least squares (PLS), Bayesian network, hidden markov model, decision tree, boosting, bagging, support vector machine, convolutional neural network, deep neural network, and recursive neural network (RNN).

24. In claim 21, A method for predicting blood sugar levels, wherein the above multiple data sets are 30 or more data sets.

25. In claim 21, A blood sugar prediction method, wherein the data set used in the step of constructing the above blood sugar prediction model is a previously stored data set of another person.

26. In claim 21, A blood sugar prediction method, wherein the data set used in the step of constructing the above blood sugar prediction model is the user's own data set.

27. In claim 17, A method for predicting blood sugar, wherein information regarding the sugar content in body fluid secreted by a user is measured at a specific point in time after the start of exercise.

28. In claim 27, A method for predicting blood sugar levels at two different specific points in time after the start of the above exercise.

29. In claim 27, A method for predicting blood sugar levels at specific points after the start of exercise, which are 20 minutes and 30 minutes after the start of exercise.

30. In claim 17, A method for predicting blood sugar, further comprising the step of transmitting and receiving information with an external device.

Images