METHOD FOR OPERATING A BLOOD PRESSURE MEASURING DEVICE AND ARRANGEMENT FOR MEASURING THE PRESSURE IN A BLOOD VESSEL

Abstract

The invention relates to a method for operating a blood pressure measuring device, comprising the method steps: performing a blood pressure measurement using the blood pressure measuring device, consisting of a cuff for placing around an extremity, a pump for inflating the sleeve and a pressure sensor for registering the cuff pressure, inflation of the cuff taking place in an inflation phase and release of the pressure in the sleeve taking place in a release phase, registering and storing the pressure profile in the cuff over time during the inflation phase and/or during the release phase in a control and storage unit, extracting a pulse-like signal component from the pressure profile over time of the measured cuff pressure during the inflation phase and/or the release phase by a signal analysis inside the control and storage unit, signal analysis of the extracted in pulse-like signal component.

Claims

1. Method for operating a blood pressure measuring device, comprising the following method steps: performing a blood pressure measurement with the blood pressure measuring device, consisting of a cuff (1) for application around an extremity, a pump (2) for inflating the cuff and a pressure sensor (3) for registering the pressure within the cuff, wherein inflation of the cuff takes place in an inflation phase (Inf) and deflation of the pressure in the cuff in a release phase (Def), registering and storing the temporal pressure curve (P.sub.meas) in the cuff (1) during the inflation phase and/or during the release phase in a control and storage unit (4), extracting a pulse-like signal component (P.sub.osc) from the temporal pressure curve of the measured cuff pressure during the inflation phase and/or the release phase by means of a signal analysis within the control and storage unit (4), signal analysis of the extracted pulse-like signal component (P.sub.osc) within the control and storage unit with the steps of: decomposition of the pulse-like signal component into data over individual periods to identify individual pulse waves (PW), combining the data of the respective individual pulse waves (PW) from values of at least one calculation function from a predetermined calculation function set to generate a respective classified individual pulse wave (PW), transforming the respective classified single pulse wave, evaluating the transformed data to determine hemodynamic parameters.

2. Method according to claim 1, characterized in that the decomposition of the pulse-like signal component into data about the individual pulse waves (PW) is carried out in the time course of the pulse-like signal component (P.sub.osc).

3. Method according to claim 1, characterized in that in combining the data of the respective individual pulse waves (PW) to generate the respective classified individual pulse wave, at least two calculation functions are combined with each other in a weighted manner.

4. Method according to claim 1, characterized in that the transformation of the respective classified single pulse wave into the period-specific amplitude and phase spectrum is carried out by means of a Fourier analysis unit.

5. Method according to claim 1, characterized in that a retransformation of the period-specific amplitude and phase spectrum of the classified individual pulse wave is performed to determine the time course of the aortal pulse wave.

6. Method according to claim 1, characterized in that the registration of the temporal pressure curve in the inflation phase and/or in the release phase is carried out digitally at a fixed device-specific sampling rate and a fixed device-specific resolution, wherein subsequent sampling to a platform-independent sampling rate is carried out in conjunction with an approximation of the measured temporal pressure curve (P.sub.meas) in the cuff.

7. Method according to claim 1, characterized in that the pulse-like signal component (P.sub.osc) is extracted via a bandpass filter to separate a non-oscillating cuff pressure (P.sub.cuff).

8. Method according to claim 1, characterized in that the separated non-oscillating cuff pressure (P.sub.cuff) is checked for compliance with a standard curve, wherein a determination of an artefact parameter for indicating a measured value-distorting influence is carried out from the degree of deviation of the non-oscillating pump trend from the standard curve.

9. Arrangement for measuring the pressure in a blood vessel, comprising a cuff (1) for application around an extremity, a pump (2) for inflating the cuff, a pressure sensor (3) for registering the temporal pressure curve (P.sub.meas) applied in the cuff, and a control and storage unit (4) for operating the pump and the pressure sensor and for storing the temporal pressure curve (P.sub.meas), wherein the control and storage unit (4) has a control and evaluation program (5) for separating pulse-like components (P.sub.osc) in the temporal pressure curve (P.sub.meas) during an inflation and/or release phase of the temporal pressure curve and for a signal analysis of the separated pulse-like components (P.sub.osc).

10. Arrangement according to claim 9, characterized in that the control and evaluation program (5) can be loaded and updated from an external storage means (6) via an interface (7) onto the control and storage unit (4).

11. Arrangement according to claim 9, characterized in that the control and storage unit (4) has an interface (8) for reading out stored data, in particular the measured temporal pressure curve (P.sub.meas), to an external evaluation unit (10) and/or a remote host.

12. Arrangement according to claim 9, characterized in that the control and storage unit (4) has a display (11) for outputting parameters obtained from the pulse-like components of the pressure curve, in particular for outputting a heartbeat frequency and/or hemodynamic parameters from the pulse wave analysis and/or a curve of an aortic pulse wave.

13. Arrangement according to claim 9, characterized in that the control and storage unit (4) has a switching facility between a first operating mode for carrying out a standard blood pressure measurement and a second operating mode for carrying out a measurement of the pulse-like components of the temporal pressure curve and for determining parameters of an aortal pulse wave.

14. Use of a method according to claim 1 for determining and analyzing an aortic pulse wave and for determining hemodynamic parameters of the pulse wave analysis, in particular an aortic blood pressure.

15. Use of an arrangement according to claim 9 for determining and analyzing an aortic pulse wave and for determining hemodynamic parameters of the pulse wave analysis, in particular an aortic blood pressure.

Description

[0034] The method and the arrangement will be explained in more detail in the following on the basis of embodiment examples. The enclosed figures serve to clarify this, wherein:

[0035] FIG. 1 shows a basic configuration of an arrangement for carrying out the method,

[0036] FIG. 2 shows an exemplary progression of the pressure within the cuff with an inflation and a release phase,

[0037] FIG. 3 shows an exemplary representation of the amplitude spectrum of the impulse response for an exemplary bandpass filter,

[0038] FIG. 4 shows an exemplary representation of the phase spectrum of the impulse response for an exemplary bandpass filter,

[0039] FIG. 5 shows an exemplary representation of a pressure signal broken down into oscillating and non-oscillating components,

[0040] FIG. 6 shows an exemplary representation of a basepoint determination on a single pulse wave,

[0041] FIG. 7 shows an exemplary pulse wave sequence with successful basepoint detection,

[0042] FIG. 8 shows an exemplary representation of the relevant pulse wave sequence range.

[0043] FIG. 1 shows a basic configuration for an arrangement for carrying out the method. The arrangement includes a cuff 1, which can be inflated by a pump 2 via a hose 2a. The cuff is applied around an extremity, for example around an upper arm. Furthermore, a pressure sensor 3 is provided, with which the time course of the pressure in the cuff is registered. The pressure sensor can be located at any position in the area of the cuff or pump system.

[0044] In the configuration presented here, the pressure sensor and the pump are combined in one housing together with further storage and data processing elements and a control and storage unit 4. The control and storage unit controls both the operation of the pump and the measured value acquisition by the pressure sensor. Especially under the control of the control and processing unit, the pump realizes e.g. a temporally linear pressure increase in the cuff during an inflation phase. During a release phase, an outlet valve not shown here, for example, causes the pressure in the cuff to drop linearly over time. These linear pressure curves are basically not necessary, but they prove to be useful and advantageous with regard to later artefact detection.

[0045] The control and storage unit 4 contains a control and evaluation program 5, which carries out the evaluation steps described in more detail below. In this example, the control and evaluation program is stored on an external storage medium 6 such as an SD card, a USB stick or another storage medium. It can be loaded into the control and storage unit 4 via an interface 7, for example an SD slot or a USB connection. Of course, it is also possible to permanently store the control and evaluation program in the control and storage unit in the form of a fixed firmware or firmware that can be updated via the interfaces.

[0046] The control and storage unit 4 contains an additional interface 8 for reading out the stored measurement data and for transferring the data to an external evaluation unit 9, for example a PC. The external evaluation unit has an evaluation program 10 with which the stored read-out data can also be analyzed. Interface 8 can be either wireless or wired. It is also possible to transfer data via a communication network to a remote host, so that remote monitoring and evaluation of blood pressure parameters is possible.

[0047] In the present case, the control and storage unit 4 also contains a display 11 for direct output of the measured blood pressure data and the temporal pressure curve in the cuff 1. Further operating elements may be provided, in particular for switching between a conventional blood pressure measurement and an operating mode in which the method according to the invention is carried out.

[0048] As mentioned above, one of the essential aspects of the method according to the invention is that the blood pressure signals during the measurement when inflating and/or deflating the cuff are both recorded and evaluated in their temporal course and also transformed for their later evaluation. As described, the apparatus consists of a cuff, a sensor and a pump. The pump inflates the cuff to a certain pressure. At the same time, the pulse waves during the process of inflating and/or deflating a cuff are recorded by monitoring the pressure curve within the cuff over time.

[0049] With the aid of special method steps described in more detail below, the pulse waves required for pulse wave analysis are then extracted from the pressure curve recorded over time during the inflation phase and/or the release phase on the arm or another limb. After subsequent processing, the signals of the extracted pulse waves are transformed to values that provide more accurate information about the aortal pulse waves. Interferences within the signal and artefacts of the measurements are detected and excluded from the analysis and/or corrected or signaled accordingly.

[0050] FIG. 2 shows an exemplary temporal signal curve of the recorded pressure in the cuff according to the procedure provided for this purpose. It can be seen that the pressure first increases in an initial inflation phase Inf and then decreases again in a subsequent release phase Def.

[0051] It can be seen that the pressure increases essentially linearly in the inflation phase and decreases essentially linearly in the release phase. Both the course of the linear pressure increase and the course of the linear pressure decrease are overlaid by an oscillatory signal component, which is shown here in the form of jagged lines in the pressure curve. This oscillatory signal component contains the information about the aortal pulse wave. According to the invention, it is extracted from the temporal pressure curve, wherein the exemplary method steps explained below are applied.

[0052] In order to record the temporal pressure curve shown in FIG. 2, a conventional blood pressure measurement is first carried out. In contrast to conventional blood pressure measurement, in which, for example, only the maximum amplitude of the oscillating signal component is registered, the method according to the invention records the temporal pressure curve P.sub.meas as a whole. This data acquisition can be done digitally in particular. The sampling rate and resolution is device-dependent. For example, a sampling rate of 40 Hz at a resolution of 12 bits is used.

[0053] Both the sampling rate and the resolution are not fixed per se, but only device-specific quantities. In order to guarantee platform independence and thus the best possible processability, the measured value quantity of the temporal pressure curve is then transformed to a uniform standard either in the control and storage unit itself or in the external evaluation unit so that its processing is platform-independent and thus uniform. For this purpose, the signal of the temporal pressure curve can be sampled up to a sampling frequency of 100 Hz, for example. Various approximation methods can be used for this sampling process. In particular, the so-called cubic spline approximation is used for this purpose. This approximation approximates two adjacent sample points.

[0054] For the cubic spline approximation, polynomials of degree 3 are determined for two adjacent sample points (x.sub.i,y.sub.i) and (x.sub.i+1,y.sub.i+1):

s.sub.i(x)=a.sub.i+b.sub.i(xx.sub.i)+c.sub.i(xx.sub.i).sup.2+d.sub.i(xx.sub.i).sup.3

[0055] Because of the continuity of the signal to be sampled, the continuity requirement s.sub.i(x.sub.i+1)=s.sub.i+1(x.sub.i+1) applies, wherein s must be continuously differentiable twice:

s.sub.i(x.sub.i+1)=s.sub.i+1(x.sub.i+1)

s.sub.i(x.sub.i+1)=s.sub.i+1(x.sub.i+1)

[0056] When a new x selected, the signal can be resampled.

[0057] If the output signal is upsampled to a sampling frequency of 100 Hz, the following applies in this case

[00001]$.Math..Math.x=.Math.\frac{1}{100}$

[0058] However, the choice of the type of spline with regard to the edges does not play a decisive role.

[0059] From this platform-independent sampled signal, the pulse waves, i.e. the oscillatory component P.sub.osc of the measured temporal pressure curve P.sub.meas, are extracted in the next step. The extraction is carried out, for example, by using an IIR bandpass filter. An FIR filter can also be used. Possible here is a Butterworth 6th order filter.

[0060] FIGS. 3 and 4 show exemplary filter characteristics. FIG. 3 shows the filter magnitude stated in dB as a function of frequency in the range from 0 to 50 Hz. FIG. 4 shows an exemplary phase response of the bandpass filter in the frequency range from 0 to 50 Hz. The bandpass range covers a frequency range from 0.5 Hz to 20 Hz. It is already adapted to the expected frequencies of the pulsating signal component. FIG. 3 shows that the magnitude of the filter characteristic curve in this range extends quasi horizontal and only decreases at higher frequencies of more than 20 Hz. The filter has the property of a zero-phase filter to avoid falsification of the phase spectrum of the processed signal.

[0061] An additive relationship can be applied to the measured temporal pressure signal. The measured pressure curve over time is the sum of the cuff pressure generated by the pump and the pulse pressure generated by the pulse wave. It therefore applies:

P.sub.meas=P.sub.cuff+P.sub.osc

P.sub.meas is the temporal pressure curve measured by pressure sensor 3, P.sub.cuff is the pump pressure generated by pump 2, i.e. the cuff pressure, and P.sub.osc are the pressure oscillations generated by the pulse wave on the arm or another limb, from which the characteristics of the aortal pulse waves are ultimately to be determined.

[0062] The bandpass range to be used for the bandpass filter can be derived from the following considerations. Correspondingly different operating conditions naturally require a basic configuration that deviates therefrom. The measurement takes place on the test person, e.g. at rest and in a sufficiently defined physiological state. Under these conditions, clearly definable heart rates in the range of 40 to 100 beats per minute can be assumed. This results in correspondingly defined cut-off frequencies for the oscillatory component P.sub.osc in the temporal pressure curve.

[0063] If Hr_Low is the low heart rate of 40 beats per minute and Hr_High is the high heart rate of 100 beats per minute, this results in a lower cut-off frequency f.sub.L and an upper cut-off frequency f.sub.H by f.sub.L=Hr_Low/60=40/600.7 Hz and f.sub.H=Hr_High/60=100/601.7 Hz respectively.

[0064] The energy of a pulse wave generated by the beating heart is distributed over a certain frequency spectrum. For example, at a heart rate of 60 beats per minute, the largest portion of the energy of the pulse wave is in the range of 1-10 Hz, distributed over 10 harmonic oscillations.

[0065] Thus, if a range of 10 harmonic oscillations is assumed and the upper limits are taken into account, a frequency range of 0.7 Hz to 10-1.7 Hz, i.e. from 0.7 to 17 Hz, results for the pulse wave. However, these filter limits are not sharply defined. An additional interval is therefore added in each case on both sides of the frequency interval, so that a range of 0.5-20 Hz is selected as the frequency interval for the bandpass filter.

[0066] The oscillatory component P.sub.osc of the measured temporal pressure signal P.sub.meas, and thus the pulse wave extracted from the measured signal, thus results from applying the filter function of the bandpass filter to the measured pressure signal:

P.sub.osc=filter(P.sub.meas)

[0067] If the pulse wave is extracted by the filter, it follows from the above additive relationship that the remaining part of the originally recorded pressure signal is thus the cuff pressure generated by the pump.

P.sub.cuff=P.sub.measP.sub.osc

[0068] The measured pressure signal P.sub.meas is thus broken down into the cuff pressure P.sub.cuff and the pulse oscillations P.sub.osc.

[0069] FIG. 5 shows an example of the oscillatory component P.sub.osc obtained from the output signal. The cuff pressure P.sub.cuff can then ideally show a non-oscillating, almost linear rise in the area of the inflation phase and a non-oscillating and at least almost linear fall in the area of the release phase, provided that the pump generates a pressure increase in the cuff that is linear over time and provided that a pressure drop is generated that is linear over time when the pressure is deflated. FIG. 5 shows that these requirements are sufficiently well met in the example presented here.

[0070] If the characteristics of the inflation or deflation, i.e. the time characteristic of the cuff pressure during inflation realized by the pump or the time characteristic of the cuff pressure during deflation realized by the deflation valve, are known in advance, these characteristics can be checked for accuracy using the separated cuff pressure function P.sub.cuff. This offers the possibility to calibrate the measuring arrangement or to determine additional artefacts in the measuring process.

[0071] If, for example, the linear characteristic of the linear drop in cuff pressure is known in advance, the actually measured and separated cuff pressure P.sub.cuff can be examined for linearity. Strong deviations of the cuff pressure P.sub.cuff from the linearity can then be interpreted as motion artefacts or other falsifying influences on the measuring process. This makes it possible to obtain an error signal which can be output to the measuring system to signalize the error. Accordingly, other non-linear characteristics can also be used to evaluate artefacts or a remaining residual.

[0072] Possible methods for the determination of the residual and a comparative test between the previously known characteristic curve and the time course of the measured cuff pressure P.sub.cuff are distance determinations of the values of the cuff pressure P.sub.cuff for the regression of the function to a known polynomial of n.sup.th degree. In a test for linearity, the polynomial is accordingly a polynomial of degree 1.

[0073] The cuff pressure P.sub.cuff is thereby subjected to a regression in one step. In a second step, the resulting regression relationship regression(P.sub.cuff) is then compared with an n.sup.th degree polynomial. The parameter of the artefact is then determined by the following relationship:

artefact=|regression(P.sub.cuff)p.sup.n|>threshold

with p.sup.n as a comparison polynomial of n.sup.th degree. This means that a measurement artefact is present precisely when the amount of the difference between the regression relationship of the cuff pressure P.sub.cuff to the comparison polynomial p.sup.n exceeds a certain predefined threshold value. In such a case, the control and storage unit will output an error signal, for example. The remaining residual can also provide further valuable information about the investigation and quality of the signals.

[0074] Following the separation of the pulse-like signal component P.sub.osc from the cuff pressure P.sub.cuff, the pulse wave signal is evaluated as such. The sub-steps carried out are first the decomposition of the pulse-like signal component into individual periods and a subsequent analysis of each individual, but at least one individual period itself, because the data contained here represent the properties of the individual pulse wave to be determined.

[0075] The identification of each individual pulse wave in the pulse-like signal component is carried out by means of a time derivative of the signal, for example by the control and storage unit with the aid of the control and evaluation program contained in this unit. The execution of the first time derivative with a subsequent basepoint determination has proven to be suitable, but is not limited to it.

[0076] With the basepoint method described here as an example, all turning points are determined with a positive slope. These are characterized in the first derivative precisely by the fact that the value of the first derivative is extreme in its place:

[00002]$w=\arg .Math..Math.\max .Math.\frac{{\mathrm{dP}}_{\mathrm{osc}}}{\mathrm{dt}}$

w are the respective positions of the positive turning points. The result is a set of all determined deflections, i.e. peaks, within the pulse-like signal component.

[0077] Then all peaks of the 1.sup.st derivative are sorted in descending order of size.

[00003]$s=\mathrm{sort}.Math..Math.\left(\frac{{\mathrm{dP}}_{\mathrm{osc}}}{\mathrm{dt}}.Math.\left(w\right)\right)$

where s is the sorted sequence of the determined turning points.

[0078] This sorted sequence s of peaks is analyzed step by step. Each peak is recorded as a turning point if its minimum height is greater than 0 and if its distance to all turning points already recorded is greater than a defined time interval with respect to a given sample rate FS.

[0079] Each peak w(i) from the sequence s(i) is therefore a turning point idx if the following conditions are simultaneously fulfilled:

(a)s(i)>0 and

[00004]$\left(b\right).Math.\mathrm{idx}=\left\{w\left(i\right)s\left(i\right)>0.Math..Math.\frac{w\left(i\right)-\mathrm{idx}\left(j\right)}{\mathrm{FS}}.Math..Math.1000>D,\mathrm{mit}.Math..Math.j<i\right\}$

[0080] Condition (b) expresses that a peak w(i) must have a certain minimum distance D to an already identified peak idx(j). This minimum distance is

(a)s(i)>0 and

[00005]$\left(b\right).Math.\mathrm{idx}=\{w\left(i\right)s\left(i\right)>0.Math..Math.\frac{w\left(i\right)-\mathrm{idx}\left(j\right)}{\mathrm{FS}}.Math..Math..Math..Math.1000>2?,\mathrm{Tii}.\mathrm{it}/<i.Math.\text{?}.Math..Math.\text{?}.Math.\text{indicates text missing or illegible when filed}.Math.$

[0081] independent of the sample rate. Here, for example, this is 350 ms.

[0082] A subsequent ascending sorting of idx puts the indexes in the correct chronological order.

[0083] In order to avoid possible noise and small disturbances in the signal and even more so in differentiation, the 1.sup.st derivative can be calculated according to the Savitzsky-Goley method if necessary.

[0084] The position of their respective basepoints is of decisive importance for the identification of the individual pulse wave in the pulsating signal component. These can be determined from the turning points.

[0085] Starting from the respective determined turning point, the basepoints of the individual pulse waves are localized, for example, with the help of intersecting tangents in the vicinity of each individual maximum in the 1.sup.st derivative of the P.sub.osc signal, i.e. in the vicinity of the previously determined turning point. FIG. 6 shows an illustration of this. Besides the determined tangent t, at the turning point w(i), the first local minimum M in the time range before the pulse wave is searched for. This local minimum M has a horizontal tangent t.sub.M. Then the intersection point S between the horizontal tangent t.sub.M and the regression line, i.e. the tangent t.sub.w, is calculated. This intersection point S, projected onto the signal, is the basepoint F of the pulse wave PW. This point can be adjusted additionally.

[0086] For each basepoint F with the coordinates F.sub.j(x, y) the following applies

x.sub.i=idx(f)i.Math.gap

wherein gap means the respective gap in the samples, and y.sub.i=P.sub.osc (x.sub.i) with i=1, . . . , n the regression points next to the turning point idx(j), wherein n is the number of regression points. Based on the coordinates (x.sub.i, y.sub.i) of the basepoints F.sub.j, coefficients a,b of a straight line equation g(x)=ax+b can now be determined. FIG. 7 illustrates a series of determined basepoints F in the pulse component of the signal, each of which includes individual pulse waves between them, some of which are marked with the reference sign PW as an example.

[0087] In a next step, a relevant part of the oscillatory signal component P.sub.osc is defined with the determined individual pulse waves PW. FIG. 8 shows an illustrative example diagram. For this purpose, a meaningful pressure value is assumed, which is related to the systolic and diastolic blood pressure value. The pressure range of the pulse signal lies in particular between a pressure value of, for example, 10 mmHg above the systolic value and 10 mmHg below the diastolic pressure value.

[0088] A heart period is then defined as a section of two consecutive basepoints F.sub.i determined as before. The pulse pressure PP of each period is determined from the difference between systole and diastole.

[0089] The average pressure can be determined by the following relationship:

[00006]$y_m=\frac{1}{T}.Math.{}_0^T.Math.\mathrm{ydt}\frac{1}{N}.Math.\underset{i=1}{\overset{N}{.Math.}}.Math.y\left(i\right)$

wherein T is the length of the time interval of period m over which averaging takes place, N is the number of samples over which averaging takes place, wherein y(i) is the amplitude of each individual registered sample i.

[0090] Subsequently, the mean pressure is subtracted from the pulse pressures of the period. This allows a so-called zero mean signal to be obtained.

[0091] Then the amplitudes of the individual pulse waves are scaled to the height of the measured pulse pressure PP.

[0092] However, not only the position and height of each individual pulse wave is of interest, but also its shape. This can be analyzed as follows:

[0093] The pulse waves in the inflation and/or release process have a different shape depending on the current cuff pressure. These forms can be classified into classes. Each class is represented by a predetermined calculation function, which reflects certain function courses of the pulse wave. Any number of calculation functions can be defined, whose values can be predefined as fixed variables in the control and storage unit, but also in the respective external evaluation devices. The number and design of the respective properties of the calculation functions can be changed flexibly.

[0094] The number of calculation functions, i.e. the number of classes, is k, wherein k is a natural number greater than 0.

[0095] Each pulse wave can then be assigned to such a class by the signal processing in the control and storage unit. Decisive for these classes are the different manifestations in the course of the individual pulse wave PW, respectively the existence of points and features of a first and a second shoulder of the individual pulse peak as well as a strength of a dicrotism or incision of the respective peak. Here, the first and second shoulder describes the strength of the double peakness of a peak, the dicrotism describes the change of the pulse wave due to the arrival of the reflected wave, while the incision indicates the expression of the dicrotism to a local minimum. Dicrotism and incision thus describe how round the individual period, i.e. the individual pulse wave, is in its course. Based on this, each individual pulse wave can be classified into shape classes.

[0096] The assignment to a class can be made according to the principle of the nearest neighbor. For example, the single pulse wave PW may have a left-side shoulder, whereas a first calculation function of class k=1 has no left-side shoulder and a second calculation function of class k=2 has a left-side shoulder. The single pulse wave thus deviates more from the shoulderless calculation function in its form than from the calculation function with the left-side shoulder. Thus the pulse wave PW belongs to class k=2 and is thus represented by the calculation function of class k=2.

[0097] Another possibility to assign a class is to perform the similarity analysis using a dynamic time equalization to make the period lengths and times invariant and allow a comparison. This comparison makes the shapes of the pulse waves comparable with each other and with the calculation functions and allows them to be assigned to classes.

[0098] For each of these classes, a class-specific algorithm can be specified, which determines the hemodynamic parameters for the class. A transfer function, consisting of a set of two subfunctions each, has proven to be suitable, but is not limited to it. These subfunctions describe the amplification of the amplitude spectrum and the shift of the phase spectrum. To determine the transformed signal, the amplitude and phase spectrum is determined from the signal. For this purpose, a Fast Fourier Transformation (FFT) is applied, for example.

[0099] The amplitude spectrum obtained therefrom is multiplied by the gain of the transfer function. The phase spectrum is added with the phase shift. The retransformation yields the aortic pulse curve of this specific class. Finally, the transformed data is evaluated to determine hemodynamic parameters.

[0100] To avoid discontinuities during transformation, the intermediate areas between classes are kept smooth. This means that if a pulse wave can be assigned to two classes, for example, each of these two classes is treated as equally probable.

[0101] If another pulse wave PW is now a little closer to one of these classes, the assignment is clear. For a coefficient {tilde over (c)} the following applies:

{tilde over (c)}=c.sub.i+(1+)c.sub.i+1,=[0,1]

C.sub.i and C.sub.i+1 are the coefficients of the two closest neighbors.

[0102] The smoothness ensures that the two periods after transformation are not completely different, regardless of the class to which the first period was assigned.

[0103] After transformation, for each scaled pulse wave determined on the arm or another limb, an aortal pulse wave with its specific hemodynamic parameters is present. This information shall be weighted accordingly. It has proven to be suitable to determine certain parameters {tilde over (x)} together with a corresponding confidence interval {tilde over (x)} at the individual aortal pulse waves, but it is not limited to this. The confidence interval {tilde over (x)} to x is calculated as follows:

[00007]$\tilde{x}=\stackrel{\_}{x}1,96.Math..Math.\frac{}{\sqrt{N}}$

[0104] Here the parameter is the standard deviation and N is the number of periods. With increasing N the confidence interval decreases, a slow inflation and deflation rate, as well as a comparatively fast heart rate increase the accuracy. At low heart rates, the speeds of the inflation phase and the release phase should therefore be adjusted accordingly.

[0105] The width of the 95% confidence interval provides information about the reliability and quality of the parameter. Based on the extracted pulse waves determined on the arm or another limb and the calculated aortic pulse waves, it is possible to determine further hemodynamic parameters, such as aortic (central) blood pressure, for a comprehensive pulse wave analysis.

[0106] The method and according to the invention and the arrangement according to the invention were explained in more detail by means of embodiment examples. Further embodiments and designs result from the subclaims and within the scope of action by the person skilled in the art.

LIST OF REFERENCE NUMERALS

[0107] 1 Cuff [0108] 2 Pump [0109] 2a Hose [0110] 3 Pressure sensor [0111] 4 Control and storage unit [0112] 5 Control and evaluation program [0113] 6 External storage medium [0114] 7 Interface [0115] 8 Additional interface [0116] 9 External evaluation unit [0117] 10 External evaluation program [0118] 11 Display [0119] Inf Inflation phase [0120] Def Release phase [0121] P.sub.meas Measured temporal pressure curve [0122] P.sub.cuff Cuff pressure [0123] P.sub.osc Oscillatory component of the pressure [0124] F Basepoint [0125] t.sub.w Turning point tangent [0126] t.sub.M Tangent in minimum