JP4322302B1 - Biological signal analyzer - Google Patents

Abstract

An object of the present invention is to provide an accurate detection technique of a pulse wave rising point that is not affected by a reflected wave in a pulse wave in order to extract abundant biological information such as a time change of a heart rate from the pulse wave.
A feature point detection unit 102 outputs feature point data including a rising point and the time and level of a direct wave peak point. The feature point detection adaptive control unit 103 accepts this feature point data as an input, and inputs it to the next feature point prediction unit 1034, the storage unit 1031 and the prediction result determination unit 1033 in the feature point detection adaptive control unit. Up to two sets of feature point data are stored for sets of rising points and direct wave peak points. The next feature point prediction unit uses the current feature point data and the feature point data up to two cycles before on the pulse waveform read from the storage unit to predict the feature points of one cycle future on the pulse waveform. To do. The control information generation unit 1035 causes the feature point detection unit to perform the feature point detection operation only within the prediction range.
[Selection] Figure 1

Description

  The present invention relates to an analysis technique for extracting accurate biological information from a biological signal.

  With the recent increase in health awareness, there is an increasing need for systems that can easily measure health conditions anywhere. As a technology that meets this need, for example, as shown in FIG. 3, a system that detects a pulse wave from a fingertip F by a pulse wave sensor 10 and extracts various biological information from the pulse wave has been researched and developed. As one method of the pulse wave sensor, a light emitting unit 11 by an infrared LED (light emitting diode) is arranged on one side (upper side in the figure) in the fingertip housing part indicated by a broken line, and the lower side is shown on the opposite side with reference to the finger. For example, a light receiving unit 12 using an infrared sensor is arranged on the side), and the amount of transmitted infrared light is measured. As the blood volume (ratio of blood) of the fingertip inserted into the fingertip accommodating portion increases, the proportion of infrared light that is partially absorbed by the fingertip increases, and the amount of transmitted infrared light decreases. Using this principle, the blood volume at the fingertip can be measured as a change in the amount of transmitted infrared light.

  FIG. 4 shows a typical waveform example of the pulse wave measured by the pulse wave sensor. As shown in FIG. 4, the pulse wave measured by the pulse wave sensor of FIG. 4 is basically generated when the heart pumps out blood, so the pulse wave cycle substantially coincides with the heartbeat cycle. . The time change of the heart rate becomes basic data for extracting abundant biological information, for example, for use in estimating the state of the autonomic nerve. Therefore, the most basic and important biological information that can be extracted from the pulse wave is the heartbeat interval time. For the measurement, it is important to accurately detect the rising point of the pulse wave. The detected signal can be analyzed by the biological signal analyzer C to obtain biological information.

  Conventionally, a processing method using a zero-cross point of time differentiation of a pulse wave signal is often used for detecting the rising point of the pulse wave. This processing procedure will be described with reference to FIGS. In the typical example of the pulse waveform (horizontal time-vertical blood volume) shown in FIG. 4, the rising point is the minimum point of the signal within the repetition period of the pulse waveform. Since this point is also a local minimum point, it can be obtained as a point where the first derivative of time is zero and the second derivative is a positive value. The configuration of a conventional example that is realized by using this method will be described with reference to the block diagram of FIG.

First, the relative change in the blood volume at the fingertip is measured using the pulse wave sensor 3011 shown in FIG. 3 described above. 5 includes a pulse wave sensor 3011 and an A / D converter 3012 that converts a pulse wave signal output from the pulse wave sensor 3011 into a digital signal. The pulse wave signal from 3012 is output. A pulse wave signal (indicated by an arrow) output from the biological signal measurement unit 301 is input to the feature point detection unit 302. In the feature point detection unit 302, the time differentiation unit 1 (3021) obtains the first-order time differentiation, and then using this result (output), the zero cross detection unit 3022 takes a time when the value becomes zero. On the other hand, the time differentiation unit 2 (3023) performs differentiation again to obtain the second time differentiation, and both results are input to the rise detection unit 3024.
The rise detection unit 3024 detects from the input information of both the cases when the zero cross is detected by the zero cross detection unit 3022 and at the same time the case where the second derivative of the pulse wave is positive, and uses that time as the rise point time. Authorize.
With the above operation, the rising point detection unit 3024 outputs the time of the determined rising point, extracts the level of the pulse wave signal corresponding to the time using the input pulse wave signal, Output as the detection result of the rising point.

  The “method of detecting feature points on a pulse wave by zero crossing” is described in, for example, Patent Document 1 below.

JP-A-10-295657 (FIG. 5 etc.)

  From the typical waveform of the pulse wave shown in the figure, it can be seen that one cycle of the pulse wave, that is, the waveform between two adjacent rising points shown in FIG. 4 is bimodal. This waveform includes the first wave W1 in which the pressure wave of blood pumping directly from the heart propagates directly through the shortest path and the pressure wave that has once propagated from the heart to the branch portion of the peripheral blood vessel or artery. The second wave W2 is generated by being reflected and propagated to the measurement site. The first wave W1 is described as a direct wave, and the second wave W2 is described as a reflected wave.

  Since the intensity of the reflected wave has individual differences and changes with time, there are cases where the waveform within one period is unimodal and bimodal. As a result, erroneous detection may occur only with the conventional method using time-differential zero crossing as described above. An example in which an actual pulse wave measurement waveform is analyzed by a conventional method will be described with reference to FIG. 6A in FIG. 6 is a pulse wave waveform actually measured, and FIG. 6B is a time differential waveform in FIG. 6A. The horizontal axis is time.

  In FIG. 6B, the rising points detected by the conventional method are marked with white circles. As shown in FIG. 6 (B), the obtained result is correctly detected in the two periods from 0 second to about 2 seconds with respect to the pulse wave period of about 1 second, but from the subsequent about 2 seconds. It can be seen that the rise of the reflected wave is erroneously detected in the period up to about 3.5 seconds. If such a rising point detection error occurs, the extraction of biological information from the pulse wave using the heartbeat interval time as well as the heart rate will all result in an incorrect result, so it is extremely difficult to solve this problem. is important.

  The object of the present invention is to influence the reflected wave in the pulse wave in order to solve the problems clarified in the above examination and to extract abundant biological information such as the time change of the heart rate from the pulse wave. It is to provide an accurate detection technique of the pulse wave rising point that is not done.

  According to one aspect of the present invention, a biological signal measuring unit for measuring a biological time series signal and a feature for detecting a feature point using the biological time series signal measured by the biological signal measuring unit. In the biological signal analyzer including the point detection unit, one or both of the feature point time output from the feature point detection unit and the value on the biological time series signal corresponding to the feature point time And a feature point detection adaptive control unit that controls the feature point detection unit so as to limit the operation of the feature point detection unit to a range calculated using the feature point data. A biological signal analyzer is provided. By controlling the operation of the feature point detection unit to be limited to the range calculated using the feature point data, the next rising point is accurately predicted based on the current and past data, and the detection accuracy is improved. Improve.

  The feature point detection adaptive control unit is configured to store a feature point of a next period based on the storage unit that stores the input feature point data, the feature point data, and past feature point data extracted from the storage unit. Next feature point prediction unit for predicting the range of feature point data, and using the output from the next feature point prediction unit, the operation of the feature point detection unit is limited to the feature point data range of the next feature point. And a control information generation unit that outputs control information to be controlled. A line segment discrimination method can be used as a rising point detection method when searching for the next feature point prediction range.

  The feature point detection adaptive control unit compares the feature point data prediction range output from the next feature point prediction unit with the feature point data input from the feature point extraction unit, and the next feature point prediction unit A prediction result determination unit that determines whether or not a next feature point is detected within the prediction range of the feature point data output from the storage unit, using the output of the prediction result determination unit, An initialization control unit that initializes a set value in the next feature point prediction unit may be included. For example, the feature point data in the storage unit can be replaced with a default value set separately, and the set value in the next feature point prediction unit can be replaced with a default value set separately.

  The feature point detection unit detects whether or not a feature point is detected within a condition of a prediction range of feature point data included in the control information output from the feature point detection adaptive control unit. A determination unit, wherein the feature point detection adaptive control unit receives a determination result from the in-prediction range detection determination unit, and the determination result is within a condition of a prediction range of the feature point data included in the control information When a feature point is not detected, an initialization control unit that initializes the storage unit in the feature point detection adaptive control unit and a set value in the next feature point prediction unit may be provided.

  The next feature point prediction unit includes a heartbeat interval time prediction unit that predicts a heartbeat interval of the next period, and a reflection delay time that predicts a reflection delay time defined as a time interval between the direct wave and the reflected wave on the next pulse wave. Including at least one of a prediction unit, and an output of at least one of the heartbeat interval time prediction unit and the reflection delay time prediction unit, and a latest pulse wave rising point And a rising point time predicting unit that predicts the time range of the next pulse wave rising point using the time.

  The next feature point predicting unit includes at least one of an amplitude predicting unit that predicts an amplitude of a pulse wave waveform of the next period, a baseline fluctuation predicting unit that predicts a baseline fluctuation of the next period, the amplitude predicting unit, A rising point level prediction unit that predicts the level range of the next pulse wave rising point using the output of at least one of the baseline fluctuation prediction units and the latest pulse wave rising point level But it ’s okay.

The next feature point prediction unit includes a heartbeat interval time prediction unit that predicts a next heartbeat interval, and a reflection delay time prediction that predicts a reflection delay time defined as a time interval between a direct wave and a reflected wave on the next pulse wave. And at least one of the heartbeat interval time predicting unit and the reflection delay time predicting unit, and the latest pulse wave rising point time Then, a rising point time prediction unit that predicts the time range of the next pulse wave rising point,
An amplitude prediction unit that predicts the amplitude of the pulse wave waveform of the next period, a baseline fluctuation prediction unit that predicts the baseline fluctuation of the next period, the amplitude prediction unit, and the baseline fluctuation prediction unit, Is used to predict the level range of the next pulse wave rise point using the output of the latest pulse wave rise point, and the information calculated by the rise point time prediction unit. And a rising point level prediction unit.

  According to the present invention, even when the reflected wave is superimposed on the pulse wave, the next rising point can be accurately predicted based on the current and past data, and the detection accuracy can be improved. Can be prevented from being erroneously detected as a rising edge, and a correct rising point of the pulse wave waveform can be detected. Therefore, it is possible to extract various biological information including a time change of the heart rate from the pulse wave, which has been a conventional problem.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

  First, a first embodiment of the present invention will be described. FIG. 1 is a functional block diagram showing a configuration example of the biological signal analyzer 100 according to the present embodiment, and FIG. 2 is a block diagram showing an internal configuration example of the next feature point prediction unit 1034 in the configuration of FIG. It is.

  The biological signal analyzer 100 shown in FIG. 1 predicts the range of the next feature point based on outputs from the biological signal measurement unit 101, the feature point detection unit 102, and the feature point detection unit 102, and the feature point detection unit 102. And a feature point detection adaptive control unit 103 for controlling

  The feature point detection adaptive control unit 103 includes a storage unit 1031 that stores an input from the feature point detection unit 102, and a next feature point prediction unit that predicts a range of the next feature point from the input and data stored in the storage unit 1031. 1034, based on the output from the next feature point prediction unit 1034, the control information generation unit 1035 that generates a signal for controlling the feature point detection unit 102, the output from the next feature point prediction unit 1034, and the feature point detection unit 102 A prediction result determination unit 1033 that compares the output of the first and second prediction points to determine a prediction result, and initialization that initializes the storage unit 1031 and the next feature point prediction unit 1034 when the prediction result in the prediction result determination unit 1033 deviates. And a control unit 1032. The next feature point prediction unit 1034 will be described in detail with reference to FIG.

  As shown in FIG. 2, the next feature point prediction unit 1034 predicts a heartbeat interval time by using the output from the input feature point detection unit 102 and the output from the storage unit 1031. Prediction unit 10341, reflection delay time prediction unit 10342 that predicts the time interval between the direct wave and the reflected wave, amplitude prediction unit 10344 that predicts the amplitude of the pulse wave, and a relatively long time at the level of the rising point Baseline fluctuation prediction unit 10343 that predicts baseline fluctuations, and outputs of heartbeat interval time prediction unit 10341 and reflection delay time prediction unit 10342, and the latest rise point time are used to predict the rise point time Rise point time prediction unit 10345, amplitude prediction unit 10344, baseline fluctuation prediction unit 10343 outputs, latest rise point level, rise point time prediction unit 103 3 is configured having a rising point level prediction unit 10346 predicts the level of the rising using rising point time prediction information calculated point, a.

Hereinafter, a specific example is a case where the measurement target in the biological measurement unit 101 is a pulse wave, and the rising point of the pulse wave waveform and the direct wave peak point are used as the feature points detected by the feature point detection unit 102. The operation of the biological signal analyzer 100 according to the present embodiment will be described in detail.
For measuring the pulse wave, as described above, the fingertip pulse wave sensor 10 shown in FIG. 3 is often used. The fingertip pulse wave sensor 10 is generally composed of an infrared light emitting unit 11 and a light receiving unit 12. The principle of the pulse wave sensor 10 is based on a method of obtaining a pulse wave by utilizing the fact that the transmittance of infrared light changes according to the blood flow volume. Therefore, you may wear | wear to parts other than a fingertip, such as an earlobe. The pulse wave data is normally output as A / D converted digital data from the biological signal measuring unit 101 (FIG. 1), and the subsequent processing is performed. When simplification of the circuit or cost reduction is necessary, it can be replaced with signal processing using an analog circuit. A typical shape of the obtained pulse wave is shown in FIG.

  Next, in the configuration shown in FIG. 1, the feature point detection unit 102 receives pulse wave waveform data from the living body measurement unit 101, and in this embodiment, detects the rising point and the peak point of the direct wave, The set of time and level is output as feature point data.

  Here, the first important point in the embodiment of the present invention is that the feature point detection unit 102 is controlled by the feature point detection adaptive control unit 103 which is also an output destination of the feature point detection unit 102. The feature point detection is performed using the information of the prediction range of the next rising point. As a simple and effective embodiment, as described in the present embodiment, a feature point detection is performed only within the prediction range. However, if more precise detection is desired, a continuous reliability distribution that decreases from the center to the periphery of the prediction range is assumed, and the detection sensitivity is increased in areas with high reliability and areas with low reliability. Then, a method may be adopted in which detection sensitivity is lowered to achieve both malfunction prevention and sensitivity improvement.

  FIG. 7 is a diagram illustrating a calculation example of the prediction range of the next rising point provided by the feature point detection adaptive control unit 103, where the horizontal axis represents time and the vertical axis represents a signal. From the feature point detection adaptive control unit 103 to the feature point detection unit 102, the next feature point prediction range in FIG. 7 (in the region RG1 to RG4 surrounded by a square indicated by a white broken line in the figure). Then, the control is performed so as to search only the range indicated by the arrow. Therefore, it can be seen that the problem of erroneously detecting a rise in waveform due to a reflected wave as a rise, which is pointed out as a problem of the conventional method in the above-mentioned “Problems to be solved by the invention”, can be avoided.

  In this specification, the rising point detection method when the feature point detection unit 102 searches the next feature point prediction range is referred to as a line segment discriminant method. The procedure for detection by the line segment discriminant method will be described below.

(1) When the pulse wave time-series data output from the biological signal measurement unit is expressed as Di (i is an integer corresponding to time), data from the latest data to the past at regular intervals from there. Take out several points. In the present embodiment, five points of data are extracted and represented by the following variables.
DSj (j = 0 to 4, where j is larger represents the latest data.)

(2) In order to determine the rise, it is calculated whether DSj satisfies all the discriminants of the following equation.
DS1-DS0 <0
DSj−DSj−1> −a (j = 2 to 4, a is a small positive constant)
DS4-DS0> 0

(3) When all the discriminants in (2) above are satisfied, it is considered that there is a rising point between DS0 and DS2, and between the data corresponding to DS0 and DS2 from the original Di The minimum value is obtained and used as the rising point.

(4) If any of the discriminants in (2) above is not satisfied, new Di data is read, DSj is taken out again, and the processing in (2) is performed.

  According to this line segment discrimination method, the rising point is detected from the data of three or more points, so that it is possible to detect against noise.

  In addition, as a method different from the above, in the case of a signal with low noise, a method using a zero-cross of time differentiation adopted in the above-described conventional technique can be used as it is.

  Next, direct wave peak detection will be described. Immediately after the rising point is detected, the time derivative of the signal is theoretically positive, but after that, if you check the slope of the signal with respect to the time axis and detect the first negative change, the peak point will be can get. The actual algorithm is the same as the rising edge detection, and when detecting the peak, it is necessary to detect the falling edge. Therefore, by inverting the sign of a in the discriminant in the rising edge detection and the inequality sign, It is possible to easily obtain a line segment discrimination algorithm for detecting a peak. Further, when detecting a peak using a zero cross of time differentiation, a method of detecting a point at which the second-order differentiation becomes negative may be used.

  In the above description, the feature point detection adaptive control unit 103 controls the feature detection unit 102 as an example of the control method of the feature detection unit 102. However, of course, from the feature point detection adaptive control unit 103, Alternatively, a method may be used in which only the information on the next feature point prediction range is output, and the feature point detection unit 102 performs control so as to autonomously search only within the prediction range. As described above, the feature point detection unit receives only the information of the prediction range (not the control signal) and performs the detection operation only during this period. The configuration of the entire system is the same. Yes (the second embodiment below performs this processing). Of course, since the functions of the entire system are the same, the resulting detection range is the same.

  Next, the internal operation of the feature point detection adaptive control unit 103 will be described with reference to FIGS. 1 and 2. In the present embodiment, the feature point detection unit 102 outputs feature point data including the rising point and the time and level of the direct wave peak point.

  The “feature point data consisting of the time and level of the rising point and the direct wave peak point” is “feature point data consisting of the time and level of the rising point and the direct wave peak point”. is there. That is, there are two sets of two-dimensional vectors.

  The feature point detection adaptive control unit 103 accepts the feature point data as an input, and inputs the feature point data to the next feature point prediction unit 1034, the storage unit 1031, and the prediction result determination unit 1033 in the feature point detection adaptive control unit 103. To do.

The storage unit 1031 stores at least feature point data used by the next feature point prediction unit 1034 from the latest to the past. In the present embodiment, up to two sets of feature point data are stored for sets of rising points and direct wave peak points.
The next feature point prediction unit 1034 uses the current feature point data and the feature point data up to two cycles before on the pulse wave waveform read from the storage unit 1031 to determine the future of one cycle on the pulse wave waveform. Predict feature points. The configuration and operation will be described below.

  The next feature point prediction unit 1034 is configured as shown in FIG. 2, and the input data includes a heartbeat interval time prediction unit 10341, a reflection delay time prediction unit 10342, a baseline fluctuation prediction unit 10343, and an amplitude prediction. Are input to the unit 10344, and each unit outputs the next heartbeat interval time, the reflection delay time, the baseline fluctuation, and the predicted amplitude value. The current feature point data, the heartbeat interval time output from each unit and the reflection delay time are input to the rising point time predicting unit 10345, and the rising point time predicting unit 10345 generates the next rising point. A predicted time range is calculated and output to the control information generation unit. On the other hand, the current feature point data, baseline fluctuation, amplitude, and information calculated by the rising point time prediction unit 10345 are input to the rising point level prediction unit 10346, and the rising point level prediction unit 10346 has the next rising point. The level range predicted to be shown is calculated and output to the control information generator.

A method for predicting the range of the next feature point data by each of the above-described units will be sequentially described below.
First, a variable representing input data is defined.
R_time_0: Latest rise time
R_level_0: Latest rising point level
R_time_i: i is a positive integer, i period rise time in the past
R_level_i: i is a positive integer, i period rising point level in the past
T_time_0: Latest direct wave peak point time
T_level_0: Latest direct wave peak point level
T_time_i: i is a positive integer, i period past direct wave peak point time
T_level_i: i is a positive integer, direct wave peak point level in the past i cycles

(1) Heartbeat interval time prediction unit As shown in FIG. 4, the heartbeat interval time is an interval time between the rising points P1 and P2 of adjacent pulse wave periodic waveforms. Therefore, the latest heartbeat interval time B0 and the heartbeat interval time B1 one cycle before are calculated from the input data of the above variables by the following equation.
B0 = R_time_0-R_time_1
B1 = R_time_1-R_time_2

The simplest prediction method is a method in which the latest heartbeat interval time B0 is output as it is as a predicted value for the next heartbeat interval, and this method can be used when priority is given to simplification of calculation over prediction accuracy. it can. However, in the present embodiment, the next heartbeat interval time Bnext is predicted by the following equation, taking into account the trend of changes in the heartbeat interval time.
Bnext = B0 + KT * (B0-B1)

  However, KT takes a value from 0 to 1 and is a value representing the degree of considering the trend.

  Further, when accuracy is required, past data may be used more than in the present embodiment, and a predicted value may be calculated using a prediction formula, generally from N + 1 data up to N cycles in the past.

(2) Reflection delay time prediction unit The reflection delay time is the peak point time interval between the direct wave and the reflected wave shown in FIG.
In this example, the predicted value Lnext of the reflection delay time is constant based on an average value of physiological measurement 0.267 [seconds].
Lnext = 0.267 [seconds]

  However, when it is allowed to make the system complicated, the feature point detection unit 102 detects the reflected wave peak point, and the reflection delay time prediction unit 10342 actually calculates the reflection delay time. It is also possible to perform prediction with higher accuracy.

(3) Rise Point Time Prediction Unit The rise point time prediction unit predicts the range of the next rise point time by the following method.
First, the center R_time_next_center of the prediction range is obtained by adding the predicted value Bnext of the next heartbeat interval time obtained by the heartbeat interval time prediction unit (1) to the latest rising point time R_time_0.
R_time_next_center = R_time_0 + Bnext

Next, the lower limit of the prediction range is obtained. The lower limit of the prediction range is the midpoint between the predicted time of the reflected wave peak point and the predicted time of the next rising point, thereby reducing the impact on false detection of the reflected wave and the detection sensitivity of the rising point Can be made compatible. Therefore, in the present embodiment, this midpoint is calculated by the following formula and is set as the lower limit R_time_next_under of the prediction range of the next rising point time.
R_time_next_under = ((T_time_0 + Lnext) + R_time_next_center) / 2

On the other hand, the upper limit R_time_next_upper of the prediction range was calculated by the following equation so that the center of the prediction range coincided with the center point.
R_time_next_upper = 2 × R_time_next_center-R_time_next_under

(4) Baseline fluctuation prediction unit In this specification, the baseline fluctuation is defined as “a component having a lower frequency than the pulse wave superimposed on the pulse wave”. Because of this component, for example, the entire pulse waveform may be a waveform having a trend in an increasing direction, such as after 2 seconds in FIG. FIG. 8 is a diagram for enlarging the portion after 2 seconds in FIG. 7 and using it as an example of a typical pulse waveform for explaining the calculation method. In the pulse waveform shown in FIG. 8, the solid line represents data that has already been measured, and the broken line represents data that will be measured in the future. In FIG. 8, it is assumed that the latest rising point detection result is point A, and the point one cycle before is point B. Assuming that the baseline variation is a linear function of time, the baseline is defined as a straight line AB connecting point A and point B. Then, the baseline slope BLA is obtained by the following equation.
BLA = (R_level_0−R_level_1) / (R_time_0−R_time_1)

In the present embodiment, the obtained BLA is predicted to be maintained in the next cycle, and the slope of the next baseline fluctuation BLNext is calculated to be the same as the latest BLA as shown in the following formula, and the baseline fluctuation prediction unit Was output to the rising point level prediction unit.
BLANnext = BLA

  However, when the system is allowed to be complicated, more accurate estimation may be performed assuming that the baseline fluctuation is a high-order function of time.

(5) Amplitude predicting unit Amplitude is defined as the level difference between the direct wave peak point and the rising point when there is no baseline fluctuation. When there is a baseline fluctuation, the baseline fluctuation is estimated, and can be estimated as a value obtained by subtracting the baseline fluctuation estimation from the level difference between the direct wave peak point and the rising point.

  The amplitude estimation will be described again with reference to FIG.

In the present embodiment, as described in (4) above, the base line is defined as a straight line connecting points A and B. Therefore, the estimated value of the amplitude can be calculated as a level difference from the baseline extension line to the point C at the time of the latest direct wave peak point indicated by the point C. As a result, the predicted value Anext of the next amplitude was calculated by the following equation and output to the rising point level prediction unit.
Next = T_level_0−R_level_0
-BLANnext x (T_time_0-R_time_0)

  However, when the system is allowed to be complicated, a past amplitude record may be held in the amplitude prediction unit 10344 and used to predict the next amplitude.

(6) Rising point level prediction section The center point of the predicted range of the next rising point level (indicated by the dotted line in FIG. 8) is that the baseline fluctuation predicted in the next period continues during the period predicted in the next period. Predicted under. Therefore, the center point R_level_next_center of the next rising point was calculated by the following equation. In the example shown in FIG. 8, the center point is indicated by point D.
R_level_next_center = R_level_0
+ BLAnext x Bnext

In addition, the upper limit of the prediction range for the next rising point level is considered to be that the level difference from the center is proportional to the amplitude, and in order to avoid the local minimum point between the direct wave and the reflected wave, the level is below this local level. There is a need to. As a calculation method satisfying the above conditions, a method of empirically adding 1/2 of the amplitude to the center point level was adopted. Therefore, the upper limit R_level_next_upper of the prediction range was calculated by the following equation. In the example of FIG.
R_level_next_upper = R_level_next_center + Next × 0.5

On the other hand, the lower limit R_level_next_under of the prediction range was calculated by the following equation so that the center of the prediction range coincided with the center point.
R_level_next_under = 2 × R_level_next_center − R_level_next_upper

  Based on the next feature point prediction range output as a result of the above processing, the control information generation unit 1035 outputs the control information so that the feature point detection unit 102 performs the feature point detection operation only within the prediction range. Generated and output to the feature point detection unit 102. With this control information, for example, in the present embodiment, a header indicating the meaning of data as “NEXT RANGE” is placed at the beginning, and then, as the next feature point prediction range, a lower limit, an upper limit, a prediction time of the prediction time It can be configured as a data string arranged like the lower limit and the upper limit of the level. As a result, a data string such as ““ NEXT RANGE ”, 11.5, 11.7, 2.0, 3.5” is obtained.

  Next, returning to the overall operation of the feature point detection adaptive control unit 103, the operations of the prediction result determination unit 1033 and the initialization control unit 1032 in the present embodiment will be described.

  The prediction result determination unit 1033 determines whether or not the feature point detection unit 102 has detected a feature point as predicted in the prediction range of the next feature point output from the next feature point prediction unit 1034. In the case of the present embodiment, the rising point of the pulse wave and the peak point of the direct wave are detected as feature points, but since the prediction is performed only on the rising point, the prediction result determination unit 1033 also has the rising point. Only deal with.

  Further, in the present embodiment, the feature point detection unit 102 performs feature point detection only within the prediction range, so the function of the prediction result determination unit 1033 is to predict the next rising point predicted by the next feature point prediction unit 1034. An operation for confirming whether or not a new feature point is detected within the range is performed.

  The prediction result determination unit 1033 outputs nothing when a new feature point is detected within the prediction range. On the other hand, if a new feature point is not detected within the prediction range and the prediction time has passed, it is considered that the prediction is not performed correctly due to a sudden change in the shape of the pulse wave waveform. Then, a control signal for controlling the storage unit 1031 and the next feature point prediction unit 1034 to enter the initialization operation is output to the initialization control unit 1032.

In response to the control signal from the prediction result determination unit 1033, the initialization control unit 1032 starts the following initialization operation.
(1) The feature point data in the storage unit 1031 is replaced with a default value set separately.
(2) Replace parameters in the next feature point prediction unit 1034 with default values set separately.
The default values of (1) and (2) above can be realized by storing them in the initialization control unit 1032 using typical parameters of pulse waves measured in advance. Alternatively, the default value may be changed based on the use history of the user.

  By using the technology according to the present embodiment by the operation of each part described above, the malfunction of the rising point detection due to the influence of the reflected wave on the pulse wave waveform, which was the initial problem, is prevented. The rising point and the direct wave peak point With this, accurate detection can be achieved.

  Next, a second embodiment of the present invention will be described. A processing procedure when a program for processing pulse wave data from the biological signal measurement unit 101 in the CPU or DSP is implemented will be described with reference to the flowcharts of FIGS. 9 and 10.

  The entire process consists of two programs. The first program is represented by the flowchart shown in FIG. 9 and processes the feature point detection and the next feature point prediction. The second program is represented by the flowchart shown in FIG. Processing the determination and initialization as necessary.

  As shown in FIG. 9, the first program first initializes parameters used when recording feature point data and calculating next feature point prediction data in step S1, and in step S2, biosignal measurement is performed. In order to read the pulse wave data from the unit 101 and use it in a second program to be described later, the time of the data that has been processed in the previous process is recorded in a recording unit or a recording medium of the apparatus. Next, in step S3, the feature point prediction range determined by the information set in the previous cycle in step S8 as described in FIG. 9 or the information set by the initialization process as described in FIG. The read pulse wave data is compared to determine whether or not the pulse wave data is within the feature point prediction range. If it is not within the feature point prediction range, the process returns to step S2 to read the next pulse wave data. On the other hand, if it is within the feature point prediction range, the feature point is searched in step S4. The method for searching for feature points can be realized by using the method described as the function of the feature point detection unit 102 in the first embodiment. In subsequent step S5, it is inspected whether or not the feature point has been detected. If no feature point is detected, the process returns to step S1 to read the next pulse wave data. On the other hand, if a feature point is detected, the process proceeds to step S6, and the detected feature point data is recorded on a recording medium such as a memory or a hard disk. In the next step S7, the next feature point prediction range is calculated from the current and past feature point data recorded on the recording medium, and in step S8, the feature point prediction range calculated in step S7 is stored in the recording unit or Record on a recording medium. Further, in step S9, it is determined whether or not there is unprocessed pulse wave data. If there is unprocessed pulse wave data (Yes), the process returns to step S2 and the process is repeated. If there is no unprocessed pulse wave data (No), the process is terminated.

In the next second program, first, in step S10, the time of the pulse wave data that has been processed by the first program is read from the area of the recording unit or recording medium of the device set by the first program, In step S11, it is determined whether or not this time is within the feature point prediction range. If it is not within the feature point prediction range (No), the process returns to step S10. If it is within the feature point prediction range (Yes), the process proceeds to step S12 to check whether a new feature point is detected within the feature point prediction time range. In other words, the detection result of the first program is continuously monitored until the end of the feature point prediction time range. When the feature point prediction time range ends, the process proceeds to step S13, and if a new feature point is detected in the feature point prediction time range as a result of the process of step S12 (Yes), Returning to step S10, if it has not been detected (No), the process proceeds to step S14 to initialize the parameters used for recording the feature point data and calculating the next feature point prediction data. Return to S10.
With the above processing procedure, the configuration according to the present embodiment can also be realized by a program.

  Next, a technique according to the third embodiment of the present invention will be described with reference to the drawings.

  FIG. 11 is a functional block diagram showing a configuration example of the biological signal analyzer 400 according to the third embodiment of the present invention.

  The difference from the first embodiment is that, in the first embodiment, the prediction result determination unit 1033 provided in the feature point detection adaptive control unit 103 is eliminated, and its function is the same as that of the feature point detection unit 102. Is provided as a prediction range detection determination unit 4022.

  Therefore, in the third embodiment, whether or not the next feature point prediction range issued by the feature point detection adaptive control unit 403 is determined inside the feature point detection unit 402, and the information of the determination result is the feature point detection adaptation. This is transmitted to the initialization control unit 4032 in the control unit 403.

  In the first place, whether the prediction range is correct or not is always confirmed when detecting the feature points. Therefore, according to the present embodiment, the entire system has the advantage of eliminating computation duplication and saving computing resources. is there.

  As mentioned above, although embodiment of this invention has been described, this invention is not limited to the example demonstrated to such embodiment, There exists a change and addition in the range which does not deviate from the summary of this invention. They are also included in the present invention. For example, in the above-described embodiment, the rise of the pulse wave waveform is detected using a pulse wave sensor, but the present invention is not limited to this, and an electrocardiographic waveform measured using an electrocardiograph. It can also be applied to a method for detecting feature points from the above. Further, it can be applied to a method of detecting feature points from a blood pressure waveform measured using a continuous blood pressure monitor. Alternatively, it can be used for an electroencephalogram signal or the like.

  The second embodiment can also be applied to a method using a program for analyzing biological information to be measured in real time. For example, when a new biological signal is measured from a living body, it is possible to activate a thread that simultaneously processes the biological signal and use it as a control structure for a program that sequentially analyzes the measured data.

  As described above, in each embodiment of the present invention, erroneous detection is prevented and accurate heart rate can be measured from the pulse wave, so that various existing physiological indices based on heart rate are accurately calculated. Also contribute to doing. Moreover, the pulse wave can be easily measured with the pulse wave sensor as described above in the description of the background art. Therefore, it can be used for portable health devices such as health management monitors that can check their health status anytime, anywhere.

  The present invention is applicable to a biological signal analyzer.

It is a functional block diagram which shows the example of 1 structure of the biosignal analyzer in the 1st Embodiment of this invention. It is a functional block diagram which shows the internal structural example of the next feature point prediction part in FIG. It is a figure which shows the example of 1 structure of the pulse wave sensor by this Embodiment. It is a figure which shows the typical waveform example of a pulse wave. It is a figure which shows an example of the biosignal analyzer by a general prior art. It is a figure which shows the example which analyzed the pulse wave measurement waveform by the conventional method. It is a figure which shows the example of calculation of the prediction range of the next rising point. It is an enlarged view of the waveform with the trend that the whole pulse waveform is increasing It is a flowchart figure which shows the flow of the program processing in the 2nd Embodiment of this invention. It is a flowchart figure which shows the flow of the program processing in the 2nd Embodiment of this invention. It is a functional block diagram which shows the example of 1 structure of the biosignal analyzer in the 3rd Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Pulse wave sensor 11 Light emission part 12 Light reception part 100 Biosignal analyzer 101 Biosignal measurement part 102 Feature point detection part 103 Feature point detection adaptive control part 1031 Storage part 1032 Initialization control part 1033 Prediction result determination part 1034 Next feature point prediction Unit 1035 Control information generation unit 10341 Heartbeat interval time prediction unit 10342 Reflection delay time prediction unit 10343 Baseline fluctuation prediction unit 10344 Amplitude prediction unit 10345 Rise point time prediction unit 10346 Rise point level prediction unit 300 Conventional biological signal analyzer 301 Biological signal Measurement unit 302 Feature point detection unit 3011 Pulse wave sensor 3012 A / D converter 3021 Time differentiation unit 1
3022 Zero cross detection unit 3023 Time differentiation unit 2
3024 Rise detection unit 400 Biosignal analyzer 401 Biosignal measurement unit 402 Feature point detection unit 403 Feature point detection adaptive control unit 4021 Feature point detection module 4022 Predictive range detection determination unit 4031 Storage unit 4032 Initialization control unit 4033 Next feature point Prediction unit 4034 Control information generation unit

Claims (4)

A living body comprising a biological signal measuring unit for measuring a biological time series signal, and a feature point detecting unit for detecting a feature point using the biological time series signal measured by the biological signal measuring unit. a signal analyzer, using a feature point time output from the feature point detection unit, the value on the biological time-series signal corresponding to the feature point time, the feature point data that includes either or both Then, in the biological signal analyzer equipped with the feature point detection adaptive control unit for controlling the operation of the feature point detection unit,
The feature point detection adaptive control unit,
A storage unit for storing the input feature point data;
A next feature point prediction unit that predicts a range of feature point data of the next feature point based on the feature point data and past feature point data extracted from the storage unit;
And a control information generating unit which outputs control information for controlling the operation of the feature point detection unit using the output from the next feature point prediction unit
The feature point detection unit detects whether or not a feature point is detected within a condition of a prediction range of feature point data included in the control information output from the feature point detection adaptive control unit. It has a judgment part,
The feature point detection adaptive control unit receives a determination result from the prediction range detection determination unit, and the determination result detects a feature point within a condition of a prediction range of feature point data included in the control information If not , a biological signal analyzer comprising: an initialization control unit that initializes the storage unit in the feature point detection adaptive control unit and a set value in the next feature point prediction unit . The next feature point prediction unit
At least of: Including any one and using the output of at least one of the heartbeat interval time prediction unit and the reflection delay time prediction unit and the time of the latest pulse wave rising point, The biological signal analyzer according to claim 1 , further comprising a rising point time predicting unit that predicts a time range of a next pulse wave rising point. The next feature point prediction unit
An amplitude prediction unit for predicting the amplitude of the next pulse waveform;
At least one of the baseline fluctuation prediction units for predicting the baseline fluctuation of the next period,
Rise that predicts the level range of the next pulse wave rise point using the output of at least one of the amplitude prediction unit and the baseline fluctuation prediction unit and the level of the latest pulse wave rise point The biological signal analyzer according to claim 1 , further comprising a point level prediction unit. The next feature point prediction unit
At least one of a heartbeat interval time prediction unit that predicts a heartbeat interval of the next period and a reflection delay time prediction unit that predicts a reflection delay time defined as a time interval between a direct wave and a reflected wave on the pulse wave of the next period Either
Using the output of at least one of the heartbeat interval time prediction unit and the reflection delay time prediction unit and the time of the latest pulse wave rise point, the time range of the next pulse wave rise point A rising point time prediction unit for predicting
At least one of an amplitude prediction unit that predicts the amplitude of the pulse wave waveform of the next period, and a baseline fluctuation prediction unit that predicts the baseline fluctuation of the next period,
Using the output of at least one of the amplitude prediction unit and the baseline fluctuation prediction unit, the latest pulse wave rising point level, and the information calculated by the rising point time prediction unit The biological signal analyzer according to claim 1 , further comprising: a rising point level prediction unit that predicts a level range of the next pulse wave rising point.

Images