WO2020228320A1 - 血压估计方法及装置 - Google Patents

血压估计方法及装置 Download PDF

Info

Publication number
WO2020228320A1
WO2020228320A1 PCT/CN2019/126258 CN2019126258W WO2020228320A1 WO 2020228320 A1 WO2020228320 A1 WO 2020228320A1 CN 2019126258 W CN2019126258 W CN 2019126258W WO 2020228320 A1 WO2020228320 A1 WO 2020228320A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
blood pressure
signal data
output
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2019/126258
Other languages
English (en)
French (fr)
Inventor
吴丹
张佳伦
马帅奇
李烨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Publication of WO2020228320A1 publication Critical patent/WO2020228320A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording for evaluating the cardiovascular system, e.g. pulse, heart rate, blood pressure or blood flow
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording for evaluating the cardiovascular system, e.g. pulse, heart rate, blood pressure or blood flow
    • A61B5/0205Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording for evaluating the cardiovascular system, e.g. pulse, heart rate, blood pressure or blood flow
    • A61B5/021Measuring pressure in heart or blood vessels
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/24Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316Modalities, i.e. specific diagnostic methods
    • A61B5/318Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/72Signal processing specially adapted for physiological signals or for diagnostic purposes
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/211Selection of the most significant subset of features
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting

Definitions

  • the present invention relates to the field of medical technology, in particular to a blood pressure estimation method and device.
  • blood pressure can reflect the physical health of the human body, and is also an important basis for evaluating the health of the heart and blood vessels. Therefore, timely attention to changes in blood pressure is related to people's health.
  • the purpose of the present invention is to provide a blood pressure estimation method and device to solve the problem of low blood pressure estimation accuracy in view of the above-mentioned shortcomings in the prior art.
  • an embodiment of the present invention provides a blood pressure estimation method, including: obtaining input parameters, where the input parameters include an encoded signal, a feature parameter, and multi-scale entropy;
  • the encoded signal is a signal in which the first signal data is encoded by a self-encoder; the first signal data includes: a first electrocardiographic signal and a first pulse wave signal; the first electrocardiographic signal is a user ECG signal, the first pulse wave signal is the pulse wave signal of the user; the characteristic parameter is the physiological signal index of the first signal data; the multi-scale entropy characterizes the complexity of the first signal data degree;
  • the input parameter is input into the blood pressure estimation model to obtain an estimated blood pressure; the estimated blood pressure has a corresponding relationship with the encoded signal, the characteristic parameter, and the multi-scale entropy respectively.
  • said acquiring input parameters includes:
  • the characteristic parameter is a physiological signal indicator of the sub-signal
  • the acquiring the multi-scale entropy according to the second output signal data includes:
  • the hyperparameter ⁇ is selected, and the second sequence is obtained through the second output signal data, and the second sequence satisfies the following formula:
  • x(i) is the i-th vector in the second output signal data, 1 ⁇ i ⁇ n, 1 ⁇ j ⁇ n/ ⁇ , 1 ⁇ 20;
  • z(i) [y(i),y(i+1),y(i+m-1)], 1 ⁇ i ⁇ n/ ⁇ -m+1;
  • z(j) [y( j),y(j+1),y(j+m-1)], 1 ⁇ j ⁇ n/ ⁇ -m+1, i ⁇ j;
  • h(i) [y(i),y(i+1),y(i+m-1),y(i+m)],
  • h(j) [y(j),y(j+1),y(j+m-1),y(j+m)], 1 ⁇ j ⁇ n/ ⁇ -m+1;
  • the multi-scale entropy According to the first average number B ( ⁇ ,m,r) and the second average number C ( ⁇ ,m,r) , the multi-scale entropy is obtained; the multi-scale entropy satisfies the following formula:
  • E(m) represents multi-scale entropy.
  • the characteristic parameters include any combination of one or more of the following: the time interval from the R peak of the electrocardiogram to the pulse wave valley point in the same cardiac cycle, and the ECG R peak to the maximum point of the pulse wave slope in the same cardiac cycle.
  • the time interval of the ECG R peak to the pulse wave peak point in the same cardiac cycle heart rate, reflectivity, systolic duration, systolic-diastolic duration ratio, diastolic duration ratio, rise time, end-systolic volume, end-diastolic volume, relative contraction Final volume, systolic-diastolic volume ratio.
  • the method further includes:
  • the second signal data includes: a second ECG signal and a second pulse wave signal;
  • the second ECG signal is the tester's ECG signal;
  • the pulse wave signal is the test Personnel’s pulse wave signal;
  • the first loss function represents the difference between the output parameter and the second signal data
  • the second signal data is input to the autoencoder to obtain the output parameter, which can be obtained by the following formula:
  • R represents the output parameter
  • the first loss function is obtained, and the first loss function can be obtained by the following formula:
  • L1 represents the first loss function
  • the inputting the input parameters into the blood pressure estimation model, before obtaining the estimated blood pressure further includes:
  • the blood pressure data being acquired at the same time as the second signal data, the blood pressure data including: systolic blood pressure sbp1 and diastolic blood pressure dbp1;
  • the blood pressure data and the output parameters are input into the blood pressure estimation model to obtain an output blood pressure;
  • the output blood pressure includes: output systolic blood pressure sbp2 and output diastolic blood pressure dbp2;
  • the second loss function represents the difference between the output blood pressure and the blood pressure data; the second loss function satisfies the following formula:
  • L2 ⁇ sbp1-sbp2 ⁇ 2 + ⁇ dbp1-dbp2 ⁇ 2 ;
  • L2 represents the second loss function
  • the present application also provides a blood pressure estimation device, including: an acquisition module and an estimation module;
  • the acquisition module is configured to acquire input parameters, and the input parameters include an encoded signal, a characteristic parameter, and multi-scale entropy; wherein the encoded signal is a signal in which first signal data is encoded by a self-encoder; the first The signal data includes: a first ECG signal and a first pulse wave signal; the first ECG signal is the user's ECG signal, the first pulse wave signal is the user's pulse wave signal; the characteristic parameter Is the physiological signal indicator of the first signal data; the multi-scale entropy represents the complexity of the first signal data;
  • the estimation module is configured to input the input parameter into a blood pressure estimation model to obtain an estimated blood pressure; the estimated blood pressure has a corresponding relationship with the encoded signal, the characteristic parameter, and the multi-scale entropy respectively.
  • the obtaining module is specifically used for:
  • the characteristic parameter is a physiological signal indicator of the sub-signal
  • the blood pressure estimation device further includes: a training module;
  • the acquisition module is also used for:
  • the second signal data includes: a second ECG signal and a second pulse wave signal;
  • the second ECG signal is the tester's ECG signal;
  • the pulse wave signal is the test Personnel’s pulse wave signal;
  • the training module is configured to obtain a first loss function according to the output parameter and the second signal data; the first loss function represents the difference between the output parameter and the second signal data; When the first loss function is less than a first preset value, it is determined that the autoencoder is successfully trained.
  • the blood pressure estimation method and device provided in this embodiment obtain the coded signal, feature parameters, and multi-scale entropy by obtaining the user's first signal data, and input the coded signal, feature parameters and multi-scale entropy as input to the blood pressure estimation Model. Since the estimated blood pressure in the blood pressure estimation model has a corresponding relationship with the coded signal, feature parameters and multi-scale entropy, the estimated blood pressure can be obtained by acquiring the user coded signal, feature parameters and multi-scale entropy according to the corresponding relationship. Using the user's first signal data to extract multiple types of data, and using multiple types of data to estimate the user's blood pressure can improve the accuracy of blood pressure estimation.
  • FIG. 1 is a schematic flowchart of a blood pressure estimation method provided by an embodiment of the application
  • FIG. 2 is a schematic flowchart of a blood pressure estimation method provided by another embodiment of this application.
  • FIG. 3 is a schematic flowchart of a method for obtaining multi-scale entropy according to an embodiment of the application
  • FIG. 4 is a schematic flowchart of a training method for an autoencoder provided by an embodiment of this application;
  • FIG. 5 is a schematic flowchart of a blood pressure estimation model training method provided by an embodiment of the application.
  • FIG. 6 is a schematic structural diagram of a blood pressure estimation device provided by an embodiment of the application.
  • FIG. 7 is a schematic structural diagram of a blood pressure estimation device provided by another embodiment of the application.
  • Fig. 1 is a schematic flowchart of a blood pressure estimation method provided by an embodiment of the application. As shown in Fig. 1, the method includes:
  • the encoded signal is a signal after the first signal data is encoded by a self-encoder.
  • the first signal data includes: a first electrocardiogram signal and a first pulse wave signal.
  • the first ECG signal is the user's ECG signal
  • the first pulse wave signal is the user's pulse wave signal.
  • the characteristic parameter is the physiological signal index of the first signal data. Multi-scale entropy characterizes the complexity of the first signal data.
  • the input parameters include three types of data: encoded signals, feature parameters, and multi-scale entropy.
  • the three types of data are all obtained through the user's first signal data. After the same signal data is extracted from different aspects, it is input into the blood pressure estimation model, which improves the accuracy of blood pressure estimation.
  • the estimated blood pressure has a corresponding relationship with the coded signal, feature parameters and multi-scale entropy respectively.
  • the blood pressure estimation method obtains the coded signal, feature parameters and multi-scale entropy by obtaining the user's first signal data, and inputs the coded signal, feature parameters and multi-scale entropy as input into the blood pressure estimation model . Since the estimated blood pressure in the blood pressure estimation model has a corresponding relationship with the coded signal, feature parameters and multi-scale entropy, the estimated blood pressure can be obtained by acquiring the user coded signal, feature parameters and multi-scale entropy according to the corresponding relationship. Using the user's first signal data to extract multiple types of data, and using multiple types of data to estimate the user's blood pressure can improve the accuracy of blood pressure estimation.
  • FIG. 2 is a schematic flowchart of a blood pressure estimation method according to another embodiment of the application.
  • S101 includes:
  • the acquired first signal data is filtered, and moving average filtering may be used to eliminate interference signals in the first signal data.
  • the first signal data is segmented according to the signal period T of the first signal wave.
  • the first ECG signal is segmented according to the heartbeat period;
  • the first pulse wave signal is segmented.
  • the first signal data is segmented, and the time length of the sub-signals that may be obtained is [0.2s, 0.4s], [ 0.4s, 0.6s], [0.6s, 0.8s], [0.8s, 1s], [1s, 1.2s].
  • alignment needs to be performed according to the interval in which the sub-signal is located, and optionally, alignment can be performed to the middle value of the two thresholds of the interval.
  • the sub-signal needs to be aligned to 0.3s. If the length of the sub-signal is within [0.2s, 0.3s], the sub-signal can be aligned to 0.3s; If the length of the sub-signal is within [0.3s, 0.4s], Ke uses the cutting method to align the sub-signal to 0.3s. The alignment of the remaining signals is similar to the sub-signals with a length interval of [0.2s, 0.4s].
  • the sub-signals with a length interval of [0.4s, 0.6s] are aligned to 0.5s.
  • the sub-signals with a length interval of [0.6s, 0.8s] are aligned to 0.7s.
  • the sub-signals with a length interval of [0.8s, 1s] are aligned to 0.9s.
  • the sub-signals with a length interval of [1s, 1.2s] are aligned to 1.1s.
  • the sub-signal includes the first ECG signal of one signal period T and the first pulse wave signal of one signal period T, so the size of the sub-signal is T*2.
  • S101-3 Use any sub-signal as an encoding input parameter of the self-encoder to obtain an encoded signal.
  • S101-4 Use the encoded signal as a decoding input parameter of the self-encoder to obtain first output signal data.
  • the sub-signal is encoded and decoded by the self-encoder, which realizes the denoising and compression of the sub-signal, which is convenient for extracting useful information.
  • S101-5 Acquire characteristic parameters according to the first output signal data.
  • the characteristic parameter is the physiological signal index of the sub-signal.
  • the characteristic parameters include any combination of one or more of the following: the time interval from the ECG R peak to the pulse wave valley point in the same cardiac cycle, the time interval from the ECG R peak to the pulse wave slope maximum point in the same cardiac cycle, the ECG R Time interval from peak to pulse wave peak point in the same cardiac cycle, heart rate, reflectivity, systolic duration, systolic-diastolic duration ratio, diastolic duration ratio, ascent time, end-systolic volume, end-diastolic volume, relative end-systolic volume, systolic-diastolic volume ratio.
  • the reflectivity satisfies the following formula:
  • R is the reflectivity
  • a is the amplitude of the pulse wave contraction peak in the sub-signal
  • b is the difference between a and the amplitude of the pulse wave diastolic peak in the sub-signal.
  • the contraction duration satisfies the following formula:
  • ST is the contraction duration
  • tn n is the time point of the pulse wave dicrotic notch in the sub-signal
  • tf n is the time point of the pulse wave trough in the sub-signal.
  • the systolic-diastolic duration ratio satisfies the following formula:
  • STR is the systolic-diastolic duration ratio
  • tf n+1 is the time point of the pulse trough value in the next sub-signal.
  • the diastolic duration ratio satisfies the following formula:
  • DTR is the diastolic duration ratio.
  • ST is the rise time
  • tp n is the time point of the pulse wave contraction peak in the sub-signal.
  • the end-systolic volume satisfies the following formula:
  • DV is the end-systolic volume
  • PPG is the pulse wave signal
  • RSV is the relative end-systolic volume.
  • the systolic-diastolic volume ratio satisfies the following formula:
  • SDVR is the systolic-diastolic volume ratio.
  • the first signal data is encoded and decoded by the self-encoder, and the second output signal data is data after denoising and dimensionality reduction of the first signal data.
  • FIG. 3 is a schematic flowchart of a method for obtaining multi-scale entropy according to an embodiment of this application.
  • S101-7 includes:
  • x(i) is the i-th vector in the second output signal data, 1 ⁇ i ⁇ n, 1 ⁇ j ⁇ n/ ⁇ . 1 ⁇ 20, ⁇ is an integer.
  • the first sequence is [1,2,3,4].
  • n 8. ⁇ takes 4
  • the second sequence obtained is [3/4,7/4], [11/4,15/4].
  • z(i) [y(i),y(i+1),y(i+m-1)], 1 ⁇ i ⁇ n/ ⁇ -m+1;
  • z(j) [y( j),y(j+1),y(j+m-1)], 1 ⁇ j ⁇ n/ ⁇ -m+1, i ⁇ j.
  • the standard deviation obtained by calculating the second output signal data is SD, and r may be 0.15*SD.
  • h(i) [y(i),y(i+1),y(i+m-1),y(i+m)], 1 ⁇ i ⁇ n/ ⁇ -m+1
  • h (j) [y(j),y(j+1),y(j+m-1),y(j+m)], 1 ⁇ j ⁇ n/ ⁇ -m+1.
  • E(m) represents multi-scale entropy.
  • Multi-scale entropy characterizes the complexity of the first signal data. The larger the value of multi-scale entropy, the greater the difference between the m+1-dimensional vector and the m-dimensional vector obtained by the first signal data, the less the data of the first signal data is repeated, and the more complex the first signal data; conversely, the multi-scale entropy The smaller the value of, the smaller the difference between the m+1-dimensional vector and the m-dimensional vector obtained by the first signal data, the more data the first signal data repeats, and the simpler the first signal data.
  • FIG. 4 is a schematic flowchart of an autoencoder training method provided by an embodiment of this application. As shown in FIG. 4, before S101 shown in FIG. 1, the method further includes:
  • the second signal data includes: a second ECG signal and a second pulse wave signal.
  • the second ECG signal is the tester's ECG signal.
  • the pulse wave signal is the pulse wave signal of the tester.
  • S202 Input the second signal data into the self-encoder to obtain output parameters.
  • the first loss function represents the difference between the output parameter and the second signal data.
  • an optional solution for training the autoencoder in Figure 4 is as follows, specifically:
  • the second signal data is the signal matrix S.
  • R represents the output parameter
  • the first loss function can be obtained by the following formula:
  • L1 represents the first loss function.
  • the coefficient in the self-encoder is adjusted according to the first loss function.
  • FIG. 5 is a schematic flowchart of a blood pressure estimation model training method provided by an embodiment of the application. As shown in FIG. 5, before S102, that is, before using the blood pressure estimation model, the training of the blood pressure estimation model is further included. for:
  • the blood pressure data includes: systolic blood pressure sbp1 and diastolic blood pressure dbp1.
  • S302 Input blood pressure data and output parameters into a blood pressure estimation model to obtain output blood pressure.
  • the output blood pressure includes: output systolic blood pressure sbp2 and output diastolic blood pressure dbp2.
  • the second signal data is input to the successfully trained autoencoder to obtain output parameters.
  • the acquired output parameters include: training coded signal, training feature parameter and training multi-scale entropy.
  • the encoded signal of the training is the signal after the second signal data is encoded by the trained autoencoder;
  • the characteristic parameter of the training is the physiological signal index of the second signal data;
  • the multi-scale entropy of the training represents the complexity of the second signal data .
  • the method of obtaining output parameters is similar to the method of obtaining input parameters in Figure 2.
  • the blood pressure estimation model may be a regressor, and a back propagation algorithm may be used to train the regressor used for blood pressure estimation.
  • S303 Obtain a second loss function according to the output blood pressure and blood pressure data.
  • the second loss function represents the difference between the output blood pressure and the blood pressure data.
  • the second loss function satisfies the following formula:
  • L2 ⁇ sbp1-sbp2 ⁇ 2 + ⁇ dbp1-dbp2 ⁇ 2 ;
  • L2 represents the second loss function.
  • the parameters of the blood pressure estimation model are adjusted according to the second loss function.
  • FIG. 6 is a schematic structural diagram of a blood pressure estimation device provided by an embodiment of this application.
  • This embodiment provides a blood pressure estimation device for executing the above-mentioned method embodiments.
  • the device Including: an acquisition module 401 and an estimation module 402.
  • the obtaining module 401 is used to obtain input parameters.
  • the input parameters include coded signals, characteristic parameters and multi-scale entropy; among them, the coded signal is the first signal data encoded by the self-encoder; the first signal data includes: the first electrocardiogram signal and the first pulse wave signal; An ECG signal is the user's ECG signal, and the first pulse wave signal is the user's pulse wave signal; the characteristic parameter is the physiological signal index of the first signal data; and the multi-scale entropy represents the complexity of the first signal data.
  • the estimation module 402 is used to input the input parameters into the blood pressure estimation model to obtain the estimated blood pressure; the estimated blood pressure respectively has a corresponding relationship with the encoded signal, the characteristic parameter and the multi-scale entropy.
  • the obtaining module 401 is specifically configured to: obtain the first signal data.
  • the first signal data is segmented to obtain multiple sub-signals. Use any sub-signal as the encoding input parameter of the self-encoder to obtain the encoded signal.
  • the encoded signal is used as the decoding input parameter of the self-encoder to obtain the first output signal data.
  • the characteristic parameter is obtained; the characteristic parameter is the physiological signal index of the sub-signal.
  • the first signal data is used as the codec input parameter of the self-encoder to obtain the second output signal data.
  • multi-scale entropy is obtained.
  • FIG. 7 is a schematic structural diagram of a blood pressure estimation device provided by another embodiment of this application.
  • the blood pressure estimation device further includes a training module 501.
  • the acquisition module 401 is also used to: acquire second signal data; the second signal data includes: a second ECG signal and a second pulse wave signal; the second ECG signal is the tester's ECG signal; the pulse wave signal is a test Person's pulse wave signal. Input the second signal data into the self-encoder to obtain output parameters.
  • the training module 501 is configured to obtain the first loss function according to the output parameters and the second signal data.
  • the first loss function represents the difference between the output parameter and the second signal data. When the first loss function is less than the first preset value, it is determined that the autoencoder training is successful.
  • the foregoing device is used to execute the method provided in the foregoing embodiment, and its implementation principle and technical effect are similar, and will not be repeated here.
  • the present invention also provides a program product, such as a computer-readable storage medium, including a program, which is used to execute the foregoing method embodiment when executed by a processor.
  • a program product such as a computer-readable storage medium, including a program, which is used to execute the foregoing method embodiment when executed by a processor.
  • the disclosed device and method may be implemented in other ways.
  • the device embodiments described above are merely illustrative, for example, the division of units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or integrated. To another system, or some features can be ignored, or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional units.
  • the above-mentioned integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium.
  • the above-mentioned software functional unit is stored in a storage medium and includes several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (English: processor) execute the method of each embodiment of the present invention Part of the steps.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (English: Read-Only Memory, abbreviated as: ROM), random access memory (English: Random Access Memory, abbreviated as: RAM), magnetic disk or optical disk, etc.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Cardiology (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Physiology (AREA)
  • Veterinary Medicine (AREA)
  • General Health & Medical Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Psychiatry (AREA)
  • Signal Processing (AREA)
  • Pulmonology (AREA)
  • Vascular Medicine (AREA)
  • Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)

Abstract

一种血压估计方法及装置,涉及医疗技术领域。装置包括:获取模块(401),用于获取输入参数,输入参数包含编码信号、特征参数和多尺度熵;其中,编码信号为第一信号数据通过自编码器编码后的信号;第一信号数据包括:第一心电信号和第一脉搏波信号;第一心电信号为用户的心电信号,第一脉搏波信号为用户的脉搏波信号;特征参数为第一信号数据的生理信号指标;多尺度熵表征第一信号数据的复杂程度;估计模块(402),用于将输入参数输入血压估计模型,获得估计血压;估计血压分别与编码信号、特征参数以及多尺度熵具有对应关系。将用户的第一信号数据进行多种类型数据的提取,将多种类型的数据均用于估计用户的血压,可提高血压估计的准确率。

Description

血压估计方法及装置 技术领域
本发明涉及医疗技术领域,具体而言,涉及一种血压估计方法及装置。
背景技术
血压作为人体重要的生理参数,能反映人体的生理健康状况,也是心脏和血管功能健康状况评估的重要依据。因此,及时关注血压的变化,关系着人们的身体健康。
在现有技术中能够,测量血压的方法众多。其中,较方便的一种方法是:采集脉搏波与血压值来训练模型,通过训练后的模型来估计血压值。
虽然,现有技术中的建模方法能够方便和快捷的估测人体血压,但其准确度有待提高。
发明内容
本发明的目的在于,针对上述现有技术中的不足,提供一种血压估计方法及装置,以解决血压估计准确率低的问题。
为实现上述目的,本发明实施例采用的技术方案如下:
第一方面,本发明实施例提供了一种血压估计方法,包括:获取输入参数,所述输入参数包含编码信号、特征参数和多尺度熵;
其中,所述编码信号为第一信号数据通过自编码器编码后的 信号;所述第一信号数据包括:第一心电信号和第一脉搏波信号;所述第一心电信号为用户的心电信号,所述第一脉搏波信号为所述用户的脉搏波信号;所述特征参数为所述第一信号数据的生理信号指标;所述多尺度熵表征所述第一信号数据的复杂程度;
将所述输入参数输入血压估计模型,获得估计血压;所述估计血压分别与所述编码信号、所述特征参数以及所述多尺度熵具有对应关系。
可选地,所述获取输入参数,包括:
获取所述第一信号数据;
将所述第一信号数据进行分段,获得多个子信号;
将任意一个所述子信号作为所述自编码器的编码输入参数,获得所述编码信号;
将所述编码信号作为所述自编码器的解码输入参数,获得第一输出信号数据;
根据所述第一输出信号数据,获取所述特征参数;其中所述特征参数为所述子信号的生理信号指标;
将所述第一信号数据作为所述自编码器的编解码输入参数,获得第二输出信号数据;
根据所述第二输出信号数据,获取所述多尺度熵。
可选地,所述根据所述第二输出信号数据,获取所述多尺度熵,包括:
获取所述第二输出信号数据,所述第二输出信号数据为第一序列x(n),n取值为正整数;
选取超参数τ,通过所述第二输出信号数据,得到第二序列,所述第二序列满足下式:
Figure PCTCN2019126258-appb-000001
其中,x(i)为第二输出信号数据中第i个向量,1≤i≤n,1≤j≤n/τ,1≤τ≤20;
将所述第二序列分为多个m维向量,所述多个m维向量为z(k),m取值为正整数,1≤k≤n/τ-m+1;
统计多个m维向量中,满足‖z(i)-z(j)‖ ≤r的向量的个数Bi,并计算第一平均个数B (τ,m,r),所述第一平均个数B (τ,m,r)满足下式:
Figure PCTCN2019126258-appb-000002
其中,z(i)=[y(i),y(i+1),y(i+m-1)],1≤i≤n/τ-m+1;z(j)=[y(j),y(j+1),y(j+m-1)],1≤j≤n/τ-m+1,i≠j;
将所述第二序列分为多个m+1维向量,所述多个m+1维向量为h(k);
统计多个m+1向量中,满足‖h(i)-h(j)‖ ≤r的向量的个数Ci,并计算第二平均个数C (τ,m,r),所述第二平均个数C (τ,m,r)满足下式:
Figure PCTCN2019126258-appb-000003
其中,h(i)=[y(i),y(i+1),y(i+m-1),y(i+m)],
1≤i≤n/τ-m+1,h(j)=[y(j),y(j+1),y(j+m-1),y(j+m)],1≤j≤n/τ-m+1;
根据所述第一平均个数B (τ,m,r)和所述第二平均个数C (τ,m,r),获得所述多尺度熵;所述多尺度熵满足下式:
Figure PCTCN2019126258-appb-000004
其中,E(m)表示多尺度熵。
可选地,所述特征参数包括下述一项或多项的任意组合:心电图R峰到同一心动周期内脉搏波谷值点的时间间隔、心电图R峰到同一心动周期内脉搏波斜率最大值点的时间间隔、心电图R峰到同一心动周期内脉搏波峰值点的时间间隔、心率、反射率、收缩时长、收缩舒张时长比、舒张时长比、上升时间、收缩末期容积、舒张末期容积、相对收缩末期容积、收缩舒张容积比。
可选地,所述获取输入参数之前,还包括:
获取第二信号数据;所述第二信号数据包括:第二心电信号和第二脉搏波信号;所述第二心电信号为测试人员的心电信号;所述脉搏波信号为所述测试人员的脉搏波信号;
将所述第二信号数据输入所述自编码器,获得输出参数;
根据所述输出参数与所述第二信号数据,获取第一损失函数;所述第一损失函数表征所述输出参数与所述第二信号数据的差值;
当所述第一损失函数小于第一预设值时,则确定所述自编码 器训练成功。
可选地,获取所述第二信号数据,所述第二信号数据为信号矩阵S;
将所述第二信号数据输入所述自编码器,获得所述输出参数,所述输出参数可通过下式获得:
R=decoder(encoder(S));
其中R表示所述输出参数;
根据所述输出参数与所述第二信号数据,获取所述第一损失函数,所述第一损失函数可通过下式获得:
L1=‖S-R‖ 2
其中,L1表示所述第一损失函数;
当所述第一损失函数小于第一预设值时,则确定所述自编码器训练成功。
可选地,所述将所述输入参数输入血压估计模型,获得估计血压之前,还包括:
获取血压数据,所述血压数据与所述第二信号数据同一时间获取,所述血压数据包括:收缩压sbp1和舒张压dbp1;
将所述血压数据与所述输出参数输入所述血压估计模型,获得输出血压;所述输出血压包括:输出收缩压sbp2和输出舒张压dbp2;
根据所述输出血压与所述血压数据,获取第二损失函数;所述第二损失函数表征所述输出血压与所述血压数据的差值;所述第二损失函数满足下式:
L2=‖sbp1-sbp2‖ 2+‖dbp1-dbp2‖ 2
其中,L2表示所述第二损失函数;
当所述第二损失函数小于第二预设值时,则确定所述血压估计模型训练成功。
另一方面,本申请还提供一种血压估计装置,包括:获取模块和估计模块;
所述获取模块,用于获取输入参数,所述输入参数包含编码信号、特征参数和多尺度熵;其中,所述编码信号为第一信号数据通过自编码器编码后的信号;所述第一信号数据包括:第一心电信号和第一脉搏波信号;所述第一心电信号为用户的心电信号,所述第一脉搏波信号为所述用户的脉搏波信号;所述特征参数为所述第一信号数据的生理信号指标;所述多尺度熵表征所述第一信号数据的复杂程度;
所述估计模块,用于将所述输入参数输入血压估计模型,获得估计血压;所述估计血压分别与所述编码信号、所述特征参数以及所述多尺度熵具有对应关系。
可选地,所述获取模块,具体用于:
获取所述第一信号数据;
将所述第一信号数据进行分段,获得多个子信号;
将任意一个所述子信号作为所述自编码器的编码输入参数,获得所述编码信号;
将所述编码信号作为所述自编码器的解码输入参数,获得第一输出信号数据;
根据所述第一输出信号数据,获取所述特征参数;其中所述特征参数为所述子信号的生理信号指标;
将所述第一信号数据作为所述自编码器的编解码输入参数,获得第二输出信号数据;
根据所述第二输出信号数据,获取所述多尺度熵。
可选地,所述血压估计装置还包括:训练模块;
所述获取模块,还用于:
获取第二信号数据;所述第二信号数据包括:第二心电信号和第二脉搏波信号;所述第二心电信号为测试人员的心电信号;所述脉搏波信号为所述测试人员的脉搏波信号;
将所述第二信号数据输入所述自编码器,获得输出参数;
所述训练模块,用于根据所述输出参数与所述第二信号数据,获取第一损失函数;所述第一损失函数表征所述输出参数与所述第二信号数据的差值;当所述第一损失函数小于第一预设值时,则确定所述自编码器训练成功。
本实施例中提供的血压估计方法及装置,通过获取用户的第一信号数据,获取编码信号、特征参数和多尺度熵,并将编码信号、特征参数和多尺度熵作为输入,输入到血压估计模型中。由于血压估计模型中估计血压均与编码信号、特征参数和多尺度熵具有对应关系,故通过对用户编码信号、特征参数和多尺度熵的获取,根据对应关系,可获得估计血压。将用户的第一信号数据进行多种类型数据的提取,将多种类型的数据均用于估计用户的血压,可提高血压估计的准确率。
附图说明
图1为本申请一实施例提供的血压估计方法流程示意图;
图2为本申请另一实施例提供的血压估计方法流程示意图;
图3为本申请一实施例提供的多尺度熵获取方法流程示意图;
图4为本申请一实施例提供的自编码器训练方法流程示意图;
图5为本申请一实施例提供的血压估计模型训练方法流程示意图;
图6为本申请一实施例提供的血压估计装置结构示意图;
图7为本申请另一实施例提供的血压估计装置结构示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅用以解释本申请,并不用于限定本申 请。
图1为本申请一实施例提供的血压估计方法流程示意图,如图1所示,该方法包括:
S101、获取输入参数,输入参数包含编码信号、特征参数和多尺度熵。
其中,编码信号为第一信号数据通过自编码器编码后的信号。第一信号数据包括:第一心电信号和第一脉搏波信号。第一心电信号为用户的心电信号,第一脉搏波信号为用户的脉搏波信号。特征参数为第一信号数据的生理信号指标。多尺度熵表征第一信号数据的复杂程度。
输入参数包含三种类型的数据:编码信号、特征参数和多尺度熵,三种类型的数据均是通过用户的第一信号数据获得。将同一个信号数据从不同的方面进行特征的提取后,输入血压估计模型,提高了血压估计的准确率。
S102、将输入参数输入血压估计模型,获得估计血压。估计血压分别与编码信号、特征参数以及多尺度熵具有对应关系。
本实施例中提供的血压估计方法,通过获取用户的第一信号数据,获取编码信号、特征参数和多尺度熵,并将编码信号、特征参数和多尺度熵作为输入,输入到血压估计模型中。由于血压估计模型中估计血压均与编码信号、特征参数和多尺度熵具有对应关系,故通过对用户编码信号、特征参数和多尺度熵的获取,根据对应关系,可获得估计血压。将用户的第一信号数据进行多种类型数据的提取,将多种类型的数据均用于估计用户的血压, 可提高血压估计的准确率。
可选地,图2为本申请另一实施例提供的血压估计方法流程示意图,如图2所示,S101,包括:
S101-1、获取第一信号数据。
可选地,对获取的第一信号数据进行滤波,可采用滑动平均滤波,以剔除第一信号数据中的干扰信号。
S101-2、将第一信号数据进行分段,获得多个子信号。
需要说明的是,对第一信号数据进行分段是按照第一信号波的信号周期T进行分段的,具体的,根据心跳周期,对第一心电信号进行分段;根据脉动周期,对第一脉搏波信号进行分段。
由于不同的用户或者同一个用户在不同的场景下的心跳周期或脉动周期不同,因此,对第一信号数据进行分段,可能获得的子信号的时间长度在[0.2s,0.4s],[0.4s,0.6s],[0.6s,0.8s],[0.8s,1s],[1s,1.2s]。对于不同长度的子信号,需要根据子信号所在的区间,进行对齐,可选地,可对齐至区间两阈值的中间值。
例如子信号的时间长度在[0.2s,0.4s]内,需要将子信号对齐至0.3s,若子信号的长度在[0.2s,0.3s]内,可采用插值的方式将该子信号对齐至0.3s;若子信号的长度在[0.3s,0.4s]内,科采用剪切的方式将该子信号对齐至0.3s。其余的信号对齐的方式与长度区间在[0.2s,0.4s]内的子信号类似。
具体的,长度区间在[0.4s,0.6s]内的子信号,对齐至0.5s。长度区间在[0.6s,0.8s]内的子信号,对齐至0.7s。长度区间在[0.8s,1s]内的子信号,对齐至0.9s。长度区间在[1s,1.2s]内的子信号,对齐 至1.1s。
由于第一信号数据包括:第一心电信号和第一脉搏波信号,子信号包括一个信号周期T的第一心电信号和一个信号周期T的第一脉搏波信号,因此子信号的大小为T*2。
S101-3、将任意一个子信号作为自编码器的编码输入参数,获得编码信号。
S101-4、将编码信号作为自编码器的解码输入参数,获得第一输出信号数据。
通过自编码器对子信号进行编码和解码,实现了对子信号的去噪和压缩,便于提取有用信息。
S101-5、根据第一输出信号数据,获取特征参数。其中特征参数为子信号的生理信号指标。
特征参数包括下述一项或多项的任意组合:心电图R峰到同一心动周期内脉搏波谷值点的时间间隔,心电图R峰到同一心动周期内脉搏波斜率最大值点的时间间隔,心电图R峰到同一心动周期内脉搏波峰值点的时间间隔,心率,反射率,收缩时长,收缩舒张时长比,舒张时长比,上升时间,收缩末期容积,舒张末期容积,相对收缩末期容积,收缩舒张容积比。
反射率满足下式:
Figure PCTCN2019126258-appb-000005
其中,R为反射率,a为子信号中脉搏波收缩峰值的幅值,b为a与子信号中脉搏波舒张峰的幅值之差。
收缩时长满足下式:
ST=tn n-tf n
其中,ST为收缩时长,tn n为子信号中脉搏波重搏切迹的时间点,tf n为子信号中脉搏波谷值的时间点。
收缩舒张时长比满足下式:
Figure PCTCN2019126258-appb-000006
其中,STR为收缩舒张时长比,tf n+1为下一子信号中脉搏波谷值的时间点。
舒张时长比满足下式:
Figure PCTCN2019126258-appb-000007
其中,DTR为舒张时长比。
上升时间满足下式:
ST=tp n-tf n
其中,ST为上升时间,tp n为子信号中脉搏波收缩峰的时间点。
收缩末期容积满足下式:
Figure PCTCN2019126258-appb-000008
其中,DV为收缩末期容积,PPG为脉搏波信号。
相对收缩末期容积满足下式:
Figure PCTCN2019126258-appb-000009
RSV为相对收缩末期容积。
收缩舒张容积比满足下式:
Figure PCTCN2019126258-appb-000010
SDVR为收缩舒张容积比。
S101-6、将第一信号数据作为自编码器的编解码输入参数,获 得第二输出信号数据。
通过自编码器对第一信号数据进行编码和解码,第二输出信号数据是对第一信号数据的去噪和降维后的数据。
S101-7、根据第二输出信号数据,获取多尺度熵。
可选地,图3为本申请一实施例提供的多尺度熵获取方法流程示意图,如图3所示,S101-7,包括:
S101-7a、获取第二输出信号数据,第二输出信号数据为第一序列x(n),n取值为正整数。
S101-7b、选取超参数τ,通过第二输出信号数据,得到第二序列,第二序列满足下式:
Figure PCTCN2019126258-appb-000011
其中,x(i)为第二输出信号数据中第i个向量,1≤i≤n,1≤j≤n/τ。1≤τ≤20,τ取整数。
例如,第一序列为[1,2,3,4]。其中n=8。τ取4,第二序列为y(1),y(2),y(3),y(4),n/τ=8/4=2,则y(1),y(2),y(3),y(4)均为两个向量。y(1)=3/4,y(2)=7/4,y(3)=11/4,y(4)=15/4。获得的第二序列为[3/4,7/4],[11/4,15/4]。
S101-7c、将第二序列分为k个m维向量,多个m维向量为z(k),m取值为正整数,
Figure PCTCN2019126258-appb-000012
可选地,以序列[3/4,7/4,11/4,15/4]为例,m取2时,
Figure PCTCN2019126258-appb-000013
将第二序列分为3个2维向量。对该第二序列逐次进行分割,分割后的两个二维向量为[3/4,7/4]、[7/4,11/4]和[11/4,15/4]。
S101-7d、统计k个m维向量中,满足‖z(i)-z(j)‖ ≤r的向量的个数Bi,并计算第一平均个数B (τ,m,r),第一平均个数B (τ,m,r)满足下式:
Figure PCTCN2019126258-appb-000014
其中,z(i)=[y(i),y(i+1),y(i+m-1)],1≤i≤n/τ-m+1;z(j)=[y(j),y(j+1),y(j+m-1)],1≤j≤n/τ-m+1,i≠j。
可选地,计算获得第二输出信号数据的标准差为SD,r可以取0.15*SD。
S101-7e、将第二序列分为多个m+1维向量,多个m+1维向量为h(k)。
S101-7f、统计多个m+1向量中,满足‖h(i)-h(j)‖ ≤r的向量的个数Ci,并计算第二平均个数C (τ,m,r),第二平均个数C (τ,m,r)满足下式:
Figure PCTCN2019126258-appb-000015
其中,h(i)=[y(i),y(i+1),y(i+m-1),y(i+m)],1≤i≤n/τ-m+1,h(j)=[y(j),y(j+1),y(j+m-1),y(j+m)],1≤j≤n/τ-m+1。
S101-7g、根据第一平均个数B (τ,m,r)和第二平均个数C (τ,m,r),获得多尺度熵。多尺度熵满足下式:
Figure PCTCN2019126258-appb-000016
其中,E(m)表示多尺度熵。多尺度熵表征第一信号数据的复杂度。多尺度熵的值越大,表示第一信号数据获取的m+1维向量和m维向量差别越大,第一信号数据重复的数据越少,第一信号数据越复杂;反之,多尺度熵的值越小,表示第一信号数据获取 的m+1维向量和m维向量差别越小,第一信号数据重复的数据越多,第一信号数据越简单。
可选地,图4为本申请一实施例提供的自编码器训练方法流程示意图,如图4所示,在图1所示的S101之前,还包括:
S201、获取第二信号数据。第二信号数据包括:第二心电信号和第二脉搏波信号。第二心电信号为测试人员的心电信号。脉搏波信号为测试人员的脉搏波信号。
S202、将第二信号数据输入自编码器,获得输出参数。
S203、根据输出参数与第二信号数据,获取第一损失函数。第一损失函数表征输出参数与第二信号数据的差值。
S204、当第一损失函数小于第一预设值时,则确定自编码器训练成功。
可选地,图4中训练自编码器的一种可选方案如下所示,具体为:
获取第二信号数据,第二信号数据为信号矩阵S。
将第二信号数据输入自编码器,获得输出参数,输出参数可通过下式获得:
R=decoder(encoder(S));
其中R表示输出参数。
根据输出参数与第二信号数据,获取第一损失函数,第一损失函数可通过下式获得:
L1=‖S-R‖ 2
其中,L1表示第一损失函数。
当第一损失函数小于第一预设值时,则确定自编码器训练成功。当第一损失函数大于第一预设值时,则根据第一损失函数调整自编码器内的系数。
可选地,图5为本申请一实施例提供的血压估计模型训练方法流程示意图,如图5所示,在S102之前,即在使用血压估计模型之前,还包括对血压估计模型的训练,具体为:
S301、获取血压数据,血压数据与第二信号数据同一时间获取,血压数据包括:收缩压sbp1和舒张压dbp1。
S302、将血压数据与输出参数输入血压估计模型,获得输出血压。输出血压包括:输出收缩压sbp2和输出舒张压dbp2。
在训练血压估计模型时,将第二信号数据输入训练成功的自编码器,获得输出参数。
获取的输出参数包括:训练的编码信号、训练的特征参数和训练的多尺度熵。其中,训练的编码信号为第二信号数据通过训练后的自编码器编码后的信号;训练的特征参数为第二信号数据的生理信号指标;训练的多尺度熵表征第二信号数据的复杂程度。获取输出参数的方法与图2中获取输入参数的方法类似。
可选地,血压估计模型可以为回归器,可利用反向传播算法对用于血压估计的回归器进行训练。
S303、根据输出血压与血压数据,获取第二损失函数。第二损失函数表征输出血压与血压数据的差值。第二损失函数满足下式:
L2=‖sbp1-sbp2‖ 2+‖dbp1-dbp2‖ 2
其中,L2表示第二损失函数。
S304、当第二损失函数小于第二预设值时,则确定血压估计模型训练成功。
当第二损失函数大于第二预设值时,则根据第二损失函数调整血压估计模型的参数。
可选地,图6为本申请一实施例提供的血压估计装置结构示意图,本实施例提供一种血压估计装置,用于执行上述方法类实施例,具体的,如图6所示,该装置包括:获取模块401和估计模块402。
获取模块401,用于获取输入参数。
输入参数包含编码信号、特征参数和多尺度熵;其中,编码信号为第一信号数据通过自编码器编码后的信号;第一信号数据包括:第一心电信号和第一脉搏波信号;第一心电信号为用户的心电信号,第一脉搏波信号为用户的脉搏波信号;特征参数为第一信号数据的生理信号指标;多尺度熵表征第一信号数据的复杂程度。
估计模块402,用于将输入参数输入血压估计模型,获得估计血压;估计血压分别与编码信号、特征参数以及多尺度熵具有对应关系。
可选地,获取模块401,具体用于:获取第一信号数据。将第一信号数据进行分段,获得多个子信号。将任意一个子信号作为自编码器的编码输入参数,获得编码信号。将编码信号作为自编码器的解码输入参数,获得第一输出信号数据。根据第一输出信 号数据,获取特征参数;其中特征参数为子信号的生理信号指标。将第一信号数据作为自编码器的编解码输入参数,获得第二输出信号数据。根据第二输出信号数据,获取多尺度熵。
可选地,在图6的基础上,增加用于模型训练的模块,具体的,图7为本申请另一实施例提供的血压估计装置结构示意图,血压估计装置还包括:训练模块501。
获取模块401,还用于:获取第二信号数据;第二信号数据包括:第二心电信号和第二脉搏波信号;第二心电信号为测试人员的心电信号;脉搏波信号为测试人员的脉搏波信号。将第二信号数据输入自编码器,获得输出参数。
训练模块501,用于根据输出参数与第二信号数据,获取第一损失函数。第一损失函数表征输出参数与第二信号数据的差值。当第一损失函数小于第一预设值时,则确定自编码器训练成功。
上述装置用于执行前述实施例提供的方法,其实现原理和技术效果类似,在此不再赘述。
可选地,本发明还提供一种程序产品,例如计算机可读存储介质,包括程序,该程序在被处理器执行时用于执行上述方法实施例。
在本发明所提供的几个实施例中,应该理解到,所揭露的装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,单元的划分,仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式,例如多个单元或组件可以结合或者可以集成到另一个系统,或一些特征可以忽略, 或不执行。另一点,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。
作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外,在本发明各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元,可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)或处理器(英文:processor)执行本发明各个实施例方法的部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(英文:Read-Only Memory,简称:ROM)、随机存取存储器(英文:Random Access Memory,简称:RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述仅是本发明的优选实施方式,应当指出,对于本技术领域的普通技术人员来说,在不脱离本发明原理的前提下,还可以做出若干改进和润饰,这些改进和润饰也应视为本发明的保 护范围。

Claims (10)

  1. 一种血压估计方法,其特征在于,包括:
    获取输入参数,所述输入参数包含编码信号、特征参数和多尺度熵;
    其中,所述编码信号为第一信号数据通过自编码器编码后的信号;所述第一信号数据包括:第一心电信号和第一脉搏波信号;所述第一心电信号为用户的心电信号,所述第一脉搏波信号为所述用户的脉搏波信号;所述特征参数为所述第一信号数据的生理信号指标;所述多尺度熵表征所述第一信号数据的复杂程度;
    将所述输入参数输入血压估计模型,获得估计血压;所述估计血压分别与所述编码信号、所述特征参数以及所述多尺度熵具有对应关系。
  2. 如权利要求1所述的血压估计方法,其特征在于,所述获取输入参数,包括:
    获取所述第一信号数据;
    将所述第一信号数据进行分段,获得多个子信号;
    将任意一个所述子信号作为所述自编码器的编码输入参数,获得所述编码信号;
    将所述编码信号作为所述自编码器的解码输入参数,获得第一输出信号数据;
    根据所述第一输出信号数据,获取所述特征参数;其中所述特 征参数为所述子信号的生理信号指标;
    将所述第一信号数据作为所述自编码器的编解码输入参数,获得第二输出信号数据;
    根据所述第二输出信号数据,获取所述多尺度熵。
  3. 如权利要求2所述的血压估计方法,其特征在于,所述根据所述第二输出信号数据,获取所述多尺度熵,包括:
    获取所述第二输出信号数据,所述第二输出信号数据为第一序列x(n),n取值为正整数;
    选取超参数τ,通过所述第二输出信号数据,得到第二序列,所述第二序列满足下式:
    Figure PCTCN2019126258-appb-100001
    其中,y(j)为所述第二序列,x(i)为第二输出信号数据中第i个向量,1≤i≤n,1≤j≤n/τ,1≤τ≤20;
    将所述第二序列分为多个m维向量,所述多个m维向量为z(k),m取值为正整数,1≤k≤n/τ-m+1;
    统计多个m维向量中,满足‖z(i)-z(j)‖ ≤r的向量的个数Bi,并计算第一平均个数B (τ,m,r),所述第一平均个数B (τ,m,r)满足下式:
    Figure PCTCN2019126258-appb-100002
    其中,z(i)=[y(i),y(i+1),y(i+m-1)],1≤i≤n/τ-m+1;
    z(j)=[y(j),y(j+1),y(j+m-1)],1≤j≤n/τ-m+1,i≠j;
    将所述第二序列分为多个m+1维向量,所述多个m+1维向量 为h(k);
    统计多个m+1向量中,满足‖h(i)-h(j)‖ ≤r的向量的个数Ci,并计算第二平均个数C (τ,m,r),所述第二平均个数C (τ,m,r)满足下式:
    Figure PCTCN2019126258-appb-100003
    其中,h(i)=[y(i),y(i+1),y(i+m-1),y(i+m)],1≤i≤n/τ-m+1,h(j)=[y(j),y(j+1),y(j+m-1),y(j+m)],1≤j≤n/τ-m+1;
    根据所述第一平均个数B (τ,m,r)和所述第二平均个数C (τ,m,r),获得所述多尺度熵;所述多尺度熵满足下式:
    Figure PCTCN2019126258-appb-100004
    其中,E(m)表示多尺度熵。
  4. 如权利要求2所述的血压估计方法,其特征在于,所述特征参数包括下述一项或多项的任意组合:心电图R峰到同一心动周期内脉搏波谷值点的时间间隔、心电图R峰到同一心动周期内脉搏波斜率最大值点的时间间隔、心电图R峰到同一心动周期内脉搏波峰值点的时间间隔、心率、反射率、收缩时长、收缩舒张时长比、舒张时长比、上升时间、收缩末期容积、舒张末期容积、相对收缩末期容积、收缩舒张容积比。
  5. 如权利要求1所述的血压估计方法,其特征在于,所述获取输入参数之前,还包括:
    获取第二信号数据;所述第二信号数据包括:第二心电信号和第二脉搏波信号;所述第二心电信号为测试人员的心电信号;所述 脉搏波信号为所述测试人员的脉搏波信号;
    将所述第二信号数据输入所述自编码器,获得输出参数;
    根据所述输出参数与所述第二信号数据,获取第一损失函数;所述第一损失函数表征所述输出参数与所述第二信号数据的差值;
    当所述第一损失函数小于第一预设值时,则确定所述自编码器训练成功。
  6. 如权利要求5所述的血压估计方法,其特征在于,获取所述第二信号数据,所述第二信号数据为信号矩阵S;
    将所述第二信号数据输入所述自编码器,获得所述输出参数,所述输出参数可通过下式获得:
    R=decoder(encoder(S));
    其中R表示所述输出参数;
    根据所述输出参数与所述第二信号数据,获取所述第一损失函数,所述第一损失函数可通过下式获得:
    L1=‖S-R‖ 2
    其中,L1表示所述第一损失函数;
    当所述第一损失函数小于第一预设值时,则确定所述自编码器训练成功。
  7. 如权利要求4所述的血压估计方法,其特征在于,所述将所述输入参数输入血压估计模型,获得估计血压之前,还包括:
    获取血压数据,所述血压数据与所述第二信号数据同一时间获 取,所述血压数据包括:收缩压sbp1和舒张压dbp1;
    将所述血压数据与所述输出参数输入所述血压估计模型,获得输出血压;所述输出血压包括:输出收缩压sbp2和输出舒张压dbp2;
    根据所述输出血压与所述血压数据,获取第二损失函数;所述第二损失函数表征所述输出血压与所述血压数据的差值;所述第二损失函数满足下式:
    L2=‖sbp1-sbp2‖ 2+‖dbp1-dbp2‖ 2
    其中,L2表示所述第二损失函数;
    当所述第二损失函数小于第二预设值时,则确定所述血压估计模型训练成功。
  8. 一种血压估计装置,其特征在于,包括:获取模块和估计模块;
    所述获取模块,用于获取输入参数,所述输入参数包含编码信号、特征参数和多尺度熵;其中,所述编码信号为第一信号数据通过自编码器编码后的信号;所述第一信号数据包括:第一心电信号和第一脉搏波信号;所述第一心电信号为用户的心电信号,所述第一脉搏波信号为所述用户的脉搏波信号;所述特征参数为所述第一信号数据的生理信号指标;所述多尺度熵表征所述第一信号数据的复杂程度;
    所述估计模块,用于将所述输入参数输入血压估计模型,获得估计血压;所述估计血压分别与所述编码信号、所述特征参数以及所述多尺度熵具有对应关系。
  9. 如权利要求8所述的血压估计装置,其特征在于,所述获取模块,具体用于:
    获取所述第一信号数据;
    将所述第一信号数据进行分段,获得多个子信号;
    将任意一个所述子信号作为所述自编码器的编码输入参数,获得所述编码信号;
    将所述编码信号作为所述自编码器的解码输入参数,获得第一输出信号数据;
    根据所述第一输出信号数据,获取所述特征参数;其中所述特征参数为所述子信号的生理信号指标;
    将所述第一信号数据作为所述自编码器的编解码输入参数,获得第二输出信号数据;
    根据所述第二输出信号数据,获取所述多尺度熵。
  10. 如权利要求8所述的血压估计装置,其特征在于,还包括:训练模块;
    所述获取模块,还用于:
    获取第二信号数据;所述第二信号数据包括:第二心电信号和第二脉搏波信号;所述第二心电信号为测试人员的心电信号;所述脉搏波信号为所述测试人员的脉搏波信号;
    将所述第二信号数据输入所述自编码器,获得输出参数;
    所述训练模块,用于根据所述输出参数与所述第二信号数据, 获取第一损失函数;所述第一损失函数表征所述输出参数与所述第二信号数据的差值;当所述第一损失函数小于第一预设值时,则确定所述自编码器训练成功。
PCT/CN2019/126258 2019-05-14 2019-12-18 血压估计方法及装置 Ceased WO2020228320A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910402224.8 2019-05-14
CN201910402224.8A CN110141201B (zh) 2019-05-14 2019-05-14 血压估计方法及装置

Publications (1)

Publication Number Publication Date
WO2020228320A1 true WO2020228320A1 (zh) 2020-11-19

Family

ID=67595353

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/126258 Ceased WO2020228320A1 (zh) 2019-05-14 2019-12-18 血压估计方法及装置

Country Status (2)

Country Link
CN (1) CN110141201B (zh)
WO (1) WO2020228320A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110141201B (zh) * 2019-05-14 2020-07-24 中国科学院深圳先进技术研究院 血压估计方法及装置
CN112382394A (zh) * 2020-11-05 2021-02-19 苏州麦迪斯顿医疗科技股份有限公司 事件处理方法、装置、电子设备及存储介质
CN112806977B (zh) * 2021-02-05 2022-03-18 复旦大学 基于多尺度融合网络的生理参数测量方法
CN116548935A (zh) * 2023-04-23 2023-08-08 华南理工大学 基于柔性有机光探测器和深度学习算法的血压测量系统
CN120531354A (zh) * 2024-02-23 2025-08-26 Oppo广东移动通信有限公司 血压估计方法及装置、电子设备、存储介质

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140236031A1 (en) * 2013-02-20 2014-08-21 Perminova Inc. Necklace-shaped physiological monitor
CN106889979A (zh) * 2016-12-30 2017-06-27 中国科学院电子学研究所 一种基于心电信号和血氧容积波的连续性血压测量方法
CN108523867A (zh) * 2018-03-28 2018-09-14 武汉麦咚健康科技有限公司 一种自校准ppg无创血压测量方法及系统
CN108720821A (zh) * 2018-05-28 2018-11-02 京东方科技集团股份有限公司 一种血压估计装置和血压估计系统
CN109310369A (zh) * 2016-06-10 2019-02-05 博能电子公司 用于估计血脉搏波特征的多传感器系统
US20190133534A1 (en) * 2016-05-11 2019-05-09 University of Maryland, Baltimore Couty Office of Technology Development Method and Apparatus for Predicting a Use for a Blood Transfusion
CN110141201A (zh) * 2019-05-14 2019-08-20 中国科学院深圳先进技术研究院 血压估计方法及装置

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10357164B2 (en) * 2014-04-24 2019-07-23 Ecole Polytechnique Federale De Lausanne (Epfl) Method and device for non-invasive blood pressure measurement
US11160461B2 (en) * 2015-06-12 2021-11-02 ChroniSense Medical Ltd. Blood pressure measurement using a wearable device
KR102486700B1 (ko) * 2015-08-11 2023-01-11 삼성전자주식회사 혈압 추정 방법 및 장치
CN105943005B (zh) * 2016-06-01 2019-08-06 合肥芯福传感器技术有限公司 基于光电绿光脉搏与心电图混合的无创血压检测装置
CN108294736A (zh) * 2017-01-12 2018-07-20 南开大学 连续血压测量系统及测量方法
US11123022B2 (en) * 2017-10-18 2021-09-21 Samsung Electronics Co., Ltd. Blood pressure estimating apparatus and blood pressure estimating method
CN107854123B (zh) * 2017-12-13 2021-02-02 清华大学深圳研究生院 一种无袖套连续血压监测方法和装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140236031A1 (en) * 2013-02-20 2014-08-21 Perminova Inc. Necklace-shaped physiological monitor
US20190133534A1 (en) * 2016-05-11 2019-05-09 University of Maryland, Baltimore Couty Office of Technology Development Method and Apparatus for Predicting a Use for a Blood Transfusion
CN109310369A (zh) * 2016-06-10 2019-02-05 博能电子公司 用于估计血脉搏波特征的多传感器系统
CN106889979A (zh) * 2016-12-30 2017-06-27 中国科学院电子学研究所 一种基于心电信号和血氧容积波的连续性血压测量方法
CN108523867A (zh) * 2018-03-28 2018-09-14 武汉麦咚健康科技有限公司 一种自校准ppg无创血压测量方法及系统
CN108720821A (zh) * 2018-05-28 2018-11-02 京东方科技集团股份有限公司 一种血压估计装置和血压估计系统
CN110141201A (zh) * 2019-05-14 2019-08-20 中国科学院深圳先进技术研究院 血压估计方法及装置

Also Published As

Publication number Publication date
CN110141201A (zh) 2019-08-20
CN110141201B (zh) 2020-07-24

Similar Documents

Publication Publication Date Title
CN110141201B (zh) 血压估计方法及装置
CN114376564A (zh) 一种基于心冲击信号的睡眠分期方法、系统、装置及介质
Kachuee et al. Cuff-less high-accuracy calibration-free blood pressure estimation using pulse transit time
Siddiqui et al. A pulse rate estimation algorithm using PPG and smartphone camera
CN111728602A (zh) 基于ppg的无接触血压测量装置
JP2023505111A (ja) 光学データからの生理学的測定のためのシステムおよび方法
CN116583216A (zh) 从光学数据测量血压的系统和方法
CN110755059B (zh) 一种血压波形监测方法及装置
CN115003215A (zh) 根据光学数据进行脉搏传播时间测量的系统和方法
Chen et al. Multi-view cross-fusion transformer based on kinetic features for non-invasive blood glucose measurement using PPG signal
CN111839488A (zh) 基于脉搏波的无创连续血压测量装置和方法
CN116509344B (zh) 一种非接触式呼吸频率监测方法、系统及设备
CN107928654A (zh) 一种基于神经网络的脉搏波信号血压检测方法
CN108364290A (zh) 对周期性生理活动的图像序列进行分析的方法、介质和系统
US20220287648A1 (en) Systems and Methods for Imputing Real-Time Physiological Signals
CN110558960A (zh) 一种基于ptt和miv-ga-svr的连续血压无创监测方法
CN113907727A (zh) 基于光电容积脉搏波描记的逐拍血压测量系统和方法
CN116012916A (zh) 远程光体积描记信号和心率检测模型构建方法及检测方法
Dey et al. Lifting wavelet transformation based blind watermarking technique of photoplethysmographic signals in wireless telecardiology
KR101016630B1 (ko) 심전도 신호 압축 방법 및 장치
CN117958777B (zh) 非侵入式血压检测系统、方法、电子设备和存储介质
CN116849635B (zh) 一种基于多通道相关图特征的脉搏疾病识别装置
WO2025175923A1 (zh) 血压估计方法及装置、电子设备、存储介质
CN116975594B (zh) 远程心冲击信号的特征提取方法和装置
Zhuang et al. Remote blood pressure measurement via spatiotemporal mapping of a short-time facial video

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19928669

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19928669

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 13.06.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19928669

Country of ref document: EP

Kind code of ref document: A1