JP4896015B2 - Real-time measurement of ventricular stroke volume variance by continuous arterial pulse contour analysis - Google Patents

Description

(Cross-reference to related applications)
No. 10 / 728,705 filed on Dec. 5, 2003, and 10 / 890,887 filed Jul. 14, 2004, which are co-pending. Some continuation applications claim priority to them, and they are incorporated by reference.

(Background of the Invention)
(Field of Invention)
The present invention relates generally to cardiac monitoring, and in particular to ventricular stroke volume variance (SVV) estimation and a system for performing this method.

(Background technology)
Stroke volume (SV), cardiac output (CO), etc. are not only for disease diagnosis, but also for “real time” monitoring of the status of both human and animal subjects, including patients. Is also an important indicator. Thus, few hospitals lack a device for monitoring one or more of these cardiac parameters. Many techniques are used-invasive and non-invasive, and a combination of both-and even more are proposed in the literature.

  Most of the techniques used to measure SV can usually be easily adapted to provide an estimate of CO as well. This is because CO is generally defined as SV times the heart rate HR normally available to monitoring devices. Conversely, most devices that estimate CO also estimate SV as a substep.

  As will be explained in more detail below, yet another cardiac parameter that is promising for providing clinically relevant information is stroke volume variance SVV. One way to estimate SVV is to simply collect multiple SV values and calculate the difference from measurement interval to measurement interval.

  One common method for measuring SV or CO is to attach a specific flow measurement device onto the catheter, and then pass the catheter through the subject, and that device is the subject's heart. Maneuver to be in or near. Some such devices have a bolus injection of either material or energy (usually heat) at the upstream location, such as the right atrium, and material injected at the downstream location, such as the pulmonary artery or Determine flow based on energy characteristics. Patents disclosing implementation of such invasive techniques (especially thermodilution) are: US Pat. No. 4,236,527 (Newbower et al., December 2, 1980); US Pat. No. 4,507,974 (Yelderman, April 2, 1985) (Patent Document 2); US Pat. No. 5,146,414 (McKown et al., September 8, 1992) (Patent Document 3); And US Pat. No. 5,687,733 (McKown et al., November 18, 1997) (Patent Document 4).

  Still other invasive devices are based on the well-known Fick technique, whereby CO is calculated as a function of arterial and mixed venous blood oxygenation.

  Invasive techniques have obvious drawbacks, one of which is, of course, cardiac catheterization, especially for the patient (especially intensive care patient) in which it is performed, with certain actual or potential Considering that they are often already in hospital because of severe symptoms, they are potentially dangerous. Invasive methods also have a less obvious drawback: some techniques, such as thermodilution, affect the accuracy of the measurement depending on how well they are satisfied. Depends on assumptions such as a uniform distribution of the generated heat. Furthermore, the very introduction of the device into the bloodstream can affect the value (eg, flow rate) that the instrument measures.

  Doppler techniques using invasive and non-invasive transducers are also used to measure flow and to calculate SV and CO from flow measurements. Not only are these systems typically expensive, their accuracy depends on accurate knowledge of the flow channel diameter and overall geometry. Such accurate knowledge, however, is rarely possible, especially under conditions where real-time monitoring is desired.

  Thus, there is a longstanding need for specific methods of determining cardiac parameters such as SV, SVV, etc. that are both non-invasive or at best minimally invasive and accurate. . One blood feature that has proven particularly promising for accurately determining such parameters, with minimal or no invasiveness, is blood pressure.

  Most known blood pressure-based systems rely on so-called pulse contouring (PCM), which is characterized by the beat-to-beat pressure waveform characteristics and the desired cardiac parameters ( Calculate one or more estimates. PCM typically builds a linear or non-linear hemodynamic model of the aorta using “Windkessel” (“Air Chamber German”) parameters (characteristic aortic impedance, compliance, and total peripheral resistance) To do. Basically, blood flow is analogous to current flow in a circuit where the impedance is in series with a resistance and capacitance (compliance) connected in parallel. The three required parameters of this model are usually “anthropometric” data, experimentally or edited by a complex calibration process, ie about the age, gender, height, weight, etc. of other patients or test subjects. Determined from any of the data. US Patent No. 5,400,793 (Wesseling, March 28, 1995) (Patent Document 5) and US Patent No. 5,535,753 (Petlucelli et al., July 16, 1996) (Patent Document) 6) is representative of a system that relies on the Windkessel circuit model to determine CO.

  PCM-based systems can monitor SV-derived cardiac parameters more or less continuously without the need to leave a catheter in the patient (usually in the right heart). In fact, some PCM systems operate using blood pressure parameter measurements taken using a finger cuff. One drawback of PCM, however, is that it is less accurate than the rather simple three-parameter model from which it is derived: in general, the complex pattern of pressure wave reflections due to multiple impedance mismatches caused by, for example, arterial bifurcations A much higher order model may be required to faithfully explain other phenomena such as Since the accuracy of the basic model is usually not good enough, many improvements have been proposed with varying complexity.

  "Method and apparatus for measuring cardiac output" disclosed by Salvatore Romano in published US Patent Application No. 2002022785A1 (February 21, 2002, "Method and apparatus for measuring cardiac output" The device ") is a PCM technique that estimates SV either invasively or non-invasively as a function of the ratio between the area under the overall pressure curve and the linear combination of the various components of impedance. Presents different attempts to improve In an attempt to account for pressure reflexes, the Romano system not only relies on an accurate estimate of the inherently noisy derivative of this pressure function, but also a numerical value for a series of experimentally determined average pressure values. It also depends on the adjustment.

  Fluid administration in hemodynamically unstable patients is often a major challenge when it comes to measuring SV, CO, or other hemodynamic parameters in real time. Accurate clinical assessment of blood volume deficiencies, such as the decision to take fluid emergency resuscitation as an initial treatment strategy, is difficult. Specifically, it is very difficult to predict whether a hemodynamically unstable patient will respond to fluid therapy with an increase in stroke volume and cardiac output. Furthermore, fluid overload can cause significant lung or heart dysfunction, while fluid deficiency can cause tissue damage resulting in fatal organ failure. Patient fluid responsiveness is the primary and most important determinant for assessing the suitability of fluid emergency resuscitation therapy and for ensuring optimal cardiac performance and organ perfusion.

  Many bedside indicators of ventricular preload are used as predictors of fluid responsiveness. Right arterial pressure (RAP) and pulmonary artery occlusion pressure (PAOP) are most commonly used in an intensive care unit (ICU) when deciding to administer fluid. Other bedside indicators of ventricular preload include right ventricular end diastolic volume (RVE V) and left ventricular end diastolic volume (LVEDA) measured transesophageally by echocardiography. Some studies and case reports, however, show that these static index values based on these heart filling pressures have poor predictors and often do not give adequate information about fluid responsiveness It was.

  Recently, several studies have shown the clinical importance of monitoring the variance observed in left ventricular stroke volume resulting from cardiovascular and lung interactions under mechanical ventilation. confirmed. These stroke volume variances (SVV) are caused by periodic increases and decreases in intrathoracic pressure due to mechanical ventilation, which leads to variance in cardiac preload and afterload. SVV has recently been extensively investigated, and several studies have shown the utility of using SVV as a predictor of fluid responsiveness in a variety of clinical situations. Several other parameters based on SVV have been found to be useful as well. In particular, it has been found that systolic pressure dispersion (SPV) has its delta-rise (Δrise) and delta-fall (Δfall) components very useful predictors of fluid responsiveness. SPV is based on changes in arterial pulse pressure due to respiratory-induced dispersion in stroke volume. Yet another parameter that has been recently investigated and shown to be an effective indicator of fluid responsiveness is pulse pressure dispersion (PPV).

  Recent developments in arterial pulsatile contour analysis have opened unique opportunities for less invasive, continuous and real-time SVV estimation. This allows clinicians to routinely use SVV in conjunction with SV and CO in their assessment of the critical care patient's hemodynamic status.

  Existing systems for measuring fluid responsiveness based on respiration-induced changes in arterial pulse pressure are almost all based on one of only a few methods. Some methods described in the literature include the following measurements of pulse pressure variance (PPV), systolic pressure variance (SPV) and stroke volume variance (SVV).

PPV estimation is based on a specific version of Equation 1 below:
PPV = 100 · [PPmax−PPmin] / [1/2 (PPmax + PPmin)] (Equation 1)
Where PP is the measured pulse pressure, and PPmax and PPmin are the maximum and minimum peak-to-peak values of the pulse pressure during one breath (inspiration-expiration) cycle, respectively.

SPV estimation is based on a specific version of Equation 2 below:
SPV = 100 · [SPmax−SPmin] / [1/2 (SPmax + SPmin)] (Equation 2)
Where SP is the measured systolic pressure and SPmax and SPmin are the maximum and minimum values of systolic pressure during one respiratory cycle, respectively.

Similarly, SVV estimation is based on a specific version of Equation 3 below:
SVV = 100 · [SVmax−SVmin] / [1/2 (SVmax + SVmin)] (Equation 3)
Here, SV is the stroke volume, and SVmax and SVmin are the maximum value and the minimum value of the stroke volume during one respiratory cycle, respectively.

  In equations 1, 2, and 3, the governing factor is the average of the maximum and minimum values of PP, SP, and SV, respectively. In other words, despite the fact that there are only two measurement points, these dominating factors are average values. Averaging this simple limit most commonly simplifies these calculations, which are typically performed manually. However, a more reliable value can be obtained by using the average of all measurements over the measurement interval, ie the first statistical product of PP, SP, and SV.

  Therefore, for each of PPV, SPV and SVV, the individual dispersion value formulas indicate the magnitude of a predetermined range of values (maximum-minimum) relative to the average of the extreme values (maximum and minimum).

  Detailed monitoring of SVV has both unique difficulties and advantages. Physiologically, SVV is based on several complex mechanisms of heart-respiration interaction. Briefly: Mechanical ventilation causes changes in the left ventricular preload, which leads to a clear distribution in left ventricular stroke volume and systolic arterial pressure. SVV monitoring allows for the prediction of left ventricular response to volume dosing, along with an accurate assessment of blood volume deficiencies, and supports subsequent decisions to perform volume emergency resuscitation in many important situations.

  In addition to the three methods listed above, there is a fourth method, commonly known as the Perel method, called the “Respiratory Systolic Dispersion Test”. This method includes airway pressure maneuvers, such as inducing variable magnitude ventilation preceded by a short apnea period. US Pat. No. 5,769,082 (Perel) describes this method. This method is not suitable for real-time monitoring due to the need for pressure steering.

The methods listed above are performed with certain CO monitoring devices such as LiDCO Ltd cardiac monitors and Pulse Medical Systems PiCCO systems (see, for example, US Pat. No. 6,315,735). (See Joeken et al.), Both of which depend on Equation 3. However, tests by the inventors show that these methods are noisy and they do not allow accurate real-time monitoring of respiratory induced changes in arterial pressure. The main reason for this noise problem in SVV estimation in these devices is how SV is calculated. For example, the PiCCO system uses a stroke-to-beat stroke volume estimation method based on a pulse contour algorithm that includes determination of specific points in the blood pressure waveform, such as a double pulse notch. Accurate detection of the double pulse notch and other points in the blood pressure signal is difficult, however, due to indefiniteness of the blood pressure waveform and its unstable nature.
US Pat. No. 4,236,527 U.S. Pat. No. 4,507,974 US Pat. No. 5,146,414 US Pat. No. 5,687,733 US Pat. No. 5,400,793 US Pat. No. 5,535,753 US Patent Application Publication No. 2002/0022785 US Pat. No. 6,315,735

  Therefore, what is needed is an operational system and method for estimating SVV in real time using, at best, minimally invasive techniques and more accurately and robustly than is currently possible. The present invention meets this need.

(Summary of the Invention)
The present invention provides a method and associated system implementation for determining cardiac parameters that are equal to or can be induced by cardiac stroke variance (SVV): determined either invasively or non-invasively. A waveform data set corresponding to the arterial blood pressure is determined and input to the processing system over a calculation interval including at least two cardiac cycles; a standard deviation value for the waveform data set is then passed over each cardiac cycle. And an estimate of the SVV is calculated as a function of this standard deviation value.

  Different methods can be used to calculate the values used in this SVV estimation. For example, a maximum standard deviation value and a minimum standard deviation value over the calculation interval can be determined, and an estimate of the SVV can then be calculated as a function of the maximum standard deviation value and the minimum standard deviation value. An average standard deviation value is also preferably calculated over the calculation interval, and the SVV estimate is then proportional to the difference between the maximum standard deviation value and the minimum standard deviation value relative to the average standard deviation value. Can be calculated. As an alternative, the processing system may calculate the standard deviation of the standard deviation value over the calculation interval, and then the ratio between the standard deviation of the standard deviation value and the average of the standard deviation values. The SVV is estimated as being proportional to.

  Blood pressure can be measured either invasively or non-invasively, for example by using a pressure transducer or finger cuff attached to the catheter. This measured arterial pressure is then converted into a waveform data set.

  In order to calculate an estimate of right ventricular end diastolic volume, the method described above according to the invention is also inversely proportional to the calculated estimate of the cardiac stroke volume variance. It has been found that it can be used.

According to one feature of the invention found to be advantageous, an approximation function that best matches the calculated standard deviation value is calculated according to a predetermined metric over at least one calculation interval. The This approximation function is then sampled at an interval-specific sampling rate to produce a set of sampled approximations, which are then low pass filtered before the SVV estimate is calculated. .

This method of approximating, resampling, and low-pass filtering at an adjustable rate is suitable for use in estimating other cardiac or hemodynamic parameters other than SVV, such as systolic pressure variance, pulse pressure variance, etc. Can be extended for. In this case, a waveform data set is generated and input to the processing system, corresponding to whatever measurement parameter is selected. A series of measurements is then generated from the waveform data set; an approximate function (or a specific function thereof) that best matches these measurements is calculated according to a predetermined metric; each of at least one calculation interval A set of sampled approximations is then generated by sampling the approximation function at interval-specific sampling rates; the sampled approximations are then low-pass filtered ; and An estimate of the quantity is low pass processed and calculated as a function of the sampled approximation.

(Detailed explanation)
(Introduction)
In its broadest sense, the present invention includes the pulse-to-pulse stroke volume variance SVV as a function of the standard deviation of the blood pressure waveform over multiple cardiac cycles. Of course, the present invention can be used to determine any other cardiac parameters that can be derived from SVV.

  The present invention can be advantageously used with any type of subject, whether human or animal. Since the most common use of the present invention is expected to be for humans in a diagnostic setting, the present invention is primarily described below in use with “patients”. However, this is by way of example only—the term “patient” is intended to encompass all human and animal subjects, regardless of setting.

  Because of its clinical importance, most implementations of the present invention are expected to generate SVV estimates based on measurements of systemic arterial blood pressure. However, it may be possible to use measurements of blood pressure taken anywhere, such as in the right pulmonary artery (although such sites may require invasive intracardiac measurements).

(Pressure waveform)
FIG. 1 shows here a single heart from the point of diastolic pressure P _dia at time t _dia0 through time t _sys of systolic pressure P _sys pressure until time t _dia1 when blood pressure once again reaches P _dia. An example of an arterial pressure waveform P (t) taken over a period is shown.

  According to the present invention, P (t), or any signal proportional to P (t), can be measured invasively or non-invasively at any point in the arterial tree. If an invasive device, in particular a pressure transducer attached to a catheter, is used, any artery can be used as a measurement point. Non-invasive transducer placement is typically dictated by the instrument itself—finger cuff, brachial pressure cuff, and earlobe clamp placement is obvious. Regardless of the device, ultimately, an electrical signal corresponding to (eg, proportional to) P (t) is generated or is generated.

  Rather than directly measuring arterial blood pressure, any other input signal that is proportional to blood pressure can be used. Any necessary scaling or transformation may then be performed at any or all of several points in the calculations described below. For example, if a particular signal other than the arterial blood pressure itself is used as input, it can be calibrated to blood pressure before that value is used in the calculations described below. In short, the fact that the present invention may use an input signal different from direct measurement of arterial blood pressure in some cases does not limit its ability to generate accurate SVV estimates. The only requirement for the present invention is that a signal or data set having a known relationship equal to (or proportional to) the patient's blood pressure over the interval of interest (including continuous) is described below. This means that it must be made available to processing systems (see below) that perform signal conditioning and various calculations.

  As is well known and as shown in FIG. 2, an analog signal such as P (t) can be digitized into a sequence of digital values using any standard analog-to-digital converter (ADC). . In other words, P (t), t0 ≦ t ≦ tf are calculated using known methods and circuits, digital form P (k), k = 0, n−1, where t0 and tf are respectively calculated. The initial and final time of the interval, and n is the number of samples of P (t) that can be included in the calculation, usually distributed uniformly over the calculation interval.

(standard deviation)
The calculation of the mean and standard deviation of a continuous or separate function or data set f is a very well-known procedure, and usually by the Greek letters μ and σ, respectively, or in most programming languages, mean and It is referred to by a function name such as std. Thus, μ (f) and mean (f) represent the average of a function or data set f over a particular interval, and σ (f) and std (f) represent its standard deviation.

Consider the calculation of the mean and standard deviation of a series of blood pressure values P = P (k) over a specific interval such as k = 1,..., N. The most common way to calculate the mean μ (P) and standard deviation σ (P) is to use an algorithm known to everyone who knows even an introductory course in statistics:
μ (P) = mean (P) = 1 / m * SUM [P (k)]
(Equation 4)
σ (P) = std (P) = sqrt {1 / (m−1) * SUM [P (i) −μ (P)] ² }
(Equation 5)
Here, sqrt represents the square root, and SUM represents the sum over the interval of i = 0,..., (M−1). Note the separate value formula for the variance (square of the standard deviation), which is usually multiplied by a factor of 1 / (m-1) instead of 1 / m for well-known statistical reasons.

  As will become clearer from the following, the present invention generates a robust estimate of SVV in real time, preferably from the calculation of the standard deviation of the pressure waveform over several cardiac cycles. The standard "textbook" formula for standard deviation (Equation 5) is preferred for well-known statistical reasons, but any formula that provides an acceptable accurate value for standard deviation, or a corresponding value. Alternatively, an algorithm can be used in the present invention instead. For example, at least in the context of measurement based on blood pressure, a rough approximation to σ (P) can be calculated by dividing the difference between the maximum and minimum measured pressure values by three. Furthermore, the absolute value of the maximum or minimum value of the first derivative of P (t) with respect to time is substantially proportional to σ (P).

  In this discussion of the invention, there are several references to calculations using “pressure” and “standard deviation” values. In the preferred embodiment of the present invention, these are in fact standard deviations as calculated using the direct measurement of blood pressure, in particular arterial blood pressure, and the most general formula given above as equation 5. . However, it should be understood that “pressure” may equally refer to indirect measurements, or measurements of specific physiological characteristics that may be correlated or corresponding to or otherwise related to pressure. Similarly, “standard deviation” can also refer to any value known to approximate this value that can be given by the usual formula; examples of such alternatives are given in the previous paragraph. Yes.

(SVV calculation)
The standard formula for calculating cardiac output (CO) is CO = SV · HR, which is simply the amount of blood that the heart pumps per minute, which is pumped per cycle. It merely represents a relationship that is equal to the amount transported (SV) multiplied by the number of cycles per minute. Given HR, the problem is, of course, assuming knowledge of a constant value SV or average SV. Based on the observation that the pulsatile nature of the pressure waveform is generated by the stroke volume of the heart into the arterial tree, one of the inventors has previously stated that SV is the arterial pressure waveform P (t). Or it can be approximated as being proportional to the standard deviation of certain other signals that are themselves proportional to P (t). Therefore, one way to estimate SV is to apply the following relationship: SV = K · σ (P) = K · std (P)
(Equation 6)
Where K is a calibration constant and from this the following is derived:
CO = K.sigma (P) .HR = K.std (P) .HR
(Equation 7)
Substituting Equation 6 into Equation 3 yields an equation for SVV that is a function of std (P):
SVV = 100 · [std (P) max−std (P) min] / [1/2 (std (P) max + std (P) min)]
(Equation 8)
Here, std (P) max and std (P) min are the maximum value and the minimum value of the standard deviation of the pressure waveform over the calculation interval, respectively. In a preferred embodiment of the invention, this calculation interval is a respiratory cycle.

As noted above, using a statistical mean of pressure-mean (P) in the denominator over the measurement interval generally provides a more accurate and robust estimate than using the average of only the maximum and minimum values. Although provided, the present invention may use any average for this denominator. Thus, the preferred form of Formula 8 is:
SVV = 100 · [std (P) max−std (P) min] / mean (std (P))
Where mean (std (P)) is not the mean μ (P) of the pressure waveform itself, but the mean of the standard deviation of the pressure waveform, but this parameter is arbitrary It can be included in the calculation of the std (P) value. Also, depending on the type of signal conditioning and conversion method used to condition the measured pressure signal P (t) and then convert it to a separate form (see below) , It may be necessary or desirable to scale equation 9. Whether or not to multiply and how to do it is obvious to those skilled in the art: for that reason, any scaling factor is not shown in Equation 9 but can be estimated .

  FIG. 3 shows a sequence of arterial pressure waveforms measured or otherwise acquired over approximately three respiratory cycles. In fact, this sequence is a data set P (k) derived from a sampled measurement of arterial pressure P (t). As mentioned above, this P (k) value can be obtained directly by invasive or non-invasive measurements, or even from a remote monitor or pre-recorded data set (possibility of this latter Of course, it does not provide real-time mottling and, in most cases, may be primarily input from certain other sources, which may be subject to comparison or experimentation).

  FIG. 4 includes points in the waveform of FIG. 3 to indicate the start of each cardiac cycle (ie, each “beat”) over the indicated calculation interval. The start of each cardiac cycle can be determined in any number of known ways using any known system that is or includes a heart rate monitor. For example, as is often the case, when the electrical activity of a patient's heart is also monitored by an electrocardiogram system (EKG), the start of each heart note is a sampled pressure value immediately after each R wave. Can occur. A pressure-based pulse rate monitor can also be used and is preferred in practice. Because they can then be better tuned with the blood pressure signal than, for example, the EKG signal.

If no such external device is present, the start of each cardiac cycle is also assumed, for example, that each beat simply starts at the time of minimum (diastolic) pressure P _dia (see FIG. 1). Or by using the Fourier transform or differential analysis from the pressure waveform P (k) itself using software alone. In these cases, the heart rate “monitor” is a software construct. In such cases, using known techniques, care should be taken to ensure that the local pressure minimum is not caused by breathing itself, but is actually a diastolic time.

  Let P (k, i) be the i-th beat-beat arterial pressure signal in the calculation interval. A standard deviation std (P (k, i)) can then be calculated as described above and a signal value (scalar) result is obtained. The standard deviation value for each of the pulsation-pulsatile arterial pressure signals during the calculation interval can be similarly calculated. In FIG. 5, each point on the standard deviation curve std (P (t)) indicates a calculated standard deviation value for one pulsation-pulsation arterial pressure signal.

Calculation of SVV according to Equation 9 requires the maximum, minimum and average values of std (P (k, i)) for each calculation interval. One way to obtain std (P) _max and std (P) _min for a given breathing cycle (FIG. 5 shows approximately three breathing cycles) is simply the maximum and minimum in this cycle. Taking a separate std (P (k, i)) value. These can be quickly identified by simply scanning these calculated values.

  During actual monitoring, noise and other factors impair the reliability of one or more std (P (k, i)) values. Deviance values and potentially unreliable std (P (k, i)) values can be detected using known algorithms such as those based on pattern matching. Interpolation may then be used to determine a replacement value for this excluded value (s). A preferred method for smoothing and filtering measured values to increase reliability, applicable even in the determination of other cardiac and hemodynamic parameters other than SVV, is described below.

Any standard algorithm can be used to distinguish std (P) _max and std (P) _min from std (P (k, i)). The simplest method is to take the largest value and the smallest value. Alternatively, these measured std (P (k, i)) points can be used as a basis for determining an approximation or smoothing function (eg, using a spline function or polynomial); In some cases, extreme values of std (P (t)) between calculated values, such as std (P) _min shown in FIG. 5, can be identified. The mean (std (P (k, i)) = mean (std (P)) of these values can be calculated using standard statistical equations (see equation 4). It also shows the value of mean (std (P)) for the first respiratory cycle, once std (P) _max and std (P) _min are determined for a given respiratory cycle, then for that respiratory cycle SVV values can be easily calculated according to either Equation 8 or Equation 9. Assuming the preferred case where mean (std (P (k))) is also calculated, using Equation 9 is more accurate and robust SVV can be calculated, and the calculated value (s) of SVV can then be displayed, stored, communicated and / or used in further calculations in any desired manner for the user. .

  As stated above, the preferred calculation interval is one respiratory cycle. The start time and end time of these cycles can be detected using any known device. Note that many patients who may need the present invention can also be ventilated, in which case the ventilator itself can provide the timing signals necessary for the present invention to separate different respiratory cycles. That.

  However, as shown in FIGS. 3 and 4, it may also be possible to detect respiratory cycle boundaries by analyzing the soot of the acquired pressure waveform itself: the least diastolic pressure (pressure waveform The “environment” local minimum) time is also usually a sign of the start of the respiratory cycle. In some situations, however, such software-based detection methods may not be more accurate than simply accepting signals from another system, especially including respiratory monitoring, especially if the patient's breathing is weak .

(Calculation interval-Alternative SVV calculation)
In the above, it was assumed that the SVV value was estimated for each respiratory cycle. This is not always necessary. Rather, using the same methodology and equations described above to calculate a single SVV value over a calculation interval greater than one respiratory cycle (even over the entire time of the monitoring session). May be possible. This, of course, does not show a period-period SVV trend, but a more accurate result for each monitoring session, and therefore a better indicator of the trend for longer periods (such as session-session or day-day) Can provide. However, it is not always necessary to calculate over "extreme" intervals. Rather, the SVV value can be calculated for any n period time with n ranging from 1 (as described above) to the upper range.

For example, assume that the calculation time is greater than one respiratory cycle. For purposes of illustration, assume that all three respiratory cycles shown in FIG. 5 constitute a single combined calculation interval. Even in this case, a set std (P (k, i)) of calculated std (P (k)) values (“points” on the curve), in fact, there are still many values. These standard deviation values have their own standard deviation and average (the center point between extreme values). According to an alternative embodiment of the invention, useful for calculation intervals over more than one respiratory cycle, SVV is therefore estimated as follows:
SVV = C · 100 · std (std (P (k, i)) / mean (std (P (k, i)) (Equation 10)
Here, C is a scaling constant determined experimentally. In one test of the present invention, the inventors determined that C was approximately 2.7, regardless of the number of respiratory cycles included in this calculation interval; C may be determined at any given implementation of the invention.

This alternative multi-period method does not require a system to determine any std (P) _max value or std (P) _min value. As a result, it is still not advantageous to determine an approximation function that interpolates std (P (k)) values over the calculation interval.

(Smoothing, resampling and filtering as needed)
Referring to FIG. 6, where the sequence of measurements X (k) of heart or hemodynamic value X is plotted as a point. In the embodiment of the invention described above, X (k) is std (P (k)), one measurement per cardiac cycle. However, the optional digital preprocessing described here can also be any other cardiac or hemodynamic value such as systolic pressure or pulsatile pressure, or blood velocity measurements obtained from Doppler ultrasound scans. Applicable to even non-pressure related parameters.

  These measurements rarely fit into a well-defined profile. This is because the underlying parameters are influenced not only by noise, but also by inherent irregularities such as non-constant heart rate, different breathing patterns, etc. In many cases, the measurement system may itself be able to alert, identify and eliminate unreliable measurements. For example, if the signal strength falls below a predetermined minimum value, the value of the parameter measured during such a time can be tagged or eliminated. In other cases, unreliable values may not be detectable without further analysis. In FIG. 6, for example, a measurement point marked X (p) deviates significantly from an obvious pattern that can be assumed to be reasonably unreliable.

  As noted above, one method of identifying deviating values is one in which the measured points were obtained from a control study performed on a patient population, or simply a confirmed profile measured in the same patient. By using a pattern match compared with the previously stored template. Any point that deviates from this template (s), measured in any known manner, by a certain threshold amount can then be eliminated.

In a preferred embodiment of the invention, for each calculated interval, according to any known metric, i.e., in any known sense, according to least squares, etc., to generate an interval profile that best matches the measurement point, An approximate function (which can be a concatenation or combination of more than one partial function) is generated. For example, polynomial functions such as spline functions (two examples: B-spline functions and Bezier curves), reduced component Fourier representations, etc. can be quickly calculated to generate a suitable approximation function. In FIG. 6, the approximate function for the first breath interval is labeled X ^* (t).

  In the example shown, the calculation interval is the respiratory cycle. Other intervals can be selected depending on the parameters of interest and need not be related to the respiratory cycle. Even when doing so, the measurement interval need not be a single period, or the whole of a plurality of single periods, which is generally meaningful if the parameters of interest are generally affected by respiration. But also generally simplify the calculation.

Once the approximate function is calculated, the preferred embodiment of the present invention low-pass filters the values per calculation interval. The theory and implementation of digital low-pass filters is well known. Several sets of sampled values are used as input to an algorithm that calculates their particular weighted linear or rational function to generate a set of transformed filtered stroke volume values. A typical digital filter is generally assumed to take a certain “distance” between samples, but most such filters also have a fixed coefficient.

However, according to this aspect of the present invention, the input value to this low pass filter can be obtained by sampling an approximate function-perhaps (although not necessarily) at a rate that is still higher than the rate at which the actual measurement was obtained. -Obtained. In FIG. 6, these sampling points are indicated by white circles on the approximate curve X ^* (t). The sampling points of the approximate curve X ^* (t) are then input into a digital low pass filter , which cut-off frequency can be selected in any known manner and of the hemodynamic or cardiac parameter to be estimated, and It may depend on the known nature of the measured value X (k). Once the desired cutoff frequency is determined, the appropriate low-pass filter coefficients can be determined in any known manner.

Note that this procedure allows for uniform spacing of these sample points even though the actual measured values may not be evenly spaced, for example due to changing heart rate. That. Furthermore, note that the sampling rate for the respiratory cycle (here selected as the calculation interval) 3 is greater than that for cycles 1 and 2. One way to select the sampling rate for each given calculation interval is to select samples that have the same number of approximate curve X ^* (t) intervals as the actual measurements present in the calculation interval. This is not necessarily required, but rather this sampling rate can be chosen as desired, for example to satisfy the Nyquist criterion, and reduces the aliasing effect for the type of digital low pass filter used.

The tests by the inventors have shown that these steps of smoothing (via approximation functions), resampling (ie sampling after sampling performed for analog-to-digital conversion), and low-pass filtering of adjustable coefficients Have shown to reject the noise of the present invention and improve the ability to generate more reliable SVV values. As stated above, this procedure can also be used for other estimated parameters other than SVV.

(System components)
FIG. 7 shows the major components of a system that implements the above method for sensing pressure and calculating SVV in accordance with the present invention. The present invention can be included within existing patient monitoring devices or implemented as a dedicated monitor. As mentioned above, pressure, or certain other input signals proportional to pressure, can be sensed in two ways: either invasive and non-invasive, or indeed both. Simply because it is expected to be the most common implementation of the present invention, the system is described as measuring arterial blood pressure against a specific other input signal that is converted to pressure.

  FIG. 7 shows both types of pressure sensing for the sake of simplicity: In most practical applications of the present invention, any one or several modifications are typically implemented. In the invasive application of the present invention, a conventional pressure sensor 100 is mounted on a catheter 110, which is inserted into an artery 120 of a body part 130 of a human or animal patient. Such arteries can be ascending aorta, or pulmonary arteries, or to reduce the level of invasiveness, arteries 120 can be peripheral, such as femoral, radial, or brachial arteries. In a non-invasive application of the present invention, a conventional pressure sensor 200, such as an optical volume pulse blood pressure probe, can be used in any conventional manner, for example, a cuff around a finger 230 or a transducer mounted on a patient's wrist. Used to attach to the outside. FIG. 3 schematically shows both types.

  Signals from sensors 100, 200 are passed as inputs to processing system 300 via any known connector, which includes one or more processors and other supporting hardware such as memory 301, And system software (not shown) for processing normal signals and executing code. The present invention can be implemented using a modified standard personal computer or it can be incorporated into a larger dedicated monitoring system. In the present invention, processing system 300 may also include or be coupled to conditioning circuitry 302 that performs normal signal processing tasks such as amplification, filtering, ranging, etc., as appropriate.

  The conditioned and sensed input pressure signal P (t) is then converted to digital form by a conventional analog-to-digital converter ADC 304, which has a clock circuit 305 or takes a time reference therefrom. As is well understood, the sampling frequency of this ADC 304 should be selected with respect to the Nyquist criterion so as to avoid aliasing of the pressure signal; this procedure is very common in the digital signal processing arts. It is well known. The output from the ADC 304 is an individual pressure signal P (k) whose value can be stored in a conventional memory circuit (not shown).

  A signal pre-processing module 306 is preferably included idiomatically, for general (interval-interval), such as motion artifact rejection, pulse beat detection (if necessary), bad beat rejection, etc. And) provide a known pre-processing such as digital filtering for noise removal. This module may also be implemented in whole or in part in hardware. As mentioned above, known circuits can be included, for example, indicating that the signal strength is too low and the delivered measurements are unreliable. Accordingly, the module 306 can also be functionally, wholly or partially disposed in front of the ADC 304.

  The value P (k) is (usually accessed from memory thereby) the software module 310 containing computer-executable code for calculating the standard deviation of P (k) over a calculation interval such as the cardiac cycle. This can be triggered by any known device or software routine 315 that detects the heart rate or at least signals the start of the cardiac cycle. Even moderately skilled programmers know how to design this software module 310.

While the above is described as preferred, if necessary, generating an approximate function (see X ^* (t) and related discussion in FIG. 6), sampling the generated function, Then there is a step of low-pass filtering the sampled value. Software modules 312, 313, 314 are included to perform these functions and can be programmed using known techniques. Of course, any or all of these modules can be combined into a single body cord: these are shown separately for clarity. Also, if the measured or calculated value used as the basis for the approximate function is not std (P (k)), this module 310 may be replaced, reprogrammed, as necessary, or Omitted and provides an appropriate value.

These std (P (k)) values are typically stored in the memory 301, but then, for each predetermined calculation / measurement interval, std (P (t)) values that are the maximum and minimum std (P (t)) values. P) A software module 320 that includes computer-executable code for detecting _max and std (P) _min and for calculating mean (std (P (k)), which is an average std (P (t)) value This can be triggered by a breathing device 325, such as a ventilator, or detected by a software module, as described above, and again, even a moderate skill programmer can Given the explanation, they know how to design the software module 320. Also, as described above, the calculation interval is If it is selected to expand over a greater respiratory cycle, or the module 320 may be omitted, or not to be at least called, maximum and minimum std (P (t)) values, std _{(P) max} and It is not always necessary to calculate std (P) _min .

These values std (P) _max , std (P) _min and mean (std (P (k))) are then passed to the SVV calculation module 330, which is for the selected interval according to Equation 8 or 9. Calculate SVV estimates: If multiple period intervals are set, this SVV calculation module 320 may operate directly on the std (P (k)) value provided by module 310, and the equation 10, the estimated value of SVV is calculated.

  As shown in FIG. 7, any or all of the software modules 306, 310, 312-314, 320, and 330 may simply be implemented as routines within a single estimation software component 370, which, of course, It can be combined with other software components of the processing system 300 as desired.

  The present invention further relates to a computer program that can be loaded into a computer unit or processing system 300 to perform the method of the present invention. Further, in accordance with the present invention, various software modules 310, 312-314, 315 (when executed in software), 320, used to perform the various calculations described above and perform the associated method steps. 330, or generally 370, can also be stored as computer-executable instructions on a computer-readable medium to allow the invention to be loaded into and executed by different processing systems.

  Once the SVV (or other cardiac or hemodynamic) estimate is calculated, it is passed to any desired output device 500, such as a monitor that the user can see, and any selected Displayed, stored, or communicated in a format. An input device 400 is also preferably included to allow the user to enter, for example, administrative and patient specific information, adjust the display, select a calculation interval, and so forth. If the user is allowed to change the calculation interval, the corresponding information must be made available to the estimation software component 370, so that equation 8 or 9 or equation When calculating the SVV according to any of 10, it may instruct the SVV estimation module to select the required input value.

(Test results)
Two different but related methods for calculating SVV from the std (P (k)) value are described above—one is that SVV is calculated over a single respiratory cycle, and another One can be used to calculate a single SVV value for a calculation interval over more than one respiratory cycle. We tested both these embodiments of the present invention against femoral artery and radial blood pressure data of animals. This data was collected from several pig experiments and these were performed in the laboratory. During the experiment, several changes (volume injection or volume extraction) were induced in the left ventricular volume. Dispersion of peripheral resistance (vasodilation or vasoconstriction) was similarly induced. Both embodiments of the present invention were found not only to provide similar SVV estimates, but these estimates were superior to the estimates obtained using prior art methods for comparison. It was found that

(Test results)
Two different but related methods for calculating SVV from the std (P (k)) value are described above—one is that SVV is calculated over a single respiratory cycle, and another One can be used to calculate a single SVV value for a calculation interval over more than one respiratory cycle. We tested both these embodiments of the present invention against femoral artery and radial blood pressure data of animals. This data was collected from several pig experiments and these were performed in the laboratory. During the experiment, several changes (volume injection or volume extraction) were induced in the left ventricular volume. Dispersion of peripheral resistance (vasodilation or vasoconstriction) was similarly induced. Both embodiments of the present invention were found not only to provide similar SVV estimates, but these estimates were superior to the estimates obtained using prior art methods for comparison. It was found that

(More output)
As stated above, the present invention can be used to estimate any cardiac parameter that can be derived from SVV, as well as SVV. In the study, we found that right ventricular end diastolic volume (EDV) is inversely proportional to SVV, for example, during constant mechanical ventilation and constant tidal volume mechanical ventilation. did. Therefore:
EDV = c / SVV
(Equation 11)
Where c is a calibration factor that may be constant. The SVV estimates provided by the present invention can therefore be used as an indirect method for estimating EDV, stroke fraction or other values known to be proportional to the degree of vascular filling.

  The calibration factor c in Equation 11 depends on several parameters such as ventricular contractility and ventricular extensibility (compliance). In general, c depends on the patient's hemodynamic state. Therefore, recalibration must be performed every time the patient's hemodynamic state changes or during periods of hemodynamic instability.

FIG. 1 is an illustrative example of a composite blood pressure curve over a cardiac cycle from one beat to a beat. FIG. 2 shows a separate time display of the pressure waveform in FIG. FIG. 3 shows a series of blood pressure signals (waveforms) over different respiratory cycles. FIG. 4 shows the beat-to-beat detection in the waveform of FIG. 3, i.e. the isolation of individual pressure waveforms per cardiac cycle. FIG. 5 shows a pressure standard deviation curve. FIG. 6 shows optional digital preprocessing including smoothing and filtering of separate cardiac or hemodynamic measurements. FIG. 7 is a block diagram showing the main components of the system according to the present invention.

Claims (44)

A method of operating an apparatus for determining a cardiac parameter equal to or derivable from a cardiac stroke volume variance (SVV) by a processing means including a standard deviation calculation module and an SVV calculation module comprising:
The method
And that said processing means, for generating a waveform data set from measurements of arterial blood pressure over a computation interval that includes at least two cardiac cycles,
And that said standard deviation calculation module calculates the standard deviation value for the waveform data set over each cardiac cycle,
The SVV calculation module comprises and calculating the estimated value of the SVV as a function of the standard deviation method. And that the scope value determination module, over the calculated distance, determines the maximum standard deviation value and the minimum standard deviation value,
The SVV calculation module, an estimate of the SVV, outermost further encompasses and calculating as a function of the large standard deviation value and said minimum standard deviation value, The method of claim 1. And the average value calculation module, over the calculated distance, to calculate the average standard deviation value,
The SVV calculation module, to the average standard deviation value, further comprising the calculating the estimated value of SVV which is proportional to the difference between the maximum standard deviation value and the minimum standard deviation value, wherein Item 3. The method according to Item 2. And the breathing device, detects the boundaries of the respiratory cycle,
The respiratory device further comprises a setting at least one respiratory cycle the calculation interval A method according to claim 1. And said standard deviation calculation module calculates the standard deviation of the standard deviation values over the computation interval,
And the average value calculating module calculates the average of the standard deviation values over the calculated out interval,
The SVV calculation module, the standard deviation further comprising the calculating the estimated value of the SVV proportional to the average the ratio of the standard deviation and the standard deviation of the values, according to claim 4 method. End diastolic volume calculation module further includes calculating the estimated value of the right ventricular end diastolic volume, as inversely proportional to the calculated estimate of the cardiac stroke volume dispersion, claim The method according to 1. And that the approximation function calculating module, which according to a predetermined metric over at least one of the calculated distance, to calculate an approximate function that best matches the standard deviation value the calculated,
The approximate function calculation module, by sampling the approximate function with specific sampling rate intervals, and generating a set of sampled approximations,
And the low-pass filter module and low-pass filtering the sampled approximation,
The SVV calculation module, an estimate of the SVV, said further encompasses and calculating as a function of the low-pass filtered sampled approximation method of claim 1. A system for determining a cardiac parameter equal to or derivable from a cardiac stroke variance (SVV) comprising:
An array for determining and inputting a waveform data set corresponding to arterial blood pressure over a calculated interval comprising at least two cardiac cycles;
A processing system and
The processing system includes:
Calculating a standard deviation value for the waveform data set over each cardiac cycle;
Calculating an estimate of the SVV as a function of the standard deviation value. The processing system includes:
Determining a maximum standard deviation value and a minimum standard deviation value over the calculation interval;
The system of claim 8, further configured to: calculate the SVV estimate as a function of the maximum standard deviation value and the minimum standard deviation value. The processing system includes:
Calculating an average standard deviation value over the calculation interval;
The system further comprises: calculating an SVV estimate that is proportional to the difference between the maximum standard deviation value and the minimum standard deviation value for the average standard deviation value. 10. The system according to 9.   9. The system of claim 8, further comprising a mechanism for detecting a respiratory cycle boundary, wherein the calculated interval is at least one respiratory cycle. The processing system includes:
Calculating a standard deviation of the standard deviation value over the calculation interval;
Calculating an average of the standard deviation values over the calculation interval;
The system of claim 11, further comprising: calculating an estimate of SVV that is proportional to a ratio of a standard deviation of the standard deviation value and an average of the standard deviation values. A pressure transducer attached to a catheter for measuring arterial pressure;
9. The system of claim 8, further comprising: a conversion circuit that converts the measured arterial pressure into the waveform data set. A non-invasive arterial pressure measuring device;
9. The system of claim 8, further comprising: a conversion circuit that converts measured arterial pressure into the waveform data set. The processing system includes:
9. The system of claim 8, further configured to calculate an estimate of right ventricular end diastolic volume as being inversely proportional to a calculated estimate of the cardiac stroke volume variance. The processing system further includes a low pass filter,
The processing system comprises:
Calculating an approximate function that most closely matches the calculated standard deviation value according to a predetermined metric over at least one of the calculation intervals;
Generating a set of sampled approximations by sampling the approximation function at an interval specific sampling rate;
9. The system of claim 8, further comprising: calculating the SVV estimate as a function of a low pass filtered sampled approximation filtered by the low pass filter. . And processing means, the measured value generating means, an approximate function calculation module, a low-pass filter module, according to an output unit, a method of operating apparatus for determining cardiac or hemodynamic stroke volume values,
The method
The processing means receives a waveform data set corresponding to a blood pressure measurement of the measurement parameter;
And the measured value generating means for generating a series of measurements from the waveform data sets,
The processing means, as a function of the standard deviation of the waveform data set over each calculation interval, for at least one calculated distance comprises at least two cardiac cycles, calculating the respective measurements of the series of measurements When,
And that the approximation function calculating module calculates an approximate function that best matches the measured value according to a predetermined metric,
The approximate function calculation module, by sampling the approximate function with specific sampling rate intervals, and generating a set of sampled approximations,
And that said low-pass filter module and low-pass filtering the sampled approximation,
Encompasses and that said processing means calculates an estimated value of該拍output quantity value as a function of the low-pass filtered sampled approximation method. The stroke volume value has a low frequency component corresponding to respiratory-induced dispersion;
The method
And the breathing device, identifying a plurality of respiratory cycles,
And the cardiac cycle detector detects a cardiac cycle,
The respiratory device further includes and setting the calculated distance to the respiratory cycle, The method of claim 17. Generating the series of measurements includes pre-processing the waveform data set;
The method of claim 17, wherein the preprocessing includes performing any of digital filtering for noise removal, motion artifact rejection, pulse beat detection, bad beat rejection.   The method of claim 17, wherein the stroke volume value is systolic pressure variance.   The method of claim 17, wherein the stroke volume value is a pulse pressure variance. A system for determining a cardiac or hemodynamic stroke value,
An array for determining and entering a waveform data set corresponding to blood pressure measurements;
A processing system and
The processing system comprises:
Generating a series of measurements from the waveform data set;
Calculating each measurement of the series of measurements as a function of the standard deviation of the waveform data set over individual calculation intervals using processing means over at least one calculation interval including at least two cardiac cycles; When,
Calculating an approximate function that most closely matches the measured value according to a predetermined metric;
Generating a set of sampled approximations by sampling the approximation function at an interval specific sampling rate;
Calculating an estimate of the stroke volume value as a function of the sampled approximation. The stroke volume value has a low frequency component corresponding to respiratory-induced dispersion;
The system
An array identifying multiple respiratory cycles;
An array for detecting a cardiac cycle, and
The processing system is
23. The system of claim 22, further configured to set the calculated interval to a respiratory cycle.   A pre-processing module included in the processing system, further comprising a pre-processing module for pre-processing the series of measured values, including pre-processing the waveform data set, wherein the pre-processing includes noise removal. 23. The system of claim 22, comprising performing any of digital filtering for: motion artifact rejection, pulse beat detection, bad beat rejection.   24. The system of claim 22, wherein the stroke volume value is systolic pressure variance.   23. The system according to claim 22, further comprising a low-pass filter, wherein the calculation of an estimate of the stroke volume value as a function of the sampled approximate value is performed after filtering by the low-pass filter.   4. The system according to claim 3, wherein any or all of the standard deviation calculation module, the range value determination module, the average value calculation module, and the SVV calculation module can be integrated into a single module. The method described.   The method according to claim 5, wherein any or all of the standard deviation calculation module, the average value calculation module, and the SVV calculation module can be integrated into a single module.   Any or all of the standard deviation calculation module, the approximate function calculation module, the low-pass filter module, the average value calculation module, and the SVV calculation module may be integrated into a single module. The method according to claim 7.   The method according to claim 18, wherein any or all of the approximate function calculation module, the sampling means, the low-pass filter module, and the processing means can be integrated into a single module. Inputting a waveform data set corresponding to measured arterial blood pressure over a calculated interval including at least two cardiac cycles;
Calculating a standard deviation value for the input waveform data set over each cardiac cycle;
Calculating an estimate of cardiac stroke volume variance (SVV) as a function of the standard deviation value, causing the processing system to equal to or derive from the SVV. A computer program that determines possible cardiac parameters. Determining a maximum standard deviation value and a minimum standard deviation value over the calculation interval;
32. The computer program according to claim 31, further causing the processing system to calculate the estimated value of the SVV as a function of the maximum standard deviation value and the minimum standard deviation value. Calculating an average standard deviation value over the calculation interval;
33. causing the processing system to further execute, for the average standard deviation value, calculating an estimated value of SVV that is proportional to a difference between the maximum standard deviation value and the minimum standard deviation value. A computer program described in 1.   Calculating the standard deviation value; determining the maximum standard deviation value and the minimum standard deviation value; calculating the average standard deviation value; and calculating an estimated value of the SVV. 34. The computer program of claim 33, any or all of which can be combined as a single software module. Calculating a standard deviation of the standard deviation value over the calculation interval set in at least one respiratory cycle;
Calculating an average of the standard deviation values over the calculation interval;
32. The computer program according to claim 31, further causing the processing system to calculate an estimated value of SVV that is proportional to a ratio of a standard deviation of the standard deviation value and an average of the standard deviation values.   32. The computer of claim 31, further causing the processing system to calculate an estimate of right ventricular end diastolic volume as inversely proportional to the calculated estimate of cardiac stroke volume variance. program. Calculating an approximate function that most closely matches the calculated standard deviation value according to a predetermined metric over at least one of the calculation intervals;
Generating a set of sampled approximations by sampling the approximation function at an interval specific sampling rate;
Low-pass filtering the sampled approximation;
32. The computer program of claim 31 further causing the processing system to calculate the SVV estimate as a function of the low-pass filtered sampled approximation. Calculating each measurement of a series of measurements as a function of the standard deviation of the waveform data set over the individual calculation intervals over at least one calculation interval including at least two cardiac cycles, the waveform data set Corresponds to blood pressure, and
Calculating an approximate function that most closely matches the measured values of the waveform data set according to a predetermined metric;
Generating a set of sampled approximations by sampling the approximation function at an interval specific sampling rate;
Low-pass filtering the sampled approximation;
Causing the processing system to calculate the cardiac output value by calculating an estimate of a cardiac or hemodynamic stroke value as a function of the low-pass filtered sampled approximation. A computer program to be determined. For each set the calculated distance to the respiratory cycle, set the specific sampling rate to satisfy the Nyquist criterion to said interval, or, the number of samples is equal to the number of cardiac cycles in the individual respiratory cycle 40. The computer program of claim 38, further causing the processing system to set to   Calculating the approximation function; generating the set of sampled approximations; setting a sampling rate specific to the interval; and low-pass filtering the sampled approximations; 40. The computer program product of claim 39, wherein any or all of calculating an estimate of the stroke volume value can be combined as a single software module. A standard deviation calculation module for calculating a standard deviation value for the waveform data set over each cardiac cycle;
And a SVV calculation module that calculates an estimated value of SVV as a function of the standard deviation value. A measurement value generating means for generating a series of measurement values from the waveform data set;
A standard deviation calculation module for calculating each measurement of a series of measurements as a function of the standard deviation of the waveform data set over the individual calculation intervals over at least one calculation interval including at least two cardiac cycles ;
An approximate function calculation module for calculating an approximate function that most closely matches the measured value according to a predetermined measurement standard;
Sampling means for generating a set of sampled approximations by sampling the approximation function at an interval specific sampling rate;
A stroke volume value calculation module that calculates an estimated value of a stroke volume value as a function of the sampled approximate value. Sampling means, for each calculated distance, set the specific sampling rate to satisfy the Nyquist criterion to said interval, or, the number of samples is equal to the number of cardiac cycles in the individual respiratory cycle The method of claim 18, further comprising: setting.   The processing system uses sampling means to set, for each calculation interval, a sampling rate specific to the interval so as to satisfy the Nyquist criterion, or the number of samples in the heart during an individual respiratory cycle. 24. The system of claim 23, further configured to set to be the same as the number of periods.

Images