Pulse wave measurement and analysis method for non-invasive blood glucose measurement

By employing two biometric devices to acquire pulse wave signals from distinct body parts and applying noise removal and machine learning, the method enhances the accuracy and reliability of blood glucose estimation.

KR1020260161972APending Publication Date: 2026-11-10SOFTWARE CONVERGENCE INST CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
KR1020250057998
Authority / Receiving Office
KR · KR
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-05-02
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing non-invasive blood glucose measurement technologies using a single PPG sensor on the wrist have low accuracy due to variations in pulse wave signals across different body parts, leading to inconsistent blood glucose estimation, and are cumbersome for continuous use.

Method used

A method utilizing two biometric measuring devices, one worn continuously and another used intermittently, to acquire pulse wave signals from different body parts, followed by noise removal and machine learning to estimate blood glucose levels, with the signals being processed through a pre-trained algorithm to enhance accuracy.

Benefits of technology

This approach enables more accurate and reliable blood glucose estimation by averaging or summing signals from multiple body parts, reducing errors and improving measurement consistency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure PAT00002_ABST
    Figure PAT00002_ABST
Patent Text Reader

Abstract

The present invention relates to a pulse wave measurement and analysis method for non-invasive blood glucose measurement, wherein pulse wave signals are acquired from two parts of a user's body, and the blood glucose level is estimated using the two acquired pulse wave signals, thereby enabling a more accurate estimation of the blood glucose level. The present invention comprises: a first pulse wave signal acquisition step in which a pulse wave signal of a user is measured in real time through a PPG sensor while a first measuring device is worn at all times on one part of the user's body; a second pulse wave signal acquisition step in which a second measuring device is attached to another part of the user's body only when measuring the pulse wave signal while not being worn on the part of the user's body, and a pulse wave signal of a specific time of the user is measured through a PPG sensor; a step of transmitting the first pulse wave signal and the second pulse wave signal acquired from different parts of the user's body to a user terminal; a step of searching for and extracting the pulse wave signal acquired at the same time among the received first pulse wave signal and the second pulse wave signal; a step of extracting only the pulse wave signal corresponding to a pre-set minimum amplitude value and maximum amplitude value range from the two extracted pulse wave signals to remove noise; and a step of extracting bio-information from the noise-removed pulse wave signal, generating input data based on the extracted bio-information, and then inputting it into a pre-trained machine learning algorithm to estimate blood glucose levels.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a pulse wave measurement and analysis method for non-invasive blood glucose measurement, wherein pulse wave signals are acquired from two parts of a user's body, and the blood glucose level is estimated using the two acquired pulse wave signals, thereby enabling a more accurate estimation of the blood glucose level. Background Technology

[0003] Diabetes is a metabolic disease characterized by hyperglycemia resulting from impaired secretion or function of insulin, which is necessary for blood sugar control within the body. Chronic hyperglycemia caused by diabetes leads to damage and dysfunction of various organs in the body; in particular, it causes microvascular complications affecting the retina, kidneys, and nerves, as well as macrovascular complications such as arteriosclerosis, cardiovascular disease, and cerebrovascular disease, thereby increasing mortality rates.

[0004] However, diabetes can worsen or increase the incidence of complications due to blood sugar control, weight loss, and medication. Therefore, diabetic patients frequently monitor their blood sugar at home to manage their blood glucose levels.

[0005] Recently, a technology for estimating blood glucose and glycated hemoglobin using a non-invasive measuring device designed to be worn on the wrist in the form of a watch has been disclosed.

[0006] Wrist-worn meters have a low degree of contact with the skin, making it difficult to measure blood glucose accurately. Furthermore, they are not popular because many users find them uncomfortable or reluctant to wear them on their wrists due to their watch-like structure. Additionally, there is a problem in that blood glucose cannot be measured continuously because they are not often worn all day or for more than 10 hours due to the cumbersome and inconvenient nature of their use.

[0007] As a conventional technology for measuring blood glucose, Korean Patent Publication No. 10-2023-0044160 (Method and apparatus for non-invasive glycated hemoglobin or blood glucose estimation using machine learning) is disclosed. The above patent is a technology that estimates the wearer's glycated hemoglobin or blood glucose levels by inputting body mass index, blood oxygen saturation, and finger thickness information from the wearer's biosignals into a machine learning model.

[0008] However, the above technology has low measurement accuracy because it acquires the pulse wave signal of the subject using a single PPG sensor. Furthermore, accuracy is low because there are differences in the pulse wave signal depending on the measurement location, even if the subject moves or it is the same body part. In particular, since the pulse wave signal differs depending on the subject's body part (e.g., wrist, finger), there is a problem where accuracy is lowered as differences occur in the estimated blood glucose value, which is the learning result, due to the pulse wave signals that differ by body part. Prior art literature

[0010] Published Patent Application No. 10-2019-0063446 (June 7, 2019. Non-invasive blood glucose prediction system, blood glucose prediction method, and blood glucose sensor) Published Patent Application No. 10-2023-0163968 (December 1, 2023. Artificial intelligence-based blood glucose prediction system and method) The problem to be solved

[0011] The present invention was devised to solve the above problems and estimates blood glucose levels using a pulse wave signal measured by a first measuring device that is constantly worn on a part of the user's body and a pulse wave signal measured by a second measuring device at a specific time at a different part of the body. By using pulse wave signals acquired from two locations rather than one part of the user's body, it is possible to estimate blood glucose levels with higher accuracy. means of solving the problem

[0013] The means of the present invention for solving the above problem comprises: a first pulse wave signal acquisition step in which a pulse wave signal of a user is measured in real time through a PPG sensor while a first measuring device is worn at all times on one part of the user's body; a second pulse wave signal acquisition step in which a second measuring device is attached to another part of the user's body only when measuring the pulse wave signal while not being worn on the part of the user's body, and a pulse wave signal of a specific time of the user is measured through a PPG sensor; a step of transmitting the first pulse wave signal and the second pulse wave signal acquired from different parts of the user's body to a user terminal; a step of searching for and extracting the pulse wave signal acquired at the same time among the received first pulse wave signal and the second pulse wave signal; a step of extracting only the pulse wave signal corresponding to a pre-set minimum amplitude value and maximum amplitude value range for noise removal from the two extracted pulse wave signals; and a step of extracting bio-information from the noise-removed pulse wave signal, generating input data based on the extracted bio-information, and then inputting it into a pre-trained machine learning algorithm to estimate blood glucose levels.

[0014] In the step of estimating the blood glucose level, two noise-removed pulse wave signals are summed to calculate an average value, and a biosignal is extracted from the average value to generate input data, which is then input into a machine learning algorithm to estimate the blood glucose level.

[0015] In addition, as another embodiment, in the step of estimating the blood glucose level, a biosignal is extracted from each of the two noise-removed pulse wave signals, and the extracted biosignal is input into a machine learning algorithm to estimate each blood glucose level, and the estimated blood glucose levels are summed to determine the average value as the final blood glucose level estimate. Effects of the invention

[0017] According to the present invention, which is configured as described above, pulse wave signals are acquired from different parts of a user's body, and bio-information is extracted from the acquired pulse wave signals to estimate blood glucose levels. This enables the derivation of more accurate results and increases the reliability of blood glucose measurement values. Brief explanation of the drawing

[0019] FIG. 1 is a system configuration diagram for non-invasive blood glucose measurement according to the present invention, FIG. 2 is a flowchart showing the process of estimating blood glucose levels using a pulse wave measurement and analysis method for non-invasive blood glucose measurement according to the present invention. Specific details for implementing the invention

[0020] Hereinafter, a pulse wave measurement and analysis method for non-invasive blood glucose measurement according to a preferred embodiment of the present invention will be described in detail with reference to the attached drawings.

[0021] As described above, the pulse wave measurement and analysis method for non-invasive blood glucose measurement according to the present invention estimates blood glucose levels without blood sampling using a biometric measuring device, and is a method that enables effective blood glucose management by measuring and analyzing the pulse wave of the subject using two biometric measuring devices to estimate blood glucose levels more accurately.

[0022] In the present invention, two biometric measuring devices are used, and one of the biometric measuring devices is a first measuring device (10) that measures and acquires pulse wave signals while being worn by the user at all times.

[0023] The first measuring device (10) is a device that measures pulse wave signals while being worn by the user being measured at all times, and can be configured, for example, in the form of a watch worn on the wrist, a ring worn on the finger, or a band attached to the skin.

[0024] Another measuring device among the biometric measuring devices is a second measuring device (15) for measuring pulse wave signals at a time when blood sugar is to be measured at any time.

[0025] The second measuring device (15) is a device for measuring a user's pulse wave signal, which is not worn by the user at all times but is used only at any specific time to measure. For example, it can be configured in a form that the user holds or grips with their hand, a form that wraps around, a form in which the palm is spread out and placed on the floor, a clamp form, or a plate form with a mounting groove formed to place the inner surface of the finger on, such as in fingerprint recognition.

[0026] Of course, the second measuring device can be configured in the same form as the first measuring device and attached to the skin of the opposite hand or another body part for measurement.

[0027] The first measuring device (10) and the second measuring device (15) transmit the acquired pulse wave signal to a user terminal (20) connected via wired or wireless communication.

[0028] The pulse wave measurement and analysis method for non-invasive blood glucose measurement according to the present invention comprises a first pulse wave signal acquisition step, a second pulse wave signal acquisition step, a step of transmitting the acquired pulse wave signal to a user terminal, a step of extracting the pulse wave signal acquired at the same time, a noise removal step, and a blood glucose level estimation step.

[0029] The first pulse wave signal acquisition step is a step of measuring and acquiring the user's pulse wave signal in real time through the PPG sensor of the first measuring device, which is worn at all times on one part of the user's body.

[0030] The second pulse wave signal acquisition step is a step in which the user's pulse wave signal is acquired from another part of the user's body only when the pulse wave signal is measured, rather than being worn on a part of the user's body; this is achieved by placing the second measuring device in close contact and measuring and acquiring the user's pulse wave signal at a specific time through the PPG sensor.

[0031] As described above, the first pulse wave signal and the second pulse wave signal acquired from different parts of the user's body are transmitted to the user terminal, and the user terminal searches for and extracts the pulse wave signal acquired at the same time among the received first pulse wave signal and second pulse wave signal.

[0032] The user terminal (20) extracts bio-information from the received pulse wave signal and inputs it into a machine learning algorithm to estimate the blood glucose level, and the estimated blood glucose level is displayed on the screen of the user terminal, allowing the user to visually check their blood glucose level.

[0033] The user terminal (20) is a terminal that is wired or wirelessly connected to the first measuring device (10) and the second measuring device (15), and is equipped with a memory for storing a measurement program, a microprocessor for executing the measurement program to perform calculations and control, etc.

[0034] For example, the user terminal is a communication-capable terminal such as a personal computer (PC), laptop, personal digital assistant (PDA), mobile communication terminal, or tablet terminal, and for convenience of use, a smart pad or smartphone carried by the wearer can be used as a mobile communication terminal.

[0035] A dedicated app (application) for connecting to a server (30) for analyzing pulse wave signals is installed on the user terminal (20), and the wearer's basic biometric information is entered into the dedicated app, and pulse wave signals detected by the first measuring device (10) and the second measuring device (15) are collected through the dedicated app.

[0036] A machine learning algorithm is installed in the server (30), and the user terminal (20) connects to the server to extract bio-information from the received pulse wave signal, inputs it into the machine learning algorithm to analyze it through machine learning and estimate the blood glucose level, and transmits the measurement information including the estimated value to the user terminal (20) to display it on the screen.

[0037] Accordingly, when a pulse wave is measured at the first measuring device (10) and the second measuring device (15) respectively while the dedicated app is running on the user terminal (20), bio-information (variables) are extracted using the measured pulse wave signal, and the extracted bio-information is input into a machine learning algorithm to estimate the blood sugar level and determine the alarm level, and the result is displayed on the screen. Based on the information displayed on the screen of the user terminal, the user can predict their health status, such as whether they have diabetes, by checking their blood sugar level.

[0038] Since the two pulse wave signals acquired from each PPG sensor of the first measuring device (10) and the second measuring device (15) contain noise, a noise removal process is performed.

[0039] When removing noise, the amplitude of the pulse wave signal is first adjusted through a scaling process, and only the pulse wave signal corresponding to a preset range of minimum and maximum amplitude values ​​is extracted from the scaled pulse wave signal.

[0040] The preset minimum and maximum amplitude values ​​are average values ​​extracted from the pulse wave signals of diabetic patients and can be set differently depending on the surrounding environment.

[0041] Bio-information is extracted from the noise-removed pulse wave signal, input data is generated based on the extracted bio-information, and then input into a pre-trained machine learning algorithm to estimate blood glucose levels.

[0042] The process of measuring blood glucose involves the steps of measuring and acquiring pulse wave signals from different body parts, transmitting them to a user terminal, removing noise, extracting bio-information, machine learning, and estimating blood glucose levels.

[0043] After estimating the blood glucose level, the alarm level corresponding to the estimated blood glucose level is calculated, and the alarm level is displayed along with the blood glucose level on the user terminal screen so that the user can identify it.

[0044] The above machine learning algorithm can use classification technique algorithms such as regression analysis, logistic regression analysis, decision tree, Bayesian classification, artificial neural network, SVM (Supporter Vector Machine), and K-nearest neighbors.

[0045] In the present invention, 10,000 biological information data sets for training were collected from diabetic patients and normal people, and the artificial neural network and logistic regression analysis algorithm were trained, respectively.

[0046] The artificial neural network and logistic regression algorithm above finally output blood glucose levels, respectively, calculate disease prediction probability through the outputted values, and determine the alarm level.

[0047] The above alert level can be classified as "Safe," "Concern," "Caution," or "Suspicious," and is determined as "Safe" when the disease prediction probability is 0% to 10%, "Concern" when the disease prediction probability is 11% to 30%, "Caution" when the disease prediction probability is 31% to 50%, and "Suspicious" when the disease prediction probability is 51% or higher.

[0048] As previously explained, in the present invention, pulse wave signals are acquired from different parts of the user's body, and bio-information is extracted from the acquired pulse wave signals to estimate blood glucose levels, thereby enabling more accurate estimation of blood glucose levels compared to conventional methods.

[0049] At this time, in the step of estimating blood glucose levels, the accuracy of estimating blood glucose levels can be improved by summing two pulse wave signals acquired at the same time and calculating the average value, and then generating input data by extracting bio-information from the average value of the pulse wave signals and performing machine learning.

[0050] In addition to this method, to improve the accuracy of blood glucose level estimation, multiple machine learning models can be constructed for each body part, and blood pressure values ​​can be estimated using pulse wave signals for each body part. The estimated blood glucose levels can then be summed to calculate the average value, which is determined as the final estimated blood glucose level.

[0051] The reason for doing this is that pulse wave signals vary by body part, and these differing signals can cause significant errors in the results (blood sugar levels) during machine learning; therefore, this measure aims to reduce such errors and improve the accuracy of the results.

[0052] For example, when the first measuring device (10) is configured in the form of a band to acquire a pulse wave signal from the wrist and the second measuring device (15) acquires a pulse wave signal from the finger, bio-information is extracted from the pulse wave signals acquired from the wrist and finger, respectively, and then generated as training data. After machine learning is performed for each part to estimate the blood glucose level, the estimated blood glucose levels are summed, and the average value is determined as the final blood glucose level. Explanation of the symbols

[0054] 10: First measuring instrument 15: Second measuring instrument 20: Measurer terminal 30: Server

Claims

Claim 1 A method for measuring and analyzing pulse waves for non-invasive blood glucose measurement, characterized by comprising: a first pulse wave signal acquisition step in which a first measuring device is worn at all times on one part of the user's body and the user's pulse wave signal is measured in real time through a PPG sensor; a second pulse wave signal acquisition step in which a second measuring device is attached to another part of the user's body only when measuring the pulse wave signal, and the user's pulse wave signal at a specific time is measured through a PPG sensor, while the device is not worn on the user's body; a step of transmitting the first pulse wave signal and the second pulse wave signal acquired from different parts of the user's body to a user terminal; a step of searching for and extracting the pulse wave signal acquired at the same time among the received first pulse wave signal and the second pulse wave signal; a step of extracting only the pulse wave signal corresponding to a preset range of minimum and maximum amplitude values ​​for noise removal from the two extracted pulse wave signals; and a step of extracting bio-information from the noise-removed pulse wave signal, generating input data based on the extracted bio-information, and inputting it into a pre-trained machine learning algorithm to estimate blood glucose levels. Claim 2 A pulse wave measurement and analysis method for non-invasive blood glucose measurement, characterized in that, in the step of estimating the blood glucose level, two noise-removed pulse wave signals are summed to calculate an average value, a biosignal is extracted from the average value to generate input data, and then input into a machine learning algorithm to estimate the blood glucose level. Claim 3 A pulse wave measurement and analysis method for non-invasive blood glucose measurement, characterized in that, in the step of estimating blood glucose levels, a biosignal is extracted from each of two noise-removed pulse wave signals, the extracted biosignals are input into a machine learning algorithm to estimate each blood glucose level, and the estimated blood glucose levels are summed to determine the average value as the final estimated blood glucose level.