JP7487873B2 - Biological signal processing device - Google Patents

Description

The present invention relates to a biosignal processing device, and more specifically, to a biosignal processing device that can obtain stable vital signals even during exercise or when the environment is changing.

Numerous technologies for measuring biosignals have been proposed to date. As an example, Patent Document 1 proposes a bioinformation analysis device that includes a bioinformation detection unit that detects a user's bioinformation, a user determination unit that determines the user's athletic ability, an information setting unit that sets analysis information according to the user's athletic ability when it is determined that the user's athletic ability meets a predetermined condition, and an analysis unit that analyzes the bioinformation based on the set analysis information, and that can appropriately measure bioinformation even for users with an athletic heart.

Patent Document 2 describes a biometric information processing device that includes a biometric information detection unit that detects biometric information related to a subject, a body movement information detection unit that detects body movement information related to the movements of the subject, a reliability determination unit that determines the reliability of the biometric information based on a predetermined criterion, a storage unit that stores a correlation between the body movement information and the biometric information, and an update unit that updates the correlation stored in the storage unit, in order to improve the accuracy of a pulse rate estimated from exercise intensity.

Patent document 3 describes a respiration rate estimation method including a first step of estimating the subject's respiration rate from a time series signal of first data related to the subject's cardiac function and a time series signal of second data related to acceleration due to the subject's respiratory movement, a second step of estimating the respiration rate estimated from the first data and the respiration rate estimated from the second data after filtering out noise using a Kalman filter, and a third step of performing a weighted averaging process on the multiple respiration rate estimates obtained in the second step.

Patent Document 4 also describes an electrocardiogram analysis device and electrode set that includes electrocardiogram measurement electrodes for measuring an electrocardiogram, noise measurement electrodes for measuring noise, a measurement unit that measures an electrocardiogram signal measured by the electrocardiogram measurement electrodes and a noise signal measured by the noise measurement electrodes, and a chest compression noise removal processing unit that analyzes the measured electrocardiogram signal and noise signal to extract a noise-removed electrocardiogram from which noise has been removed, thereby reducing the effects of noise due to chest compression.

International Publication No. 2016/088307 International Publication No. 2014/196119 International Publication No. 2017/090732 JP 2014-076117 A

Biometric information is measured (detected) not only when at rest, but also while working or exercising, but the above-mentioned conventional technologies mainly focus on pulse waves and respiratory rates, and even in the case of electrocardiograms, the effects of body movement in an AED are addressed, and do not take into account strenuous exercise, exercise, work, and the large changes in the environment that accompany them.

The objective of the present invention is to make it possible to obtain stable vital information during daily life and exercise, and even when there are changes in body movement or the environment.

In order to solve the above problems, the present invention has several characteristic configurations. First, the first invention is characterized by comprising an ECG measurement unit that acquires an electrocardiogram signal from the body surface of a living body, an R-wave extraction unit that extracts R-waves contained in the electrocardiogram signal from their characteristics and identifies the position of the R-wave peak, an SNR calculation unit that calculates the SNR (signal-to-noise ratio) of the R-wave section identified by the R-wave extraction unit, and an output unit that determines and outputs the R-wave interval from the position of the R-wave peak and the SNR of the R-wave section.

The first invention further includes a signal calculation unit that uses the R-wave time extracted by the R-wave extraction unit to align and average the positions of each R-wave to calculate the signal amount for calculating the SNR in the SNR calculation unit, a wave removal unit that removes the original electrocardiogram signal such as PQRST from the electrocardiogram signal including noise measured by the ECG measurement unit, a noise calculation unit that obtains the noise amount from the noise waveform extracted by the wave removal unit in order to calculate the SNR in the SNR calculation unit, an SNR comparison unit that compares the SNR calculated by the SNR calculation unit with a predetermined threshold value in each R-wave section, and a frequency analysis unit, and the output unit calculates the RRI by taking the difference in time between two consecutive R-wave peaks in the SNR comparison unit in a section excluding a section where the SNR is less than the threshold value, and outputs the calculation result, and the frequency analysis unit calculates and outputs the fluctuation of the RRI.

In the first invention, the frequency analysis unit uses the RRIs that remain at indefinite intervals and performs frequency analysis based on the least squares method of a sine wave.

In the second invention, the ECG measurement unit is characterized by processing signals from a plurality of ECG electrodes, and further includes an SNR/electrode comparison unit that compares the SNRs calculated by the SNR calculation unit between the ECG electrodes and selects the best SNR, and an SNR/threshold comparison unit that compares the best SNR selected by the SNR/electrode comparison unit with a predetermined threshold and provides the comparison result to the output unit.

Preferably, the device further includes a weight synthesis unit that multiplies each section of the raw waveform of each ECG electrode measured by the ECG measurement unit by the SNR of that section calculated by the SNR calculation unit and adds the polarity of each ECG electrode, and an R-wave re-extraction unit that re-extracts an R-wave from the composite wave synthesized by the weight synthesis unit and provides the R-wave to the output unit.

The third invention further includes a case as a storage container having therein a signal processing circuit including the ECG measurement unit, the R-wave extraction unit, the SNR calculation unit, and the output unit, and is characterized in that some of the plurality of ECG electrodes are electrically connected to the ECG measurement unit via terminals provided on the upper part of the case, and the remaining part of the plurality of ECG electrodes are disposed on the bottom surface of the case and connected to the ECG measurement unit via wiring within the case.

In the third invention, preferably, a heat flow measuring unit having a plurality of thermometers for measuring the heat flow between the inside of the living body and the environment is provided within the case.

The third invention is characterized in that the bottom of the case is covered with an insulating sheet, two split ECG electrodes are arranged at a predetermined distance on the outer surface of the insulating sheet, the thermometer is arranged on the inner surface of the insulating sheet at a position between the two ECG electrodes, and the electrode surfaces of the ECG electrodes and the sheet surface of the insulating sheet between them are on the same plane.

In addition, according to a preferred embodiment of the third invention, the portion of the outer surface of the insulating sheet where the ECG electrodes are arranged is made thinner by the thickness of the ECG electrodes.

The fourth invention is characterized by having clothing worn on the upper body of a living body, and the ECG electrodes and heat flow pads for measuring the heat flow between the living body and the environment being provided at predetermined parts of the clothing.

In the fourth invention, the heat flow pad is provided with a heat flow measuring means consisting of an insulating material and two thermometers arranged on its upper and lower surfaces.

In the fourth invention, the ECG electrodes are provided at locations on the garment that correspond to areas where the electrocardiogram has a predominant positive potential and at locations that correspond to areas where the electrocardiogram has a predominant negative potential.

The fourth invention includes an embodiment in which a spring material is disposed between the fabric of the clothing and the ECG electrodes to maintain contact between the ECG electrodes and the skin.

It is also preferable that a spring material be disposed between the clothing fabric and the insulating material to keep the thermometer on the underside in contact with the skin.

As the insulating material, a gas mat containing a number of gas bubbles arranged on a flat surface is preferably used.

The fourth invention also includes an embodiment in which the heat flow pad is provided with two of the heat flow measuring means, the gas mat is divided into a first gas mat for one of the heat flow measuring means and a second gas mat for the other heat flow measuring means, and the first gas mat and the second gas mat contain gases with different thermal conductivities.

The fifth invention is characterized by comprising a heat flow measuring unit that measures the heat flow between the inside of the living body and the environment, a deep temperature calculating unit that calculates the deep temperature of the living body based on temperature information from the heat flow measuring unit, a noise calculating unit that calculates noise contained in the deep temperature, and an output unit that calculates and outputs a new deep temperature from the noise calculated by the noise calculating unit and the deep temperature calculated by the deep temperature calculating unit.

The fifth invention is characterized in that the noise is obtained from the output of a high-pass filter with the deep temperature input.

The fifth invention, in another aspect, is characterized in that it includes a heat flow measuring unit that measures the heat flow between the inside of the living body and the environment, an environmental temperature measuring unit that measures the environmental temperature, a false signal generating unit that creates a false signal from the environmental temperature measured by the environmental temperature measuring unit, a deep temperature calculating unit that calculates the deep temperature of the living body from the heat flow measured by the heat flow measuring unit, and an output unit that calculates and outputs a new deep temperature from the false signal generated by the false signal generating unit and the deep temperature calculated by the deep temperature calculating unit.

The sixth invention is characterized by comprising an ECG measuring section for acquiring an electrocardiogram signal from the body surface of the living body, a heat flow measuring section for measuring the heat flow between the inside of the living body and the environment, an acceleration measuring section for measuring the acceleration of the living body, an environmental measuring section for measuring the environmental condition surrounding the living body, an ECG-SNR comparing section for calculating the signal-to-noise ratio of the electrocardiogram signal and comparing it with a predetermined first threshold, a deep temperature-noise comparing section for calculating the noise contained in the heat flow measured by the heat flow measuring section and comparing it with a predetermined second threshold, a threshold generating section for changing the first threshold and/or the second threshold based on the acceleration measured by the acceleration measuring section and the environmental condition measured by the environmental measuring section, and an output section for outputting a notification signal based on the comparison results from the ECG-SNR comparing section and the deep temperature-noise comparing section.

The present invention makes it possible to obtain stable vital information during daily life and exercise, even when there are changes in body movement or the environment.

1 is a schematic diagram showing a basic configuration of a first embodiment of the present invention; FIG. 2 is a schematic diagram showing a specific configuration of the first embodiment. FIG. 4 is a waveform diagram showing an example of an electrocardiogram waveform measured by an ECG measurement unit. FIG. 4 is a waveform diagram for explaining an example of signal processing in an R-wave extraction unit. 5 is a waveform diagram for explaining an example of signal processing in a signal calculation unit. FIG. 5A to 5D are waveform diagrams for explaining an example of signal processing in the wave removal unit. 5 is a waveform diagram for explaining an example of signal processing in a noise calculation unit. 6 is a waveform diagram for explaining an example of signal processing in an SNR calculation unit. 5 is a schematic diagram for explaining an example of signal processing in an SNR comparison unit, an output unit, and a frequency analysis unit. FIG. 4 is a schematic diagram showing a first aspect of a second embodiment of the present invention. FIG. 4 is a schematic diagram for explaining the operation of the first embodiment. 4 is a wave diagram for explaining details of the SNR comparison means in the first aspect. FIG. 4 is a schematic diagram showing a second aspect of the second embodiment of the present invention. FIG. 4 is a schematic diagram for explaining the operation of the second embodiment. 13A is a schematic perspective view showing a first aspect of a sensor structure having a plurality of ECG electrodes according to a third embodiment of the present invention, and FIG. 13B is a view taken along an arrow A in FIG. 13A (bottom view). 5A is a schematic perspective view showing a second embodiment of the sensor structure, and FIG. 5B is a bottom view taken along the line B in FIG. 5A. FIG. 4 is a schematic perspective view showing a third embodiment of the sensor structure. FIG. 13 is a schematic perspective view showing a fourth embodiment of the above-mentioned sensor structure further equipped with a deep body temperature measurement function; FIG. 14 is a side view taken along the line C in FIG. 13; and FIG. 15 is a bottom view taken along the line D in FIG. FIG. 13 is a schematic exploded perspective view showing a first example of a case applied to the fourth aspect. FIG. 13 is a schematic exploded perspective view showing a second example of a case applied to the fourth aspect. FIG. 13 is a schematic exploded perspective view showing a third example of a case applied to the fourth aspect. 13A and 13B are schematic diagrams showing the front and rear views, respectively, of a garment having ECG electrodes and a heat flow pad for detecting heat flow according to a fourth embodiment of the present invention. 1A is a schematic plan view showing wiring on the above-mentioned wafer, FIG. 1B is a perspective view, and FIG. 1C is a schematic cross-sectional view taken along line E-E in FIG. 1A is a plan view showing ECG electrode pads provided on the garment, and FIG. 1B is a schematic cross-sectional view taken along line FF in FIG. 16(a) and 16(b) are schematic cross-sectional views taken along line GG in FIG. 16(a), each showing an ECG electrode pad provided on the garment. FIG. 1A is a plan view showing a heat flow pad of the SHF type provided on the above-mentioned wafer, and FIG. 1B is a schematic cross-sectional view taken along line HH in FIG. 18A and 18B are plan views similar to those of FIG. 18A and a schematic cross-sectional view taken along line II in FIG. 18A, showing a heat flow pad of the SHF type provided on the above-mentioned wafer. FIG. 1A is a plan view showing a DHF type heat flow pad provided on the above-mentioned garment, and FIG. 1B is a schematic cross-sectional view taken along line J-J in FIG. 20(a) is a plan view similar to FIG. 20(a), and FIG. 20(b) is a schematic cross-sectional view taken along line K-K in FIG. 20(a), showing a DHF type heat flow pad provided on the above-mentioned wear. 1A is a schematic perspective view showing a first example of an insulating material applied to the heat flow pad; FIG. 1B is a schematic perspective view showing a second example; and FIG. 1C is a schematic perspective view showing a third example. FIG. 13 is a schematic diagram showing a first aspect according to a fifth embodiment of the present invention. FIG. 23 is a schematic diagram showing a second aspect of the fifth embodiment. FIG. 4 is a schematic diagram for explaining the operation of the second embodiment. FIG. 23 is a schematic diagram showing a third aspect of the fifth embodiment. FIG. 11 is a schematic diagram for explaining the operation of the third embodiment. FIG. 23 is a schematic diagram showing a fourth aspect of the fifth embodiment. FIG. 13 is a schematic diagram for explaining the operation of the fourth embodiment. FIG. 13 is a schematic diagram showing a fifth aspect of the fifth embodiment. FIG. 13 is a schematic diagram showing a sixth aspect of the fifth embodiment. FIG. 13 is a schematic diagram for explaining the operation of the sixth embodiment. FIG. 13 is a schematic diagram showing a basic configuration of a sixth embodiment of the present invention. FIG. 23 is a schematic diagram showing a specific configuration of the sixth embodiment.

Next, several embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited to these embodiments.

[First embodiment] Calculation of SNR of ECG wave and its use As shown in FIG. 1, the basic configuration of the biological signal processing device according to the first embodiment includes an ECG measurement unit 110, an R wave extraction unit 120, an SNR calculation unit 130, and an output unit (RRI output unit) 140.

To give an overview of the role of each component, the ECG (Electrocardiogram) measurement unit 110 measures (acquires) electrocardiogram signals from the body surface. The R-wave extraction unit 120 extracts R-waves contained in the electrocardiogram signal from their characteristics and identifies the position of the R-wave peak.

The SNR calculation unit 130 calculates the signal-to-noise ratio (SNR) of the electrocardiogram signal. As the signal, the amplitude from minimum to maximum of the electrocardiogram signal can be used, or the effective value can be used. Noise is a component in the acquired electrocardiogram signal other than the thermal power caused by the beating of the heart, and is caused by the shift and contact change between the electrode and the skin caused by movement or motion, changes in contact potential, static electricity, changes in conductance due to sweat, etc., myoelectricity, induction from an AC power source, etc.

The output unit (RRI output unit) 140 applies signal processing based on the SNR obtained by the SNR calculation unit 130 to the R wave and its surrounding section obtained by the R wave extraction unit 120, and outputs the result. As part of this signal processing, for example, the heart rate, heart rate variability (HRV, Heart Rate Variability), the shape of the PQRST wave, and vital information derived from these are displayed, notified, or made available for use by other programs.

Next, as shown in FIG. 2, the specific configuration of this biosignal processing device includes an ECG measurement unit 110, an R-wave extraction unit 120, an SNR calculation unit 130, and an output unit (RRI output unit) 140, as well as a signal calculation unit 121, a wave removal unit 122, a noise calculation unit 123, an SNR comparison unit 131, and a frequency analysis unit 150.

To explain the operation of each part in more detail, the ECG measurement unit 110 amplifies the signal from the electrodes (ECG electrodes) for electrocardiogram detection attached to the body surface, applies an appropriate filter, and outputs it (see Figure 3a). The filters include a high-pass filter (HPF) and a low-pass filter (LPF).

The high-pass filter removes fluctuations in the baseline of the electrocardiogram signal. The low-pass filter removes inductive noise from AC power lines and electromyographic noise caused by chest muscle movements, etc. The passband determined by the high-pass and low-pass filters is usually set to pass a few Hz to around 40 Hz, and in noisy environments, around 10 Hz to 20 Hz.

Narrowing the passband increases the effect of suppressing noise, but it also reduces the reproducibility of the original electrocardiogram signal emitted when the heart beats. In addition, noise can get into the narrowed passband due to exercise or other movements in daily life, making it impossible to completely remove it.

One of the factors that causes noise due to movement is the contact potential that occurs between the ECG electrodes and the skin. The contact potential can be reduced by using silver/silver chloride (Ag/Agcl) as the electrode material, but it is difficult to reduce it to zero.

When movement causes a shift in the contact between the ECG electrodes and the skin, or when the contact area changes, this finite potential changes and becomes noise. Impulse-like potential changes have a wide spectrum and fall within the passband mentioned above.

Belts and clothing used to attach ECG electrodes to the body are usually made of fabric, making it difficult to completely eliminate static electricity. Static electricity also generates impulse-like potentials, causing similar problems.

When the body sweats due to exercise, a layer of sweat may form on the surface of the skin, or may be absorbed by a belt or clothing, resulting in a conductivity of about 0.1 mS (resistivity of about 10 kΩ). If this conductivity changes due to exercise, the same problem can occur.

These phenomena are random processes that vary depending on the vector of movement, the position of the ECG electrodes, the degree of contact with the skin, etc., and appear as noise (artifacts) in certain sections, as shown in Figure 3a.

Here, the interval is defined as the time domain for each R wave, which is divided at the midpoint between the peak time of one R wave and the peak time of the next R wave. As shown in Figure 3a, there are intervals with little noise and intervals with a lot of noise.

Next, referring to FIG. 3b, the R-wave extraction unit 120 extracts R-waves. One method for extracting R-waves is to set a threshold value for the amplitude of the QRS wave and identify anything above the threshold as a QRS wave. As an alternative method, a threshold value can be set for characteristics such as the slope or length of the waveform.

In addition, since electrocardiogram signals are generally repetitive waveforms of the same shape, the position of the R wave can be identified using autocorrelation functions, template matching, wavelets, etc.

Many methods are known for extracting these R waves, but they have limitations when there is noise that resembles the original electrocardiogram signal, or when the original electrocardiogram signal is buried in the noise. In addition, by applying the features of the present invention, which selects and rejects R wave sections based on the SNR and selects and combines good signals from multiple electrodes while using these conventional R wave detection methods, it is possible to obtain a more effective output.

Referring to FIG. 3c, the signal calculation unit 121 calculates the signal amount for SNR calculation. As described above, the amplitude from the minimum to the maximum of the electrocardiogram signal can be used as the signal, or the effective value can be used.

For example, by averaging the positions of the R waves using the R wave times extracted by the R wave extraction unit 120, it is possible to calculate a signal amount with less noise. A moving average can also be used as an averaging method. Since the QRS wave is usually the wave with the maximum amplitude of the original electrocardiogram signal, the peak to peak of this QRS wave can also be used as the signal amplitude.

In some cases, it may be possible to determine the signal quality using only the noise without performing signal calculation in the signal calculation unit 121. However, in the case of ECG, there may be individual differences in amplitude, differences in amplitude may occur depending on the position of the ECG electrodes, and furthermore, the gain of the amplifier may be automatically adjusted by AGC (Auto Gain Control). Therefore, it is preferable to perform signal calculation in the signal calculation unit 121.

Referring to FIG. 3d, the wave removal unit 122 removes waves such as PQRST contained in the original electrocardiogram signal (electrocardiogram waveform). In this example, as shown in FIG. 3d(a), the QRS wave period and the T wave period are removed. The periods of other waves such as P waves and U waves can also be removed, but these P waves, U waves, etc. often do not have a significant effect on the calculation of the noise amount, so they may be left as they are.

The top row of Figure 3d (b) shows the actual electrocardiogram waveform before the PQRST wave is removed, and the bottom row shows the waveform after the PQRST wave has been removed. Approximately 100 sections are overlapped, aligned with the peak positions of the R waves. Since the data was acquired in a stationary state, the amount of noise is small, with an amplitude of -8 to +6 LSB, and an effective value of approximately 3 LSB rms. Incidentally, the amplitude of the QRS wave is approximately 350 LSB in this example, and the signal-to-noise ratio (SNR) is 100 or more (signal amplitude to noise effective value).

Figure 3d (c) is an example obtained during exercise; the top row shows the actual electrocardiogram waveform before the PQRST waves are removed, and the bottom row shows the waveform after the PQRST waves have been removed. Approximately 100 sections are overlapped, aligned with the peak positions of the R waves. The noise level has an amplitude of -600 to +600 LSB, and an effective value of approximately 100 LSB rms. The signal is approximately 300 LSB, so the SNR is approximately 3.

As another example of wave removal, as shown in FIG. 3d(d), the R-wave positions of each section may be aligned to obtain a moving average, and the moving average may be subtracted from the electrocardiogram signal including noise; such an embodiment is also included in the present invention.

Next, referring to FIG. 3e, the noise calculation unit 123 obtains, for example, the effective value of the noise waveform extracted as described above as the noise amount. The noise amplitude can also be used as the noise amount. If the noise amount for each section is, for example, the noise amount Ni for the latest section, the noise amount Ni for the section immediately preceding that, Ni-1, the noise amount Ni for the section immediately preceding that, ..., the noise amount Ni for the latest section is obtained in sequence and added after the section immediately preceding Ni-1.

The SNR calculation unit 130 calculates the SNR from the signal amount and the noise amount. The signal amount can be a moving average or an instantaneous value for each section, and the noise amount can be an effective value for each section. An example of the SNR calculated in this way is shown in Figure 3f. The heart rate is also shown for reference. The first half of the figure, in which the heart rate is low, is when the person is sitting (sitting), and the second half, in which the heart rate is high, is when the person is exercising on an exercise bike (registered trademark) (when exercising).

When the SNR falls below about 20, the frequency of R-wave extraction errors (false detection), such as the R-wave extraction time being shifted or noise being extracted as an R-wave, increases. There is also a risk of missed detection, where the actual R-wave cannot be extracted.

In the first half of the session, when the patient is sitting, the SNR fluctuates but is generally above 30, and the frequency of false positives and missed detections is low. In the second half of the session, when the patient is exercising, the SNR frequently drops below 20, and false positives and missed detections occur frequently. False positives and missed detections not only reduce the accuracy of heart rate calculations, but also reduce the accuracy of HRV, which is calculated using the R-wave interval (RRI: R-R Interval). In some cases, it may not be possible to output heart rate or HRV.

The SNR for each section varies greatly. Even in the first half of the sitting position, there are sections where the body is completely still and sections where it is not, so the SNR varies from over 200 to a few tens. Even in the second half of the exercise, as the aero bike (registered trademark) is constantly pedaled, there are sections where the SNR is very poor and sections where it is reasonably good.

The length of the interval (R-R interval) varies depending on the heart rate, but is around 0.3 to 1 second. Exercise bike (registered trademark) exercise results in a 1 second period when pedaling at 60 rpm, for example. If there is a period of time within the 1 second period of motion vector change, for example around 0.5 seconds, during which no artifacts occur in the ECG electrodes, then this means that an interval with high SNR exists.

Referring to FIG. 3g, the SNR comparator 131 compares the SNR with a threshold value for each section. For example, a value of 20 or less, which is the SNR at which the aforementioned false detection occurs, can be used as the threshold value.

A decrease in the section SNR makes it impossible to extract electrocardiogram features, resulting in missed detections, but missed detections are actually a problem with feature extraction and are not related to the threshold. Setting a high threshold in an attempt to reduce false positives increases the risk of causing problems such as deleting sections that were normally detected (missed detections due to the threshold).

The probability of missed detections due to the threshold value increases when the SNR, at which R-wave detection is possible to a certain extent, is around 10. In order not to discard normal detections, it is preferable to set the threshold value at around 10 or less.

If the SNR comparison result is the latest section Ri and the previous section Ri-1, the latest section Ri is found and added to the results before Ri-1 in sequence. If it is equal to or greater than the threshold, it is judged as OK, and if it is less than the threshold, it is judged as NG.

The RRI output unit 140 outputs the result of the RRI calculation. The RRI is calculated by taking the difference in time between two consecutive R-wave peaks. At this time, the period in which the SNR comparison unit 131 judges the result to be NG is not used.

For example, if section Ri-2 is NG, the R wave extraction result in section i-2 is not used. This means that the RRI is not calculated between sections i-1 and i-2, and further between sections i-2 and i-3.

The NG section often includes missed R-wave detections, and it is often incorrect to skip section i-2 and calculate the RRI between section i-1 and section i-3, so this calculation is not performed either.

The frequency analysis unit 150 calculates and outputs the fluctuation of the RRI, that is, the HRV. Conventionally, there are examples in which the RRI is subjected to spline interpolation and resampled at regular time intervals, but applying spline interpolation to an RRI that includes gaps can result in spurious vibrations, which is not desirable.

It is preferable to perform frequency analysis based on the least squares method of a sine wave using the RRI with an indefinite interval, without performing spline interpolation or resampling. An example of such frequency analysis is the Lomb-Scargle periodogram.

In addition, by aligning the position of the R-wave peaks in the raw ECG waveform in a section with a high SNR and averaging several sections, it is possible to obtain a waveform that is faithful to the original ECG waveform with little error due to noise. Averaging using ECG waves that are disturbed by artifacts during exercise will result in large errors in the wave information.

The peak values and slopes of PQRSTU waves and the like reflect the characteristics of the living body, the state of the autonomic nervous system, hormones, blood oxygen concentration, potassium concentration, and the concentrations of other substances. For example, in an experiment simulating high-altitude mountain climbing, it has been reported that a decrease in blood oxygen concentration causes changes in ST waves. By obtaining accurate ECG wave information, it is possible to grasp such a living body's state.

[Second embodiment] Synthesis of multiple electrode signals using ECG SNR The second embodiment includes a first aspect shown in FIG. 4 and a second aspect shown in FIG. 7, both of which have multiple ECG electrodes 1, 2, ....

-First aspect-
First, the first aspect will be described. In addition to the ECG measurement section 210, the R-wave extraction section 220, and the SNR calculation section 230, the first aspect includes an SNR/electrode comparison section 231, an SNR/threshold comparison section 232, an RRI output section 240, and a frequency analysis section 250.

Of these, the ECG measurement unit 210, the R-wave extraction unit 220, the SNR calculation unit 230, the RRI output unit 240, and the frequency analysis unit 250 may have the same configuration as the ECG measurement unit 110, the R-wave extraction unit 120, the SNR calculation unit 130, the RRI output unit 140, and the frequency analysis unit 150 described in the first embodiment above, respectively.

The ECG electrodes 1, 2, ... are attached to the skin of a living body to detect electrocardiograms. The electrode materials preferably used are Ag/AgCl or Ag, which have a small contact potential, or conductive plastic.

The artifacts are superimposed on the original electrocardiogram signal, depending on the vector of the movement, the location of the ECG electrodes, the muscle movement, the degree of contact with the skin, etc. The muscles and vectors used vary depending on the type of movement (for example, from basic training such as sit-ups, push-ups, and running to training using machines, competitive sports, stretching, yoga, dance, etc. that involve free movement), so the way artifacts appear varies between multiple ECG electrodes.

An ECG measurement unit 210 may be provided for each ECG electrode, but it can be reduced to one by switching between high-resolution ADCs (analog-to-digital converters) in a time-division (interleaved) manner.

The effects of artifacts can be suppressed by combining signals from multiple ECG electrodes 1, 2, ... according to the SNR. The ECG signal can be amplified by synchronously adding (adding in alignment) the original ECG signal.

Because the ECG signal is unknown, it is difficult to completely separate the ECG from the artifacts. However, because the artifacts are caused by the mechanism described above, they have a waveform and spectrum that differ from the original ECG signal (PQRST wave).

Many of these are waves that are band-limited due to impulse-like or step-function-like changes in contact potential caused by the impact of exercise, and are different in nature from the original electrocardiogram signal. The SNR calculation unit 230 quantifies the wave energy of the artifacts that spread across each section.

The R-wave extraction unit 220 and the blocks following it can have the same configuration as in the first embodiment described above, but can be made into a single system by switching the processing electrodes at regular intervals.

The SNR calculation unit 230 calculates the SNR at each time for each ECG electrode 1, 2, ..., and obtains a time series of section SNR as shown in Figure 5 for each ECG electrode 1, 2, .... In Figure 5, SNR1,i-2: SNR1,i-1: SNR1,i is the time series of section SNR for ECG electrode 1, and SNR2,i-2: SNR2,i-1: SNR2,i is the time series of section SNR for ECG electrode 2.

The SNR/electrode comparison unit 231 compares the SNR for each ECG electrode at each time and selects the best SNR.

The SNR/threshold comparison unit 232 compares the selected SNR with a threshold, and if it is equal to or greater than the threshold, it outputs the R-wave time extracted from the ECG electrode with the best SNR at each time. If it is less than the threshold, it outputs a blank. In FIG. 5, Ri, Ri-1, Ri-2, etc. are the above outputs stored in a memory, etc.

The RRI output unit 240 calculates and outputs the R-wave interval (R-R Interval) based on the stored R-wave time. As in the first embodiment, a blank space means that the value is less than the threshold value, so the RRI before and after the blank space is not calculated.

The frequency analysis unit 250 uses the RRI to perform an irregular frequency analysis based on least-squares approximation of a sine wave, similar to the first embodiment.

The details of the SNR/threshold comparison unit 232 are described below with reference to Figure 6.

Figure 6 (a) shows a relatively simple example, where the timing of the section switching is synchronized for ECG electrode 1 and ECG electrode 2, and an artifact is present in section i-2 of ECG electrode 1.

The SNR/electrode comparison unit 31a determines that the SNR of ECG electrode 1 is less than that of ECG electrode 2. If the SNR of ECG electrode 2 is equal to or greater than the threshold value, the SNR/threshold comparison unit 31b selects ECG electrode 2 and adopts its R-wave time (if it is equal to or less than the threshold value, neither is adopted). That is, as follows.

(1) If threshold<SNR1,i-2<SNR2,i-2, the R-wave time extracted from ECG electrode 2 is used.
(2) If SNR1,i-2<threshold value<SNR2,i-2, the R-wave time extracted from ECG electrode 2 is used.
(3) If SNR1,i-2<SNR2,i-2<threshold, then neither is adopted.

Figure 6(b) shows a slightly more complicated case, where the lengths of the sections are different for ECG electrode 1 and ECG electrode 2. For ECG electrode 1, artifacts are extracted as R waves in sections i-2 and i-1. Section i-2 of ECG electrode 2 correctly detects the original R wave, but it straddles sections i-2 and i-1 of ECG electrode 1.

Generally, the QRS period is the most important period for extracting ECG features, so it is best to focus on this narrow period. By narrowing it down, it may be possible to narrow down the number of comparisons to one, but as in this example, there may still be two. In this case, it is best to carry out the following process.

In other words, if SNR1,i-2<SNR1,i-1<SNR2,i-2, SNR2,i-2 is used. This is because the SNR is low in both sections of ECG electrode 1.

Other than the above, SNR2, i-2 is not used. The reason is that there is an extraction with a high SNR on ECG electrode 1, and there is a possibility that artifacts exist on ECG electrode 2.

-Second aspect-
As shown in Fig. 7, in order to perform weighted synthesis according to the SNR, a weight synthesis section 233 and an R-wave re-extraction section 234 are provided instead of the SNR/inter-electrode comparison section 231 and the SNR/threshold comparison section 232 in the first embodiment. The other configurations are the same as those in the first embodiment. Fig. 8 is a schematic diagram for explaining the operation of each section.

The weight synthesis unit 233 multiplies the ECG raw waveform of each ECG electrode 1, 2, ... by the SNR of the corresponding section at each time. For example, the ECG raw waveform of ECG electrode 1 is divided into sections v1,i:v1,i-1:v1,i-2, ... by provisional R-wave detection.

The length of each section is determined by the R-R interval as described above, and is, for example, about 0.3 to 1 second. For example, if the ADC sampling rate is 128 sps, each section will contain 38 to 128 pieces of time series data (x1, x2, ...).

In other words, each interval is a set of the above time series data, for example, interval v1,i = {x1, x2, ...}. Multiplying this by the SNR gives SNR1,i x v1,i = {SNR1,i x x1, SNR1,i x x2, ...}.

Furthermore, the weight synthesis unit 233 sums the polarities of the ECG electrodes 1, 2, ... together. For example, if the signals to be synthesized in the latest section are v and i, then
v,i = p1 x SNR1, i x v1,i + p2 x SNR2, i x v2,i +...
p1, p2, ... are the polarities of the ECG electrodes and take the values of +1 or -1.

Normally, once the placement of the ECG electrodes is determined, the polarity of the PQRST wave does not change, so there is no need to change the polarity for each section. In other words, the same values can be used for p1, p2, ... even if the section changes.

On the other hand, there is a phenomenon in which the shape of the PQRST wave changes depending on the location of attachment of multiple ECG electrodes. For example, in a 12-lead electrocardiograph, the S wave is larger than the R wave at the V1 electrode. In this case, it is advisable to determine the composite polarity so that the QRS amplitude is maximized, for example.

The R-wave re-extraction unit 234 re-extracts R-waves from the combined signal v,1:v,i-1:v,i-2.... The SNR of the combined wave is approximately the sum of the SNRs of each ECG electrode 1, 2,... before combination, and R-wave extraction can be performed with an improved SNR.

As described above, during exercise, there are sections where the SNR is very poor and sections where it is relatively good, and this depends on the motion vector and the location of the electrodes. Therefore, by combining weights across multiple ECG electrodes 1, 2, ..., it is possible to reduce the number of sections where the SNR is very poor. This leads to a reduction in false detections and missed detections of RRI.

Third Embodiment
Here, a sensor structure having multiple ECG electrodes for electrocardiogram detection will be described. Normally, increasing the number of ECG electrodes means increasing the number of terminals for connection to a case (container) that houses a signal processing circuit, etc., and the electrode area also inevitably becomes larger, but in the present invention, we have devised a way to minimize these points.

-First aspect-
9 shows an example in which ECG electrodes 3 are also provided on the bottom surface of the case 310, minimizing an increase in the number of terminals and an increase in the occupied area. In this example, the case 310 is fixed to the skin surface via a belt 301. An adhesive tape may be used instead of the belt 301. Although not shown, the case 310 has a built-in signal processing circuit and the like.

As shown in Figures 9(a) and (b), this example has three ECG electrodes: ECG electrodes 1 and 2 arranged on both sides of the case 310, and ECG electrode 3 provided on the bottom surface of the case 310 (the surface that comes into contact with the skin).

As explained above, ECG electrodes 1 and 2 are preferably electrodes made of Ag/AgCl or conductive plastic, and these ECG electrodes 1 and 2 are connected to a signal processing circuit etc. within case 310 via terminals 321a and 321b provided on the top surface of case 310.

The ECG electrode 3, like the ECG electrodes 1 and 2, is preferably an electrode made of Ag/AgCl or conductive plastic and is provided on the bottom surface of the case 310, but a conductive gel may also be used for the ECG electrode 3.

The ECG electrodes 1, 2, and 3 are attached so as to be in contact with the skin, and an insulating cover 302 is placed to cover the terminal portions of the ECG electrodes 1 and 2. It is preferable that the insulating cover 302 is placed so as to protrude 10 mm or more from the end of the case 310. This is to minimize the decrease in insulation resistance between the ECG electrodes 1 and 2 due to sweat.

Figures 10(a) and (b) show an example in which the ECG electrode 3 placed on the bottom surface of the case 310 is divided into two ECG electrodes 3a and 3b, thereby increasing the number of electrodes.

Figure 11 shows an example in which, in addition to the above-mentioned ECG electrodes 1, 2, and 3, four more ECG electrodes 4 to 7 are provided. Of these, ECG electrodes 4 and 5 are placed on belt 301 so as to contact the sides of the body, and ECG electrodes 6 and 7 are placed on belt 110 so as to contact the back of the body.

The sides of the body act as fulcrums for belts and clothing fabric, and are areas that are relatively unlikely to shift during movement. The combination of the back, front, and sides of the body increases the probability of generating a state with fewer artifacts at any given moment in an ECG electrode when the body is stretched forwards, backwards, left, or right.

-Second aspect-
12(a) to 12(c), the second embodiment includes a plurality (e.g., three) of ECG electrodes 1 to 3 for detecting electrocardiograms, and thermometers 11a and 11b for measuring deep body temperature. When it is not necessary to distinguish between the thermometers 11a and 11b, they are collectively referred to as the thermometer 11.

The case 310 contains a substrate 330 and two thermometers 11a and 11b, which measure the heat flow between the skin and the environment. One thermometer, 11a, is placed on the bottom surface of the case 310, and the other thermometer, 11b, is provided on the substrate 330 away from the thermometer 11a.

As shown in the figure, one heat flow can be measured using one thermometer pair (thermometers 11a, 11b) (SHF: Single Heat Flux). Also, two thermometer pairs can be used to measure two heat flows (DHF: Dual Heat Flux). In this case, the ECG electrodes 3 are provided on the bottom surface of the case 310 so as to be in contact with the skin.

-Third aspect-
In this third embodiment, an example of connection between the substrate 330, the thermometer 11, and the ECG electrodes 1 to 3 will be described. Fig. 13a shows a case where the ECG electrode 3 is located on the entire bottom surface of the case 310, and Fig. 13b shows a case where the ECG electrode 3, which is a bottom electrode, is divided into ECG electrode 3a and ECG electrode 3b.

In either case, the top cover 311 of the case 310 is provided with terminals 321a and 321b that are electrically connected to the ECG electrodes 1 and 2 attached to the belt 301 (or wear). When there is no need to distinguish between the terminals 321a and 321b, they are collectively referred to as terminal 321.

The terminals 321a and 321b are connected to the substrate 330 inside the case 310 by wiring L1 and L2, respectively. An insulating sheet 312 is attached to the bottom surface of the case 310 with the ECG electrodes 3 attached to the outside and the insulating sheet 312 is attached to the inside.

The first thermometer 11a is placed on the insulating sheet 312. The second thermometer 11b is placed on the substrate 330. The thermometers 11a and 11b are preferably placed in the center of the case 310. The second thermometer 11b may be placed on the underside of the substrate 330.

The first thermometer 11a can be omitted if the second thermometer 11b is given the function of a radiation thermometer so that the substrate temperature and the bottom surface temperature can be measured simultaneously.

Through holes 313 are provided in the insulating sheet 312, and the ECG electrodes 3 (bottom electrodes) are connected to the substrate 330 by wiring L3 via the through holes 313. The first thermometer 11a is also connected to the substrate 330 by appropriate wiring. Flexible printed wiring can also be used as the above wiring.

Plastic can be used as the insulating sheet 312. In order to increase sensitivity, it is preferable to make the insulating sheet 312 thin and bring the temperature of the first thermometer 11a as close to the skin temperature as possible, thereby increasing the temperature difference with the second thermometer 11b.

Furthermore, in order to reduce the thickness of the case 310, it is better for the insulating sheet 312 to be thin. If a high-strength plastic material is used, the thickness can be reduced to about 1 mm. By using the insulating sheet 312 to provide strength for the bottom of the case 310, the thickness of the bottom ECG electrode 3, which uses expensive silver, can be reduced. Sheet-shaped Ag/AgCl can be used for the bottom ECG electrode 3.

Also, the ECG electrode 3 on the bottom can be made of a metal with a high Young's modulus, such as iron, plated with Ag/AgCl. By using the ECG electrode 3 to provide strength for the bottom of the case 310, the thickness of the insulating sheet 312 can be made thinner, and the temperature of the first thermometer 11a can be made closer to the skin temperature.

In the example of FIG. 13b, when the ECG electrode 3 on the bottom surface is divided into two ECG electrodes 3a and 3b, the surface that comes into contact with the skin can be made flat by thinning both side portions 312a and 312b of the insulating sheet 312 on which the divided electrodes 3a and 3b are arranged by the thickness of the electrodes 3a and 3b.

The area directly below and around the first thermometer 11a is an important surface for measuring heat flow, and any gap between this surface and the skin will cause errors in the heat flow measurement. Naturally, contact with the skin is also important for the ECG electrodes 3a and 3b, so it is important that there are no steps on the bottom surface of the case 310.

Note that, as a relative measure, the thickness of the central portion 312c of the insulating sheet 312 between the ECG electrodes 3a and 3b may be increased by the thickness of the electrodes 3a and 3b to prevent a step from occurring on the bottom surface of the case 310.

-Fourth aspect-
In the fourth embodiment, as shown in FIG. 13c, the ECG electrodes 3 on the bottom surface are divided into four, and the insulating sheet 312 is used as a first insulating sheet, and a second insulating sheet 317 is further provided on the bottom surface of the case 310 in addition to this.

As in the third embodiment, the first thermometer 11a and the first insulating sheet 312 are placed on the bottom surface of the case 310. Four ECG electrodes 3a to 3d, which are obtained by dividing the ECG electrode 3 into four parts, are placed on the skin side (bottom side) of the first insulating sheet 312.

A dummy electrode D is placed around the first thermometer 11a so that no gap is left under the first thermometer 11a. Note that an electrically insulating dummy sheet of the same thickness may be used instead of the dummy electrode D. A second insulating sheet 317 is placed underneath it.

The second insulating sheet 317 has an area larger than the bottom surface of the case 310, and its peripheral portion extends beyond the bottom surface of the case 310. The second insulating sheet 317 is made of an insulating material such as plastic.

Conductive gels 315a-315d and non-conductive gel 316 are placed on the bottom surface of second insulating sheet 317. An opening 318 is provided in the center of second insulating sheet 317, and the divided ECG electrodes 3a-3d on the case side are electrically connected to conductive gels 315a-315d, respectively. Also, dummy electrode D and non-conductive gel 316 are in thermal contact. Second insulating sheet 317 can be disposable.

By making the conductive gels 315a to 315d and the non-conductive gel 316 sticky, the device can be attached to the skin, including the case 310. This has the advantage that it can be put on and taken off without removing clothing.

Fourth Embodiment
14(a) and (b) show an embodiment of a garment having a plurality of ECG electrode pads and heat flow pads to be worn on the upper body of a human body, in which an example of garment W provided with ECG electrode pads E, heat flow pads Th, and a signal processing circuit P is shown. Fig. 14(a) shows the front side of the garment W, and Fig. 14(b) shows the back side of the garment W.

As shown in FIG. 14(a), in the case of the human body, there are areas on the front of the body, slightly below the ribs, on the left and right, where the electrocardiogram has a prominent (high) positive potential. Furthermore, there are areas at the solar plexus and above and below it, where the electrocardiogram has a prominent (high absolute value) negative potential. By providing ECG electrode pads E in this area, the signal strength can be increased.

The ECG electrode pad placed in the part where the electrocardiogram has a prominent positive potential is designated E(+), and the ECG electrode pad placed in the part where the electrocardiogram has a prominent negative potential is designated E(-). When there is no need to distinguish between them, they are simply referred to as ECG electrode pads E.

If the ECG electrode pad E is too large, it is undesirable because it may cause a short circuit between the high potential area and the surrounding areas with relatively low potential. On the other hand, arranging many small ECG electrode pads E puts a strain on signal processing and wiring.

Because the area of the high positive potential portion is relatively small, it is preferable to place one ECG electrode pad (E+) on each side, and three ECG electrode pads (E-) in the central portion with the high negative potential, for example. This allows a large signal to be obtained, mitigating the above problems and improving the SNR by using multiple electrodes.

Referring to FIG. 14(b), the back of the body has a similar potential distribution to the front, although it is smaller. Although this will increase the number of electrodes, wiring, and signal processing, the above effect can be further improved by using this part (the back of the body) appropriately.

The heat flow pad Th is composed of a thermometer 11 that measures heat flow (described later) and a thermal insulator 432, and directs heat flow vertically to the body surface. The reason for arranging multiple heat flow pads Th is that there is a temperature difference deep inside the body (for example, the upper and lower parts of the abdomen) and there are artifacts (causes of error) in heat flow measurement, and this must be reduced.

Gaps between the skin and the heat flow pad Th can cause artifacts, so it is preferable to attach it to as flat a surface as possible. It is preferable to avoid areas such as the pectoral muscles, chest fat, female breasts, and the shoulder blades on the back, as these areas impair flatness.

From the above viewpoints, it is preferable to place the heat flow pad Th around the sternum C, which is relatively flat at the top, and below the ribs, which is relatively flat at the bottom. In the example of FIG. 14(a), the ECG electrode pad E and the heat flow pad Th are arranged so as not to overlap, but it is also possible to store the ECG electrode pad E and the heat flow pad Th in a single package, combining the functions of both.

Subcutaneous fat reduces heat flow and increases the thermal time constant. It also reduces the sensitivity of the heat flow meter and slows down its response speed. Furthermore, based on the analogy of the electric field lines of a parallel plate capacitor, the greater the distance between the parallel plates, the more the surrounding flux bends (three-dimensional effect), and the greater the distance from the depths to the heat flow meter, the more susceptible it becomes to the effects of environmental temperature.

On the front of the body, the area around the sternum C has relatively little subcutaneous fat, so for people with a thick layer of subcutaneous fat, it is a good idea to choose the area around the sternum C. Also, the back of the body often has less subcutaneous fat than the front, so you can also choose the area on the back mentioned above.

It is preferable to secure a large area for the ground electrode G, since artifacts also occur. In order not to disturb the electrocardiogram potential distribution described above, it is preferable to secure a large area on the back side, where the potential is relatively small, especially around the lower neck and around the lower back, where the potential is also small.

The signal processing circuit P is equipped with a wireless circuit for communicating with the outside world. In order to extend the communication distance, it is preferable to raise the height of the antenna above the ground. Furthermore, depending on the type of exercise, the user may need to place their back on the ground, so it is preferable for the user to have the option of choosing not to have the signal processing circuit get in the way of their exercise.

From these perspectives, it is preferable to arrange the signal processing circuit P appropriately around the neck on the back, or even around the neck on the front. The signal processing circuit P can be made thin and small thanks to technologies such as SiP (System in Package) and COP (Chip on Board), and the signal processing and battery can be arranged in separate locations and connected by wiring.

A thin laminate battery can be detachably arranged, for example, at the position shown in B in Figure 14a. This allows the thickness and area of the signal processing circuit P to be reduced. The weight and thickness around the neck can be reduced.

Next, referring to Figures 15(a) to (c), a method is described for reducing the effects of static electricity, electromagnetic induction, electrostatic coupling, and sweat, which are factors that cause artifacts, for the GND (ground) wiring 410 that is wired on the garment W to connect the ECG electrode pads E(+), E(-) and each heat flow pad Th to the signal processing circuit P.

Figure 15(c) shows the details of the GND wiring 410. According to this, the GND wiring 410 is formed by sequentially laminating a coated conductor 411 serving as a signal line, a first insulating sheet 412a, a GND conductor 413 serving as a GND electrode, and a second insulating sheet 412b on the garment fabric 401.

It is preferable that the wear fabric 401 has characteristics that make it difficult to become charged with static electricity while maintaining comfort, breathability, and a holding ability with compression. However, it is difficult to reduce static electricity to zero, so measures against static electricity are necessary. It is also necessary to reduce inductive noise from alternating current power sources such as AC 100V power sources, and static electricity noise caused by capacitive coupling between the human body and surrounding objects.

By using the coated conductor 411 as the signal line, when the wear fabric 401 absorbs sweat, it is possible to prevent short-circuiting with other wiring caused by salt in the sweat. In the heat flow pad Th, it is preferable to use two coated conductors as a stranded wire (twisted pair). By making the signal line from the thermometer (e.g., a thermistor) a twisted pair, common mode noise can be more effectively removed when the signal is processed as a differential signal in the signal processing circuit P.

The sandwich structure in which the GND conductor 413 is disposed between the first insulating sheet 412a and the second insulating sheet 412b prevents the GND conductor 413 from shorting out to other GND wiring through the clothing fabric 401 due to sweat. It is preferable that the GND conductor 413 is grounded at one point at the signal processing circuit P.

The GND conductor 413 is made wide enough to fully cover the coated conductor 411, which is the signal line. For example, a width of about 10 mm is preferable for suppressing induction noise and static electricity noise.

The coated conductor 411, GND conductor 413, and insulating sheets 412a, 412b are preferably arranged on the environment side (the side opposite the skin) of the clothing fabric 401 in order to suppress external inductive noise and electrostatic noise, but they can also be arranged on the skin side of the clothing fabric 401 depending on the required SNR and the design of the clothing.

Next, the structure of the ECG electrode pad E will be described with reference to Figures 16(a) and (b). The ECG electrode pad E has a laminate of an ECG electrode 421, a first insulating sheet 422a, and a spring material 423 on the side of the wear fabric 401 that faces the skin, and a laminate of a second insulating sheet 422b, a GND conductor 424, and a third insulating sheet 422c on the environmental side of the wear fabric 401 (the side opposite the skin).

The ECG electrodes 421 are placed so that they are in contact with the skin surface. As mentioned above, the ECG electrodes 421 can be made of Ag/AgCl, a metal mainly composed of Ag, or conductive plastic. There is a trade-off in the electrode area as mentioned above, and it is preferable that the area be around 100 to 500 square mm.

A first insulating sheet 422a is placed on top of the ECG electrode 421 so that it overhangs the ECG electrode. This is to prevent the ECG electrode 421 from shorting out with other ECG electrodes 421 via the garment W when the garment fabric 401 absorbs sweat. When sweating, it is preferable to ensure that the resistance value with other ECG electrodes is 100 kΩ or more, and the overhanging length is preferably about 5 to 10 mm.

The spring material 423 acts as a spring to keep the ECG electrode 421 in contact with the skin. The skin surface has unevenness, and the height of the peaks and valleys changes with breathing and movement. Some parts of the skin surface have a concave shape, and the convex parts around it support the garment fabric.

If the ECG electrodes 421 hanging from the garment fabric 401 are left as is, they will come off the skin and the contact pressure will change. This will cause the aforementioned chemical potential to change, which will be a major cause of artifacts.

The spring material 423 serves to press the ECG electrodes 421 against the skin with a constant pressure even when the skin surface is recessed. Functionally, it is a structure in which many small springs are arranged in a planar shape. Examples of such materials that can be used include sponge, cushion, foam material, and bubble cushioning material. Plastics such as cotton and polyurethane can also be used. It is preferable that the thickness is several mm to about 10 mm when no pressure is applied. The top and bottom relationship of the spring material 423 and the insulating sheet 422a may be reversed.

To minimize the intrusion of electrostatic noise and inductive noise into the ECG electrode 421, a GND area is provided on the opposite side of the ECG electrode 421, sandwiching the garment fabric 401. As with the GND wiring 410, a structure in which a conductor 424 serving as GND is sandwiched between two insulating sheets (second and third) 422b, 422c is arranged to protrude from the ECG electrode 421.

It is preferable that the GND conductor 424 protrudes from the periphery (four sides) of the ECG electrode 421 by about 5 to 10 mm. A GND laminate in which the GND conductor 424 is sandwiched between insulating sheets 422b and 422c can also be placed on the spring material 423 and under the garment fabric 401.

Referring to Figures 17(a) and (b), the coated conductor 411 in the GND wiring 410 is connected to the ECG electrode 421, and the ECG electrode 421 is connected to the signal processing circuit P via the coated conductor 411.

The coated conductor 411 is pulled out from the environmental side of the garment fabric 401, penetrating the garment fabric 401, the spring material 423, and the insulating sheet 422a, and is connected to the ECG electrode 421 via the contact 425.

The contact 425 may be a rivet-like metal fitting that mechanically clamps the ECG electrode 421 from above and below. Since the coated conductor 411 penetrates the insulating sheet 422a, it is preferable to provide a gasket 426 at the penetration point of the coated conductor 411 in the insulating sheet 422a so that moisture such as sweat that has soaked into the clothing fabric 401 does not reach the ECG electrode 421. The gasket 426 may be made of adhesive or a ring-shaped plastic.

Next, the heat flow pad Th will be described with reference to Figures 18(a) and (b). A first insulating sheet 431a is placed so as to contact the skin surface, and a heat insulating material 432 equipped with thermometers 11a and 11b is placed on top of the first insulating sheet 431a.

One thermometer, 11a, is placed on the underside (skin side) of the insulating material 432, and the other thermometer, 11b, is placed on the environmental side (opposite the skin side) of the insulating material 432. It is preferable that both thermometers 11a and 11b are placed in overlapping positions in the center of the insulating material 432. The insulating sheet 431a prevents sweat from the skin from reaching the thermometer 11a on the underside.

The thermometers 11a and 11b measure the heat flow from the skin to the environment. The temperature difference generated by multiplying the thermal resistance of the insulation 432 by the heat flow is detected by the thermometers 11a and 11b, and the heat flow can be calculated (heat flow meter using SHF).

Normally, skin temperature is around 30°C, but when a gap is created, heat exchange with the ambient temperature occurs instantly, and even when contact is restored, it takes tens to hundreds of seconds for the temperature to return to thermal equilibrium. During this time, deep temperature measurements will be subject to large errors.

A spring material 433 is disposed between the garment fabric 401 and the insulating material 432. Similar to the spring material 423 in the ECG electrode pad E, the spring material 433 has the effect of preventing loss of contact between the skin and the heat flow measurement structure (insulating material + thermometer).

In addition, by placing a GND laminate consisting of a GND conductor 434 sandwiched between second and third insulating sheets 431b, 431c on the environmental side of the garment fabric 401, it is possible to suppress the inclusion of noise that is superimposed on the thermometer signal (resistance change in the case of a thermistor).

Next, the wiring connection to the thermometers 11a and 11b will be described with reference to Figures 19(a) and (b). Note that the first insulating sheet 431a, the heat insulating material 432 having the thermometers 11a and 11b, and the spring material 433 are omitted from Figure 19(b).

The coated conductor 411 of the GND wiring 410 that has passed from the signal processing circuit P through the environment side of the wear fabric 401 is passed to the skin side to connect to the thermometers 11a and 11b. The wiring is a twisted pair wiring 411a for the thermometer 11a and a twisted pair wiring 411b for the thermometer 11b, which are connected to the thermometer 11a on the lower side of the insulating material 432 and the thermometer 11b on the upper side of the insulating material 432, respectively. Note that it is also possible to connect one terminal of the thermometer 11a to one terminal of the thermometer 11b, making the wiring three-wire. It is also possible to use a three-wire stranded structure.

Next, an example of the configuration of a heat flow pad Th using the DHF method that has two sets of heat flow meters will be described with reference to Figures 20(a) and (b). In this case, four thermometers 11a to 11d and two heat insulating materials 432a and 432b are used. The rest of the configuration may be the same as the heat flow pad described in Figure 18.

Of these, thermometers 11a and 11b are arranged on the underside and upper side of one of the insulating materials 432a as a thermometer pair TP1, and thermometers 11c and 11d are arranged on the underside and upper side of the other insulating material 432b as a thermometer pair TP2.

In this way, by providing two pairs of thermometers TP1 and TP2, two heat flows can be measured. By changing the thermal resistance of the insulating materials 432a and 432b, heat flows of different magnitudes can be created, and the unknown thermal resistance under the skin can be calculated to calculate the core body temperature.

The wiring connections to the thermometers 11a to 11d will be described with reference to Figures 21(a) and (b). Note that the first insulating sheet 431a, the heat insulating material 432a with the thermometers 11a and 11b, the heat insulating material 432b with the thermometers 11c and 11d, and the spring material 433 are not shown in Figure 21(b).

The thermometers 11a to 11d are each connected to the signal processing circuit P via two sets of wiring 410A made of coated conductors 411 directed to the thermometer pair TP1, and two sets of wiring 410B made of coated conductors 411 directed to the thermometer pair TP2.

If one terminal of each thermometer is combined into one wire, a total of five wires can be used. Also, the number of wires can be reduced by providing each heat flow pad with a semiconductor switch, for example, to selectively switch the thermometers.

Next, referring to Figures 22(a) to (c), we will explain the insulation material suitable for the heat flow pad for DHF described in Figures 20 and 21.

First, in the example of FIG. 22(a), a number of air bubbles (in this example, hollow cylindrical air bags) B are arranged as air bubble cushioning material on a bottom film 441 of a size (area) equivalent to the combined size of the insulating materials 432a and 432b, and the gas enclosed in the air bubbles Ba in one insulating material 432a and the gas enclosed in the air bubbles Bb in the other insulating material 432b are different.

Furthermore, the top film 442 and the side film 443 are bonded together to form an airtight gas mat, and the inside of the gas mat is divided by a partition film 444 into a gas chamber Ra on the insulating material 432a side and a gas chamber Rb on the insulating material 432b side, and a different gas is sealed in each of them.

The gas sealed in the bubble Ba on the insulating material 432a side and the gas chamber Ra may be the same gas, and the gas sealed in the bubble Bb on the insulating material 432b side and the gas chamber Rb may be the same gas.

By changing the thermal conductivity of the gas sealed in the bubble Ba and gas chamber Ra on the insulating material 432a side and the gas sealed in the bubble Bb and gas chamber Rb on the insulating material 432b side, it is possible to construct insulating materials 432a and 432b with two types of thermal resistance.

Gases with high thermal conductivity (each value is in W/mK) can be hydrogen at 0.3, helium at 0.14, etc. In contrast, gases with low thermal conductivity (each value is in W/mK) can be air at 0.026, argon at 0.016, and carbon dioxide at 0.015.

The overall area of the insulating material 432, including the insulating materials 432a and 432b, is determined so as to reduce the three-dimensional effect caused by the subcutaneous thickness described above. The shape of the insulating material 432 may be cylindrical. For example, the length and width in plan view should be approximately several tens of mm (several tens of mm in diameter), preferably 30 mm or more. The subcutaneous thickness is usually 5 mm or more, which has the effect of reducing the effects of changes in environmental temperature.

The thickness of the insulating material 432 should be several mm, preferably 2 mm or more. By making it thicker, the sensitivity of the heat flow measurement increases, but the thickness of the garment increases, which reduces comfort and heat dissipation. Also, making it too thin is not preferable because there is a risk of the bottom and top of the foam coming into contact when the body surface is displaced, and fluctuations in thickness are more likely to appear as errors.

As mentioned above, if a gap occurs between the heat flow pad Th and the skin, it will result in a large error. The foam B deforms to fit the contours of the body. It is pressed against the surface of the body by the compression of the spring material or clothing, preventing gaps from occurring even during exercise. Even if the shape of the foam B changes microscopically, it is sufficient as long as the change in thermal resistance of the insulating material 432 as a whole is within the allowable error range. Gas has a lower thermal conductivity than liquids and solids, which increases the sensitivity of the heat flow meter.

A similar structure may be one in which adjacent bubbles B are separated by a partition, as shown in FIG. 22(b). It is also possible to enclose a cluster of small beads, rather than gas, inside the bubbles B or the film.

As shown in FIG. 22(c), synthetic resin sheets 451, 452 with different thermal resistances may be used as the insulating material 432. The synthetic resin is preferably a material that can change shape, such as a rubber material or a foam material. By changing the composition, additives, and foaming ratio of these materials, it is possible to provide two types of thermal resistance.

Fifth Embodiment
-First aspect-
The fifth embodiment is an embodiment for quantifying and correcting temperature artifacts, and the first aspect has, as its basic configuration, a heat flow measuring unit 500, a deep temperature calculating unit 510, a noise calculating unit 520, and a deep temperature output unit 540, as shown in Figure 23.

The heat flow measurement unit 500 measures one or more heat flows flowing from the living body toward the surrounding environment using the SHF method, DHF method, or the like. The deep temperature calculation unit 510 calculates the deep temperature using the heat flow, subcutaneous thermal resistance, and the temperature of the epidermis (body surface). The noise calculation unit 520 calculates the noise superimposed on the deep temperature. The deep temperature output unit 540 processes the deep temperature according to the calculated noise and outputs the result.

-Second aspect-
Preferably, as a second aspect, as shown in FIG. 24, a noise comparison section 530 is further provided after the noise calculation section 520, and data of the noise period calculated by the noise calculation section 520 is removed.

In this example, the heat flow measurement unit 500 employs the DHF method, and as explained above in FIG. 20(b), it is equipped with two pairs of thermometers TP1 and TP2, and as shown in FIG. 25, the thermometer pairs TP1 and TP2 measure two heat flows Ith1 and Ith2, and the skin surface side thermometers included in the thermometer pairs TP1 and TP2 measure skin temperatures Tsk1 and Tsk2.

The deep body temperature calculation unit 510 calculates Rthbody, which is the internal thermal resistance from the deep tissues of the living body to the body surface, from (Tsk2-Tsk1)/(Ith1-Ith2), and then obtains the time series of the deep body temperature Tc of the living body from (Ith1 x Rthbody + Tsk1) or (Ith2 x Rthbody + Tsk2).

The noise calculation unit 520 has a high-pass filter (HPF) and receives the deep temperature Tc as input and outputs components above a specific frequency. As the high-pass filter, for example, a first-order filter can be set, and a frequency of about 0.001 Hz can be set as the specific frequency (cutoff frequency).

This allows temperature fluctuation components with periods of 160 seconds or less to pass through as calculated noise with almost no attenuation. The amount of noise can be output as the moving average of the squares, in other words, the effective value.

In addition, temperature noise may appear in certain body postures. For example, when lying on one's back facing left with the heat flow pad Th as a heat flow meter attached to the left side of the abdomen, if the tension of the belt fixing the heat flow pad Th loosens or the body surface is displaced, a gap appears between the skin and the heat flow pad Th, causing large temperature noise, as observed by the inventors.

Changes in body posture can be detected by an acceleration sensor. Since temperature noise occurs later than posture changes, once a specific posture is detected, it is possible to generate pseudo-temperature noise that is easy to remove and output it as calculated noise.

The noise comparison unit 530 compares the temperature noise calculated by the noise calculation unit 520 with a predetermined threshold value. The threshold value can be determined based on the required temperature error, but can be set to a value of, for example, several tens to several hundreds of m°C.

The deep temperature output unit 540 outputs the deep temperature Tc if, for example, the effective noise value is less than the threshold value. In other words, the output of the deep temperature Tc is stopped during the period when the calculated noise is equal to or greater than the threshold value. The portion where the output is stopped is blank, but by displaying, for example, as a user interface (UI) by connecting the points just before and just after the blank, it is possible to prevent the user, or other programs or systems, from making erroneous judgments due to noise.

Note that an error will occur in the line connecting the threshold values, so the effect of the error can be reduced by masking the noise generation period by extending it forward and backward.

-Third aspect-
In the third aspect, as shown in FIG. 26 , the heat flow measuring unit 500 includes a plurality of heat flow measuring units 500a, 500b, ..., a deep temperature calculation unit 510, a noise calculation unit 520, a noise/measurement unit comparison unit 531, a noise/threshold comparison unit 532, and a deep temperature output unit 540.

The deep temperature calculation unit 510, the noise calculation unit 520, and the deep temperature output unit 540 may be the same as those in the first and second aspects. In addition, the heat flow measurement units 500a, 500b, ... employ the DHF method.

Referring to the operation explanation diagram in Figure 27, there may be multiple deep temperature calculation units 510, but here, each heat flow measurement unit 500a, 500b, ... is connected to one calculation circuit by sequential switching, and as described above, the deep temperatures Tca, Tcb, ... of each part are calculated based on the heat flow from each heat flow measurement unit 500a, 500b, ... and data such as skin temperature.

The noise calculation unit 520 calculates the effective temperature noise values Nca, Ncb, ... for each deep temperature, as in the second aspect above.

The noise/measurement unit comparison unit 531 compares the temperature noise effective values between each heat flow measurement unit and selects the smallest temperature noise effective value.

The noise/threshold comparison unit 532 compares the minimum temperature noise effective value with a predetermined threshold value.

If the minimum temperature noise effective value is less than the threshold value, the deep temperature output unit 540 outputs the deep temperature Tc extracted by the measurement unit. If the minimum temperature noise effective value is equal to or greater than the threshold value, it outputs a blank.

As with the synthesis of ECG signals, the effects of artifacts can be reduced by synthesizing signals from multiple heat flow measurement units according to the SNR. There may be an offset in the deep temperature between multiple heat flow pads. In that case, the offset can be recorded when there is little noise and corrected when synthesizing.

-Fourth aspect-
28 and 29, in the fourth embodiment, a plurality of deep temperature calculation values are weighted and synthesized according to noise. Therefore, a weight synthesis unit 533 is used instead of the noise/inter-measurement unit comparison unit 531 and the noise/threshold comparison unit 532 in the third embodiment. The heat flow measurement units 500a, 500b, ..., the deep temperature calculation unit 510, and the noise calculation unit 530 perform the same processing as in the third embodiment.

The weight synthesis unit 533 multiplies the deep temperatures Tca, Tcb, ... measured by each heat flow measurement unit 500a, 500b, ... by the reciprocal of the calculated noise, adds up the multiplication results of each heat flow measurement unit, and then divides by the sum of the reciprocals of each heat flow measurement unit. The deep temperature output unit 540 outputs the above calculation result as the synthesized Tc. Tc is expressed by the following formula.
Tc=(1/Nca×Tca+1/Ncb×Tcb)/(1/Nca+1/Ncb)

As a result, the measurement values of each heat flow measurement unit 500 (500a, 500b, ...) are evaluated and combined if the noise is small, allowing the results of each heat flow measurement unit to be used effectively. The combined SNR is approximately the sum of the SNRs of each heat flow measurement unit.

-Fifth aspect-
As shown in FIG. 30, the fifth aspect has, as its basic configuration, a heat flow measuring unit 500, an environmental temperature measuring unit 501, a false signal generating unit 502, a deep temperature calculating unit 503 and a deep temperature output unit 540, and corrects the false temperature signal.

The heat flow measurement unit 500 measures one or more heat flows flowing from the living body toward the surrounding environment using the SHF method or the DHF method. The environmental temperature measurement unit 501 measures the environmental temperature. The false signal generation unit 502 creates a false signal from the environmental temperature. The deep temperature calculation unit 503 calculates the deep temperature using the heat flow, subcutaneous thermal resistance, and epidermal temperature. The deep temperature output unit 540 processes the deep temperature according to the generated false signal and calculates and outputs a new deep temperature.

-Sixth aspect-
As shown in FIG. 31, the sixth embodiment further comprises a pseudo signal remover 504 in addition to the components of the fifth embodiment, and its operation will be described with reference to the schematic diagram of FIG.

The heat flow measurement by the heat flow measuring unit 500 measures the heat flow Ith that flows from the living body toward the environment, and is therefore affected by changes in environmental temperature. The three-dimensional effect due to the subcutaneous thickness mentioned above is one example. This can cause large false signals in the deep temperature Tc.

The environmental temperature measuring unit 501 is a thermometer that measures the environmental temperature, and can be installed, for example, at the top of the case 310 described in the previous embodiment, or on the surface of the belt 301 or the garment W. It is also possible to use the temperature acquired at a location close to the environmental side of the heat flow measuring unit 500, but since a delay occurs if there is a thermal time constant, it is preferable to acquire the temperature at a location with a small thermal time constant.

The fake signal generator 502 acquires the relationship between the environmental temperature and the fake signal in advance, and creates a fake signal from the current environmental temperature according to the acquired results. The fake signal has a certain time response with the environmental temperature as input. This relational expression is acquired in advance by experiments, etc.

For example, we consider the hot plate to represent the deep tissue and the silicone rubber to represent the subcutaneous tissue, and conduct an experiment in which the temperature of the hot plate is kept constant while the environmental temperature is changed. Since the thicker the silicone rubber, the larger the false signal generated, we obtain the time response transfer coefficient of the false signal while changing the thickness of the silicone rubber.

Changes in the environmental temperature cause spurious signals to appear in the raw deep temperature calculation results, but the spurious signal removal unit 504 removes the spurious signals by subtracting them from the raw deep temperature calculation results. The deep temperature output unit 540 outputs the above processing results to a user or administrator, or to other programs or systems.

Sixth Embodiment
In the sixth embodiment, the measurement function is diagnosed from noise measurement of each unit. As a basic configuration for this purpose, as shown in Fig. 33, the device includes a vital sign measurement unit 610, an environment measurement unit 620, a noise diagnosis unit 630, and an output unit 640.

The biosignal measurement unit 610 measures electrical signals generated by the living body, as well as signals obtained by attaching the device to the living body according to the state of the living body. The environment measurement unit 620 measures the state of the environment surrounding the living body. The noise diagnosis unit 630 acquires noise contained in each measurement unit and diagnoses the state of the measurement function. The output unit 640 outputs the diagnosis results.

Next, a specific configuration will be described with reference to FIG. 34. The biosignal measurement unit 610 includes an ECG measurement unit 611, a heat flow measurement unit 612, and an acceleration measurement unit 613, which respectively measure the ECG, heat flow, and acceleration emitted by the living body. Meanwhile, the environment measurement unit 620 measures the temperature, humidity, atmospheric pressure, wind speed, atmospheric composition, latitude, longitude, etc. surrounding the living body.

Note that the ECG measurement unit 611 and the heat flow measurement unit 612 are not unique to the sixth embodiment, and the ECG measurement unit 611 may have the same configuration as the ECG measurement unit 110 described above, and the heat flow measurement unit 612 may have the same configuration as the heat flow measurement unit 500 described above.

The noise diagnosis unit 630 includes an ECG/SNR comparison unit 631, a deep temperature/noise comparison unit 632, a movement determination unit 633, an environment determination unit 634, and a threshold generation unit 635.

The ECG/SNR comparison unit 631 compares the SNR of the ECG electrodes with the threshold generated by the threshold generation unit 635. As mentioned above, the ECG electrodes are preferably made of Ag/AgCl or conductive plastic, but as measurements are repeated, they react with precipitates from the body, oxidize, or become physically damaged. In addition, the belt 301 and clothing W that support the ECG electrodes absorb sweat, and the insulating sheet deteriorates or becomes damaged.

These can result in increased noise even when stationary, or signal degradation resulting in poor SNR. Furthermore, motion artifacts can be larger and more frequent than usual.

Substances deposited by the living body can be removed to some extent by washing the ECG electrodes, belts, and clothing, and their original functions may be restored. Even with heat flow measurements, the insulation may deform or become physically damaged over time.

Also, if the calibrated state at the time of shipment is changed, errors may occur even with the same calibration parameters. Also, if sweat gets into the signal wiring from the thermometer to the signal processing circuit, an error may occur in the thermometer's indication. If the GND pattern is broken, the desired shielding effect cannot be obtained, and noise may increase.

Other causes of noise and SNR degradation can be improper wearing technique. Differences in body size and belt or garment size can result in the ECG electrodes and heat flow pads not making contact with the skin or making incomplete contact.

In particular, if the device is worn away from the heart, such as on a belt, the ECG signal can become extremely small. In addition, the area around the shoulder blades undulates significantly with arm movement, and movement can cause the ECG electrodes or heat flow pads to lose contact with the skin. The chest is affected by myoelectric signals from the pectoral muscles and unevenness caused by fat and breasts. Twists or wrinkles in the belt or clothing can also cause noise and degradation of the SNR.

Measurements may also be unstable at the beginning. Depending on the material of the ECG electrodes, deposits from the skin may reduce the chemical potential between the skin and the ECG electrodes, and the SNR may remain low until deposits accumulate. This may be improved by first applying water to the ECG electrodes.

When measuring heat flow, depending on the environmental temperature, large fluctuations such as overshoot and undershoot may occur when the device is attached.

Taking these factors into consideration, it is preferable to appropriately set the threshold generating unit 635 to a first threshold value "threshold value at the start of wearing," a second threshold value "threshold value when stationary," a third threshold value "threshold value during exercise," a fourth threshold value "threshold value due to high or low environmental temperature," etc.

For example, when exercising, SNR and noise degradation are likely to occur in the ECG and heat flow in the first place. The threshold value needs to be raised compared to when stationary, but it is also necessary to detect SNR and noise degradation caused by deterioration of components such as ECG electrodes, heat flow pads, belts, and clothing, which only appears during exercise. By detecting the amount of exercise while observing heart rate, deep temperature, acceleration, environmental temperature, etc., and changing the threshold value as appropriate, it is possible to determine whether the device needs to be reattached, cleaned, or replaced.

When stationary, the SNR and noise are relatively stable. By recording the SNR and noise when stationary each time a measurement is made and observing the changes during daily use, it is also possible to judge the deterioration of the aforementioned components over time. If cleaning is encouraged at a certain frequency and the deterioration continues without recovery even after cleaning, observing such changes over time will enable accurate judgment of the deterioration that requires replacement.

Since the SNR and noise deteriorate at startup and when the environmental temperature changes, it is advisable to determine the first through fourth thresholds mentioned above by combining the values of the ECG, heat flow pad, acceleration sensor, and environmental sensor.

The output unit 640 notifies the user or administrator, instructing them to re-install the part or encouraging them to clean or replace it. Regarding part replacement, a notification can be sent to the customer support department to encourage the support department to take action.

Furthermore, if there are problems such as frequent reattachment, customer support can be encouraged to provide assistance, such as checking the size or applying water to the ECG electrodes. These tasks may be performed by the user's administrator.

By looking at the operation history of each sensor unit, such as the ECG measurement unit 611 and heat flow measurement unit 612, particularly the history when stationary, it is possible to determine and predict changes over time, as described above, and this information can be included in notifications to users, administrators, and customer support.

1, 2, ... ECG electrode 11 (11a, 11b) Thermometer 110, 210 ECG measurement unit 120, 220 R-wave extraction unit 121 Signal calculation unit 122 Wave removal unit 130, 230 SNR calculation unit 131 SNR comparison unit 140, 240 RRI output unit 150, 250 Frequency analysis unit 231 SNR/electrode comparison unit 232 SNR/threshold comparison unit 233 Weight synthesis unit 234 R-wave re-extraction unit 301 Belt 310 Case 330 Substrate 401 Clothing fabric 410 GND wiring 411 Coated conductor (signal line)
421 ECG electrode 423 Spring material 432 Thermal insulation material 500 (500a, 500b, ...) Heat flow measurement unit 510 Deep temperature calculation unit 520 Noise calculation unit 530 Noise comparison unit 531 Noise/measurement unit comparison unit 532 Noise/threshold comparison unit 533 Weight synthesis unit 540 Deep temperature output unit

Claims (2)

a heat flow measuring unit for measuring a heat flow between the inside of the living body and the environment;
A deep temperature calculation unit that calculates a deep temperature of the living body based on the temperature information from the heat flow measurement unit;
A noise calculation unit that calculates noise contained in the deep temperature;
a noise comparator for comparing the noise with a predetermined threshold;
a deep temperature output unit that outputs the deep temperature in response to a comparison result of the noise comparison unit ;
The noise calculation unit obtains the noise from the output of a high-pass filter to which the deep temperature is input, and calculates and outputs an effective noise value using a moving average of squares;
The noise comparison unit compares the effective noise value output from the noise calculation unit with a predetermined threshold value,
The deep temperature output unit outputs the deep temperature when the RMS noise value is less than the threshold value, and stops outputting the deep temperature when the RMS noise value is equal to or greater than the threshold value . a heat flow measuring unit for measuring a heat flow between the inside of the living body and the environment;
An environmental temperature measuring unit for measuring an environmental temperature;
a false signal generating unit that generates a false signal from the environmental temperature measured by the environmental temperature measuring unit based on a false signal of a deep body temperature measured in advance in an experiment in which the environmental temperature is changed ;
A deep temperature calculation unit that calculates a deep temperature of a living body from the heat flow measured by the heat flow measurement unit;
an output unit that calculates and outputs a new deep temperature from the false signal generated by the false signal generating unit and the deep temperature calculated by the deep temperature calculating unit;
A biological signal processing device comprising:

Images