Blood pressure manometer, blood pressure measuring method, blood pressure measuring program, learning model construction method, and learning model construction program

The blood pressure monitor decomposes waveforms into multiple feature quantities using Wave-U-Net, addressing noise-induced inaccuracies through a machine learning model, ensuring accurate blood pressure measurement without data collection or calibration.

JP2024163785APending Publication Date: 2024-11-22OMRON CORP
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
JP2023079664
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-12
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Blood pressure measurement accuracy is compromised by noise in pulse waveforms, which is influenced by cuff wrapping and positioning, necessitating costly data collection and calibration for parameter settings.

Method used

A blood pressure monitor that decomposes cuff and pulse waveforms into multiple feature quantities using Wave-U-Net, enabling accurate blood pressure measurement through a learning model constructed with machine learning, without requiring data collection or calibration.

Benefits of technology

Achieves high accuracy in blood pressure measurement by learning global and local features, effectively mitigating noise impacts, thus eliminating the need for data collection and calibration.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024163785000001_ABST
    Figure 2024163785000001_ABST
Patent Text Reader

Abstract

To provide a blood pressure manometer capable of performing blood pressure measurement as accurately as possible without performing data collection for setting parameters of a blood pressure measurement algorithm or calibration of a measurement result.SOLUTION: A blood pressure manometer includes: a cuff wound around the arm of a person to be measured; detection part for detecting a cuff pressure that the cuff receives from the arm; a decomposition part for acquiring a first waveform indicating a time-series change in a feature amount related to a blood pressure of the person to be measured on the basis of the cuff pressure, and decomposing the first waveform into a plurality of second waveforms indicating two or more different feature amounts; a learning model constructed by machine learning with the second waveforms as an explanatory variable and with the blood pressure actually measured by an auscultation method as an objective variable, for a plurality of subjects; a measuring part for inputting the plurality of second waveforms to the learning model and measuring the blood pressure of the person to be measured; and an output part for outputting the blood pressure measured by the measuring part.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical Field]

[0001] The present invention relates to a sphygmomanometer, a blood pressure measurement method, a blood pressure measurement program, a learning model construction method, and a learning model construction program. [Background technology]

[0002] Health conditions are known through blood pressure measurement. Attempts are being made to improve the accuracy of blood pressure measurement by using machine learning, including deep learning. For example, methods of measuring blood pressure using deep learning are being considered, in which the shape of each pulse or the shape of the envelope of the pulse waveform is input into a model (see Non-Patent Documents 1-3). [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Ahmadreza Argha, Branko G. Celler, Nigel H. Lovell, Artificial Intelligence Based Blood Pressure Estimation From Auscultatory and Oscillometric Waveforms: A Methodological Review, "IEEE REVIEWS IN BIOMEDICAL ENGINEERING", (USA), 2022, VOL.15, p.152-p.168 [Non-patent document 2] AHMADREZA ARGHA, JI WU, STEVEN W. SU, AND BRANKO G. CELLER, "Blood Pressure Estimation From Beat-by-Beat Time-Domain Features of Oscillometric Waveforms Using Deep-Neural-Network Classification Models", [online], August 6, 2019, IEEE, [searched on March 28, 2023], Internet <URL: https: / / ieeexplore.ieee.org / document / <8789404>

Non-Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0004] When measuring blood pressure using the oscillometric method, blood pressure is measured based on the waveform of the pulse wave, so the accuracy of blood pressure measurement decreases if the pulse wave contains noise. The noise contained in the pulse wave can increase depending on the way the cuff is wrapped and the position where it is wrapped. Therefore, to suppress the effects of this noise, calibration of the measured blood pressure and data collection for setting parameters in the blood pressure measurement algorithm are carried out. However, even if calibration or parameter setting is carried out, there is no guarantee that measurement errors due to noise can be eliminated, and data collection is also costly.

[0005] One aspect of the disclosed technology aims to provide a sphygmomanometer, a blood pressure measurement method, a blood pressure measurement program, a learning model construction method, and a learning model construction program that can perform blood pressure measurements with the highest possible accuracy without collecting data to set parameters for a blood pressure measurement algorithm or calibrating the measurement results. [Means for solving the problem]

[0006] One aspect of the disclosed technology is exemplified by the following blood pressure monitor: This blood pressure monitor includes a cuff that is wrapped around the arm of a person to be measured, a detection unit that detects the cuff pressure that the cuff receives from the arm, and The apparatus includes a decomposition unit that acquires a first waveform indicating a time-series change in a feature related to the blood pressure of the subject based on the cuff pressure and decomposes the first waveform into a plurality of second waveforms indicating two or more different feature values; a learning model constructed by machine learning with the second waveform of the subject as an explanatory variable and blood pressure actually measured by auscultation as a target variable; a measurement unit that inputs the plurality of second waveforms into the learning model to measure the blood pressure of the subject; and an output unit that outputs the blood pressure measured by the measurement unit.

[0007] According to this sphygmomanometer, a first waveform showing time-series changes in a feature value related to the subject's blood pressure is decomposed into multiple second waveforms showing two or more different feature values. In other words, each of the second waveforms can be said to represent the first waveform from multiple perspectives. Therefore, even if the first waveform contains noise, the noise may have a significant effect on the second waveform related to a certain feature value, but the noise may have a small effect on the second waveform related to another feature value. Therefore, according to this sphygmomanometer, blood pressure measurement can be performed with the highest possible accuracy without collecting data or calibrating the measurement results to set parameters for the blood pressure measurement algorithm.

[0008] Here, in this blood pressure monitor, the first waveform includes a cuff pressure waveform showing a time series change in the cuff pressure, the decomposition unit decomposes the cuff pressure waveform into a plurality of third waveforms showing two or more different feature quantities, the learning model is constructed using the third waveform for the subject as the explanatory variable, and the measurement unit may input the third waveform into the learning model to measure the blood pressure of the person being measured.

[0009] In addition, in this blood pressure monitor, the first waveform may include a pulse waveform showing time-series changes in the pulse wave of the subject obtained based on time-series changes in the cuff pressure, the decomposition unit may decompose the pulse waveform into a plurality of fourth waveforms showing two or more different feature amounts, the learning model may be constructed using the fourth waveform for the subject as the explanatory variable, and the measurement unit may input the fourth waveform into the learning model to measure the blood pressure of the subject.

[0010] Furthermore, in this blood pressure monitor, the first waveform may include a cuff pressure waveform showing a time series change in the cuff pressure and a pulse waveform showing a time series change in the pulse wave of the subject obtained based on the time series change in the cuff pressure, the decomposition unit may decompose the cuff pressure waveform into a plurality of third waveforms showing two or more different feature amounts and decompose the pulse waveform into a fourth waveform obtained from the time series change in the cuff pressure, the learning model may be constructed using the third waveform and the fourth waveform for the subject as the explanatory variables, and the measurement unit may input the third waveform and the fourth waveform to the learning model to measure the blood pressure of the subject.The decomposition unit may also include Wave-U-Net.

[0011] The disclosed technology can also be understood from the aspects of a blood pressure measurement method, a blood pressure measurement program, a learning model construction method, and a learning model construction program. [Effects of the Invention]

[0012] According to the disclosed technology, blood pressure measurement can be performed with the highest possible accuracy without collecting data to set parameters for the blood pressure measurement algorithm or calibrating the measurement results. [Brief explanation of the drawings]

[0013] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of a sphygmomanometer according to an embodiment. [Figure 2] FIG. 2 is a diagram schematically illustrating a state in which the blood pressure monitor is attached to the wrist of a subject. [Figure 3] FIG. 3 is a diagram illustrating a process performed by the waveform decomposition unit according to the embodiment. [Figure 4] FIG. 4 is a first diagram schematically illustrating processing by the blood pressure measurement unit according to the embodiment. [Figure 5] FIG. 5 is a second diagram schematically illustrating the processing of the blood pressure measurement unit according to the embodiment. [Figure 6] FIG. 6 is a first diagram schematically illustrating construction of a blood pressure model according to an embodiment. [Figure 7]FIG. 7 is a second diagram schematically illustrating construction of a blood pressure model according to an embodiment. [Figure 8] FIG. 8 is a third diagram schematically illustrating construction of a blood pressure model according to an embodiment. [Figure 9] FIG. 9 is a fourth diagram schematically illustrating construction of a blood pressure model according to an embodiment. [Figure 10] FIG. 10 is a fifth diagram schematically illustrating construction of a blood pressure model according to an embodiment. [Figure 11] FIG. 11 is a diagram illustrating an example of a processing flow of the sphygmomanometer according to the embodiment. [Figure 12] FIG. 12 is a diagram illustrating a verification result of the blood pressure measurement accuracy of the sphygmomanometer according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0014] <Application example> An application example of the present invention will be described. A blood pressure monitor 100 according to the application example is a device that measures the blood pressure of a subject by wrapping a cuff 120 around the subject's upper arm or wrist. In the blood pressure monitor 100, the measurement unit 114 inputs at least one of a time series change in the cuff pressure and a time series change in the pulse wave to a blood pressure model 117, and acquires the blood pressure of the subject from the blood pressure model 117.

[0015] The measurement unit 114 controls the pressure of the cuff 120 to acquire time-series changes in the feature quantity related to the blood pressure of the subject from the pressure sensor 113. Examples of time-series changes in the feature quantity related to the blood pressure of the subject include time-series changes in the cuff pressure and time-series changes in the waveform of an OscilloMetric Waveform (OMW).

[0016] The measurement unit 114 has a waveform decomposition unit 115 and a blood pressure measurement unit 116. The waveform decomposition unit 115 decomposes the time series changes in the feature quantities related to the subject's blood pressure into a plurality of second waveforms (mode waveforms). The mode waveforms include waveforms obtained by decomposing the time series changes in the feature quantities related to the subject's blood pressure based on waveform characteristics such as high and low frequency and large and small amplitude. The mode waveforms may include, for example, a waveform that globally indicates the entire time series changes in the feature quantities related to the subject's blood pressure, or a waveform that locally indicates a part of the time series changes in the feature quantities related to the subject's blood pressure. The waveform decomposition unit 115 includes, for example, Wave-U-Net.

[0017] The blood pressure measurement unit 116 inputs the multiple mode waveforms decomposed by the waveform decomposition unit 115 to the blood pressure model 117, and acquires the blood pressure of the subject from the blood pressure model 117. The blood pressure measurement unit 116 causes the output unit 118 to output the acquired blood pressure.

[0018] The blood pressure model 117 is a learning model that outputs the blood pressure of a subject when multiple mode waveforms are input. The blood pressure model 117 is constructed by machine learning using multiple mode waveforms obtained by decomposing the time-series changes in features related to the subject's blood pressure as explanatory variables and blood pressure measured by auscultation as a target variable. The machine learning includes deep learning.

[0019] According to this application example, the mode waveform includes a waveform that indicates a global change in the time-series change of the feature quantity related to the subject's blood pressure and a waveform that indicates a local change. Therefore, the blood pressure model 117 can uniformly learn both the global and local features of the time-series change of the feature quantity related to the subject's blood pressure. Such blood pressure model 117 can output the subject's blood pressure with high accuracy even if the time-series change of the feature quantity related to the subject's blood pressure contains noise. In other words, according to this application example, even if the time-series change of the feature quantity related to the subject's blood pressure acquired by the measurement unit 114 contains noise, high-accuracy blood pressure measurement can be achieved without collecting data to set parameters for the blood pressure measurement algorithm or calibrating the measurement results.

[0020] <Embodiment> Hereinafter, an embodiment will be described with reference to the drawings. FIG. 2 is a diagram showing an example of the configuration of a blood pressure monitor 100 attached to the wrist H1 of a person being measured. The blood pressure monitor 100 includes a main body 110 and a cuff 120. The blood pressure monitor 100 measures the blood pressure of the person being measured by wrapping the cuff 120 around the wrist H1 of the person being measured, for example. The position where the cuff 120 is wrapped is not limited to the wrist H1, and may be the upper arm of the person being measured.

[0021] Cuff 120 is formed in a belt shape and has a bag inside which air is supplied. When air is supplied to the bag inside cuff 120 while cuff 120 is wrapped around the upper arm or the like of the person being measured, the pressure exerted by the cuff on the upper arm or the like increases. Furthermore, when air is inhaled from the bag inside cuff 120 while cuff 120 is wrapped around the upper arm or the like of the person being measured, the pressure exerted by the cuff on the upper arm or the like decreases.

[0022] The main body 110 includes, for example, a processor and a storage unit. The processor executes a program stored in the storage unit to implement various processing units, such as a start switch 111, a control unit 112, a pressure sensor 113, a measurement unit 114, a blood pressure model 117, and an output unit 118. The measurement unit 114 includes a waveform decomposition unit 115 and a blood pressure measurement unit 116. In other words, the sphygmomanometer 100 can be regarded as a computer having a cuff 120. Note that the various processing units, such as the start switch 111, the control unit 112, the pressure sensor 113, the measurement unit 114, the blood pressure model 117, and the output unit 118, may be dedicated hardware circuits. When the start switch 111 is turned on, the sphygmomanometer 100 is started up and blood pressure measurement is initiated.

[0023] The control unit 112 controls the pressure applied to the upper arm or the like of the subject by the cuff 120 by supplying air to the cuff 120 and inhaling air from the cuff 120. When the activation switch 111 is turned on, the control unit 112 starts controlling the pressure of the cuff 120.

[0024] Pressure sensor 113 is a sensor that detects the pressure (cuff pressure) caused by cuff 120. Pressure sensor 113 outputs the detected cuff pressure to measurement unit 114. The cuff pressure includes, for example, the pressure of cuff 120 pressing against the upper arm or the like, and pressure vibrations caused by the pulsation of blood flowing in the artery of the upper arm or the like.

[0025] The measurement unit 114 measures blood pressure based on the pressure detected by the pressure sensor 113. The measurement unit 114 acquires, from the pressure sensor 113, a time series change in the cuff pressure of the cuff 120, which is increased to a pressure higher than the systolic blood pressure and then gradually decreased to a pressure lower than the diastolic blood pressure. The measurement unit 114 also extracts, from the time series change in the cuff pressure acquired from the pressure sensor 113, a time series change in pressure vibration caused by the pulsation of blood flow in the artery. The waveform decomposition unit 115 and the blood pressure measurement unit 116 process the extracted waveform showing the time series change in the cuff pressure (hereinafter also referred to as the cuff pressure waveform) and the waveform showing the time series change in the pressure vibration (hereinafter also referred to as the pulse waveform), respectively.

[0026] The waveform decomposition unit 115 decomposes each of the cuff pressure waveform and the pulse waveform into a plurality of mode waveforms. Fig. 3 is a diagram schematically illustrating the processing of the waveform decomposition unit 115 according to the embodiment. For example, the waveform decomposition unit 115 decomposes the cuff pressure waveform W11 into 14-channel mode waveforms, each with a data length of 16,384 bits, to generate a mode waveform group W12. The waveform decomposition unit 115 also decomposes the pulse waveform W21 into 14-channel mode waveforms, each with a data length of 16,384 bits, to generate a mode waveform group W22.

[0027] The waveform decomposition unit 115 includes, for example, Wave-U-Net. Wave-U-Net is a learning model used, for example, for waveform decomposition of voice. Voice contains non-periodic components in addition to periodic components, and it is considered that, like voice, the cuff pressure waveform and pulse waveform also contain non-periodic components in addition to periodic components. For this reason, in this embodiment, the waveform decomposition unit 115 that decomposes the cuff pressure waveform and pulse waveform includes Wave-U-Net. In addition, the waveform decomposition unit 115 may include, for example, any one of DEMUCS, D3NET, Conv-TasNet, Wavenet, Sams-Net, and Meta-TasNet instead of Wave-U-Net.

[0028] Here, as explained in the application examples, the mode waveform may include a waveform obtained by decomposing the time-series change in a feature quantity related to the subject's blood pressure based on waveform characteristics such as high and low frequency and amplitude. The mode waveform may also include a waveform that globally indicates the entire time-series change in a feature quantity related to the subject's blood pressure, or a waveform that locally indicates a portion of the time-series change in a feature quantity related to the subject's blood pressure. An example of a global waveform is the envelope of a waveform that indicates the time-series change in a feature quantity related to the subject's blood pressure. An example of a locally indicated waveform is a waveform that is an extracted (pulled out) portion of a waveform that indicates the time-series change in a feature quantity related to the subject's blood pressure.

[0029] The blood pressure measurement unit 116 measures the blood pressure of the subject based on the mode waveforms decomposed by the waveform decomposition unit 115. FIGS. 4 and 5 are diagrams schematically illustrating the processing of the blood pressure measurement unit 116 according to an embodiment. The blood pressure measurement unit 116 applies a one-dimensional convolutional neural network (1D-CNN) to the mode waveform group W12 input from the waveform decomposition unit 115, replacing it with intermediate features W13 of 30 channels, each with a data length of 1025 bits. The blood pressure measurement unit 116 further applies 1D-CNN to the intermediate features W3, replacing it with intermediate features W14 of 5 channels, each with a data length of 35 bits.

[0030] Furthermore, the blood pressure measurement unit 116 applies a one-dimensional convolutional neural network (1D-CNN) to the mode waveform group W22 input from the waveform decomposition unit 115, replacing it with 30-channel intermediate feature values ​​W23, each with a data length of 1025 bits. The blood pressure measurement unit 116 further applies 1D-CNN to the intermediate feature values ​​W23, replacing it with 5-channel intermediate feature values ​​W24, each with a data length of 35 bits. The blood pressure measurement unit 116 inputs the intermediate feature values ​​W14 and W24 to the blood pressure measurement unit 116, and acquires the blood pressure of the subject as an output of the blood pressure measurement unit 116.

[0031] The blood pressure model 117 is a learning model that outputs blood pressure when an intermediate feature value W14 generated from the cuff pressure waveform and an intermediate feature value W24 generated from the pulse waveform are input. In other words, the blood pressure model 117 is a learning model that outputs blood pressure when a feature value of the cuff pressure waveform and a feature value of the pulse waveform are input. The blood pressure model 117 is stored in, for example, a storage unit contained in the main body 110.

[0032] The output unit 118 outputs the blood pressure measured by the blood pressure measuring unit 116. The output unit 118 is, for example, a liquid crystal display (LCD), a plasma display, or the like. The output unit 118 may be a display panel (PDP), an inorganic electroluminescence (EL) panel, or an organic EL panel. The output unit 118 may be a printer or a speaker.

[0033] <Construction of Blood Pressure Model 117> 6 to 10 are diagrams schematically illustrating the construction of a blood pressure model 117 according to an embodiment. The construction of the blood pressure model 117 will be described with reference to FIGS. 6 to 10. In the construction of the blood pressure model 117, multiple subjects (E1, E2, . . . , EN) are prepared. Note that, although multiple subjects are prepared in this embodiment, the number of subjects prepared may be one. In step T1, the measurement unit 114 acquires a cuff pressure waveform P11 and a pulse waveform P21 of subject E1.

[0034] In step T2, the waveform decomposition unit 115 divides the cuff pressure waveform P11 into 1 In step T3, the waveform decomposition unit 115 decomposes the pulse waveform P21 into 14-channel mode waveforms, each with a data length of 16384 bits, to generate a mode waveform group P22.

[0035] In step T4, the blood pressure measurement unit 116 applies 1D-CNN to the mode waveform group P12 input from the waveform decomposition unit 115, replacing it with 30-channel intermediate features P13, each with a data length of 1025 bits. In step T5, the blood pressure measurement unit 116 applies 1D-CNN to the intermediate features P13, replacing it with 5-channel intermediate features P14, each with a data length of 35 bits.

[0036] In step T6, the blood pressure measurement unit 116 applies 1D-CNN to the mode waveform group P22 input from the waveform decomposition unit 115, replacing it with 30-channel intermediate features P23, each with a data length of 1025 bits. In step T7, the blood pressure measurement unit 116 applies 1D-CNN to the intermediate features P23, replacing it with 5-channel intermediate features P24, each with a data length of 35 bits.

[0037] In step T8, a blood pressure model 117 is generated by machine learning using the intermediate feature value P14 generated from the cuff pressure waveform and the intermediate feature value P24 generated from the pulse waveform as explanatory variables and the blood pressure as a response variable. The blood pressure used as the response variable is, for example, blood pressure measured by auscultation. The processes from steps T1 to T8 are also performed for each of subjects E2 to EN.

[0038] <Processing flow> 11 is a diagram showing an example of a processing flow of the sphygmomanometer 100 according to the embodiment. Hereinafter, an example of a processing flow of the sphygmomanometer 100 will be described with reference to FIG.

[0039] In step S1, activation switch 111 is turned on to activate sphygmomanometer 100. In step S2, control unit 112 supplies air to cuff 120 to pressurize the wrist of the subject. After inflating cuff 120 to a pressure higher than the systolic blood pressure, control unit 112 gradually reduces the pressure in cuff 120.

[0040] In step S3, the measurement unit 114 acquires a cuff pressure waveform and a pulse waveform based on the cuff pressure detected by the pressure sensor 113. Furthermore, the waveform decomposition unit 115 decomposes the waveform acquired in step S3 into a plurality of mode waveforms. In step S4, the blood pressure measurement unit 116 measures the blood pressure of the subject using the blood pressure model 117 based on the mode waveforms decomposed in step S3.

[0041] In step S5, the measurement unit 114 determines whether or not the blood pressure measurement of the subject has been completed. The measurement unit 114 may determine that the blood pressure measurement is completed when, for example, both the systolic blood pressure and the diastolic blood pressure of the subject have been measured. If the blood pressure measurement is completed (YES in step S5), the process proceeds to step S6. If the blood pressure measurement is not completed (NO in step S5), the process proceeds to step S2.

[0042] In step S6, the control unit 112 deflates the air from the cuff 120. In step S7, the output unit 118 outputs the blood pressure measured in step S4.

[0043] <Verification> The blood pressure measurement accuracy of the sphygmomanometer 100 according to the embodiment described above has been verified, and will be described below. FIG. 12 is a diagram illustrating the verification results of the blood pressure measurement accuracy of the sphygmomanometer 100 according to the embodiment. In FIG. 12, the blood pressure measured by the sphygmomanometer 100 and the oscillometric The blood pressure measured by the Rick method is compared.

[0044] In this verification, blood pressure measurements were performed multiple times for each of multiple subjects to obtain measurement results. In this verification, the obtained measurement results were evaluated using Standard Deviation of Error (SDE). Referring to Figure 12, the "embodiment" showed higher accuracy than the "oscillometric method" for both diastolic blood pressure (DBP) and systolic blood pressure (SBP).

[0045] <Effects of the embodiment> In this embodiment, the cuff pressure waveform and pulse waveform are each decomposed into multiple mode waveforms. The mode waveforms include waveforms that show global changes in time-series changes in the feature quantities related to the subject's blood pressure and waveforms that show local changes. Therefore, the blood pressure model 117 can uniformly learn both global and local features of the time-series changes in the feature quantities related to the subject's blood pressure. This blood pressure model 117 can output the subject's blood pressure with high accuracy even if the cuff pressure waveform or pulse waveform contains noise. In other words, this embodiment allows high-accuracy blood pressure measurement without performing calibration or parameter setting of the blood pressure measurement algorithm, even if the cuff pressure waveform or pulse waveform contains noise due to the position of the cuff 120 or the subject's posture.

[0046] <Modification> In the embodiment described above, both the cuff pressure waveform and the pulse waveform are input to the waveform decomposition unit 115, but either the cuff pressure waveform or the pulse waveform may be input to the waveform decomposition unit 115. In this case, the blood pressure model 117 may be constructed by using, as explanatory variables, the intermediate feature generated from the waveform used for blood pressure measurement, out of the intermediate feature generated from the cuff pressure waveform and the intermediate feature generated from the pulse waveform.

[0047] The embodiments and modifications disclosed above can be combined with each other.

[0048] <Computer-readable recording medium> An information processing program that causes a computer or other machine or device (hereinafter referred to as a computer, etc.) to realize any of the above functions can be recorded on a computer-readable recording medium. Then, by having the computer, etc. read and execute the program from this recording medium, the function can be provided.

[0049] Here, a computer-readable recording medium refers to a recording medium that stores information such as data and programs electrically, magnetically, optically, mechanically, or chemically and that can be read by a computer. Among such recording media, those that are removable from a computer include, for example, flexible disks, magneto-optical disks, Compact Disc Read Only Memory (CD-ROM), Compact Disc-Recordable (CD-R), Compact Disc-Rewriteable (CD-RW), Digital Versatile Disc (DVD), Blu-ray Disc (BD), Digital Audio Tape (DAT), 8mm tape, flash memory, external hard disk drives, and solid state drives (SSDs). Furthermore, recording media that are fixed to a computer include built-in hard disk drives, SSDs, ROMs, etc.

[0050] <Appendix 1> A cuff (120) to be wrapped around the subject's arm (H1); a detection unit (113) that detects the cuff pressure that the cuff (120) receives from the arm (H1); a decomposition unit (115) that acquires a first waveform (W11, W21) that indicates a time-series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and decomposes the first waveform (W11, W21) into a plurality of second waveforms (W12, W22) that indicate two or more different feature amounts; A learning model (117) constructed by machine learning using the second waveform (W12, W22) of the subject as an explanatory variable and blood pressure measured by auscultation as a target variable; a measurement unit (116) that inputs the second waveforms (W12, W22) into the learning model to measure the blood pressure of the subject; and an output unit (118) that outputs the blood pressure measured by the measurement unit. Sphygmomanometer (100). <Appendix 2> The first waveform (W11, W21) includes a cuff pressure waveform (W11) showing a time series change in the cuff pressure, The decomposition section (115) Decomposing the cuff pressure waveform (W11) into a plurality of third waveforms (W12) showing two or more different feature amounts; the learning model (117) is constructed using the third waveform (P14) for the subject (E1...EN) as the explanatory variable; The measurement unit (116) inputs the third waveform (W12) into the learning model (117) to measure the blood pressure of the subject. 10. The blood pressure monitor (100) of claim 1. <Appendix 3> the first waveform (W11, W21) includes a pulse waveform (W21) showing a time-series change in the pulse wave of the subject, which is acquired based on a time-series change in the cuff pressure; The decomposition section (115) The pulse waveform (W21) is decomposed into a plurality of fourth waveforms (W22) each showing two or more different characteristic amounts, the learning model (117) is constructed using the fourth waveform (W24) for the subject (E1...EN) as the explanatory variable; The measurement unit (116) inputs the fourth waveform (W22) into the learning model (117) to measure the blood pressure of the subject. 3. The blood pressure monitor (100) according to claim 1 or 2. <Appendix 4> the first waveforms (W11, W21) include a cuff pressure waveform (W11) showing a time series change in the cuff pressure, and a pulse waveform (W21) showing a time series change in the pulse wave of the subject obtained based on the time series change in the cuff pressure; The decomposition section (115) The cuff pressure waveform (W11) is decomposed into a plurality of third waveforms (W12) each showing two or more different characteristic amounts, and the pulse waveform (W21) is decomposed into a fourth waveform (W22) obtained from a time-series change in the cuff pressure; the learning model (117) is constructed using the third waveform (P14) and the fourth waveform (P24) for the subject (E1...EN) as the explanatory variables; the measurement unit (116) inputs the third waveform (W12) and the fourth waveform (W22) into the learning model (117) to measure the blood pressure of the subject; 4. A blood pressure monitor (100) according to any one of claims 1 to 3. <Appendix 5> The decomposition unit (115) includes Wave-U-Net. 5. A blood pressure monitor according to any one of appendices 1 to 4. <Appendix 6> a detecting step of detecting a cuff pressure received from the arm (H1) of the subject by a cuff (120) wrapped around the arm (H1); a decomposition step of acquiring a first waveform (W11, W21) showing a time-series change in a feature quantity related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform (W11, W21) into a plurality of second waveforms (W12, W22) showing two or more different feature quantities; a measuring step of measuring the blood pressure of a subject (E1...EN) by inputting a plurality of the second waveforms (W12, W22) into a learning model (117) constructed by machine learning using the second waveforms (W12, W22) of the subject (E1...EN) as explanatory variables and blood pressure measured by auscultation as a target variable; an output step of outputting the blood pressure measured in the measurement step, How to measure blood pressure. <Appendix 7> a detecting step of detecting a cuff pressure received from an arm (H1) of a subject (E1...EN) by a cuff (120) wrapped around the arm (H1); a decomposition step of acquiring a first waveform (W11, W21) showing a time-series change in a feature quantity related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform (W11, W21) into a plurality of second waveforms (W12, W22) showing two or more different feature quantities; a measuring step of inputting a plurality of the second waveforms (W12, W22) into a learning model (117) constructed by machine learning using the second waveforms (W12, W22) as explanatory variables and blood pressure measured by auscultation as a target variable, and measuring the blood pressure of the subject; an output step of outputting the blood pressure measured in the measurement step; Blood pressure measurement program. <Appendix 8> a detecting step of detecting a cuff pressure received from an arm (H1) of a subject (E1...EN) by a cuff (120) wrapped around the arm (H1); a decomposition step of acquiring a first waveform (W11, W21) showing a time-series change in a feature quantity related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform (W11, W21) into a plurality of second waveforms (W12, W22) showing two or more different feature quantities; a construction step of constructing a learning model (117) by machine learning using the second waveforms (W12, W22) as explanatory variables and the blood pressure measured by the auscultation method as a target variable, the computer executing How to build a learning model. <Appendix 9> a detecting step of detecting a cuff pressure received from the arm (H1) of the subject by a cuff (120) wrapped around the arm (H1); a decomposition step of acquiring a first waveform indicating a time-series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform (W11, W21) into a plurality of second waveforms (W12, W22) indicating two or more different feature amounts; a construction step of constructing a learning model (117) by machine learning using the second waveforms (W12, W22) as explanatory variables and the blood pressure measured by the auscultation method as a target variable, A program for building learning models. [Explanation of symbols]

[0051] 100··Sphygmomanometer 110··Main unit 111··Start switch 112 Control Unit 113 Pressure Sensor 114··Measurement section 115...Waveform decomposition section 116 Blood Pressure Measurement Unit 117··Blood Pressure Model 118··Output section 120··Cuff H1 Wrist

Claims

1. A cuff to be wrapped around the subject's arm; A detection unit that detects a cuff pressure received by the cuff from the arm; a decomposition unit that acquires a first waveform indicating a time series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and decomposes the first waveform into a plurality of second waveforms indicating two or more different feature amounts; A learning model constructed by machine learning using the second waveform of the subject as an explanatory variable and blood pressure measured by auscultation as a target variable; a measurement unit that inputs a plurality of the second waveforms into the learning model to measure the blood pressure of the subject; and an output unit that outputs the blood pressure measured by the measurement unit. Blood pressure monitor.

2. the first waveform includes a cuff pressure waveform indicating a time series change in the cuff pressure; The decomposition unit includes: decomposing the cuff pressure waveform into a plurality of third waveforms exhibiting two or more different characteristic quantities; The learning model is constructed using the third waveform for the subject as an explanatory variable; the measurement unit inputs the third waveform into the learning model to measure the blood pressure of the subject.

2. The blood pressure monitor according to claim 1.

3. the first waveform includes a pulse waveform indicating a time series change in a pulse wave of the subject, the time series change being obtained based on the time series change in the cuff pressure, The decomposition unit includes: The pulse waveform is decomposed into a plurality of fourth waveforms each showing two or more different characteristic amounts, The learning model is constructed using the fourth waveform for the subject as an explanatory variable; the measurement unit inputs the fourth waveform into the learning model to measure the blood pressure of the subject; 2. The blood pressure monitor according to claim 1.

4. the first waveform includes a cuff pressure waveform indicating a time series change in the cuff pressure, and a pulse waveform indicating a time series change in a pulse wave of the subject obtained based on the time series change in the cuff pressure, The decomposition unit includes: The cuff pressure waveform is decomposed into a plurality of third waveforms each showing two or more different characteristic amounts, and the pulse waveform is decomposed into a fourth waveform obtained from a time series change of the cuff pressure; The learning model is constructed using the third waveform and the fourth waveform for the subject as explanatory variables; the measurement unit inputs the third waveform and the fourth waveform into the learning model to measure the blood pressure of the subject.

2. The blood pressure monitor according to claim 1.

5. The decomposition unit includes Wave-U-Net, The blood pressure monitor according to any one of claims 1 to 4.

6. a detection step of detecting a cuff pressure received from a subject's arm by a cuff wrapped around the subject's arm; A decomposition process is performed to obtain a first waveform indicating a time series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and to decompose the first waveform into a plurality of second waveforms indicating two or more different feature amounts. Tep and a measuring step of inputting a plurality of the second waveforms into a learning model constructed by machine learning using the second waveform of the subject as an explanatory variable and blood pressure measured by the auscultation method as a target variable, and measuring the blood pressure of the subject; and an output step of outputting the blood pressure measured in the measurement step, How to measure blood pressure.

7. a detection step of detecting a cuff pressure received from a subject's arm by a cuff wrapped around the subject's arm; a decomposition step of acquiring a first waveform indicating a time series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform into a plurality of second waveforms indicating two or more different feature amounts; a measuring step of inputting a plurality of the second waveforms into a learning model constructed by machine learning using the second waveform as an explanatory variable and a blood pressure measured by an auscultation method as a target variable, and measuring the blood pressure of the subject; and an output step of outputting the blood pressure measured in the measurement step. Blood pressure measurement program.

8. a detection step of detecting a cuff pressure received from a subject's arm by a cuff wrapped around the subject's arm; a decomposition step of acquiring a first waveform indicating a time series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform into a plurality of second waveforms indicating two or more different feature amounts; A construction step of constructing a learning model by machine learning using the second waveform as an explanatory variable and blood pressure measured by the auscultation method as a target variable, the computer executing the construction step. How to build a learning model.

9. a detection step of detecting a cuff pressure received from a subject's arm by a cuff wrapped around the subject's arm; a decomposition step of acquiring a first waveform indicating a time series change in a feature amount related to the blood pressure of the subject based on the cuff pressure, and decomposing the first waveform into a plurality of second waveforms indicating two or more different feature amounts; A construction step of constructing a learning model by machine learning using the second waveform as an explanatory variable and blood pressure measured by the auscultation method as a target variable. A program for building learning models.

Citation Information

Cited By

  • Blood pressure measurement device, blood pressure measurement method, program

    JP2026147665A

  • Blood pressure measurement device, blood pressure measurement method, and program

    WO2026186324A1