GB2552455A

GB2552455A - Blood monitoring

Info

Publication number: GB2552455A
Application number: GB1610539.7A
Authority: GB
Inventors: Clemson Philip
Original assignee: Digital and Future Technologies Ltd
Current assignee: Digital and Future Technologies Ltd
Priority date: 2016-06-16
Filing date: 2016-06-16
Publication date: 2018-01-31
Anticipated expiration: 2036-06-16
Also published as: GB2552455B8; GB201610539D0; GB2552455B

Abstract

Apparatus for calculating blood pressure, comprising: a sensor unit comprising at least one sensor, such as an ECG (electrocardiogram) or PPG (photoplethysmographic) sensor, configured to obtain a set of one or more signals indicative of heart activity. The one or more signals comprising a blood pressure waveform represent variation in blood pressure over time in response to the heart activity. A processor coupled to the sensor unit is configured to analyse the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; select a reference feature from the set of one or more signals having a known relationship to the reflection feature; derive a time interval between the selected reference feature and the reflection feature and calculate a measurement of blood pressure using the time interval. The reference feature may be indicative of a systolic wave, dicrotic wave or be the R-peak of an ECG signal.

Description

(54) Title of the Invention: Blood monitoring

Abstract Title: Measuring blood pressure through reflected pressure waves (57) Apparatus for calculating blood pressure, comprising: a sensor unit comprising at least one sensor, such as an ECG (electrocardiogram) or PPG (photoplethysmographic) sensor, configured to obtain a set of one or more signals indicative of heart activity. The one or more signals comprising a blood pressure waveform represent variation in blood pressure over time in response to the heart activity. A processor coupled to the sensor unit is configured to analyse the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; select a reference feature from the set of one or more signals having a known relationship to the reflection feature; derive a time interval between the selected reference feature and the reflection feature and calculate a measurement of blood pressure using the time interval. The reference feature may be indicative of a systolic wave, dicrotic wave or be the R-peak of an ECG signal.

1/9

2/9

\/

3/9

ο

4/9 vcLujus |er sih cyparo.A\ato^

3oj

Sansa r urn b ©£>leu\s sib op^{- 0Λί}- ^ό!~~ OwO^ S>_Qft0U ln<U«E^ V bU^U press lira, UciM_uf-Orv^

30$

Lotobb reJLckorv $to&re rndUcolwsb kJ-UcWL prcas/Ure, l*o.vuu

3OS

S-al&ob r^-titAUx |Ao£urt ka^*^ knou^ balatioftskip io ra/tkJwv UhL·

J^zOUNb L<w Wizvwtz Ub Jjj^lr-' SjrfjuA-JldL·	¹' —
-^xotorc Ο·ί5υ	(ocofxi, rejCtciu?^ ^jtebur«x

CdcuJofcc /VjbCixucufKzi «^-- kfooi. jOr-esX-une. u_s;^ iaUtvoL	—>

3Oj

311

OoUpwb volut/ k|®O(t fUosx^b/MAfc

3/J

P| gu^e/

5/9

6/9

-ο

-NnsS'W

C?r

7/9

8/9

to c

9/9

GJ

$ o

BLOOD MONITORING

BACKGROUND

This disclosure relates to a blood monitoring system. In particular, aspects of the present disclosure relate to an apparatus for obtaining a measurement of a user’s blood pressure.

The pressure of a person’s blood is not constant in time but instead takes the form of a temporal series of pressure pulses caused by respective contractions, or beats, of the heart. The pressure pulse is a wave that propagates through the arteries of the person’s arterial tree at a pulse wave velocity (PWV). The contour, or shape of the pulse wave may be referred to as the pressure pulse waveform. A typical example of a pressure pulse waveform 100 is shown in figure 1A. The waveform is shown comprising a systolic peak, a dicrotic peak and a dircotic notch located between the systolic and dicrotic peaks. The systolic peak is the peak of a systolic wave forming part of the waveform, and the dicrotic peak is the peak of a dicrotic wave forming part of the waveform. Depending on where the pulse wave is measured on a user’s body, the pressure waveform 100 may include reflections of the pressure pulse. The reflections may be generated by changes in the peripheral vascular resistance at positions where the blood vessels branch from larger arteries to smaller arterioles and capillaries. Reflected waves are often smaller than the pressure pulses (as the incident waves are not fully reflected) and may not always be directly visible in the waveform.

A patient’s blood pressure (either systolic, diastolic, or both) is often used as an indication of a patient’s cardiovascular health. High blood pressure may increase a patient’s risk of heart attack, heart failure or other severe conditions such as kidney disease and stroke. Measurement of a person’s blood pressure is therefore prevalent within medical communities.

One way to measure a patient’s blood pressure is through the use of a blood pressure cuff. Typically, the cuff comprises a flexible bladder connected to a pump. The cuff also comprises a pressure gauge for measuring the pressure in the bladder. To measure a person’s blood pressure, the cuff is usually placed around the patient’s arm so that it sits directly over an artery above the crook of the elbow. The tester then inflates the bladder using the pump so that the cuff cuts off blood flow through the arm. Air from the bladder is then slowly released, allowing blood to resume flow through the arm (the flow of blood can be detected through pulsations in the pressure reading of the gauge). The pressure at which blood resumes flow through the arm is measured as the systolic pressure. As the pressure within the bladder continues to decrease, the measured pulses become increasingly faint until they are no longer detectable. The pressure reading of the gauge when the pulses are no longer detectable is measured as the diastolic pressure.

Though pressure cuffs are capable of providing accurate blood pressure readings, they suffer from the drawback of requiring a fairly long set-up time. In addition, it is often not possible for a user to measure their blood pressure unassisted, meaning the use of the blood pressure cuff within a clinical environment may require dedicated personal, decreasing the efficiency of the process. A corollary of measuring the blood pressure within a clinical environment is the greater likelihood that the person feels under additional stress, which may in turn leads to a greater number of incorrect readings of resting blood pressure. Cuffs may also be uncomfortable for the user.

SUMMARY

According to a first aspect of the present disclosure there is provided an apparatus for calculating blood pressure, comprising: a sensor unit comprising at least one sensor configured to obtain a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity; and a processor coupled to the sensor unit and configured to: analyse the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; select a reference feature from the set of one or more signals having a known relationship to the reflection feature; derive a time interval between the selected reference feature and the reflection feature; and calculate a measurement of blood pressure using the time interval.

The reference feature may be a feature of the blood pressure waveform.

The reference feature may be indicative of an incident pressure wave and the reflection feature is indicative of a corresponding reflected pressure wave.

The reference and reflection features may be indicative of either: i) a systolic wave and systolic wave reflection respectively; or ii) a dicrotic wave and dicrotic wave reflection respectively.

The reference and reflection features may be indicative of the systolic wave and systolic wave reflection respectively and the processor may be configured to use the time interval between the reference and reflection features to calculate a measurement of systolic blood pressure.

The reference and reflection features may be indicative of the dicrotic wave and dicrotic wave reflection respectively and the processor may be configured to use the time interval between the reference and reflection features to calculate a measurement of diastolic blood pressure.

The processor may be configured to analyse the blood pressure waveform to locate first and second reflection features indicative of first and second reflected pressure waves; select first and second reference features from the blood pressure waveform having known relationships to the first and second reflection features respectively; and derive a first time interval between the first reference feature and first reflection feature; and a second time interval between the second reference feature and second reflection feature.

The first reference feature may be indicative of an incident pressure wave and the first reflection feature may be indicative of a corresponding reflected pressure wave; and the second reference feature may be indicative of a second incident pressure wave and the second reflection feature may be indicative of a corresponding reflected pressure wave.

The processor may be configured to use the first interval to calculate a first measurement of blood pressure and the second time interval to calculate a second measurement of blood pressure.

The first reference and reflection features may be indicative of a systolic wave and systolic wave reflection respectively and the second reference and reflection features may be indicative of a dicrotic wave and dicrotic wave reflection respectively; the processor may be configured to calculate a measurement of systolic pressure from the first time interval and to calculate a measurement of diastolic pressure from the second time interval.

The reference feature and reflection feature may be a maxima in the blood pressure waveform and/or the time derivative of the blood pressure waveform.

The sensor unit may comprise at least one sensor configured to obtain an electrical signal representing electrical activity of the heart and the processor may be configured to select the reference feature from the electrical signal.

The at least one sensor is an electrocardiograph (ECG) sensor configured to obtain an ECG signal and the processor may be configured to select as the reference feature the R-peak of the ECG signal.

The reflection feature may be indicative of a reflected systolic wave or a reflected dicrotic wave.

The reflection feature may be indicative of a reflected systolic wave and the processor may be configured to calculate a measurement of systolic blood pressure using the time interval between the selected reference feature and the reflection feature.

The reflection feature may be indicative of a reflected dicrotic wave and the processor may be configured to calculate a measurement of diastolic blood pressure using the time interval between the selected reference feature and the reflection feature.

The processor may be configured to locate a first reflection feature indicative of a first reflected pressure wave of the blood pressure waveform and a second reflection feature indicative of a second reflected pressure wave of the blood pressure waveform; and to derive a first time interval between the selected reference feature of the electrical signal and the first reflection feature, and to derive a second time interval between the selected reference feature of the electrical signal and the second reflection feature.

The processor may be configured to use the first interval to calculate a first measurement of blood pressure and the second interval to calculate a second measurement of blood pressure.

The processor may be further configured to analyse the blood pressure waveform to locate an incident feature indicative of an incident pressure wave and to derive a time interval between the selected reference feature of the electrical signal and the incident feature.

The processor may be configured to calculate the measurement of blood pressure using the time interval between the reference feature of the electrical signal and the reflection feature and the time interval between the reference feature of the electrical signal and the incident feature.

The incident feature may be indicative of a systolic wave and the reflection feature may be indicative of a systolic wave reflection.

The sensor unit may be configured to obtain said blood pressure waveform by taking a plurality of singular blood pressure waveform readings and averaging those waveform readings.

The sensor unit may further comprise at least one electrocardiograph (ECG) sensor configured to obtain an ECG signal and synchronise the waveform readings with Rwaves of the ECG signal to thereby use the R-wave as a reference event for averaging the waveform readings.

The processor may be configured to locate reflection features that appear as inflection points in the blood pressure waveform by calculating the first derivative of the blood pressure waveform and to locate the reflection features from extrema within the profile of the first derivative of the blood pressure waveform.

The processor may be configured to locate the reflection feature in the pressure waveform by performing an extremum-detection algorithm comprising the steps of:

i) searching the blood pressure waveform for an extremum indicative of a reflected pressure wave;

ii) calculating the first derivative of the blood pressure waveform if an extremum indicative of the reflected pressure wave is not found;

iii) searching the first derivative of the blood pressure waveform for an extremum indicative of a reflected pressure wave; and iv) locating the reflection feature from identified extrema.

The processor may be configured to determine as the location of the reflection feature the location of an identified extremum.

The at least one sensor configured to obtain the blood pressure waveform may comprise at least one of: a PPG sensor; an acoustic sensor; an arterial catheter.

According to a second aspect of the present disclosure there is provided a method of obtaining a measurement of a user’s blood pressure, comprising: obtaining a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity; analysing the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; selecting a reference feature from the set of one or more signals having a known relationship to the reflection feature; deriving a time interval between the selected reference feature and the reflection feature; and calculating a measurement of blood pressure using the time interval.

According to a third aspect of the present disclosure there is provided a method of calculating blood pressure, comprising: receiving a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity; analysing the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; selecting a reference feature from the set of one or more signals having a known relationship to the reflection feature; deriving a time interval between the reflection feature and the selected reference feature; and calculating a measurement of blood pressure using the time interval.

According to a fourth aspect of the present disclosure there is provided an apparatus for calculating blood pressure, comprising an input for receiving a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity; and a processor configured to: analyse the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; select a reference feature from the set of one or more signals having a known relationship to the reflection feature; derive a time interval between the selected reference feature and the reflection feature; and calculate a measurement of blood pressure using the time interval.

BRIEF DESCRIPTION OF DRAWINGS

The present invention will now be described by way of example with reference to the accompanying drawings. In the drawings:

Figure 1A shows an example of a blood pressure waveform.

Figure 1B shows a series of example blood pressure waveforms and their component waves for differing patient conditions.

Figure 2 shows a schematic illustration of an apparatus for calculating a measure of a user’s blood pressure.

Figure 3 shows a flowchart illustrating a series of steps for obtaining a measurement of a user’s blood pressure.

Figure 4 is a schematic illustration of an ECG signal and blood pressure waveform for a series of cardiac cycles.

Figure 5A shows an example of an obtained blood pressure waveform and the component incident and reflected waves.

Figure 5B shows a first-order derivative of the blood pressure waveform shown in figure 5A.

Figure 6 shows an example of a device that is configured to calculate a value of blood pressure from a set of one or more inputs.

Figure 7 shows an example of an apparatus for calculating a value of blood pressure.

DETAILED DESCRIPTION

Figure 5A shows an illustrative example of a user’s blood pressure waveform 501. The component waves of the waveform are shown beneath the waveform 501 and indicated by the solid and dashed lines. The waveform 501 is formed of a plurality of component waves, including a systolic wave 503 and a dicrotic wave 505. Depending on where the pressure waveform is measured, it may additionally comprise one or more reflected pressure waves. These reflected pressure waves may include a reflected systolic wave 507 and/or reflected dicrotic wave 509. As described above, reflected pressure waves may arise when there is a change in the peripheral vascular stiffness and so may only be visible for certain measurement sites on the user’s anatomy, e.g. in the hands and/or wrists.

Thus, in general, the pressure waveform may comprise, or be composed of, a set of incident pressure waves (the systolic wave and dicrotic wave) and a set of one or reflected pressure waves (the systolic wave reflection and/or dicrotic wave reflection). The pressure waveform 501 may be formed from the superposition of these component waves.

One aspect of the present disclosure is directed to analysing a pressure waveform to locate a feature indicative of a reflected pressure wave and to use the temporal relationship between that feature and a reference feature to calculate a value of blood pressure. In one set of examples, the reference feature is itself a feature of the blood pressure waveform, for example indicative of an incident pressure wave. In another set of examples, the reference feature is a feature of a separate signal to the pressure waveform, such as an ECG signal.

The inventors have realised that the temporal location of reflected pressure waves in the pressure waveform may be a reliable biomarker for blood pressure. By calculating a suitable time interval from the reflected pressure wave, a value of blood pressure may be calculated. One way to model the relationship between time intervals and blood pressure is through a set of parameters specified by the Moens-Korteweg equation. As will be described in more detail below, the time interval may be measured non-invasively, meaning that the apparatus may obtain a measure of a user’s blood pressure in a non-invasive manner.

The Moens-Korteweg equation provides an estimate of the pulse wave velocity (i.e. the speed at which the pressure pulse propagates through the blood) by modelling the arteries as elastic tubes. Analysis can show that the pulse wave velocity (PWV) may be dependent on the blood pressure according to the equation:

2rp (1) where E is the Young’s modulus of the arteries, h is the arterial thickness, r is the arterial radius, p is the density of the blood, γ is a constant and P is the blood pressure. This equation can be re-arranged in terms of the blood pressure as:

= - [2 ln(PW) + ln(2rp) - ln(£/i)] (2)

It has been appreciated that, in practice, the PWV is difficult to measure nonintrusively, and that an improved approach is to substitute the PWV for the pulse transit time (PTT) and a corresponding length / travelled by the wave over that time, i.e. PWV = //PTT.

The equation for the blood pressure then becomes:

In + ln(2rp) - ln(£7i) = i [2 ln(Z) - 2 ln(P7T) + ln(2rp) - ln(£7i)] (3)

The values of /, r, p, E, h and γ are constants for a given user (though may vary from user to user). Thus equation (3) demonstrates that the blood pressure is inversely proportional to the value ln(PTT). Thus, obtaining an accurate measure of the pulse transit time, or time interval, between two reference points along an artery can be used to obtain a measure of a user’s blood pressure. Described below is an apparatus that is configured to calculate a time interval from the temporal location of a reflected pressure wave and a reference feature and to use that time interval to calculate a measurement of the user’s blood pressure., The reference feature may be a feature of the pressure waveform or a feature of a separate signal indicative of heart activity.

Figure 2 shows a schematic diagram of an example apparatus 200 for obtaining a measurement of a user’s blood pressure. The device comprises a sensor unit (denoted generally at 201) comprising a set of one or more waveform sensors 207 for obtaining a pressure waveform. The waveform sensors could for example be photoplethysmogram PPG sensors, optical sensors, LED sensors, laser sensors, fluorescent sensors, acoustic sensors, piezo sensors, ultrasound sensors etc. The sensor unit may comprise one waveform sensor or a plurality of waveform sensors. The waveform sensors may be adapted for engagement with a user.

The apparatus 200 further comprises a processor 205 coupled to the sensor unit 201, including the one or more waveform sensors 207. Processor 205 is coupled to memory 203. The memory has stored thereon, in non-transitory form, computer program code that when executed causes the processor 205 to operate in the manner described herein.

The apparatus 200 may optionally further comprise an input 209 for receiving values of a set of physiological parameters of a user, for example their age, height, weight, sex, historical or ongoing medical conditions etc. The input could for example be in the form of a user interface (e.g. a touchscreen). Alternatively, the input could be in the form of a transceiver, socket etc. Information received through the input is communicated to the processor 705.

Operation of the apparatus 200 to obtain a measurement of a user’s blood pressure will now be described with reference to figure 3.

At step 301, the apparatus receives values for a set of physiological parameters. These values may be received via input 209. The values may be input into the apparatus by a user, for example through a user interface. The physiological parameters could include parameters such as the user’s age, height, weight, sex, arm length etc.

The parameters may be those that have a relationship to blood pressure that can be estimated through one or more mathematical models. For example, the distance between a site of measurement of a pressure waveform and the reflection site of the pressure wave may influence the pulse transit time (and thus possibly the calculation of the user’s blood pressure), and so the user’s height and/or arm length may be a useful parameter. As another example, age may be an indicator of arterial stiffness, another parameter known to influence blood pressure.

At step 303, the sensor unit 201 obtains a set of one or more signals indicative of heart activity. This set of signals includes a blood pressure waveform. It may comprise additional signals (e.g. ECG signals) as will be described below. As used herein, the term blood pressure waveform refers to any wave, or signal, which is proportional to the change in blood pressure over time. Different waveforms may thus be obtained from different sensors and/or different types of sensors, and be represented by different signals, but those signals may all delineate the change in blood pressure as a function of time and all preserve the same characteristic features and shapes (e.g. the systolic wave, systolic wave reflection, dicrotic wave and dicrotic wave reflection).

As one example, the sensor unit 201 may comprise a set of PPG sensors, with the device being arranged so that the sensors obtain a measurement of the user’s blood pressure waveform in their fingertips by illuminating the skin and tracking changes of light absorption. The PPG sensors thus track changes in the blood volume at the site of measurement (in this case the fingertips) over time. Since the blood volume at a given site is directly proportional to the blood pressure at that site, the blood-volume signal measured by the PPG sensors is an example of a blood pressure waveform.

The apparatus 200 may obtain the user’s blood pressure waveform signal in a number of ways. In a simple example, the apparatus may obtain (via the one or more sensors of the sensor unit 201) a single measurement of the waveform (e.g. from one fingertip and for one pulse) and take this to be the obtained waveform signal. In certain situations, it may be feasible to obtain a measurement of a user’s pressure waveform using only a single reading. For example, for certain users a component reflected wave of the pressure waveform may coincide with an incident pressure wave of the waveform at a point where the gradient of the incident pressure wave is relatively flat, and so the reflected wave may be visible in the resultant waveform. This is illustrated in figure 1B for the systolic wave reflection, which shows an example waveform 105 formed as the superposition of component incident pressure wave 107 and reflected pressure wave 109. The reflected pressure wave occurs approximately at the dicrotic notch (where the gradient is relatively low) and thus the systolic wave reflection is visible in the resultant waveform (indicated at 111).

However, for other users the reflected pressure waves may not be directly visible in the waveform. This may be due to the patient’s age, ongoing medical condition, or other reasons. For example, if the systolic wave reflection coincides with an incident pressure wave at a point where the gradient of the incident wave is relatively high, then the systolic wave reflection may be less visible in the resultant waveform and may more likely appear as a kink in the waveform rather than a peak. This is illustrated by the waveform 113, which is composed of incident pressure wave 115 and reflected pressure wave 117. Here, the systolic peak reflection occurs in the region between the systolic peak and dicrotic notch (a region where the gradient magnitude is relatively high), and so is not directly visible in the resultant waveform but instead appears as a small kink 119.

The reflection peaks (i.e. peaks in the reflected pressure waves) may in many cases be smaller than the peaks of the corresponding incident pressure waves. For example, the systolic peak reflection is typically smaller than the corresponding systolic peak. This can make the peaks of the reflected waves more difficult to observe, particularly for sensors suffering from measurement noise (as may be the case for PPG sensors in a practical implementation). This may make it difficult to obtain an accurate measurement of a user’s blood pressure waveform using a single reading.

One way to improve the quality of the blood pressure waveform signal measurement is for the sensor unit 201 to measure a plurality of individual pressure waveform readings and take an average of those measured waveforms. This approach tends to cause random fluctuations in the individual measurements caused by noise to cancel out, while the characteristic, or deterministic features of the individual waveforms (e.g. the systolic wave peak and corresponding reflection; dicrotic wave peak and corresponding reflection) remain. This approach can reduce the noise in the measured pressure waveform and thus improve the signal-to-noise ratio. If the signalto-noise ratio is improved sufficiently then points indicative of the reflected wave(s) may be visible in the waveform.

In order to the correctly average a set of individual waveforms, it is important that the waveforms are added in phase, i.e. each waveform is centred, or positioned, with respect to a reference point. The peak of the systolic wave is one example of a reference point used to centre each waveform. The inventor has appreciated that the R-peak of an electrocardiograph (ECG) signal is another useful reference point.

An ECG signal records the electrical activity of the heart as a function of time. The signal typically comprises a temporal series of waves, with one wave being produced for each cardiac cycle. Each ECG wave typically comprises a QRS complex caused by the depolarization of the ventricles.

The R-wave of the QRS complex is a (typically) large upward spike that is particularly narrow (i.e. localised in time) and often has a large amplitude relative to the noise in an ECG signal. It has been appreciated that these characteristics of the R-wave make it particularly suitable for use as a reference point with which to synchronise measured pressure waveforms as the measurement noise is less likely to adversely affect the ability to detect it.

The sensor unit 201 of apparatus 200 may therefore further comprise a set of one or more sensors 211 for measuring the user’s ECG signal.

An illustration of how the sensor unit 201 may synchronise the measured pressure waveforms with R-waves of the QRS complex to average the waveforms will now be described with reference to figure 4.

Figure 4 shows a temporal series of ECG waves 401, 403 and 405, and a temporal series of blood pressure waveforms 407, 409 and 411. Each ECG wave is shown comprising a QRS complex. The figure shows ECG and pressure waveforms for three cardiac cycles. The ECG wave and pressure waveform for each cardiac cycle is shown on a separate graph (denoted 413, 415 and 417), with each graph containing both the ECG wave and pressure waveform for a given cardiac cycle. Though in practice the ECG waves will be detected by one type of sensor (e.g. an ECG sensor) and thus appear on a single graph, and the pressure waveforms will be detected by another type of sensor (e.g. a PPG sensor) and so appear on a different single graph, the schematic format shown in figure 4 has been chosen to aid understanding, as will be apparent from the following. It is also noted that, for clarity, each graph is shown with respect to a common time axis.

The time difference between the R-wave of the first and second cardiac cycles is denoted ΔΤι, and the time difference between the R-waves of the second and third cardiac cycles is denoted ΔΤ₂.

To average the individual waveform readings 407, 409, 411, the sensor unit 201 uses the R-wave of a measured ECG signal as a reference point for the pressure waveform. The processor may use the knowledge of the temporal offset between the R waves of each of a plurality of ECG signals (in this example three) to thereby temporally align the plurality of measured waveforms. That is, because the sensor unit has synchronised the blood pressure waveforms with the ECG signals, it can centre each of the measured waveforms using the corresponding R-waves as the reference event.

For example, the sensor unit may exploit the knowledge of the values of ΔΤι and ΔΤ2 to temporally offset the waveform 409 by a value of ΔΤ1 and to temporally offset the waveform 411 by the value (ΔΤ1+ΔΤ2). This centres the waveforms with respect to a common reference event (the R-wave of the ECG signal) and allows the sensor unit 201 to average the waveform readings 407, 409 and 411 to generate an averaged pressure waveform signal. As described above, the averaged waveform may have an improved signal-to-noise ratio compared to an individual waveform reading due to the cancellations of noise fluctuations in the individual signal readings as a result of the averaging.

Configuring the sensor unit 201 to average a plurality of waveform readings using the R-wave of the ECG signal as the reference event to centre each waveform reading is advantageous because the features of the R-wave (its relatively high signal to noise ratio, and its temporal localisation) mean its detection is less likely to be affected by measurement noise compared to using a characteristic point of the pressure waveform as the reference event.

In the example described above, the sensor unit obtained a measurement of the user’s blood pressure waveform by averaging three waveform readings. It will be appreciated that the number of waveform readings was chosen to be three merely for the purpose of illustration and that in a practical implementation any suitable number of waveform readings may be averaged to obtain a measurement of the waveform. For example, the sensor unit 201 may collect readings for a predetermined amount of time, and then use the number of readings (i.e. the number of ECG and pressure waveform readings) collected in that time to calculate the average waveform measurement. A predetermined time of 30 seconds may be a convenient amount of time. This would enable a relatively large amount of pulse waveform readings to be collected (approximately 30 for a patient with a heart rate of 60 beats per minute) whilst also allowing a measurement of the pressure waveform to be obtained relatively quickly from the perspective of the patient using the apparatus.

The blood pressure waveform obtained by the sensor unit 201 is then output to the processor 205.

At step 305, the processor 205 analyses the obtained blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave, and at step 307 the processor selects a reference feature from the set of one or more signals obtained at step 303 having a known relationship to the reflection feature.

Examples of steps 305 and 307 will now be described in more detail with reference to figure 5A, which shows an example waveform measurement obtained by the sensor unit 201. The waveform is denoted at 501, and is formed as the superposition of four component waves: the systolic wave 503; dicrotic wave 505; systolic wave reflection 507 and the dicrotic wave reflection 509. The resultant pressure waveform formed as the superposition of these four component waves is shown at 501.

The systolic wave comprises a peak at time Ti; the dicrotic wave comprises a peak at time T₃; the systolic wave reflection comprises a peak at time T2 and the dicrotic wave reflection comprises a peak at time T4.

In this example the systolic wave reflection and dicrotic wave reflection are not directly visible in the waveform 501. As discussed above, this is a fairly common scenario for certain patient demographics, even when the waveform being analysed is one formed from averaging multiple waveform readings.

In a first set of examples, the reference feature selected at step 307 is also a feature of the blood pressure waveform. That is, the processor 205 identifies a reflection feature of the pressure waveform indicative of a reflected pressure wave, and a reference feature of the pressure waveform.

In one such example, the processor 205 analyses the blood pressure waveform 501 to locate corresponding points of a component incident pressure wave and component reflected pressure wave. That is, the processor 205 locates a reflection feature indicative of a reflected pressure wave of the pressure waveform, and selects as the reference feature an incident feature indicative of a corresponding incident pressure wave of the pressure waveform. More specifically, the processor may select as the reference feature a feature that is indicative of the systolic wave 503 and locate a reflection feature indicative of the corresponding reflected systolic wave 507. Alternatively, the processor may select as the reference feature a feature indicative of the dicrotic wave 505 and locate a reflection feature indicative of the corresponding reflected dicrotic wave 509.

The reflection and/or reference features may be characteristic points of the component pressure waves, i.e. points indicative of, or representing, a characteristic feature of the waveform and its component waves. For example, the reflection and/or reference features may be peaks, or more generally extrema, in respective component waves of the pressure waveform. In other words, the reflection feature may be a peak of a reflected pressure wave and the reference feature may be a peak of an incident pressure wave. The processor may be configured to select as a reference feature a characteristic point of a component incident pressure wave and to locate as the reflection feature a corresponding characteristic point in a component reflected pressure wave. The points may be corresponding in the sense that both points identify, or represent, the same characteristic feature in the respective component waves.

If the reflection feature and reference feature both happen to be visible in the waveform 501 (e.g. both appear as extrema in the waveform), then location of these features by the processor 205 may be relatively straightforward, for example by execution of a peak-detection algorithm. However, in some cases one or both of the reflection and reference features are not visible in the waveform. For example, as has been discussed it is common for one or both of the reflected waves (the systolic wave reflection and dicrotic wave reflection) to not be visible in the waveform.

The inventor has appreciated that by taking higher order time-derivatives of the obtained pressure waveform 501, characteristic features not readily visible in the original waveform may be more easily detected. For example, a point appearing in the original waveform as an inflection point will appear as an extrema in the first-order time derivative of the waveform. Location of the extrema in the first-order derivative will therefore correspond to location of the inflection point in the original waveform. It has also been found that characteristic features of the reflected wave not visible in the original waveform may appear as small peaks (i.e. extrema) or inflection points in the first-order time derivative.

Figure 5B shows at 511 the first-order time derivative of the example pressure waveform 501. It is noted that the systolic wave peak and dicotic wave peak, which both appear as extrema in the original waveform 501 at times Ti and T3 respectively, are zero-crossings in the derivative waveform, shown at 513 and 515 respectively. The peak of the systolic wave reflection, which is not visible in the original waveform, is shown as an inflection point 517, and the peak of the dicrotic wave reflection (also not visible in the original waveform), is shown as the small peak 519.

Thus by taking the first-order derivative of the original pressure waveform, the processor 205 can use a peak-detection algorithm to temporally locate features of the reflected wave that are not visible in the original waveform. The processor may further be configured to take the second-order time derivative of the pressure waveform to further locate features of the reflected wave that do not appear as extrema in the firstorder derivative. For example, for the example waveforms in figures 5A and 5B, if the processor were to additionally take the second derivative of the waveform 501, the peak of the systolic wave reflection (which appears as inflection point 517 in the firstorder derivative) would appear as an extremum and thus be capable of being located by performing a peak search.

The processor 205 may therefore advantageously be configured to differentiate the obtained waveform 501 in order to locate one or more features of the pressure waveform.

The processor 205 may locate the reference feature and reflection feature by performing an extremum-detection algorithm that first attempts to locate desired points in the component waves by searching the original waveform for extrema. The extrema may be indicative of a characteristic feature of a component incident wave and a corresponding characteristic feature of a component reflected wave. To give some examples, the processor may search the original waveform for extrema indicative of the systolic peak and the systolic peak reflection; alternatively the processor may search the original waveform for extrema indicative of the dicrotic wave peak and the peak of the dicrotic wave reflection.

If extrema are located that are indicative of both an incident and reflected pressure wave, those extrema are taken as the desired features of the waveform. If an extremum is located that is indicative of one of the component waves but not the other (i.e. only one of the reflection and reference feature is identified), or no indicative extrema are identified, then the algorithm proceeds to the next step and the processor 205 performs a first-order differentiation of the waveform.

Features not found in the original waveform (e.g. a reflection feature indicative of the systolic or dicrotic wave reflection) are then searched for by looking for extrema in the first-order derivative of the waveform. This step would enable features appearing as inflection points in the original waveform to be located. For the example waveforms shown in figure 5, taking the first-order derivative would allow the dicrotic peak reflection to be identified, for example.

If the missing features are not found in the first-order derivative, the processor may calculate the second order derivative. Features not found in either the original waveform or the first-order derivative are then searched for in the second-order derivative by searching for extrema in the second-order derivative. This step would enable points that appear as inflection points in the first-order derivative to be located. For example, the inventor has appreciated that for certain users the peak of the systolic wave reflection appears as an inflection point in the first-order derivative of the pressure waveform. For such waveforms, the peak of the systolic wave reflection appears in the second-order time derivative as a maxima and so may be more easily located.

The temporal locations of the reference and reflection features are then identified from the identified extrema (from the original waveform and/or higher order time derivatives).

If the extremum-detection algorithm is being used to locate both the reflection feature and reference feature, the processor 205 may perform additional processing to determine which extrema corresponds to the reference feature and which extrema corresponds to the reflection feature. One approach would be to take the take the extrema having the lower temporal value as the reference feature and the extrema having the higher temporal value as the reflection feature (because the reflection features occurs later in time than the reference feature).

If multiple extrema are identified for a given n’th order derivative of the waveform (where n = 0, 1, 2...), then the processor 205 may perform additional processing to select one of the extrema as the reference feature and one as the reflection feature. For example, the processor may use the temporal positions of the extrema and/or the magnitudes of the extrema to select the reference and reflection features from the set of identified extrema.

Performing the above-described algorithm may offer advantages compared to either searching only the original waveform without taking the derivative, or simply taking one or more higher-order derivatives by default without first searching the original waveform for extrema. This is because if an extrema indicative of a characteristic feature of the waveform (e.g. the systolic peak) is present in the original waveform, then any located extrema in a higher-order derivative of the waveform will not correspond to the correct temporal position of that characteristic feature. For example, referring to the waveform in figures 5A and 5B, the systolic wave peak occurs at time Ti in the original waveform. Because the peak is present in the original waveform, the point corresponding to that peak is a zero-crossing in the first-order derivative 511. The neighbouring peak in the first-order derivative corresponds to the peak flow of the systolic peak and occurs at time T5/T1. Thus searching for extrema in the derivative waveform 511 would not correctly locate the systolic peak. In contrast, first searching the original waveform for the appropriate extrema before calculating a higher-order derivative increases the likelihood that the correct temporal location of the extrema can be located.

As mentioned above, patient’s may display a wide variety of blood pressure waveforms depending on age, medical condition etc. It has been recognised that due to the variety of pressure waveforms that may be exhibited by patients, it may be suboptimal to process waveforms for each patient in the same fixed manner. By performing the above-described extremum-detection algorithm, the correct temporal locations of characteristic features of the component waves can be determined for a range of different pressure waveforms thus making the apparatus compatible with a wide patient demographic.

Once a reflection feature has been identified and a reference feature selected, then at step 309 the processor 205 calculates a time interval between those features.

For example, referring again to figures 5A and 5B, if the processor 207 identifies the peak of the dicrotic wave reflection as the reflection feature and selects as the reference feature the peak of the dicrotic wave, then the processor calculates the time interval ΔΤ-τ = T₄ - T₃ between those features. Alternatively, if the processor identifies the peak of the systolic wave reflection as the reflection feature and selects the peak of the systolic wave as the reference feature, the processor calculates the time interval ΔΤζ = T₂ - Ti between those features. If the processor has located both the peak of the dicrotic wave reflection and the systolic peak reflection as reflection features, and selected both the systolic peak and dicrotic wave peak as reference features, it may calculate both time intervals ΔΤ-τ and ΔΤ₂\

Once at least one time interval has been calculated, then at step 311 the processor 205 uses the time interval(s) to calculate a measurement of the user’s blood pressure.

It has been appreciated that incident pressure waves of the waveform (which are typically used to calculate a pulse transit time for the purposes of calculating blood pressure) depend on the same physiological parameters as the reflected pressure waves of the waveform. Thus it has been realised that the time interval between an incident wave and reflected wave, or more accurately, between corresponding features of those waves, can be used to calculate a transit time that depends on the same physiological parameters as a conventional pulse transit time. This means the time interval between corresponding features of incident and reflected pressure waves of the waveform can be used to calculate a value of the user’s blood pressure, for example using the parameters specified in the Moens-Korteweg equation.

In particular the reflected systolic peak travels during systole (when the aortic valve is still open and blood is being ejected from the ventricles), whereas the dicrotic wave reflection travels during diastole (after the valve has closed). It has been realised that this means the time interval between the systolic peak and systolic peak reflection is dependent on, or a biomarker for, the systolic pressure, and the time interval between the dicrotic wave peak and peak of the dicrotic wave reflection is dependent on, or a biomarker for, diastolic pressure. Thus the time intervals calculated at step 309 can be used to calculate a value of the user’s diastolic pressure, systolic pressure, or both.

The user’s blood pressure may be calculated in a number of ways. For example, the modified Moens-Korteweg equation shown in equation 3 shows one way that a pulse transit time, or time interval, can be related to the blood pressure through a set of the user’s physiological parameters. The processor 205 may therefore execute an algorithm using as an input the calculated time interval from step 309 in order to estimate the remaining parameters specified in the Moens-Korteweg equation to arrive at a value of the user’s blood pressure.

The time interval between corresponding points of the component waves may in certain circumstances represent the time taken for the wave to travel along the patient’s arterial tree from the point of measurement (e.g. fingertips of the patient in the above example) to the reflection site and back again. Thus the ‘transit time’ used to derive the patient’s blood pressure may be equal to the time interval between the corresponding points of the component waves, in which case the length / in equation (3) may be equal to twice the distance between the point of measurement and the reflection site. In other examples, the ‘transit time’ may be equal to half the time interval, in which case the length / in equation (3) may be equal to the distance between the point of measurement and the reflection site.

As discussed above, the parameters of the Moens-Korteweg equation (except the pulse transit time/time interval) are constants for a given individual. Though the values of these parameters could be measured, this may require fairly extensive testing which is not particularly convenient. This is especially true if the apparatus 200 was implemented as a self-service device within a medical facility.

However, there is a set of the user’s physiological parameters that are more readily known. Examples include the user’s age, height, weight, sex etc. Further examples include anatomical parameters, for example the size of the user’s hands, feet, a distance or length between two given points of the patient’s anatomy etc. Values of these parameters may be input to the apparatus (e.g. through the input 209). The processor 205 could then execute an algorithm that uses the values of the input parameters and various mathematical models that model the relationship between the input parameters and the physiological parameters that link the pulse transit time to the blood pressure. In this way the processor 205 can calculate a value of the user’s blood pressure using the measured time interval from step 307 and the values of the physiological parameters input into the apparatus without having to obtain measurements of all the parameters specified in the Moens-Korteweg equation.

For example, the processor 205 may execute a machine-learning algorithm. The machine learning algorithm may be predetermined. The machine learning algorithm may enable the processor 205 to calculate an accurate value of the user’s blood pressure using the calculated time interval and input physiological parameters by exploiting knowledge of a learning data set. That learning data set may be collected a priori. The learning data set may establish relationships between a pulse transit time or time interval, a learning set of physiological parameters and measured blood pressure for a plurality of test subjects. That is, during a data gathering stage, the blood pressure for a set of test subjects is measured (e.g. using a conventional cuff). For each subject, the values of a set of physiological and/or anatomical parameters is also measured. A subset of these parameters may be the same as the parameters input by the user into the apparatus 200. The gathered data can then be analysed to establish relationships between the parameters and measured blood pressure. During operation of the apparatus 200 (i.e. after the data gathering stage), the processor 205 can use then use those established relationships and models to derive a value of the user’s blood pressure when presented with the values of the physiological parameters for that user and a measured time interval.

The algorithm executed by processor 205 to calculate the user’s blood pressure may be in the form of a decision tree. At each branch of the decision tree, an item of observed data (e.g. a parameter value input by the user via input 209) may be compared against a threshold value, or binned, and a decision taken based on that comparison or bin value. As the tree is traversed the set of observed parameters (including the calculated time interval) are mapped to conclusions, or predictions about the value of the target data item (the user’s blood pressure).

It will be appreciated that a machine-learning algorithm is just one example of the type of algorithm the processor 205 may employ to calculate the blood pressure. The processor may execute any suitable type of model-fitting algorithm to derive the parameters needed to calculate the blood pressure from the time intervals. These model-fitting algorithms may exploit knowledge of one or more mathematical models that model the relationship between one or more physiological parameters. These models may then be used to derive values of the parameters needed to calculate blood pressure from the time interval (e.g. the parameters specified in equation (3)) from the values of the parameters input by the user.

The value of the blood pressure calculated by the processor 205 may be the user’s diastolic pressure, systolic pressure, or both. It may be, or may include, a timeaveraged value of the user’s blood pressure. For instance, the apparatus may calculate a value of the user’s systolic pressure by taking the time interval between the systolic peak and the systolic peak reflection. If instead the time interval between the dicrotic notch and dicrotic notch reflection was measured, then a value of the user’s diastolic pressure can be calculated.

At step 313, the apparatus outputs a value of the user’s blood pressure. The apparatus may for example display a value for the blood pressure on a display screen.

Using the time interval between corresponding features of the incident and reflected pressure waves of the waveform to calculate blood pressure may offer several advantages over conventional approaches for calculating blood pressure. For example, one potential drawback with the more conventional approach of calculating blood pressure from a pulse transit time using the R-peak of the ECG signal and the systolic peak is that it neglects the differences in the pre-ejection time (i.e. the time lag between the R-peak and the beginning of the ejection stroke), both for different beats for a given subject and also between different subjects. Whilst it is possible to improve the accuracy of this method by using a timing reference that corresponds to the opening and/or closing of the aortic valve, this may require other sensors (e.g. microphones) attached to the patient’s chest. In contrast, by measuring the time interval between an incident and reflected pressure wave using a single sensor unit, there are no unknown delays in the calculated time interval and hence variations in the pre-ejection time do not affect the value of the time interval. Thus the time interval between features of the incident and reflected waves may result in a more accurate blood pressure calculation compared to the convention approach of calculating a pulse transit time from the R-peak of the ECG signal.

A further potential advantage is that two time intervals can be calculated from the pressure waveform: a first time interval between the systolic peak and systolic peak reflection and a second time interval between the peak of the dicrotic wave and the peak of the dicrotic wave reflection. This means that one or both of the systolic pressure and diastolic pressure can be calculated from the user’s pressure waveform. In contrast, measuring a pulse transit time from the R-peak of the ECG signal to the systolic peak generally only allows the systolic pressure to be measured. In addition, the use of the R-peak in this manner measures the transit time of the systolic peak at the very beginning of systole. This is a transient period in the pressure waveform typically characterised by a pressure gradient which has been found to vary between different subjects. For example, patients whose waveforms exhibit an anacrotic notch may have a different pressure gradient in the initial stages of systole compared to patients whose waveforms don’t exhibit an anacrotic notch. Thus using a time interval between the incident and reflected waves may provide more accurate blood pressure calculations for a wider patient demographic compared to the more conventional approach of using a pulse transit time calculated from the R-peak and systolic peak.

In the examples described above, the reference feature selected at step 303 was a feature of the blood pressure waveform. In an alternative arrangement the reference feature selected at step 303 may be a feature of a separate signal to the pressure waveform. For example, the sensor unit 201 may comprise one or more sensors for measuring the electrical activity of the heart, and the processor may be configured to select a reference feature from the electrical signal obtained from those sensors. The reflection feature may still be located from the pressure waveform signal in the manner of any of the examples previously described.

The sensors may be configured to obtain a measurement of the user’s ECG signal. The sensor unit could for example comprise an ECG sensor (as previously described), an electric potential sensor or some other type of heart rate sensor for measuring the electrical activity of the heart. If the electrical signal is an ECG signal, for example, then the reference feature selected at step 303 may be the R-peak of the ECG signal.

The reflection feature of the pressure waveform may then be located in the same manner as any of the examples described above. For example, the reflection feature may be located in a pressure waveform signal formed from averaging a plurality of pressure waveform readings. Those pressure waveform readings may be synchronised with the R-waves in the manner described above to enable the waveforms to be correctly aligned before averaging.

The reflection feature identified may be indicative of the reflected systolic wave, or reflected dicrotic wave. Alternatively, first and second reflection features indicative of both reflected waves may be located in the pressure waveform. This would enable two time intervals to be calculated (a first time interval between the R-peak and reflection feature indicative of the reflected systolic wave; and a second time interval between the R-peak and reflection feature indicative of the reflected dicrotic wave) and thus both the systolic and diastolic pressure may be calculated. The reflection feature indicative of the reflected systolic wave may be the systolic peak reflection. The reflection feature indicative of the reflected dicrotic wave may be the peak of the dicrotic wave reflection.

The reflection features may again be located in the pressure waveform using any of the techniques described herein. For example, the processor 205 may execute an extremum-detection algorithm to locate the reflection features. This algorithm may comprise the same set of steps as the extremum-detection algorithm described above, with the exception that only reflection features of the pressure waveform need be located, rather than both reflection features and reference features. Thus, briefly, the processor may analyse the pressure waveform to locate one or more reflection features indicative of respective reflected pressure waves. If the features are found in the pressure waveform, the algorithm ends. If one or both of the reflection features are not found in the original waveform, the processor takes the first-order time derivative of the waveform and searches that for the missing reflection features. If the reflection features are found in the first-order derivative of the waveform, the algorithm ends. If one or both of the reflection features are not found in the first-order derivative, the processor takes the second-order time derivative of the waveform and searches that for the missing reflection features.

With the R-peak serving as the reference feature, the time intervals calculated at step 309 may include one or both of: i) the time interval between the R-peak and a reflection feature indicative of the reflected systolic wave; ii) the time interval between the Rpeak and a reflection feature indicative of the reflected dicrotic wave.

Although in this set of examples the refection feature and reference feature belong to different signals, the processor 205 may exploit a known temporal relationship between the two signals to enable the time interval to be calculated. For example, the pressure waveform signal may be synchronised to the ECG signal (and in particular to the R-wave of the ECG signal) as described above.

As was described above, the inventor has realised that a reflected pressure wave (e.g. the systolic wave reflection) is governed by the same physical parameters as an incident pressure wave, and thus the time between the R-peak and reflected pressure wave varies in accordance with blood pressure. With reference to equation (3), if the parameters apart from the transit time/time interval are taken to be constant (which is a reasonable assumption for a given user) then it can be seen that the time interval (taken here to be the time interval between the R-peak and the systolic and/or dicrotic reflection) is inversely proportional to the blood pressure.

Thus the time interval(s) may be used to calculate a value of the user’s blood pressure in a similar manner to what was described above, i.e. using a parameter estimation technique to calculate values of parameters needed to calculate the blood pressure from the time interval. The processor may for example execute a machine-learning algorithm or model-fitting algorithm to derive the values of the other parameters specified in equation (3).

Unlike the approach in which the time interval is calculated using an incident pressure wave as a reference, the time interval between the R-wave and reflection feature is dependent on the user’s pre-ejection time. This means the resultant blood pressure calculated using the R-wave as a reference feature may be less accurate than one calculated using an incident pressure wave as a reference feature. However, there may still be advantages in using the time interval between the R-wave and reflected wave to calculate blood pressure compared to the conventional approach of using the time interval between the R-wave and systolic peak.

For a given measurement site, a reflected pressure wave has travelled further than its corresponding incident pressure wave. This means that a given change in blood pressure ΔΡ causes a larger corresponding change in the time of measurement of the reflected waves compared to the incident waves. Thus using the time interval between the R-wave and a reflected wave may result in a more precise blood pressure calculation compared to using the time interval between the R-wave and incident pressure waves.

In one convenient arrangement, the R-waves of the measured ECG signal may be used as both the reference event to centre multiple pressure waveform readings for averaging as part of step 303 and also as the reference feature used to calculate the time interval in step 309.

In another related example, the processor may be configured to select as the reference event the R-wave of the ECG signal, but to locate within the pressure waveform both a reflection feature indicative of a component reflected pressure wave and an incident feature indicative of a component incident pressure wave. For example the processor may locate features indicative of both the systolic wave and the systolic wave reflection; or features indicative of both the dicrotic wave and dicrotic wave reflection; or some combination thereof. As before, those features may be peaks of the component waves.

The processor may then calculate a first time interval between the R-wave and the reflection feature and a second time interval between the R-wave and the incident feature. The processor may then use both time intervals to calculate a measurement of blood pressure.

As a specific example, the processor may analyse the pressure waveform to locate both the systolic peak and the systolic peak reflection, and calculate a first time interval between the R-wave and the systolic peak and a second time interval between the Rwave and the systolic peak reflection. The processor may then use both time intervals to calculate a value of blood pressure. As another example, the processor may locate both the peak of the dicrotic wave and the peak of the dicrotic wave reflection, and calculate a first time interval between the R-wave and dicrotic peak and a second time interval between the R-wave and the dicrotic peak reflection. Again the processor may use both time intervals to calculate a value of blood pressure.

The processor may use both the first and second time intervals to calculate the blood pressure in a number of ways.

In one example, the processor may use the relationship between the first time interval, second time interval and blood pressure to calculate a value of the pre-ejection time. It has been appreciated that the relationship between the first time interval, second time interval and blood pressure is dependent on the pre-ejection time. The relationship between the values of the first time interval, second time interval and blood pressure can be modelled as a curve within a three dimensional (3D) space spanned by those parameters, with the curve being specific to a certain pre-ejection time. Thus the relationship between the values of the first time interval, second time interval and blood pressure for a different pre-ejection time may be modelled as a different curve in the 3D space. For a given set of values of the first and second time interval and the blood pressure that defines a point in the 3D space, the processor may be configured to identify the curve on which that point lies and therefore determine the pre-ejection time. The data modelling the relationship between the first and second time intervals and the blood pressure for various pre-ejection times may be accessed the processor. It may for example be stored in memory 203.

In another example, the processor may use the two time intervals to derive a third time interval between corresponding points of the component waves (the third time interval may for example be the time between the systolic wave and the systolic wave reflection). The third time interval may then be used to calculate blood pressure as described above. Information on either or both of the first and second time intervals may be used by the processor in addition to the third time interval to improve the accuracy of the blood pressure calculation. For example, the first and second time intervals and their relationship to each other may be used to glean information on the user’s pre-ejection time (e.g. as described in the preceding paragraph) which may improve the accuracy of the blood pressure calculation.

In another example, the processor may plot, track or otherwise use the ratio of the first and second time intervals to improve the accuracy of the blood pressure calculation. For instance, the ratio of the second interval over the first interval is smaller for a larger pre-ejection time. The processor may therefore use this ratio as part of the algorithm calibration to calculate the blood pressure.

It will be appreciated from these examples that the processor may use information on the time interval between the R-wave and the systolic peak and/or dicrotic peak as part of the blood pressure calculation to potentially obtain a more precise result.

The apparatus 200 described above comprises a sensor unit for obtaining the one or more signals indicative of heart activity. In an alternative arrangement, there may be provided a device that is configured to calculate a measurement of blood pressure from a set of one or more signals (in any of the ways described herein), but that does not itself comprise a sensor unit for obtaining the signals. That is, the operations of obtaining the set of one or more signals indicative of heart activity through a sensor unit, and the processing steps to derive a value of blood pressure from those signals, may be split across different devices.

Figure 6 shows a device 601 configured to calculate a value of blood pressure. The device comprises a processor 603 coupled to an input 609, a memory 605 and a transceiver 607.

The device 601 is configured to receive the set of one or more signals indicative of heart activity comprising the blood pressure waveform. The device may receive those signals via transceiver 607, or input 609. The input could take the form of, for example, a wireless transceiver unit; a network socket (e.g. an internet or Ethernet socket) or a user interface (e.g. touchscreen, mouse, keypad etc.).

Once the set of one or more signals have been received, they are communicated to the processor, which may then perform steps 305, 307, 309 and 311 according to any of the examples described herein. The processor may perform those steps in response to executing computer program code stored in non-transitory form in the memory 605.

The device 600 may be configured to receive the set of one or more signals over a wireless communication network. That is, the processing of the signals by device 600 may be done remotely of the device that acquired, or measured, the signals. Device 600 could be, or be implemented within, a server. This may be advantageous as it might enable the sensor device (the device or devices that obtain the signal measurements) to have relatively little processing power.

The apparatus in figure 2 was described as comprising a set of one or more sensors (e.g. electric potential sensors) for measuring a subject’s ECG signal. It will be appreciated that the inclusion of the ECG sensors is optional, and in other examples the apparatus may not include ECG sensors.

The apparatus 200 may take various forms of physical manifestation. The apparatus may be a single device, and the processor for processing the obtained waveforms may be located within that device locally to the sensor unit. For example, the device may take the form of a station, or console, and both the waveform sensor and processor are located within a housing of that station/console. Alternatively the single device may take the form of a smartphone, smartwatch, laptop, computer console tablet etc. In other examples, the processor that processes the obtained waveforms may be located remotely from the sensor unit. For example, the sensor unit may be located within a first constituent device of the apparatus and the processor may be located within a second constituent device of the apparatus. The sensor unit could for example be located within, or otherwise form part of, a smartphone, smartwatch, tablet, laptop or other electronic device. The processor could be located within a laptop, desktop computer, or located in a server (e.g. be cloud-based). In such examples the first constituent device may be configured to communicate data collected from the sensor unit to the second constituent device via a wired and/or wireless communication network. That communication network could for example be a near-field communication network such as Zigbee, Bluetooth, NFC etc. Alternatively data collected from the sensors could be communicated to the device housing the processor via one or more communications protocols such as TCT/IP, UDP etc. One or both of the constituent devices may comprise a transceiver for communicating data therebetween.

In some of the examples above the sensor unit 201 performed a degree of processing on measured waveform readings to calculate an average waveform. That processing may be performed by a sensor processor housed within, or forming part of, the sensor unit. The sensor processor may be a separate processor from the processor 205.

Alternatively the analysis of the waveforms and the calculation of the blood pressure may be performed by a single processor (i.e. there may not be a separate sensor processor and general processor as in the example shown in figure 2).

The sensor unit 201 may obtain a measurement of a single pressure waveform and locate corresponding features of the component waves of that waveform in order to calculate a value of blood pressure. In other examples, the sensor unit may obtain a plurality of waveform measurements for a single user. For example, the sensor unit may obtain a first measured waveform from one sensor of the sensor unit (e.g. by taking the average of multiple individual waveform readings) and a second measured waveform from another sensor of the sensor unit. The sensor unit may as such comprise a plurality of sensors for obtaining a respective measurement of the user’s pressure waveform. The sensors may be arranged so as to engage a respective fingertip of the user. The sensors may be arranged to engage a set of one or more fingertips on one or both hands of the user. Alternatively the sensors may be arranged to engage different positions along a user’s arm; different arms; the legs or feet of the user or a combination thereof.

Each obtained waveform may be analysed in the manner described above to locate respective sets of corresponding points and calculate time intervals between those points and thus respective values of the user’s blood pressure (i.e. a value of blood pressure is calculated from each obtained waveform). In this way the blood pressure can be measured across different parts of the body, for example left arm, right arm, within the radial or ulnar artery etc.

Obtaining multiple values of the user’s blood pressure at different parts of their anatomy may be useful for detecting a potential underlying health issue. For example, if a blood pressure value calculated from a waveform measurement from the user’s left hand differs from a blood pressure value calculated from a waveform measurement from the user’s right hand by more than a threshold amount, the apparatus may output an alert indicating the detection of a potential health issue. That alert may be a visual and/or audible alert. It may alternatively be an electronic alert, for instance an electronic communication to a medical professional such as the subject’s registered doctor.

One example physical arrangement of apparatus 200 is shown in figure 7.

Figure 7 shows an apparatus 700 that comprises a sensor unit 701 comprising a plurality of PPG sensors 703, 705, 707 and 709. The sensor unit further comprises a sensor processor 721 coupled to each sensor. The sensors are set into a surface 711 of the device. In this example, the surface takes the form of an electrically insulated mat. The surface 711 is located within a user-interaction region 713 of the apparatus.

As shown, the apparatus comprises a set of indicia (denoted generally at 715 by the dashed markings) that indicate the position that a user is to place their fingertips when using the apparatus. The indicia are positioned so that, when a user’s fingertips are placed at the positions marked by the indicia, the user’s fingertips sit over (e.g. directly above) the PPG sensors 703, 705, 707 and 709. Arranging the PPG sensors to interface with a user’s fingertips is advantageous because it enables the device to be used with a variety of different anthropometries.

The apparatus further comprises processor 205, a memory 203 and a user interface 723 for receiving values of a set of physiological parameters of a user.

The sensor unit 701 further comprises a set of sensors 717 and 719 for measuring the user’s ECG signal. The sensors could for example be electric potential, or ECG sensors. The sensors may be located within the user-interaction region of the apparatus. They may be embedded in the surface 711 of the device. For example, the sensors may be positioned with respect to the indicia 715 such that when the user’s fingertips are placed on the indicia, one or more of the fingertips sits atop the electric potential sensors. This is a convenient arrangement because it would enable the apparatus to take both pressure waveform readings and ECG signal readings simultaneously by having the user place their fingertips on the indicia 715.

Apparatus 700 may be in the form of a console, or station, for example.

In the examples described herein, the reflection and reference features of the pressure waveform and/or electric potential signal were extrema. It will be appreciated that, whilst this is a convenient approach due to the relative ease with which extrema may be detected, other features of the pressure waveform and/or electric potential signal may be chosen to calculate the time interval and thus blood pressure. For example, the features may correspond to a terminal end of the waves, e.g. the reflection feature may be the point indicating the beginning or trailing end of a reflected pressure wave. Similarly, the reference feature may be indicative of the terminal end of an incident pressure wave, or ECG signal. Alternatively the reference feature may be a different component wave of the QRS complex. The reflection feature may indicate any suitable part of a reflected pressure wave, and the reference feature may indicate any suitable part of the ECG signal, pressure waveform or some other signal indicative of heart activity.

It will be appreciated that the flowchart shown in figure 3 does not depict a strict temporal order of steps. For example, the processor may locate the reflection feature before selecting the referencing feature, or vice versa.

The applicant hereby discloses in isolation each individual feature described herein and any combination of two or more such features, to the extent that such features or combinations are capable of being carried out based on the present specification as a whole in the light of the common general knowledge of a person skilled in the art, irrespective of whether such features or combinations of features solve any problems disclosed herein, and without limitation to the scope of the claims. The applicant indicates that aspects of the present invention may consist of any such individual feature or combination of features. In view of the foregoing description it will be evident to a person skilled in the art that various modifications may be made within the scope of the invention.

Claims

1. Apparatus for calculating blood pressure, comprising:

a sensor unit comprising at least one sensor configured to obtain a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity; and a processor coupled to the sensor unit and configured to: analyse the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; select a reference feature from the set of one or more signals having a known relationship to the reflection feature; derive a time interval between the selected reference feature and the reflection feature; and calculate a measurement of blood pressure using the time interval.

2. Apparatus as claimed in claim 1, wherein the reference feature is a feature of the blood pressure waveform.

3. Apparatus as claimed in claim 2, wherein the reference feature is indicative of an incident pressure wave and the reflection feature is indicative of a corresponding reflected pressure wave.

4. Apparatus as claimed in claim 3, wherein the reference and reflection features are indicative of either: i) a systolic wave and systolic wave reflection respectively; or ii) a dicrotic wave and dicrotic wave reflection respectively.

5. Apparatus as claimed in claim 4, wherein the reference and reflection features are indicative of the systolic wave and systolic wave reflection respectively and the processor is configured to use the time interval between the reference and reflection features to calculate a measurement of systolic blood pressure.

6. Apparatus as claimed in claim 4, wherein the reference and reflection features are indicative of the dicrotic wave and dicrotic wave reflection respectively and the processor is configured to use the time interval between the reference and reflection features to calculate a measurement of diastolic blood pressure.

7. Apparatus as claimed in claim 2, wherein the processor is configured to analyse the blood pressure waveform to locate first and second reflection features indicative of first and second reflected pressure waves; select first and second reference features from the blood pressure waveform having known relationships to the first and second reflection features respectively; and derive a first time interval between the first reference feature and first reflection feature; and a second time interval between the second reference feature and second reflection feature.

8. Apparatus as claimed in claim 7, wherein the first reference feature is indicative of an incident pressure wave and the first reflection feature is indicative of a corresponding reflected pressure wave; and the second reference feature is indicative of a second incident pressure wave and the second reflection feature is indicative of a corresponding reflected pressure wave.

9. Apparatus as claimed in claim 7 or 8, wherein the processor is configured to use the first interval to calculate a first measurement of blood pressure and the second time interval to calculate a second measurement of blood pressure.

10. Apparatus as claimed in any of claims 7 to 9, wherein the first reference and reflection features are indicative of a systolic wave and systolic wave reflection respectively and the second reference and reflection features are indicative of a dicrotic wave and dicrotic wave reflection respectively; the processor being configured to calculate a measurement of systolic pressure from the first time interval and to calculate a measurement of diastolic pressure from the second time interval.

11. Apparatus as claimed in any of claims 2 to 10, wherein the reference feature and reflection feature are maxima in the blood pressure waveform and/or the time derivative of the blood pressure waveform.

12. Apparatus as claimed in claim 2, wherein the sensor unit comprises at least one sensor configured to obtain an electrical signal representing electrical activity of the heart and the processor is configured to select the reference feature from the electrical signal.

13. Apparatus as claimed in claim 12, wherein the at least one sensor is an electrocardiograph (ECG) sensor configured to obtain an ECG signal and the processor is configured to select as the reference feature the R-peak of the ECG signal.

14. Apparatus as claimed in claim 12 or 13, wherein the reflection feature is indicative of a reflected systolic wave or a reflected dicrotic wave.

15. Apparatus as claimed in claim 14, wherein the reflection feature is indicative of a reflected systolic wave and the processor is configured to calculate a measurement of systolic blood pressure using the time interval between the selected reference feature and the reflection feature.

16. Apparatus as claimed in claim 14, wherein the reflection feature is indicative of a reflected dicrotic wave and the processor is configured to calculate a measurement of diastolic blood pressure using the time interval between the selected reference feature and the reflection feature.

17. Apparatus as claimed in any of claims 12 to 16, wherein the processor is configured to locate a first reflection feature indicative of a first reflected pressure wave of the blood pressure waveform and a second reflection feature indicative of a second reflected pressure wave of the blood pressure waveform; and to derive a first time interval between the selected reference feature of the electrical signal and the first reflection feature, and to derive a second time interval between the selected reference feature of the electrical signal and the second reflection feature.

18. Apparatus as claimed in claim 17, wherein the processor is configured to use the first interval to calculate a first measurement of blood pressure and the second interval to calculate a second measurement of blood pressure.

19. Apparatus as claimed in any of claims 12 to 18, wherein the processor is further configured to analyse the blood pressure waveform to locate an incident feature indicative of an incident pressure wave and to derive a time interval between the selected reference feature of the electrical signal and the incident feature.

20. Apparatus as claimed in claim 19, wherein the processor is configured to calculate the measurement of blood pressure using the time interval between the reference feature of the electrical signal and the reflection feature and the time interval between the reference feature of the electrical signal and the incident feature.

21. Apparatus as claimed in claim 20, wherein the incident feature is indicative of a systolic wave and the reflection feature is indicative of a systolic wave reflection.

22. Apparatus as claimed in any preceding claim, wherein the sensor unit is configured to obtain said blood pressure waveform by taking a plurality of singular blood pressure waveform readings and averaging those waveform readings.

23. Apparatus as claimed in claim 22, wherein the sensor unit further comprises at least one electrocardiograph (ECG) sensor configured to obtain an ECG signal, the sensor unit being configured to synchronise the waveform readings with R-waves of the ECG signal to thereby use the R-wave as a reference event for averaging the waveform readings.

24. Apparatus as claimed in any preceding claim, wherein the processor is configured to locate reflection features that appear as inflection points in the blood pressure waveform by calculating the first derivative of the blood pressure waveform and to locate the reflection features from extrema within the profile of the first derivative of the blood pressure waveform.

25. Apparatus as claimed in any preceding claim, wherein the processor is configured to locate the reflection feature in the pressure waveform by performing an extremum-detection algorithm comprising the steps of:

26. Apparatus as claimed in claim 25, wherein the processor is configured to determine as the location of the reflection feature the location of an identified extremum.

27. Apparatus as claimed in any preceding claim, wherein the at least one sensor configured to obtain the blood pressure waveform comprises at least one of: a PPG sensor; an acoustic sensor; an arterial catheter.

28. A method of obtaining a measurement of a user’s blood pressure, comprising: obtaining a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity;

analysing the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave;

selecting a reference feature from the set of one or more signals having a known relationship to the reflection feature;

deriving a time interval between the selected reference feature and the reflection feature; and calculating a measurement of blood pressure using the time interval.

29. A method of calculating blood pressure, comprising:

receiving a set of one or more signals indicative of heart activity, the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity;

deriving a time interval between the reflection feature and the selected reference feature; and calculating a measurement of blood pressure using the time interval.

30. Apparatus for calculating blood pressure, comprising an input for receiving a set of one or more signals indicative of heart activity,

5 the one or more signals comprising a blood pressure waveform representing variation in blood pressure over time in response to the heart activity; and a processor configured to: analyse the blood pressure waveform to locate a reflection feature indicative of a reflected pressure wave; select a reference feature from the set of one or more signals having a known relationship to the reflection

10 feature; derive a time interval between the selected reference feature and the reflection feature; and calculate a measurement of blood pressure using the time interval.

Intellectual

Property

Office

Application No: Claims searched:

GB1610539.7

1-30