JP2012045304A - Blood pressure estimation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a blood pressure estimation device which can perform highly precise blood pressure estimation.SOLUTION: A digital computing unit 16 uses a pulse wave signal and an electrocardiogram signal to estimate blood pressure. Further, the quality value q1 of the pulse wave signal and the quality value q2 of the electrocardiogram signal are computed to determine whether a sensor signal is good or not based on the quality values q1, q2. Moreover, the correlation value of noises of the respective sensor signals is calculated from a cross-correlation function, and whether the noises of the pulse wave signal are living body-originated or contact-originated is determined from the noise correlation value. Then, when all of the sensor signals are good and when all of the sensor signals are bad and the correlation value of noise is higher than a predetermined value, a blood pressure quality flag is put up. When the blood pressure quality flag is put up, a blood pressure value estimated by a blood pressure estimation logic 24 is stored. When a predetermined number of times of estimated blood pressure values are received, the average value of stored blood pressure values is calculated.

Description

  The present invention relates to a blood pressure estimation device that estimates blood pressure based on an electrocardiogram signal and a pulse wave signal.

  Conventionally, blood pressure is generally measured using a cuff. However, there are problems such as giving a sense of pressure to the person to be measured by tightening with the cuff, and inability to continuously measure blood pressure.

  On the other hand, in recent years, many methods for estimating blood pressure using an electrocardiogram signal or a pulse wave signal (electrocardiogram waveform, pulse wave waveform) have been proposed. For example, if blood pressure is measured using volume pulse waves, it can be measured using light, so the device can be downsized, the pain of the subject can be eliminated, and continuous blood pressure measurement is possible Become.

  Furthermore, many techniques for measuring blood pressure using both an electrocardiogram signal and a pulse wave signal have been proposed (see Patent Documents 1 and 2). By using both an electrocardiogram signal and a pulse wave signal, blood pressure can be measured with high accuracy.

  By the way, in order to accurately measure blood pressure based on an electrocardiogram signal or a pulse wave signal, it is necessary to measure those signals with high accuracy. In order to solve such a problem, it has been proposed to evaluate the presence of variation (noise) in the feature amount of each signal used for blood pressure estimation. For example, one that uses the peak of the first-order differential pulse wave (see Patent Document 3), one that uses variations in feature quantities for five beats (see Patent Document 4), and uses variation based on the difference between heart rate and pulse rate A thing (refer patent document 5) etc. are proposed conventionally.

JP 2008-279185 A Japanese Patent Laid-Open No. 08-140948 JP 2001-61795 A JP 10-137202 A Japanese Patent Laid-Open No. 11-47104

  In order to monitor the health and physical condition of the human body, it is necessary to sense all the information generated in the living body. Since biological signals such as pulse waves have unique shapes and frequency characteristics, it is generally easy to distinguish them from exogenous noise due to poor contact. However, depending on conditions, the biological signal may exhibit very similar properties to those extrinsic noises (= exogenous noise caused by contact). For example, a pulse wave signal at the time of heart failure or arrhythmia is a representative example when it is difficult to distinguish from exogenous noise. Until now, it was possible to cope only by removing both the extrinsic noise and the endogenous noise-like signal caused by the living body by the noise removal processing. As a result, important pulse wave information related to the disease was missed. Under such a background, there has been a strong demand for capturing external noise caused by a contact, but removing a noise-like signal caused by a living body without removing it.

  However, in the conventional method, since the noise-like signal caused by the living body cannot be distinguished from the extrinsic noise caused by the contact, (1) the two noises are removed together, or (2) the two noises are not removed together. I could only choose either to use it. As a result, there has been a problem that blood pressure estimation accuracy decreases. This tendency was particularly noticeable in subjects with heart disease.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to provide a blood pressure estimation device that can perform highly accurate blood pressure estimation.

  The invention according to claim 1, which has been made to solve the above-described problem, relates to a blood pressure estimation device that estimates blood pressure based on a biological signal (an electrocardiogram signal and a pulse wave signal). The blood pressure estimation device includes a signal acquisition unit, a noise determination unit, and a blood pressure estimation unit.

  The signal acquisition means acquires an electrocardiogram signal and a pulse wave signal detected from the living body. The noise determination unit determines whether or not the ratio of the magnitude of noise in the entire signal is greater than or equal to a predetermined value for each of the electrocardiogram signal and the pulse wave signal acquired by the signal acquisition unit. In the following description, a signal includes noise means that the ratio of noise in the entire signal is a predetermined value or more.

  The blood pressure estimation means estimates blood pressure using the biological signals when the electrocardiogram signal and the pulse wave signal acquired by the signal acquisition means satisfy a predetermined condition. The predetermined condition means that the noise determination means determines that the ratio of the magnitude of noise of both the electrocardiogram signal and the pulse wave signal is not equal to or greater than a predetermined value, or the electrocardiogram signal and the pulse wave signal. In either case, it is determined that the ratio of the magnitude of noise is greater than or equal to a predetermined value.

  The electrocardiogram signal and the pulse wave signal used for blood pressure estimation by the blood pressure estimation unit are acquired by the signal acquisition unit in a period corresponding to the electrocardiogram signal generated by the signal acquisition unit in a predetermined period. Pulse signal. The period for acquiring the pulse wave signal may be the same as the period for acquiring the electrocardiogram signal, or is delayed by a certain time from the electrocardiogram signal in consideration of the delay of the pulse wave signal with respect to the electrocardiogram signal. It is good also as a period.

  The signal acquisition means described above acquires a biological signal detected by sensors such as a heart rate sensor and a pulse wave sensor. If noise is superimposed on this biomedical signal, both the electrocardiographic signal and the pulse wave signal will affect both the electrocardiogram signal and the pulse wave signal if the noise is a noise-like signal originating from the living body. Noise appears on the screen.

  On the other hand, the noise caused by contact (measurement state between the sensor and the human body) is caused by the sensor signal (hereinafter referred to as the sensor signal on the side having a problem in the measurement state) of the heartbeat sensor that acquires the electrocardiogram signal and the pulse wave sensor that acquires the pulse wave signal In addition, noise appears only in a signal acquired by a sensor (also referred to as a sensor signal). From this, when large noise is included only in either one of the electrocardiogram signal and the pulse wave signal, it can be determined that the signal is not a noise signal caused by the living body.

  Therefore, in the blood pressure estimation device having the above-described configuration, when it can be determined that any of the electrocardiogram signal and the pulse wave signal includes extrinsic noise caused by contact, blood pressure estimation based on those sensor signals is not performed. That is, since blood pressure estimation can be performed based on a sensor signal that is highly likely not to include external noise caused by contact, blood pressure estimation can be executed with high accuracy.

  The invention according to claim 2, which has been made in order to solve the above-described problem, relates to a blood pressure estimation device that estimates blood pressure based on a biological signal. The blood pressure estimation apparatus includes a signal acquisition unit, a noise determination unit, a correlation calculation unit that calculates a correlation value of noise, and a blood pressure estimation unit.

  The signal acquisition means acquires an electrocardiogram signal and a pulse wave signal detected from the living body. The noise determination unit determines whether or not the ratio of the magnitude of noise in the entire signal is greater than or equal to a predetermined value for each of the electrocardiogram signal and the pulse wave signal acquired by the signal acquisition unit.

The correlation calculation means calculates a correlation value of noise between the electrocardiogram signal and the pulse wave signal.
The blood pressure estimation means estimates blood pressure using the biological signals when the electrocardiogram signal and the pulse wave signal acquired by the signal acquisition means satisfy a predetermined condition. The predetermined condition means that the noise determination means determines that the ratio of the magnitude of noise of both the electrocardiogram signal and the pulse wave signal is not equal to or greater than a predetermined value, or the electrocardiogram signal and the pulse wave signal. In either case, it is determined that the ratio of the magnitude of noise is greater than or equal to a predetermined value, and the correlation value calculated by the correlation calculation means is greater than or equal to a predetermined threshold.

  The electrocardiogram signal and the pulse wave signal used for blood pressure estimation by the blood pressure estimation unit are acquired by the signal acquisition unit in a period corresponding to the predetermined period and the electrocardiogram signal acquired by the signal acquisition unit in a predetermined period. Pulse signal.

  In the blood pressure estimation device configured as described above, in the same way as the blood pressure estimation device according to claim 1, when a large noise is included only in one of the electrocardiogram signal and the pulse wave signal, a noise state due to the living body is detected. It is prevented that it is determined that the signal is not a signal and the blood pressure is estimated based on such a sensor signal.

  Furthermore, when noise is included in both the electrocardiogram signal and the pulse wave signal, it is determined whether or not the signal is a noise-like signal caused by a living body by correlating those noises. The grounds are as follows. If the noise that appears in any sensor signal is a noise-like signal derived from a living body, the correlation between the noises becomes high. On the other hand, if any one or both of the sensor signals includes extrinsic noise caused by contact, the correlation between the noises becomes low. In other words, it is possible to determine whether or not the signal is a noise-like signal caused by a living body by taking a correlation of noise.

  Therefore, in the blood pressure estimation device having the above-described configuration, not only when one of the electrocardiogram signal and the pulse wave signal includes noise, but also when both sensor signals include large noise, an external factor caused by contact Since blood pressure estimation can be performed based on a sensor signal that is highly likely not to contain sexual noise, blood pressure estimation can be performed with high accuracy.

The invention described in claim 3 is the blood pressure estimation device according to claim 1 or 2, further having the following characteristics.
For each of the electrocardiogram signal and the pulse wave signal, the noise determination means compares the power of the component in the frequency band including the main part of the signal from the living body with the power of the component in a band other than the frequency band, When the ratio of the power in the frequency band is less than a predetermined value, it is determined that the ratio of the magnitude of noise to the signal is greater than or equal to a predetermined value for each of the electrocardiogram signal and the pulse wave signal.

  The frequency band including the main part of the biological signal can be determined experimentally and statistically optimally. Therefore, if the power of the component other than the frequency band is increased, it can be determined that the signal is noisy. Therefore, the blood pressure estimation apparatus having the above configuration can appropriately determine the noise level of the signal acquired by the sensor.

  Since most of the components of the electrocardiogram signal and pulse wave signal are concentrated below 20 to 30 Hz, the power of the component below it is compared with the power of the component below it with the vicinity of 30 Hz (for example, 40 Hz) as a threshold. By doing so, it is possible to appropriately determine the magnitude of the ratio of noise included in the signal.

Block diagram showing blood pressure estimation device Explanatory diagram showing the concept of blood pressure estimation Diagram showing bias due to power frequency of biological signal Map used for sensor signal pass / fail judgment Flowchart explaining the processing procedure of blood pressure estimation processing Flowchart explaining processing procedure of error estimation processing Explanatory drawing which shows the modification of the acquisition method of a quality value Explanatory drawing which shows an example of noise amount and pp

Embodiments of the present invention will be described below with reference to the drawings.
[Example]
(1) System Configuration of Blood Pressure Estimation Device As shown in FIG. 1, the blood pressure estimation device 1 of the present embodiment includes a pulse wave sensor 12, a heart rate sensor 14, a digital computing unit 16, and a display device 18. ing.

  The pulse wave sensor 12 is an optical sensor provided with a well-known light emitting element (LED) or light receiving element (PD). For example, the pulse wave sensor 12 irradiates light on a fingertip of a subject and uses the reflected wave to generate a pulse wave ( Volume pulse) can be detected. A pulse wave signal used for blood pressure estimation described later can be obtained from the pulse wave sensor 12. In addition, the said structure shows an example as a pulse wave sensor, and is not limited to the said structure.

  The heart rate sensor 14 is a general electrocardiograph having a plurality of electrodes used for obtaining an electrocardiographic signal. An electrocardiographic signal used for blood pressure estimation to be described later can be obtained from the heartbeat sensor 14. In addition, the said structure shows an example as a heart rate sensor, and is not limited to the said structure.

  The digital computing unit 16 is an electronic control unit centering on a well-known microcomputer comprising a CPU, ROM, RAM, I / O, and a bus line connecting these components, and is based on the pulse wave sensor 12 and the heart rate sensor 14. The blood pressure is estimated based on this signal, and the display device 18 is controlled.

The display device 18 is configured by a display such as a liquid crystal that displays information such as a blood pressure value estimated by the digital computing unit 16, a speaker that outputs the contents thereof by voice, and the like.
Next, the function of the digital arithmetic unit 16 of this embodiment will be described with reference to the block diagram of FIG. 1 and the explanatory diagram of FIG. The digital calculator 16 includes an AD converter 22, a blood pressure estimation logic 24, a signal quality calculation logic 26, a signal quality determination logic 28, a noise correlation calculation logic 30, a noise correlation determination logic 32, and a flag determination logic 34. And an average value output logic 36.

The AD converter 22 A / D converts the sensor signals acquired from the pulse wave sensor 12 and the heart rate sensor 14.
The acquired sensor signal has a raw waveform as shown in FIG. The symbol 100 shown on the waveform means that a lot of noise is included in the section (ΔT).

  The blood pressure estimation logic 24 is a logic that estimates blood pressure using a pulse wave signal and an electrocardiogram signal. There are various methods for estimating blood pressure using a pulse wave signal and an electrocardiogram signal. For example, the methods described in Japanese Patent Application Laid-Open Nos. 2008-279185 and 08-140948 are used. Can be considered.

The signal quality calculation logic 26 cuts out the pulse wave signal acquired by the pulse wave sensor 12 and the electrocardiogram signal acquired by the heart rate sensor 14 for a certain period ΔT (for example, 5000 ms), and the quality value q1 of the pulse wave signal in that period. And a quality value q2 of the electrocardiogram signal. The quality value here is a value determined by the ratio of the magnitude of noise in the entire sensor signal, and the pulse wave signal and the electrocardiogram signal are obtained by using a specific method for obtaining the quality values q1 and q2 in the case of the electrocardiogram signal. Explained as an example. As shown in FIG. 3, it is known that most of the QRS group which is the main part of the electrocardiographic waveform is concentrated at a frequency of 30 Hz or less. Table 1 shows the relationship between the biological signal and noise.

  As described above, the main frequency band in the biological signal is concentrated in a low range of 20 to 30 Hz or less. On the other hand, the extrinsic noise caused by contact and the endogenous noise-like signal caused by the living body appear in a wide band. In particular, it appears strongly in a high band of 40 Hz or higher. Utilizing the above properties, this logic calculates the ratio pr between the low frequency component power and the high frequency component power, which is the main frequency band of the sensor signal, as a signal quality value using 40 Hz as a threshold value.

  Specifically, the Fourier transform of the sensor signal X is obtained by the following equation (1), and the power ratio between the low frequency portion and the high frequency portion of the converted coefficient is calculated by the equation (2).

  Since high frequency component / low frequency component = noise power / signal power, the larger the noise, the smaller the obtained pr value, and the smaller the noise, the larger the obtained pr value. In this way, the quality values q1 and q2 are acquired.

The signal quality determination logic 28 determines the quality of the sensor signal based on the quality values q1 and q2 acquired by the signal quality calculation logic 26.
The specific quality determination is determined using the map shown in FIG. If the value of pr (that is, q1, q2) is equal to or greater than a predetermined value, it is determined that the sensor signal is good, and if the value of pr is less than the predetermined value, it is determined that the sensor signal is defective. As a result of the determination, FlagPULSE, which is a flag corresponding to the pulse wave signal, is set if the pulse wave signal is good, and FlagECG, which is a flag corresponding to the electrocardiogram signal, is set if the electrocardiogram signal is good. In FIG. 2, 1 is shown when the flag is raised, and 0 is shown when the flag is not raised.

  Here, Table 2 shows the relationship between the quality of each sensor signal and the condition caused by noise.

  In Table 2, “cause of contact” means that noise is generated depending on the measurement state of the sensor and the human body, that is, whether the sensor is in proper contact with the human body (or an error of the sensor itself). The cause means that an abnormal waveform that is not normally observed increases based on the human body itself such as a disease of the human body or an elderly person, and a signal that can be judged as noise is generated as compared with a normal waveform. Note that “◯” indicates that the noise is less than a predetermined ratio, and “X” indicates that the noise is greater than or equal to the predetermined ratio.

  As shown in Table 2, when only one of the pulse wave signal and the heart rate signal is defective, it can be said that there is an exogenous noise caused by contact in the output of the sensor on the defective side. Further, when both sensor signals are good, it can be said that the measurement is correctly performed without the external noise caused by the contact and the noise-like signal caused by the living body.

  Here, when both of the sensor signals are defective, when a noise-like signal caused by a living body is generated (when a patient is a sick person or an elderly person), both sensors cause exogenous noise caused by contact. A case can be considered. Therefore, in this case, it is determined by the noise correlation calculation logic 30 and the noise correlation determination logic 32 whether it is caused by contact or biological.

  The noise correlation calculation logic 30 is set when neither the FlagPULSE nor the FlagECG is set in the signal quality determination logic 28, that is, when a sensor signal having a lot of noise in both the pulse wave signal and the electrocardiogram signal is obtained. The correlation value of noise of each sensor signal is obtained by a cross correlation function.

An example of correlation value calculation will be described. Specifically, the similarity between the two waveforms of the signal X obtained from the pulse wave sensor 12 and the signal Y obtained from the heart rate sensor 14 is defined by the Pearson product moment coefficient r shown in the following equation (3). This r is used as a correlation value. In the equation (3), n is the data length of the signals X and Y (= the number of data), and the bars on the symbols X and Y mean to take an average value. The relationship between the two waveforms X and Y is determined as follows according to the value of r.
If r = 1, X = Y. If r = −1, X = −Y. If r = 0, there is no “waveform similarity”.

  The noise correlation determination logic 32 determines from the correlation value obtained by the noise correlation calculation logic 30 whether the pulse wave signal and the heartbeat signal, both of which are defective, are due to a living body or contact.

  In the case where both are noise-like signals due to living bodies, the noise of the sensor signals acquired from both sensors has the same cause of occurrence, so the correlation value of the noise in the two sensor signals is high (close to 1). Show. On the other hand, when the noises of the pulse wave signal and the heartbeat signal are both extrinsic noises caused by contact, the correlation between the noises of the two sensor signals is low (close to 0).

  Based on this, the noise correlation determination logic 32 determines that the noise signal is a biological signal when the correlation value of noise between the sensor signals acquired from both sensors is higher than a predetermined threshold. On the other hand, when the correlation value of the noise of the sensor signals acquired from both sensors is not higher than a predetermined value, it is determined that the external noise is caused by contact.

  The flag determination logic 34 determines a blood pressure quality flag from the flags (FlagPULSE, FlagECG) set by the signal quality determination logic 28 and the determination result by the noise correlation determination logic 32. The blood pressure quality flag is a flag indicating whether or not signals acquired from the pulse wave sensor 12 and the heart rate sensor 14 are appropriate for measuring blood pressure.

  Specifically, when FlagPULSE and FlagECG are both set in the signal quality determination logic 28, and when neither FlagPULSE nor FlagECG is set, the noise correlation determination logic 32 determines the noise correlation value. If it is determined that the blood pressure is higher than a predetermined level, a blood pressure quality flag is set.

  On the other hand, when only one of FlagPULSE and FlagECG is set in the signal quality determination logic 28, and when either FlagPULSE or FlagECG is not set, and the noise correlation determination logic 32 sets the noise correlation value to a predetermined value. If it is determined that it is not higher than this, the blood pressure quality flag is not raised.

In FIG. 2, 1 is set when the blood pressure quality flag is set, and 0 is set when the blood pressure quality flag is not set.
The average value output logic 36 stores the blood pressure value estimated by the blood pressure estimation logic 24 when the blood pressure quality flag is set. If the blood pressure quality flag is not set, the blood pressure value estimated at that time is discarded. Then, when an estimated value of the blood pressure value is received a predetermined number of times, an average value of the stored blood pressure values is calculated. In FIG. 2, five blood pressure quality flags are set, and all estimated blood pressure values when the blood pressure quality flag is set are added and divided by 5, thereby obtaining an average value.
(2) Processing by Blood Pressure Estimation Device Hereinafter, blood pressure estimation processing executed by the digital calculator 16 included in the blood pressure estimation device 1 will be described with reference to the flowchart shown in FIG. This process is executed, for example, when a start switch (not shown) is pressed in a state where the pulse wave sensor 12 and the heart rate sensor 14 of the blood pressure estimation device 1 are attached to the subject, or whenever a predetermined time elapses.

In this blood pressure estimation process, first, a pulse wave signal and an electrocardiogram signal detected by the pulse wave sensor 12 and the heart rate sensor 14 are cut out and acquired for a certain period (for example, 5000 ms) (S1).
Next, blood pressure estimation is executed based on the pulse wave signal and electrocardiogram signal acquired in S1 (S2). Here, blood pressure estimation is executed by the blood pressure estimation logic 24.

  Next, signal quality calculation is performed based on the pulse wave signal and the electrocardiogram signal acquired in S1 (S3). Here, the quality value q1 of the pulse wave signal and the quality value q2 of the electrocardiogram signal are calculated by the signal quality calculation logic 26.

  Next, the signal quality is determined (S4). Here, the quality of each sensor signal is determined by the signal quality determination logic 28 based on q1 and q2 calculated in S3, and FlagPULSE is set if the pulse wave signal is good, and FlagECG if the electrocardiogram signal is good. Stand up.

  Next, it is determined whether or not any flag is set in S4 (S5). If none of the flags are set (S5: YES), that is, if a sensor signal having a lot of noise in both the pulse wave signal and the electrocardiogram signal is obtained, the noise of those sensor signals is caused by the living body. The process proceeds to S6 in order to determine whether there is a contact cause or not.

On the other hand, when at least one of the flags is set (S5: NO), the process proceeds to S7.
Subsequently, a noise correlation calculation is performed (S6). Here, the noise correlation calculation logic 30 determines the correlation value of the noise of each sensor signal using a cross-correlation function.

  Next, blood pressure quality is determined (S7). Here, first, when both FlagPULSE and FlagECG flags are set in S4, the blood pressure quality flag is set assuming that each sensor signal is appropriate for estimating blood pressure. If only one of FlagPULSE and FlagECG is set, the blood pressure quality flag is not set.

  When both FlagPULSE and FlagECG flags are not set, based on the correlation value obtained in S6, if the correlation value is higher than a predetermined threshold value, a blood pressure quality flag is set and a predetermined threshold value is set. If it is less than, the blood pressure quality flag is not raised.

The process of S7 is realized using the noise correlation determination logic 32 and the flag determination logic 34.
Next, the blood pressure value estimated by the blood pressure estimation logic 24 is stored in the RAM of the digital calculator in association with the blood pressure quality flag set in S7 (S8).

Next, it is determined whether or not the processing loop of S1 to S8 has been executed a predetermined number of times (6 in this embodiment) or more (S9). If it is less than the predetermined number (5 or less) (S9: NO), the process returns to S1, and the processes of S1 to S8 are executed again. On the other hand, if the processing loop of S1 to S8 is equal to or greater than the predetermined number (S9: YES), the average of the blood pressure values stored in the RAM (the blood pressure values estimated based on the sensor signal when the blood pressure quality flag is set) The value is calculated (S10) and output to the display device 18 (S11). Thereafter, this process is terminated.
(3) Effects of the Invention In the blood pressure estimation device 1 of the present embodiment, neither the acquired pulse wave signal nor the electrocardiogram signal contains exogenous noise caused by contact (the ratio of the magnitude of noise to the entire signal is When it can be determined that it is not equal to or greater than a predetermined value, the blood pressure value is estimated using those sensor signals, so that blood pressure estimation can be performed with high accuracy.

In addition, when both sensor signals contain noise, only the sensor signal including extrinsic noise caused by contact is determined in order to determine whether it is caused by a living body or contact by the correlation value of noise of each sensor signal. The blood pressure can be estimated by appropriately excluding.
(4) Correspondence Relationship The processing of S1 executed by the digital computing unit 16 corresponds to the processing by the signal acquisition means of the present invention, the processing of S3 and 4 corresponds to the processing by the noise determination means, and the processing of S6 is correlated. This corresponds to the processing by the calculation means, and the processing for calculating the blood pressure value to be output to the display device 18 in S2, 7 to 10 corresponds to the processing by the blood pressure estimation means.
(5) Modifications While the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and can take various forms as long as they belong to the technical scope of the present invention. Needless to say.

  For example, in the above-described embodiment, the sensor signal quality values q1 and q2 are exemplified only for determining whether or not the blood pressure value estimated by the blood pressure estimation logic 24 is used. The error correction value of the blood pressure value estimated by the blood pressure estimation logic 24 may be derived using the value. A processing procedure of error estimation processing for calculating an error correction value will be described with reference to a flowchart shown in FIG. This process is executed in parallel with the blood pressure estimation process shown in FIG.

In this error estimation process, first, a pulse wave signal and an electrocardiogram signal are cut out and acquired from the pulse wave sensor 12 and the heart rate sensor 14 (S21).
Next, signal quality calculation is performed (S22). Here, the quality value q1 of the pulse wave signal and the quality value q2 of the electrocardiogram signal are calculated by the signal quality calculation logic 26.

Next, the absolute value Δq of the difference between the quality values q1 and q2 calculated in S22 is obtained (S23).
Next, noise correlation calculation is performed (S24). Here, the noise correlation calculation logic 30 determines the noise correlation value n of each sensor signal using a cross-correlation function.

Next, error estimation is performed (S25). Here, the error correction value ε is obtained by the following equation.
ε = (α1 × Δq) + (α2 × (1-n))
Here, α1 and α2 are weighting factors, and optimal values are determined experimentally and statistically. N is a correlation value obtained in S24, and takes a value of 0 ≦ n ≦ 1.

  That is, the error correction value ε is large when Δq is large, that is, when there is a difference in the ratio of the magnitude of noise between the pulse wave signal and the electrocardiogram signal, and when the correlation value of noise of each sensor signal is small. , A large value.

  Note that the estimated value of blood pressure using the error correction value ε is represented by X + ε, where X is the blood pressure value estimated by the blood pressure estimation logic 24. By calculating the error correction value ε in this manner, blood pressure can be estimated with higher accuracy.

  Moreover, in the said Example, in order to acquire quality value q1, q2, the structure which calculates ratio pr of the low frequency component power of a sensor signal and high frequency component power was illustrated. However, the quality values q1 and q2 may be acquired by a method other than this method as long as a value based on the ratio of the magnitude of noise in the entire sensor signal can be calculated as the quality value.

  For example, the standard deviation of the sensor signal may be used as the signal quality value. Since the standard deviation is an index representing signal variation, it can be determined that the noise is large when the variation is large, and the noise is small when the variation is small.

  As another method different from the method described above, there is a method for obtaining a signal quality value by using a ratio of amplitude values of signal waveforms. Specifically, the following relationship is used. The method will be described with reference to FIGS. Assume that a signal as shown in FIG. 7A is obtained from the pulse wave sensor 12 or the heart rate sensor 14. (1) First, the obtained signals are rearranged by height to obtain the signal sequence of FIG. (2) Next, the top N pieces of data regarding the height (for example, N = 3, point 6, point 5, point 2) are selected. (3) Then, an average value Av1 of the top N selected data and an average value Av2 of all data (point 1 to point 6) are obtained. (4) Finally, the ratio pp = Av1 / Av2 is obtained, and the pp is used as the signal quality value. For a signal with a lot of noise, pp becomes small, and for a signal with a little noise, pp takes a large value.

  The following equations (4) and (5) show the calculation formula when the peak of the biological signal is the R wave height of the electrocardiogram waveform. Table 3 shows the relationship between pp and noise level (clean / noisy) in that case.

  An example of the relationship between the amount of noise and pp is shown in FIGS. (A)-(C) show the signal (the number of data is 7000) for the electrocardiogram waveform for 7 seconds, and (D)-(F) show the result of rearranging it by the amplitude value. (A) and (D) show the case where the noise is zero, (B) and (E) show the case where noise is added thereto, and (C) and (F) show the case where a stronger noise is added. .

  The ratio Av1 ÷ Av2 between the average value Av1 of the top 100 data points obtained with the number of extractions N = 100 and the average value Av2 of all data (7000) was compared to the right. In order from the top, pp (D) = 5.7515, pp (E) = 4.6373, pp (F) = 3.8661. As described above, if the signal is clean, the value of pp increases, If it is noisy, the value of pp becomes small.

  Further, in the above embodiment, in the blood pressure estimation process shown in FIG. 5, after the blood pressure estimation by the blood pressure estimation logic 24 is performed, the quality of the signal based on the sensor signal quality values q1 and q2 is determined, and the estimated blood pressure value is determined. Although the configuration for determining whether or not to use is exemplified, the configuration for performing blood pressure estimation by the blood pressure estimation logic 24 after the determination of the quality of the sensor signal may be used, and the processing for estimating the blood pressure by the blood pressure estimation logic 24 is blood pressure estimation. The configuration may be executed as another process executed in parallel with the process.

DESCRIPTION OF SYMBOLS 1 ... Blood pressure estimation apparatus, 12 ... Pulse wave sensor, 14 ... Heart rate sensor, 16 ... Digital calculator, 18 ... Display apparatus, 22 ... AD converter, 24 ... Blood pressure estimation logic, 26 ... Signal quality calculation logic, 28 ... Signal quality Determination logic, 30 ... noise correlation calculation logic, 32 ... noise correlation determination logic, 34 ... flag determination logic, 36 ... average value output logic, 100 ... symbol

Claims (3)

Signal acquisition means for acquiring an electrocardiogram signal and a pulse wave signal detected from a living body;
For each of the electrocardiogram signal and the pulse wave signal acquired by the signal acquisition means, a noise determination means for determining whether a ratio of the magnitude of noise in the entire signal is a predetermined value or more,
Both the electrocardiogram signal acquired by the signal acquisition unit in a predetermined period by the noise determination unit and the pulse wave signal acquired by the signal acquisition unit in a period corresponding to the predetermined period. When it is determined that the noise magnitude ratio is not equal to or greater than a predetermined value, and when both the electrocardiogram signal and the pulse wave signal are determined to have a noise magnitude ratio equal to or greater than a predetermined value, A blood pressure estimation means for estimating blood pressure based on the electrocardiogram signal and the pulse wave signal. Signal acquisition means for acquiring an electrocardiogram signal and a pulse wave signal detected from a living body;
For each of the electrocardiogram signal and the pulse wave signal acquired by the signal acquisition means, a noise determination means for determining whether a ratio of the magnitude of noise in the entire signal is a predetermined value or more,
Correlation calculating means for calculating a correlation value of noise between the electrocardiogram signal and the pulse wave signal;
Both the electrocardiogram signal acquired by the signal acquisition unit in a predetermined period by the noise determination unit and the pulse wave signal acquired by the signal acquisition unit in a period corresponding to the predetermined period. When it is determined that the noise magnitude ratio is not equal to or greater than a predetermined value, and both the electrocardiogram signal and the pulse wave signal are determined to have a noise magnitude ratio equal to or greater than a predetermined value, and the correlation A blood pressure estimator that estimates blood pressure based on the electrocardiogram signal and the pulse wave signal when the correlation value calculated by the calculator is equal to or greater than a predetermined threshold value. . The noise determination means compares, for each of the electrocardiogram signal and the pulse wave signal, the power of the component in the frequency band including the main part of the signal from the living body and the power of the component in the band other than the frequency band. Then, when the ratio of the power in the frequency band is less than a predetermined value, for each of the electrocardiogram signal and the pulse wave signal, it is determined that the ratio of the magnitude of noise to the signal is greater than or equal to a predetermined value. The blood pressure estimation apparatus according to claim 1, wherein the blood pressure estimation apparatus is characterized.