Determining blood pressure

Abstract

According to one aspect, a computer-implemented method for determining a blood pressure of a subject is provided. The method includes receiving (101) an envelope dataset for the subject. The envelope dataset includes measurements of blood flow in a body part of the subject for various degrees of restriction applied to the body part. The method then includes using (103) a set of probability density functions, PDFs, to determine, for various pairs of systolic and diastolic blood pressure values, the respective likelihoods of obtaining measurements in the envelope dataset. The method finally includes determining (105) that the systolic and diastolic blood pressure of the subject is the pair of systolic and diastolic blood pressure values ​​that results in the highest likelihood for the envelope dataset.

Description

[Technical field]

[0001] This disclosure relates to determining blood pressure in a subject, and in particular to computer-implemented methods, computer program products, and apparatus for determining blood pressure from measurements of blood flow. [Background technology]

[0002] A common method of measuring blood pressure non-invasively involves temporarily restricting blood flow in an artery of a limb (e.g., an arm or leg) by applying external pressure, and then determining measurements of blood flow as the amount of pressure applied to the limb changes. If the external pressure is great enough to completely occlude an artery, even at the peak of systolic blood pressure, there will be no blood flow in parts of the limb distal to the site of restriction. If the external pressure is great enough to occlude an artery during a portion of the cardiac cycle, blood flow through the limb will be highly pulsatile and turbulent rather than laminar. If the external pressure is too low to occlude the artery at any point, blood flow will be laminar and moderately pulsatile. Blood flow restriction is typically achieved by placing an inflatable cuff around one of the patient's limbs, then inflating and deflating the cuff to achieve the desired degree of blood flow restriction.

[0003] There are at least three methods for determining blood flow with varying degrees of restriction. The most common method is the oscillometric method, which works by detecting and measuring small variations in pressure in the cuff caused by changes in the arterial volume due to blood flow. This method is frequently used because it is inexpensive and does not require any sensors or electrical connections in the cuff. The pressure sensor can be located in the blood pressure monitor along with the pump used to inflate the cuff. The drawback of this method is that it is susceptible to artifacts caused by patient movement, since even small disturbances of the cuff will cause pressure changes inside of it that are the same magnitude as the vibrations caused by blood flow.

[0004] Another commonly used method is photoplethysmography (PPG), which uses a light source and photoreceptors to measure minute changes in light absorbance caused by blood flow in the distal portion of the limb. This method is more tolerant in the presence of motion artifacts, but requires the additional connection of a light source and light sensor.

[0005] A third commonly used method uses a microphone to detect sounds made by turbulent flow in blood vessels, the so-called Korotkoff sounds. These sounds are a very good indicator of when the degree of restriction is just large enough to occlude that particular artery segment, but they are also susceptible to the effects of ambient noise. This method is widely used when performing manual non-invasive blood pressure measurements using a stethoscope, cuff, and pressure sensor / gauge to pick up the Korotkoff sounds.

[0006] Other methods of determining blood flow include using Doppler ultrasound to measure the flow velocity within an artery.

[0007] To determine a patient's blood pressure, it is necessary to measure the degree of pulsatility of blood flow at each of the various degrees of restriction (i.e., at various cuff pressure values ​​when using an inflatable cuff). This measurement provides a measurement set including value pairs of the degree of restriction and the corresponding blood flow measurement. In the case of oscillometric measurements, the measurement set would be a static cuff pressure and cuff pressure oscillation amplitude pair (static cuff pressure, cuff pressure oscillation amplitude). The measurement set of value pairs is referred to herein as an "envelope dataset" or a "measured envelope." In the case of an envelope dataset including oscillometric measurements, the envelope dataset may be referred to as an oscillation table.

[0008] Conventional techniques allow for the estimation of systolic and diastolic pressure values ​​from the envelope data set, provided the envelope range is sufficient (i.e., blood flow measurements are obtained for a range of degrees of restriction).

[0009] A currently used method for calculating blood pressure values ​​from measured envelope data is to use interpolation to find reference points in the envelope and derive blood pressure from the locations of the reference points, as described, for example, in U.S. Pat. No. 4,984,577, entitled "Oscillometric non-invasive method for measuring blood pressure and apparatus for automated oscillometric blood pressure measuring." Another method fits a model function to a set of envelope points and derives blood pressure values ​​from parameters obtained by a curve fitting operation, as described, for example, in U.S. Pat. No. 5,704,362, entitled "Method for oscillometric blood pressure determination employing curve fitting." Summary of the Invention [Problem to be solved by the invention]

[0010] One drawback of the conventional method is that it tends to rely heavily on some points of the measured envelope, especially the maximum value of blood flow, and ignore information contained in other points. In particular, blood pressure calculations usually rely heavily on the maximum amplitude and location of the measured envelope. As a result, a single erroneous measurement (especially an erroneous maximum amplitude) can potentially cause a large change in the calculated blood pressure value. This means that the method tends to be non-robust in the case of erroneous envelope points caused by motion artifacts or similar factors.

[0011] Another drawback is that some current methods lack a way to incorporate prior information into the blood pressure calculation, which might include knowledge that certain blood pressure ranges are more likely than others (e.g., diastolic pressure is more likely to be in the range 70...90 mmHg than outside this range), that certain pulse pressure values ​​are more likely than others, or that the patient's blood pressure is more likely to be close to previous measurements rather than significantly different from previous measurements.

[0012] Therefore, there is a need for improved techniques for determining systolic and diastolic blood pressures from an envelope data set that is less dependent on a particular or single measurement point of the envelope data set. [Means for solving the problem]

[0013] Aspects and embodiments of the disclosed techniques are provided below.

[0014] It should be noted that some embodiments also provide a way to include one or more types of prior information in the calculation of blood pressure. In such embodiments, the prior information can be taken into account by using maximum a posteriori estimation techniques from Bayesian statistics.

[0015] According to a first particular aspect, a computer-implemented method for determining a subject's blood pressure is provided. The method includes receiving an envelope data set of the subject, the envelope data set including measurements of blood flow in the subject's body part for each of different degrees of restriction applied to the body part, using a set of probability density functions (PDFs) to determine respective likelihoods for obtaining measurements in the envelope data set for different pairs of systolic and diastolic blood pressure values, and determining the subject's systolic and diastolic blood pressures as the pair of systolic and diastolic blood pressure values ​​that results in the highest likelihood for the envelope data set. The first aspect thereby provides a technique for determining measurements of systolic and diastolic blood pressure without relying too much on a particular or single measurement point in the envelope data set. As a result, the technique is less susceptible to errors resulting from outlying measurements and other errors in the envelope data set.

[0016] In some embodiments, using the set of PDFs comprises determining, for each measurement of the envelope dataset, a respective probability for obtaining the measurement of the envelope dataset for various pairs of systolic and diastolic blood pressure values, and determining, for each pair of systolic and diastolic blood pressure values, a likelihood for obtaining the measurement of the envelope dataset by combining the respective determined probabilities for obtaining the measurement of the envelope dataset for the pair of systolic and diastolic blood pressure values. This embodiment provides that the systolic and diastolic blood pressures are determined according to the joint probability of each measurement of the resulting envelope dataset.

[0017] In such an embodiment, combining the respective probabilities includes multiplying the respective probabilities together or summing the logarithms of the respective probabilities. An embodiment that includes summing the logarithms has an advantage when identifying the systolic and diastolic pressure values ​​as those having the highest sum.

[0018] In some embodiments, using the set of PDFs comprises determining the respective likelihoods of obtaining a measurement of the envelope data set for all possible pairs of systolic and diastolic blood pressure values. This embodiment has the advantage that no optimization algorithm is required to determine the systolic and diastolic blood pressure values, and by exhaustively searching all possible pairs of systolic and diastolic blood pressure values ​​it may be possible to quantify the quality of the determined systolic and diastolic blood pressure values, for example by determining the prominence of the most likely one.

[0019] In an alternative embodiment, using the set of PDFs and determining the systolic and diastolic blood pressures comprises using an iterative optimization algorithm to determine the highest likelihood of the envelope data set. Such an embodiment has the advantage that the systolic and diastolic blood pressure values ​​can be determined more quickly and / or efficiently than using an exhaustive search through all possible combinations of systolic and diastolic blood pressures.

[0020] In some embodiments, using the set of PDFs includes determining a blood flow scaling factor based on one or more highest measurements of blood flow in the envelope dataset, and using the blood flow scaling factor to align blood flow values ​​in the set of PDFs with blood flow measurements in the envelope dataset. In such embodiments, it is possible to normalize the set of PDFs to the envelope dataset, or to normalize the envelope dataset to the set of PDFs. In such embodiments, determining the blood flow scaling factor includes determining a blood flow scaling factor based on a function that median represents a number of highest measurements of blood flow in the envelope dataset. Using multiple measurements of blood flow to determine the blood flow scaling factor makes the scaling less susceptible to outlying measurements and other errors.

[0021] In some embodiments, the method further comprises generating a set of PDFs. In such embodiments, generating the set of PDFs comprises generating the set of PDFs from a prior distribution for the systolic and diastolic blood pressures. In such embodiments, the prior distribution includes or is based on one or more of the following: information on physiologically possible values ​​of the systolic and diastolic blood pressures, information on physiologically possible differences between the systolic and diastolic blood pressure values, one or more previous values ​​of the systolic and diastolic blood pressures of the subject, one or more characteristics of the subject, one or more characteristics of the sensor used to measure the blood flow, and one or more characteristics of the device used to apply the restriction to the body part. In this way, the PDFs can be constructed such that the likelihood of the subject's systolic and diastolic blood pressure being considered to be outside the normal physiological range of the subject or population is relatively low unless strongly supported by the measured envelope data set.

[0022] In some embodiments, the set of PDFs includes PDFs scaled with a PDF scaling factor.

[0023] In some embodiments, varying degrees of restriction are applied to the body part using a cuff that is inflated and / or deflated to multiple different pressure steps or continuously inflated or deflated. In such embodiments, the envelope data set is a set of measurement pairs, which include measurements of blood flow in the body part and the respective pressures in the cuff at which the blood flow measurements are made. In such embodiments, the blood flow measurements are the amplitude of pressure oscillations in the cuff or measurements of pressure oscillations in a fluid or gel filled pad below the cuff. Alternatively, in such embodiments, the blood flow measurements are obtained from a photoplethysmography, PPG signal from a PPG sensor placed distal to the cuff on the body part. Alternatively, in such embodiments, the blood flow measurements are obtained from an acoustic signal from a sound or ultrasound sensor placed distal to the cuff on the body part.

[0024] According to a second aspect, a computer program product is provided, comprising a computer readable medium having computer readable code embodied therein, the computer readable code being configured to cause the computer or processing unit to perform the method according to the first aspect or any embodiment of the first aspect when executed by a suitable computer or processing unit. According to a third particular aspect, an apparatus is provided configured to determine a blood pressure of a subject. The apparatus comprises a processing unit configured to receive an envelope dataset of the subject, comprising measurements of blood flow of a body part of the subject for each of different degrees of restriction applied to the body part, use a set of probability density functions (PDFs) to determine respective likelihoods for obtaining measurements of the envelope dataset for different pairs of systolic and diastolic blood pressure values, and determine the systolic and diastolic blood pressures for the subject as being the pair of systolic and diastolic blood pressure values ​​that results in the highest likelihood for the envelope dataset. The third aspect thereby provides an apparatus for determining measurements of systolic and diastolic blood pressure without significant reliance on a particular or single measurement point of the envelope dataset. As a result, the system is less susceptible to errors resulting from outlying measurements and other errors in the envelope data set.

[0025] In some embodiments, the processing unit is configured to use the set of PDFs by determining, for each measurement in the envelope dataset, for various pairs of systolic and diastolic blood pressure values, respective probabilities of obtaining the measurement in the envelope dataset, and by combining, for each pair of systolic and diastolic blood pressure values, the determined respective probabilities of obtaining the measurement in the envelope dataset, this embodiment enables the systolic and diastolic blood pressures to be determined according to the joint probabilities of each measurement in the resulting envelope dataset.

[0026] In such an embodiment, the processing unit may be configured to combine the respective probabilities by multiplying them together or by summing the logarithms of the respective probabilities. An embodiment involving summing the logarithms has an advantage in case the systolic and diastolic pressure values ​​are identified as those whose sum is the highest. In some embodiments, the processing unit is configured to use the set of PDFs by determining for all possible pairs of systolic and diastolic pressure values ​​the respective likelihood of obtaining a measurement of the envelope data set. This embodiment has the advantage that no optimization algorithm is required to determine the systolic and diastolic pressure values, and that by exhaustively searching through all possible pairs of systolic and diastolic pressure values, it may be possible to quantify the quality of the determination of the systolic and diastolic pressure values, for example by determining the prominence of the highest probability.

[0027] In an alternative embodiment, the processing unit is configured to use the set of PDFs to determine the systolic and diastolic blood pressures by using an iterative optimization algorithm to determine the highest likelihood of the envelope data set. Such an embodiment has the advantage that the systolic and diastolic blood pressure values ​​can be determined more quickly and / or efficiently than using an exhaustive search through all possible combinations of systolic and diastolic blood pressures.

[0028] In some embodiments, the processing unit is configured to use the set of PDFs by determining a blood flow scaling factor based on one or more highest measurements of blood flow in the envelope dataset and using the blood flow scaling factor to align blood flow values ​​in the set of PDFs with the measurements of blood flow in the envelope dataset. In such embodiments, it is possible to normalize the set of PDFs to the envelope dataset or normalize the envelope dataset to the set of PDFs. In such embodiments, the processing unit may be configured to determine the blood flow scaling factor based on a function that median-represents a number of highest measurements of blood flow in the envelope dataset. Using multiple measurements of blood flow to determine the blood flow scaling factor makes the scaling less susceptible to outlying measurements and other errors.

[0029] In some embodiments, the processing unit is further configured to generate the set of PDFs. In such embodiments, the processing unit may be configured to generate the set of PDFs by generating the set of PDFs from a prior distribution for the systolic and diastolic blood pressures. In such embodiments, the prior distribution includes or is based on one or more of the following: information on physiologically possible values ​​of the systolic and diastolic blood pressures, information on physiologically possible differences between the systolic and diastolic blood pressure values, one or more previous values ​​of the systolic and diastolic blood pressures of the subject, one or more characteristics of the subject, one or more characteristics of the sensor used to measure the blood flow, and one or more characteristics of the device used to apply the restriction to the body part. In this way, the PDFs may be constructed such that the likelihood that the systolic and diastolic blood pressures of the subject are considered to be outside the normal physiological range of the subject or population is relatively low unless strongly supported by the measured envelope dataset.

[0030] In some embodiments, the set of PDFs includes PDFs scaled with a PDF scaling factor.

[0031] In some embodiments, various degrees of restriction are applied to the body part using a cuff that is inflated and / or deflated to a number of different pressure steps or is continuously inflated or deflated. In such embodiments, the envelope data set is a set of measurement pairs, which include measurements of blood flow in the body part and the respective pressures in the cuff at which the blood flow measurements are made. In such embodiments, the blood flow measurements are the amplitude of pressure oscillations in the cuff, or measurements of pressure oscillations in a fluid or gel filled pad below the cuff. Alternatively, in such embodiments, the blood flow measurements are obtained from a photoplethysmography, PPG signal from a PPG sensor placed on the body part distal to the cuff. Alternatively, in such embodiments, the blood flow measurements are obtained from an acoustic signal from a sound or ultrasound sensor placed on the body part distal to the cuff.

[0032] In some embodiments, the processing unit is configured to receive the envelope data set from a pressure sensor measuring the pressure in the cuff. In such embodiments, the device may further comprise the pressure sensor or the pressure sensor may be separate from the device. In alternative embodiments, the processing unit is configured to receive a measure of the degree of restriction applied to the body part from a pressure sensor measuring the pressure in the cuff and a measure of blood flow in the body part from a blood flow sensor, such as a PPG sensor, a sound sensor, or an ultrasound sensor. In such embodiments, the device may further comprise the pressure sensor and / or the blood flow sensor or the pressure sensor and / or the blood flow sensor may be separate from the device.

[0033] These and other aspects will be apparent from and elucidated with reference to the embodiments described hereinafter.

[0034] Exemplary embodiments will now be described, by way of example only, with reference to the following drawings: [Brief description of the drawings]

[0035] [Figure 1]1 is a graph illustrating an exemplary envelope data set. [Diagram 2] FIG. 1 is a block diagram of an apparatus that can be used to implement the techniques described herein. [Diagram 3] 1 is a flow chart illustrating a method for determining blood pressure, according to various embodiments. [Figure 4] FIG. 1 shows a set of probability density functions (PDFs) for normalized oscillation amplitude over a range of cuff pressure values, at fixed systolic and diastolic blood pressures. [Diagram 5] FIG. 11 is a map illustrating the likelihood of obtaining an envelope data set for various combinations of systolic and diastolic blood pressures. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0036] As mentioned above, the disclosed techniques allow the systolic and diastolic blood pressures to be determined from an envelope dataset with less reliance on specific or single measurement points of the envelope dataset than conventional techniques.

[0037] The envelope dataset includes measurements of blood flow in a body part for various degrees of restriction applied to the body part. For example, if a cuff is used to apply pressure to the subject's upper arm, the blood flow is measured downstream of the cuff, i.e., in a part of the arm distal to the cuff (e.g., the lower arm or wrist). The envelope dataset thus consists of a series of "measurement pairs" or "measurement points", each pair including a measurement of blood flow and a measurement of the pressure applied when the blood flow measurement was obtained (e.g., the static pressure in the cuff that is restricting the blood flow). In the case of oscillometric measurements, the envelope dataset includes pairs of static cuff pressure and cuff pressure oscillation amplitude (static cuff pressure, cuff pressure oscillation amplitude). A set of measurement pairs is referred to herein as an "envelope dataset" or "measured envelope". The form of blood flow measurements may vary depending on the type of sensor used to measure blood flow. The sensor should specifically measure pulsatile blood flow. Measurements of blood flow may be obtained, for example, by sensors that measure or detect changes in volume (such as plethysmography, including photoplethysmography (PPG), oscillometric methods, or pressure sensing using liquid or gel-filled pads placed between an inflatable cuff and a body part) or by sensors that measure flow velocity (such as Doppler ultrasound, or, to some extent, acoustic detection of Korotkoff sounds).

[0038] The graph in Figure 1 shows a portion of an exemplary envelope data set including measurements of oscillation amplitude in mmHg versus cuff pressure (also in mmHg) for a series of cuff pressures ranging from 70 mmHg to 160 mmHg. The subject's systolic blood pressure was 132 mmHg and diastolic blood pressure was 74 mmHg, as measured using a reference blood pressure measurement technique. It will be appreciated that the envelope data set may also include measurements of blood flow at cuff pressures outside the range shown in Figure 1.

[0039] In a technique for determining / estimating systolic and diastolic blood pressures, the measurement pairs of the envelope dataset are considered to be samples of a probability distribution that depends on the variables "systolic blood pressure" and "diastolic blood pressure". The technique uses available measurement pairs of the envelope dataset, and optionally any prior information about the distribution of systolic and diastolic blood pressures, to find systolic and diastolic blood pressure values ​​that maximize the likelihood of observing the measured envelope dataset.

[0040] In this method, more specifically, each individual observation x corresponding to a measurement pair of the envelope data set is i Given a set of x, estimate the vector of population parameters θ (corresponding to the combination of systolic and diastolic blood pressures) by finding the value of θ that maximizes the joint probability of observation x from the sampling distribution f, and optionally the prior distribution g(θ).

number

number

[0041] The joint probability can be calculated as the product of the individual probabilities. However, in certain embodiments, it can be considered that the measurements of the envelope data set are not statistically independent, in which case the calculation of the joint probability can be more complicated than simply multiplying the individual probabilities. In this case, instead of multiplying the probabilities, the maximum can also be found by maximizing the sum of the logarithms of the individual probabilities, which avoids a long series of multiplications that are not well behaved numerically.

number

[0042] Empirically, f should represent the relationship that when the cuff pressure is outside the diastolic-systolic range, the blood flow is more likely to be non-pulsatile or weakly pulsatile, and within the diastolic-systolic range, the flow is more likely to be strongly pulsatile. In some preferred embodiments, f can be constructed as a composition of smooth functions such as logistic functions or bell curves, which facilitates the use of standard optimization algorithms to find the maximum. In some embodiments, the construction of f also takes into account some information available from the envelope data set itself, such as the range of minimum and maximum oscillation amplitudes.

[0043] Once f is constructed, one can use one of several possible methods to find the global maximum of f over the set of valid systolic and diastolic pairs. Since the set of possible systolic and diastolic pressure value pairs is limited in size, some embodiments provide for an exhaustive search of the entire set with moderate computational power by calculating likelihood values ​​for all possible pairs and finding the maximum. Standard optimization algorithms such as interior point methods or gradient search can also be used. Finally, depending on the choice of f, analytical solutions may be available.

[0044] Thus, the presented technique minimizes the reliance on individual measurement points of the envelope data set when calculating the systolic and diastolic blood pressures. The technique also efficiently uses more or all of the available information, including measurement pairs of the envelope data set with no or little blood flow measurements (e.g., no or little oscillation amplitude). While the information contained in such measurement points is usually ignored in "maximum amplitude algorithm" type approaches, in the disclosed technique such measurement points will contribute information and the maximum of the likelihood function will become more prominent. For example, a measurement pair of the envelope data set with a cuff pressure X and an amplitude close to zero will make the blood pressure range not including X more likely and the blood pressure range including X more unlikely.

[0045] FIG. 2 is a block diagram of a system 2 according to various embodiments for determining the blood pressure of a subject according to the techniques described herein. The system 2 forms a non-invasive blood pressure (NIBP) monitoring system 2, i.e. a system for non-invasively measuring the blood pressure of a subject. The blood pressure is determined from an envelope data set that includes measurements of blood flow in a body part of the subject for each of various degrees of restriction applied to the body part, as mentioned above. In the embodiment shown in FIG. 2, a cuff 4 is used to apply the restriction to the body part. However, it will be understood that other types of devices can be used to apply the restriction to the body part and change the blood flow in the body part. For example, a tourniquet can be used that is electrically, mechanically, pneumatically or hydraulically actuated to tighten and change the applied pressure.

[0046] The system 2 of FIG. 2 comprises a cuff 4, a pump 6 connected to the cuff 4 (e.g., via a connecting tube 7), and a cuff pressure sensor 8 for measuring the pressure in the cuff 4. The cuff pressure sensor 8 outputs a cuff pressure signal representative of or related to the pressure in the cuff 4 over time. The cuff pressure signal provides a measurement of the various degrees of restriction applied to the body part of the envelope data set. The cuff 4 is to be placed around the subject's body part of interest, for example around a limb, such as an arm or leg, and the pump 6 is controllable to selectively inflate the cuff 4. The pump 6 can also selectively deflate the cuff 4 and / or a valve (not shown) is provided to allow the cuff 4 to be deflated. The cuff pressure sensor 8 measures the pressure in the cuff 4 at least when the cuff 4 is providing a predetermined stimulus to the artery located inside the cuff 4 (this can be during inflation of the cuff 4, during deflation of the cuff 4, or while the pressure in the cuff 4 is held at a particular level).

[0047] The cuff 4, the pump 6 and the cuff pressure sensor 8 can be considered to form a measuring device 10. In addition to the measuring device 10, the system 2 shown in Fig. 1 also comprises an apparatus 12, operating according to the techniques described herein, for determining the subject's blood pressure from an envelope data set provided by the measuring device 10. The apparatus 12 is thus configured to receive a cuff pressure signal from the cuff pressure sensor 8. In some embodiments, the apparatus 12 is configured to control the operation of the pump 6, thereby initiating inflation / deflation of the cuff 4 at the appropriate times. Although the apparatus 12 is shown in Fig. 2 separately from the measuring device 10, it will be understood that in some implementations the measuring device 10 can be part of the apparatus 12 and vice versa.

[0048] The apparatus 12 may be in the form of, or may be part of, a computing device such as a server, desktop computer, laptop, tablet computer, smartphone, smart watch, or other type of device typically found in a clinical environment, such as a patient monitoring device (e.g., a monitoring device located at a patient's bedside in a clinical environment) used to monitor (and optionally display) various physiological characteristics of a subject / patient.

[0049] The system 2 comprises a blood flow sensor for measuring blood flow (or pulsatile blood flow) in a body part of a subject. In some embodiments, the measurements of blood flow in the body part forming the envelope data set are measurements of the amplitude of pressure oscillations in the cuff 4 due to blood flow. In such embodiments, the measurements of blood flow are obtained from the cuff pressure signal, and thus the cuff pressure sensor 8 is used as the blood flow sensor. This type of blood flow measurement is also called oscillometric measurement or oscillometric method.

[0050] In an alternative embodiment, the measurements of blood flow in the body part forming the envelope data set are obtained using a blood flow sensor in the form of a PPG sensor 14 providing a PPG signal. This type of blood flow measurement is also called photoplethysmography. Thus, in such an embodiment, the system 2 comprises a cuff pressure sensor 8 and one or more PPG sensors 14. The one or more PPG sensors 14 may be part of the measurement device 10. The PPG sensor 14 should be placed on the subject's body distal to the cuff 4 and outputs a PPG signal related to the blood flow through that part of the body. The PPG sensor 14 comprises one or more optical sensors and typically one or more light sources, as known to those skilled in the art. The PPG signal output from the PPG sensor 14 is a raw measurement signal from the optical sensor, and the PPG signal may be, for example, an analog or digital signal representing light intensity over time.

[0051] In another alternative embodiment, the measurements of blood flow in the body part forming the envelope data set are obtained using a blood flow sensor in the form of a pressure sensor that measures the pressure or pressure oscillations in a pad filled with fluid or gel, placed between the inflatable cuff 4 and the body part. This type of sensor provides a hydraulic connection between the body part and the cuff 4 and provides a better quality signal than a pneumatic connection between the body part and the air-filled cuff 4. The pressure sensor used to measure the pressure / pressure oscillations in the pad may be of the type typically used to invasively measure pressure in an artery. An example of such a blood flow sensor, also known as a "shell cuff", is described in "Clinical Evaluation of a High-fidelity Upper Arm Cuff to Measure Arterial Blood Pressure during Noncardiac Surgery" by Josef Briegel, MD et al., Anesthesiology, November 2020, Vol. 133, pp. 997-1006. The pressure sensor may be part of the measurement device 10. The pressure sensor may output a pressure signal related to the pressure in the pad due to pressure exerted by the cuff 4 and the blood flow of the body part passing through the pad and under the cuff 4. The pressure signal output from the pressure sensor is the raw measurement signal from the pressure sensor, the pressure signal may be an analog or digital signal, e.g., representing the pressure / pressure oscillations in the pad over time.

[0052] In other alternative embodiments, the measurements of blood flow in the body part forming the envelope data set are obtained using a blood flow sensor in the form of one or more microphones 16 (to measure sounds of blood flow, including Korotkoff sounds caused by turbulent flow in arteries) or one or more ultrasonic sensors 18 (to measure blood flow velocity from Doppler shift). This type of blood flow measurement is also called auscultatory measurement. Thus, in such an embodiment, the system 2 comprises a cuff pressure sensor 8 and one or more microphones 16 or ultrasonic sensors 18. The one or more microphones 16 or the one or more ultrasonic sensors 18 may be part of the measurement device 10. The sensor 16, 18 should be placed on the subject's body distal to the cuff 4 and output a sound signal related to the blood flow passing through that part of the body. The sound signal output from the microphone 16 or ultrasonic sensor 18 is the raw measurement signal from the sensor 16, 18, the sound signal may be, for example, an analog or digital signal representing sound / ultrasound over time.

[0053] Those skilled in the art will appreciate that other types of (non-invasive) sensors can be used to obtain measurements of blood flow in a body part, and the disclosed techniques are not limited to the sensors described above.

[0054] The device 12 comprises a processing unit 22, which may be configured to control the operation of the device 12 and to implement or execute the methods described herein. The processing unit 22 may be implemented in many ways by software and / or hardware to perform the various functions described herein. The processing unit 22 comprises one or more microprocessors or digital signal processors (DSPs) programmed using software or computer program code to perform the required functions and / or to control the components of the processing unit 22 to achieve the required functions. The processing unit 22 may be implemented as a combination of dedicated hardware (e.g., amplifiers, preamplifiers, analog-to-digital converters (ADCs) and / or digital-to-analog converters (DACs)) to perform some functions and processors (e.g., one or more programmed microprocessors, controllers, DSPs, and associated circuitry) to perform other functions. Examples of components that may be used in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, DSPs, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), hardware for implementing neural networks, and / or so-called artificial intelligence (AI) hardware accelerators (i.e., processors or other hardware designed specifically for AI applications that can be used in conjunction with a main processor).

[0055] Processing unit 22 is coupled to a memory unit 24 that may store data, information, and / or signals used by processing unit 22 in controlling the operation of device 12 and / or in implementing or executing the methods described herein. In some implementations, memory unit 24 stores computer readable code that may be executed by processing unit 22 to cause processing unit 22 to perform one or more functions, including the methods described herein. In certain embodiments, the program code may be in the form of an application for a smartwatch, smartphone, tablet, laptop, or computer. Memory unit 24 may comprise any type of non-transitory machine-readable medium, such as a cache or system memory, including volatile and non-volatile computer memory, such as random access memory (RAM), static RAM (SRAM), dynamic RAM (DRAM), read only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), and electrically erasable PROM (EEPROM), and memory unit 24 may be embodied in the form of a memory chip, an optical disk (such as a compact disk (CD), digital versatile disk (DVD), or Blu-ray disk), a hard disk, a tape storage means, or a solid-state device, including a memory stick, a solid-state drive (SSD), a memory card, and the like.

[0056] In some embodiments, the device 12 includes a user interface 26 that includes one or more components that enable a user of the device 12 to input information, data, and / or commands into the device 12 and / or that enable the device 12 to output information or data to a user of the device 12. Information that may be output from the user interface 26 may include systolic and diastolic blood pressures determined from an envelope data set in accordance with the techniques described herein. The user interface 26 may include any suitable input components, including, but not limited to, a keyboard, a keypad, one or more buttons, switches or dials, a mouse, a trackpad, a touch screen, a stylus, a camera, a microphone, etc., and / or the user interface 26 may include any suitable output components, including, but not limited to, a display screen, one or more lights or light elements, one or more loudspeakers, a vibration element, etc.

[0057] It will be appreciated that actual implementations of the apparatus 12 will comprise additional components to those shown in Figure 2. The apparatus 12 also comprises a power source, such as a battery, or components to enable the apparatus 12 to be connected to a mains power source. The apparatus 12 also comprises interface circuitry to enable data connection to and / or exchange with other devices, including any one or more of the measuring device 10 (or a separate cuff pressure sensor 8), the sensors 14, 16, 18, a server, a database, and a user device.

[0058] The flow diagram of Fig. 3 illustrates a method for determining blood pressure according to an embodiment of the technique described herein. Various steps of the method may be performed by the device 12, specifically the processing unit 22. In that regard, computer program code may be provided that, when executed by the processing unit 22, causes the processing unit 22, and thus the device 12, to perform the method described below. The method of Fig. 3 may generally be implemented by a processor, a processing unit, or a computer. One or more of the steps of the method of Fig. 3 may use or require the use of any of the cuff 4 (or other blood flow restriction device), the pump 6, the cuff pressure sensor 8 (or other "degree of restriction" sensor and / or blood flow sensor), and the PPG sensor 14 (if present), the microphone 16, and the ultrasonic sensor 18, as further mentioned below.

[0059] In step 101 of the method of Fig. 3, an envelope dataset of a subject is received. The envelope dataset comprises measurements of blood flow in a body part of the subject for various degrees of restriction applied to the body part, as mentioned above. The envelope dataset is received in step 101, for example directly from the associated sensor 8, 14, 16, 18 as the blood flow measurements are obtained, in which case the envelope dataset is formed of the received measurements. Step 101 alternatively comprises retrieving a previously obtained envelope dataset from a memory, for example the memory unit 24.

[0060] In step 103, a set of probability density functions (PDFs) is used to determine the respective likelihoods of obtaining an envelope data set measurement for various pairs of systolic and diastolic blood pressure values. As mentioned above, PDFs are used to quantify the likelihood of a particular envelope data set occurring for various pairs of systolic and diastolic blood pressure values. The PDFs in the set are functions of one or more variables: a blood flow measurement (e.g., oscillation amplitude), a degree of restriction (e.g., cuff pressure), systolic blood pressure, and diastolic blood pressure. In one embodiment, the PDFs in the set are functions of five variables: a blood flow measurement (e.g., oscillation amplitude), a degree of restriction (e.g., cuff pressure), systolic blood pressure, diastolic blood pressure, and maximum oscillation amplitude. In one embodiment, the PDFs in the set are functions of six variables: a blood flow measurement (e.g., oscillation amplitude), a degree of restriction (e.g., cuff pressure), systolic blood pressure, diastolic blood pressure, maximum oscillation amplitude, and a parameter representing prior information. Because the blood flow measurement is the key unknown variable obtained by measurement, the PDFs are used to quantify the likelihood that a particular blood flow measurement at a particular degree of restriction corresponds to various systolic and diastolic blood pressures. Thus, each PDF in the set of PDFs can be considered to be for a particular value of the degree of restriction and a particular pair of systolic and diastolic blood pressure values.

[0061] Using the terminology of equations (1) and (2), the set of PDFs quantifies how likely it is to observe the set of measurement pairs x and the parameter vector θ (corresponding to a combination of systolic and diastolic blood pressures) that form the envelope data set. The set of PDFs can be viewed as being formed of a series of subsets of PDFs, where each subset of PDFs corresponds to a particular pair of systolic and diastolic blood pressure values ​​(e.g., 120 mmHg systolic and 80 mmHg diastolic blood pressure), and each PDF in the subset is the PDF for that pair of systolic and diastolic blood pressure values ​​for a particular cuff pressure value (e.g., 100 mmHg).

[0062] An example subset of PDFs f(x i |θ) is shown in FIG. 4. This subset of PDFs is composed of a respective PDF for each cuff pressure value. Based on this exemplary subset of PDFs, it is very likely to measure a low (or zero) oscillation amplitude far outside the interval of cuff pressures from 80 mmHg to 120 mmHg, and very unlikely to measure a high oscillation amplitude. The opposite is true for cuff pressures within the interval of 80 mmHg to 120 mmHg. Normalization is discussed further below, and in some embodiments, the oscillation amplitude normalization is performed using a number of highest magnitude oscillation amplitudes, for example using the median of a number of highest N magnitude oscillation amplitudes. The normalization is therefore not based only on the maximum observed oscillation amplitude, so that the maximum observed oscillation amplitude will be normalized to a value greater than 1. This is taken into account by the set of PDFs also covering normalized amplitudes greater than 1. For a "good quality" measurement, there will usually be another envelope point close to the maximum value. Therefore, even if normalizing using the median, the maximum oscillation amplitude will be close to 1. However, in artifactual measurements, the maximum value may be an outlier, with a normalized value significantly larger than 1. This large value is equally unlikely for any combination of systolic / diastolic pressures, and will therefore have little effect on the location of the maximum likelihood.

[0063] Thus, in step 103, the set of PDFs is used to determine the likelihood of pairs of systolic and diastolic blood pressure values ​​for the received envelope data set. In other words, step 103 provides the likelihood that the subject has various pairs of systolic and diastolic blood pressure values ​​given the received envelope data set. Each likelihood determined in step 103 is also referred to herein as a "joint likelihood" since it represents the likelihood of all measurement pairs in the envelope data set occurring for a given systolic and diastolic blood pressure value pair.

[0064] A particular embodiment of step 103 has a two-step process for determining the respective likelihoods. In a first step, for each measurement pair of the envelope data set, respective probabilities of obtaining the measurement pair of the envelope data set for various pairs of systolic and diastolic blood pressure values ​​are determined. That is, for each measurement pair, a first probability is determined for a first pair of systolic and diastolic blood pressure values, a second probability is determined for a second pair of systolic and diastolic blood pressure values, etc. In a second step, for each pair of systolic and diastolic blood pressure values ​​evaluated in the first step, a likelihood of obtaining the envelope data set is determined by combining the probabilities for that pair of systolic and diastolic blood pressure values ​​determined in the first step. In the second step, the probabilities can be combined by multiplying the probabilities together or by summing the logarithms of the respective probabilities. Summing the logarithms of the respective probabilities may be recommended as this avoids a series of multiplications that may occur when multiplying probabilities, which may be numerically poorly behaved.

[0065] In some embodiments, step 103 may include determining the likelihood of each of the envelope data sets for all possible pairs of systolic and diastolic blood pressure values, i.e., all possible combinations of systolic and diastolic blood pressure values ​​are evaluated in step 103. This exhaustive search is possible because the size of the set of pairs of systolic and diastolic blood pressure values ​​is relatively small.

[0066] In alternative embodiments, step 103 may comprise applying an iterative optimization algorithm to determine the respective likelihoods. Such embodiments may typically be used when steps 103 and 105 (described below) are performed in parallel or otherwise simultaneously. In such embodiments, it is not necessary to perform an exhaustive search and determine the likelihoods for all possible pairs of systolic and diastolic blood pressures.

[0067] Returning to the example subset of PDFs in FIG. 4, a highly probable measurement pair under this distribution would increase the likelihood of a blood pressure of 120 / 80 mmHg generating the set of measured envelope points, while a less probable measurement pair (e.g., a small oscillation amplitude at a cuff pressure of 100 mmHg, or a large amplitude at a cuff pressure of 150 mmHg) would decrease the likelihood of a blood pressure of 120 / 80 mmHg generating a measurement pair in the envelope data set.

[0068] Once the respective likelihoods have been determined in step 103, in step 105 the subject's systolic and diastolic blood pressures are determined to be the pair of systolic and diastolic blood pressure values ​​that yield the highest likelihood for the envelope data set.

[0069] In an embodiment of step 103 in which an exhaustive search of possible systolic and diastolic blood pressure pairs is performed, step 105 may comprise determining the subject's systolic and diastolic blood pressures as those pairs that yielded the highest likelihood for the received envelope data set. In an embodiment of step 103 in which an iterative optimization algorithm is used to determine the respective likelihoods, the algorithm may be applied iteratively over the pairs of systolic and diastolic blood pressure values ​​to find the highest likelihood value for the envelope data set. Those skilled in the art will know a variety of different iterative optimization algorithms that can be used to implement steps 103 / 105, but details are not presented here. Nevertheless, techniques such as sequential quadratic programming, interior point methods, or active constraint methods may be used in steps 103 / 105. Depending on the way the likelihood function is constructed, there may be multiple local maxima. In this case, a single run of a local optimization algorithm may reach one of the local maxima rather than the global maximum. A global optimization algorithm may also be used.

[0070] In embodiments where an exhaustive or substantially exhaustive search is performed (e.g. over all plausible combinations of systolic and diastolic pressure values), the method may further comprise a step of determining the prominence or significance of the maximum likelihood. The prominence or significance may be determined by examining a neighborhood of a given size around the maximum. The algorithm may calculate the average or minimum likelihood in the neighborhood and calculate the ratio of this value to the maximum. If the ratio is large, the maximum is considered prominent. If the ratio is small, the maximum is not very distinguishable / prominent / significant. The prominence or significance of the maximum may provide a quantification of the quality of the determination of the systolic and diastolic pressure values. In some embodiments, if the received envelope data set does not provide a sufficiently prominent or significant maximum, the systolic and diastolic pressure values ​​may be ignored, possibly as unreliable or not sufficiently accurate. The exact definition of prominence may depend on the function used to construct the likelihood function. The values ​​can be determined empirically (e.g., using actual envelopes that are known to be good or bad) as well as by simulation (e.g., using synthetic envelopes generated as inputs that represent both known good and bad envelopes), where statistical methods including machine learning can be used to determine optimal thresholds for ignoring / retaining systolic / diastolic pressure values.

[0071] FIG. 5 is a map showing the likelihood of obtaining the envelope data set shown in FIG. 1 for various combinations of systolic and diastolic blood pressures. The likelihoods shown in FIG. 5 are the log-likelihoods of the envelope data set shown in FIG. 1. Thus, FIG. 5 represents the output of an exhaustive search of all possible pairs of systolic and diastolic blood pressures for the envelope data set of FIG. 1 in step 103. From FIG. 5 it can be seen that the highest likelihood for this envelope data set occurs for a systolic blood pressure of 132 mmHg and a diastolic blood pressure of 74 mmHg, and that the determined likelihood drops significantly (even to 0) for very small differences in these values ​​(the likelihood is, for example, close to 0 outside the range of 110-140 mmHg for systolic blood pressure and close to 0 outside the range of 60-80 mmHg for diastolic blood pressure).

[0072] Although not shown in FIG. 3, the systolic and / or diastolic blood pressure values ​​determined in step 105 can be output, for example, visually and / or audibly, to the subject or another user of the method / apparatus 12, such as a physician or other healthcare provider, and / or can be output electronically, for example, in the form of a signal sent to a device or apparatus that stores the systolic and / or diastolic blood pressure values ​​in a patient record for the subject. Each PDF in the set of PDFs can be formed with a smooth function, such as a logistic function and / or a Gaussian bell curve. The use of a smoothing function can facilitate the use of an iterative optimization algorithm to find the maximum likelihood. In a preferred embodiment, the set of PDFs can be constructed such that certain combinations of systolic and diastolic blood pressures have zero probability associated with them. For example, all pairs of systolic and diastolic blood pressures where the diastolic pressure is higher than the systolic pressure should have zero probability (from the definition of systolic and diastolic pressures).

[0073] The exemplary subset of PDFs in FIG. 4 was generated by approximating the expected oscillation amplitude as a Gaussian bell curve, with cuff pressure as a parameter. The location of the maximum on the oscillation amplitude axis, following the axis of cuff pressure, represents this bell curve. A family of bell curves with a maximum mean value is used to define a set of probability density functions, with each bell curve in the family corresponding to a particular pair of systolic and diastolic blood pressure values. Each individual curve is scaled so that the area under the curve is 1 (since a real Gaussian bell curve is defined over the real numbers, but pressure and oscillation amplitude cannot be negative). This is a rough approximation, but even in this form it is a good detector for the envelope data set obtained from 120 / 80 blood pressure.

[0074] Instead of approximating the envelope shape as a Gaussian bell curve, other bell-shaped functions could be used, including the extreme (non-smooth) case of a function that is 0 outside the diastolic / systolic interval and 1 inside the diastolic / systolic interval, which would still result in a detector for a given blood pressure.

[0075] To form the subset of probability density functions, a bell-shaped distribution can be used in the diastolic / systolic interval, while outside this interval a scaled bell-shaped distribution or a one-sided distribution such as an exponential distribution can be used.

[0076] It will be appreciated that in certain embodiments, the functions used to form the PDFs in the set may be selected as a compromise between computational complexity and modeling accuracy: a simple set of basis functions that works "good enough" is preferred over a computationally complex model that provides minimal additional benefit.

[0077] In some embodiments, step 103 comprises scaling the envelope data set or the set of PDFs based on the envelope data set. This scaling is performed using a scaling factor, called the "blood flow scaling factor". The blood flow scaling factor can be used to align the blood flow values ​​of the set of PDFs with the blood flow measurements of the envelope data set. That is, the blood flow scaling factor can be used to normalize the blood flow values ​​of each PDF to the envelope data set or to normalize the blood flow measurements of the envelope data set. Alternatively, the blood flow scaling factor can be used to align the blood flow values ​​of the set of PDFs with the blood flow measurements of the envelope data set by distributing the blood flow scaling factor to both the blood flow values ​​of the set of PDFs and the blood flow measurements of the envelope data set. For example, the blood flow scaling factor can be distributed by multiplying one of (i) the blood flow values ​​of the set of PDFs and (ii) the blood flow measurements of the envelope dataset by the square root of the blood flow scaling factor and dividing the other of (i) the blood flow values ​​of the set of PDFs and (ii) the blood flow measurements of the envelope dataset by the square root of the blood flow scaling factor. The blood flow scaling factor can be based on one or more maximum measurements of blood flow of the envelope dataset. The magnitude of the maximum measurement of blood flow of the envelope dataset typically depends on several factors that may vary between subjects and between measurements of the same subject, such as the subject's physical characteristics, the size of the cuff, the type of cuff, how tightly the cuff is attached to the body part, etc. Thus, the PDF can be fitted with a blood flow scaling factor by using the magnitude of the maximum measurement of blood flow to derive the blood flow scaling factor of the PDF. In some embodiments, the blood flow scaling factor is determined based on the largest magnitude of the blood flow measurement of the envelope dataset. For example, the blood flow measurement that is the maximum amplitude can be scaled to 1 and all other amplitudes can be scaled accordingly.However, since the largest magnitude blood flow measurement may lead to significant errors if it is an artifact, the blood flow scaling factor is preferably determined from multiple largest magnitude blood flow measurements so that the scaling is less susceptible to outlying measurements and other errors. The blood flow scaling factor may be determined from a function of multiple largest magnitude measurements, where the function may be a representative value such as the mean, mode, or median (the median being the preferred option over the mean and mode). The number of measurements used to determine the blood flow scaling factor may be between 2 and 10, e.g., 3, 5, or 7. The blood flow scaling factor is tolerant to one of the maxima being an artifact if the median is taken and the three largest magnitude measurements are used, the blood flow scaling factor is tolerant to two of the maxima being an artifact if the median is taken and the five largest magnitude measurements are used, etc. Determining the blood flow scaling factor from multiple largest magnitude blood flow measurements also contributes to the blood pressure measurement technique being less dependent on a single measurement point of the envelope data set.

[0078] In some embodiments, the method may further comprise generating a set of PDFs. The set of PDFs may be generated as outlined above with respect to FIG. 4. As mentioned above, in some embodiments, prior information regarding the distribution of systolic and diastolic blood pressures may be used or taken into account when generating the PDFs. The prior information is in the form of one or more prior distributions, g(θ). The prior information is based on static information, i.e., information that does not change or does not change quickly. Static information may include, for example, characteristics of the subject (e.g., age, height, weight, medical condition, medications, etc.), characteristics of the sensor used to measure blood flow (e.g., an indication of offset and / or noise present in measurements using that sensor), characteristics of the device (e.g., cuff) used to apply the restriction to the body part, information about physiologically likely values ​​of systolic and diastolic blood pressure, and information about physiologically likely differences between systolic and diastolic blood pressure values. In the last two examples (mentioned above), the probability of occurrence of certain combinations of systolic and diastolic blood pressures is zero, and the probability of occurrence of other combinations of systolic and diastolic blood pressures is near zero or otherwise low. For example, the diastolic blood pressure cannot be higher than the systolic blood pressure, so the probability of any pair where the diastolic blood pressure is higher than the systolic blood pressure should be zero. Certain values ​​of systolic and diastolic blood pressures at both extremes, e.g. very low and very high values ​​outside the known physiological range, may also have zero or near zero probability. Similarly, certain values ​​of pulse pressure (the difference between systolic and diastolic blood pressure) are less likely and should have a near zero or otherwise low probability of occurrence. Similarly, certain values ​​of systolic and diastolic blood pressure, e.g. values ​​within the common physiological range, may be more likely and have a probability in the PDF that makes this more likely. Similarly, certain values ​​of pulse pressure are more likely and the PDF should be constructed accordingly. With this prior information taken into account in the PDF, the likelihood that a subject's systolic and diastolic blood pressure will be considered to be outside the normal physiological range is relatively low, unless strongly supported by a measured envelope dataset.

[0079] In some embodiments, the prior information also or alternatively includes dynamic information, i.e., changing or rapidly changing information. The dynamic information may include one or more previous values ​​of the subject's systolic and diastolic blood pressure. This dynamic information can be used to generate a PDF such that new values ​​of the systolic and diastolic blood pressure detected using the envelope dataset are more likely to be close to the subject's previous values ​​of the systolic and diastolic blood pressure. The exact impact of the subject's previous values ​​on the PDF may vary depending on how recent the previous values ​​are, since blood pressure changes relatively slowly. Thus, if the previous values ​​are more recent, the systolic and diastolic blood pressures are more likely to indicate the current values ​​than if the previous values ​​are older.

[0080] Although the strict mathematical definition of a PDF requires that the integral over the range is 1, it will be appreciated that the above method can be implemented using a PDF scaled by some PDF scaling factor such that the integral over the range is greater than (or less than) 1. It will also be appreciated that step 103 uses a set of PDFs accordingly. That is, step 103 comprises using a set of PDFs, i.e., using a set of PDFs whose probability axes are scaled by a scaling factor. For example, referring to FIG. 4, the probability density axis can be scaled such that the highest probability density provided by the set of PDFs is 3. The PDF scaling factor can be any value, e.g. 2, 3, etc., or can be selected to improve the performance of the algorithm implementing steps 103 and 105.

[0081] Thus, a technique is provided for determining systolic and diastolic blood pressures from an envelope data set that is less dependent on a particular measurement point or a single measurement point of the envelope data set than conventional techniques.

[0082] Variations to the disclosed embodiments can be understood and produced by those skilled in the art from a study of the drawings, the disclosure, and the appended claims, when practicing the principles and techniques described herein. In the claims, the word "comprises" does not exclude other elements or steps, nor does the singular element exclude a plurality. A single processor or other unit performs the functions of several items recited in the claims. The mere fact that certain means are recited in mutually different dependent claims does not indicate that a combination of these means cannot be used to advantage. The computer program may be stored or distributed in a suitable medium, such as an optical storage medium or a solid-state medium, provided integrally with or as part of other hardware, or distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. Any reference signs in the claims should not be interpreted as limiting the scope.

Claims

1. 1. A computer-implemented method for determining blood pressure in a subject, the method comprising: receiving an envelope data set for the subject comprising measurements of blood flow in the body part of the subject for each of different degrees of restriction applied to the body part; using a set of probability density functions to determine respective likelihoods of obtaining said measurements in said envelope data set for various pairs of systolic and diastolic blood pressure values; determining the systolic and diastolic blood pressures for the subject as being the pair of systolic and diastolic blood pressure values ​​that yields the highest likelihood for the envelope data set; The method comprising:

2. 2. The method of claim 1, wherein using the set of probability density functions comprises determining a respective likelihood for obtaining the measurements in the envelope data set for every possible pair of the systolic and diastolic blood pressure values.

3. 2. The method of claim 1 , wherein using the set of probability density functions and determining the systolic and diastolic blood pressures comprises using an iterative optimization algorithm to determine the maximum likelihood for the envelope data set.

4. using the set of probability density functions determining a blood flow scaling factor based on one or more highest measurements of blood flow in the envelope data set; using a blood flow scaling factor to align the blood flow values ​​of the set of probability density functions with the measurements of blood flow in the envelope data set.

4. The method according to claim 1 , further comprising:

5. 5. The method of claim 4, wherein determining the blood flow scaling factor comprises determining the blood flow scaling factor based on a function that represents a median of a plurality of the highest measurements of blood flow in the envelope data set.

6. The method of claim 1 , further comprising generating the set of probability density functions from a prior distribution on systolic and diastolic blood pressure.

7. The prior distribution is Information regarding physiologically probable values ​​of systolic and diastolic blood pressure; information regarding physiologically possible differences between systolic and diastolic blood pressure values; one or more previous values ​​of the subject's systolic blood pressure and diastolic blood pressure; one or more characteristics of the subject; One or more characteristics of a sensor used to measure blood flow; and One or more characteristics of a device used to apply the restriction to the body part. The method of claim 6 comprising or based on one or more of:

8. A computer readable medium having computer readable code embodied thereon, the computer readable code, when executed by a suitable computer or processing unit, causing the computer or processing unit to perform a method according to any one of claims 1 to 7.

9. 1. An apparatus for determining a blood pressure of a subject, the apparatus comprising a processing unit, the processing unit comprising: receiving an envelope data set for the subject comprising measurements of blood flow in the body part of the subject for each of different degrees of restriction imposed on the body part; using a set of probability density functions to determine respective likelihoods of obtaining the measurements in the envelope data set for various pairs of systolic and diastolic blood pressure values; The apparatus determines a systolic blood pressure and a diastolic blood pressure for the subject as being the pair of systolic and diastolic blood pressure values ​​that results in a highest likelihood for the envelope data set.

10. 10. The apparatus of claim 9, wherein the processing unit uses the set of probability density functions to determine a respective likelihood for obtaining the measurements in the envelope data set for every possible pair of the systolic and diastolic blood pressure values.

11. 10. The apparatus of claim 9, wherein the processing unit uses the set of probability density functions to determine systolic and diastolic blood pressures by using an iterative optimization algorithm to determine the maximum likelihood of the envelope data set.

12. The processing unit comprises: determining a blood flow scaling factor based on one or more highest measurements of blood flow in the envelope data set; using the blood flow scaling factor to align the blood flow values ​​of the set of probability density functions with the measurements of blood flow of the envelope data set; 12. Apparatus according to claim 9, wherein the set of probability density functions is used by:

13. The apparatus of claim 12 , wherein the processing unit determines the blood flow scaling factor based on a function representative of a median of the highest measurements of blood flow in the envelope data set.

14. The processing unit further comprises:

14. Apparatus according to any one of claims 9 to 13, wherein the set of probability density functions is generated from a prior distribution on systolic and diastolic blood pressure.

15. The prior distribution is Information regarding physiologically probable values ​​of systolic and diastolic blood pressure; information regarding physiologically possible differences between systolic and diastolic blood pressure values; one or more previous values ​​of the subject's systolic blood pressure and diastolic blood pressure; one or more characteristics of the subject; One or more characteristics of a sensor used to measure blood flow; and 15. The apparatus of claim 14, comprising or based on one or more of one or more characteristics of a device used to apply the restriction to the body part.

Images