| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
TRANSLATIONAL PHYSIOLOGY
Department of Electrical and Computer Engineering, Michigan State University, East Lansing, Michigan
Submitted 28 November 2005 ; accepted in final form 4 March 2006
ABSTRACT
We present an evaluation of a novel technique for continuous (i.e., automatic) monitoring of relative cardiac output (CO) changes by long time interval analysis of a peripheral arterial blood pressure (ABP) waveform in humans. We specifically tested the mathematical analysis technique based on existing invasive and noninvasive hemodynamic data sets. With the former data set, we compared the application of the technique to peripheral ABP waveforms obtained via radial artery catheterization with simultaneous thermodilution CO measurements in 15 intensive care unit patients in which CO was changing because of disease progression and therapy. With the latter data set, we compared the application of the technique to noninvasive peripheral ABP waveforms obtained via a finger-cuff photoplethysmography system with simultaneous Doppler ultrasound CO measurements made by an expert in 10 healthy subjects during pharmacological and postural interventions. We report an overall CO root-mean-squared normalized error of 15.3% with respect to the invasive hemodynamic data set and 15.1% with respect to the noninvasive hemodynamic data set. Moreover, the CO errors from the invasive and noninvasive hemodynamic data sets were only mildly correlated with mean ABP (
= 0.41, 0.37) and even less correlated with CO ( = –0.14, –0.17), heart rate ( = 0.04, 0.19), total peripheral resistance ( = 0.38, 0.10), CO changes ( = –0.26, –0.20), and absolute CO changes ( = 0.03, 0.38). With further development and successful prospective testing, the technique may potentially be employed for continuous hemodynamic monitoring in the acute setting such as critical care and emergency care.
Doppler ultrasound; Finapres; pulse contour analysis; radial artery catheterization; thermodilution
On the other hand, peripheral arterial blood pressure (ABP), which is related to CO through the arterial tree, may be measured reliably and continuously via radial artery catheterization, which is less invasive than pulmonary artery catheterization. Indeed, this relatively safe procedure is performed in a majority (e.g., 50–80%) of all critically ill patients (33). Moreover, over the past few decades, totally noninvasive methods have been developed and refined to continuously measure peripheral ABP on the basis of finger-cuff photoplethysmography (16) and arterial tonometry (17). These noninvasive methods are even available as commercial systems at present (see, for example, the Finometer and Portapres, Finapres Medical Systems, The Netherlands and the T-Line Blood Pressure Monitoring System, Tensys Medical, San Diego, CA).
Beginning with the seminal work of Frank in 1899 (13) and Erlanger and Hooker in 1904 (12), numerous cardiovascular researchers have sought analysis techniques to compute CO to within a constant scale factor from the contour of ABP waveforms so as to permit continuous (i.e., automatic and without the need for an operator), quantitative measurements of relative changes in CO and expand the clinical monitoring of CO. (Note that some researchers have also sought to calibrate the proportional CO estimates via empirical formula.) Although a wide variety of "pulse contour analysis" techniques have been proposed, they are all conceptually the same to the extent that the waveform analysis is performed only over time scales within a cardiac cycle (e.g., see Fig. 1) (26). However, over such short time scales, peripheral ABP waveforms are dominated by highly complex waves propagating back and forth in the distributed arterial tree (28). Thus the previous analysis techniques have generally proven to be too inaccurate (in terms of estimating proportional CO) for clinical use.
|
In this paper, we present an evaluation of the technique in humans based on previously published invasive and noninvasive hemodynamic data sets (10, 24, 25, 27). With these data, we were specifically able to compare the application of the technique to 1) invasive peripheral ABP waveforms obtained via radial artery catheterization with reference thermodilution measurements in 15 intensive care unit (ICU) patients in which CO was changing because of disease progression and therapy, and 2) noninvasive peripheral ABP waveforms obtained via a commercial finger-cuff photoplethysmography system with reference Doppler ultrasound measurements made by an expert in 10 healthy subjects in which CO was altered through pharmacological and postural interventions.
METHODS
Mathematical Analysis Technique
Our technique for monitoring relative CO changes by mathematically analyzing a peripheral ABP waveform was introduced in Ref. 26 and is described in detail therein. Here, we review the background, underlying concepts, and implementation of the technique from a physiological perspective while including its mathematical steps.
Background and underlying concepts.
The technique that we have developed builds on the previous pulse contour analysis work of Bourgeois et al. (6) as well as Osborn et al. (31). These investigators assumed that the arterial tree could be well represented by a two-element windkessel model accounting for the lumped compliance of the large arteries (arterial compliance, AC) and the total peripheral resistance (TPR) of the small arteries (see electrical analog in Fig. 1). They further assumed that TPR does not vary within a diastolic interval [as justified by the relatively slow local and autonomic nervous mechanisms responsible for modulating TPR (3, 14)] and that AC is approximately constant over a wide pressure range and on the time scale of days to months (see experimental evidence in Refs. 5, 6, 15, 26, 31, and DISCUSSION). On the basis of these assumptions, these investigators predicted that ABP should decay like a pure exponential during each diastolic interval with a time constant equal to the product of TPR and AC (windkessel time constant, 
). Thus their pulse contour analysis involved first fitting a monoexponential function to each diastolic ABP interval to measure and then dividing the time-averaged ABP with to compute CO to within a constant scale factor equal to 1/AC (see Fig. 1).
This pulse contour analysis proved to be successful when applied to ABP waveforms measured centrally in the aorta, because the diastolic interval of these waveforms can resemble an exponential decay after incisura (see Fig. 2). However, central ABP is rarely measured clinically because of the risk of blood clot formation and embolization. Moreover, in readily available peripheral ABP waveforms, an exponential diastolic decay is usually not apparent (see Fig. 2). Indeed, it is well known that the contour of the arterial pulse changes significantly as it traverses through the arterial tree (30). The reason is that the arterial tree is not simply a lumped system as the windkessel model suggests, but rather a complicated distributed system with impedance mismatches throughout due to vessel tapering, bifurcations, and caliber changes. Thus the diastolic (and systolic) intervals of peripheral ABP waveforms are corrupted by complex wave reflections that occur at each and every site of impedance mismatch. [Note that the complexity of these sites and their varying distances from the aorta result in reflected waves with large phasic differences, which can tend to mitigate the cumulative effects of these waves on the central ABP waveform, i.e., destructive interference (28).] The above pulse contour analysis therefore cannot be applied to readily measured peripheral ABP waveforms.
|
Implementation steps.
Our technique therefore mathematically analyzes a digitized peripheral ABP waveform over long time intervals (
6 min) to determine the pure exponential decay that would eventually result if pulsatile activity abruptly ceased. More specifically, the ABP response to a single, solitary cardiac contraction [h(t) in Fig. 3 ] is estimated from the ABP waveform. Then, the windkessel time constant is determined by fitting a monoexponential function to the tail end of this response once the faster wave reflections have vanished (Fig. 3). Finally, proportional CO is computed via Ohm's law. Figure 3 illustrates the technique, which is specifically implemented in four steps as follows.
|
Then, the impulse response function [h(t)] is estimated that, when convolved with x(t), best fits the (unfiltered) ABP waveform [y(t)] in the least squares sense. [Note that, whereas 6-min intervals of x(t) and y(t) are utilized to estimate h(t), only a few seconds of x(t) and y(t) are shown in Fig. 3 for the purpose of clarity.] By definition, the estimated h(t) represents the (scaled) ABP response to a single, solitary cardiac contraction. The impulse response function is specifically estimated according to the following autoregressive exogenous input equation:
 | (1) |
k}, h(t) is computed as follows:
 | (2) |
(t) is the unit impulse function.
Next, is determined over the interval of h(t) ranging from 2 to 4 s after the time of its maximum value on the basis of the following exponential equation:
 | (3) |
are estimated through the least squares minimization of the unmeasured residual error w(t). This optimization problem is solved in closed form by log transformation of h(t) (21). In theory, accurate determination of the windkessel time constant is achieved by virtue of h(t) coupling the long time scale or beat-to-beat variations in x(t) to y(t) (see Fig. 4). That is, the technique is not just trivially extrapolating the ABP waveform at the end of diastole.
|
. Note that the above mathematical steps can easily be implemented in near real time with only a delay on order of a few seconds with a standard home personal computer. Invasive Human Hemodynamic Data Set
The hemodynamic data utilized to evaluate the mathematical analysis technique with respect to human invasive peripheral ABP waveforms were obtained from the MIMIC (Multi-parameter Intelligent Monitoring for Intensive Care) database, which is described in detail elsewhere and freely available on the web (23–25). Briefly, this database includes 72 ICU patient records, typically ranging from 24 to 48 h in duration, that were archived from patient monitors in the medical, surgical, and cardiac intensive care units of the hospital formerly known as the Beth Israel Hospital, Boston, MA. Each of these records consists of continuous waveforms sampled at 125 Hz, such as invasive peripheral ABP via radial artery catheterization and surface ECG leads, as well as 1-min trends, such as thermodilution CO, mean ABP (MAP), and heart rate (HR). Sixteen of the 72 patient records were applicable to the present evaluation study, as they included 1) radial ABP waveforms, and 2) more than one reference thermodilution CO measurement. Within each of these records, CO was changing because of disease progression and therapy.
On the basis of these 16 MIMIC patient records, we created a data set for technique evaluation as follows. First, we downloaded from these records all of the distinct, 1-min thermodilution CO measurements and 6-min contiguous segments of the corresponding radial ABP waveforms (from 2.5 min preceding the 1-min CO measurements to 2.5 min after these measurements). Then, we visually examined each of the radial ABP waveforms and extracted the longest contiguous, artifact-free segment from each of these waveforms (see DISCUSSION). Finally, we excluded from the study all radial ABP waveforms that were less than 5 min in duration, had a significant linear trend (
20 mmHg change), or represented the only waveform segment within a patient record (because the technique estimates changes in CO). Note that radial ABP waveforms with significant linear trends were removed from the study, because the corresponding thermodilution measurements, which strictly represent average CO over intervals of typically less than 1 min, are unlikely to provide an adequate reference CO to the entire 5- to 6-min period of unsteady ABP. A total of 101 pairs of simultaneous measurements of artifact-free, invasive radial ABP waveforms and reference thermodilution CO values from 15 ICU patients [10 men and 5 women; age: 67 ± 12 yr (mean ± SD)] remained for technique evaluation. Table 1 summarizes the clinical class and hemodynamic data for each of these patients.
|
The hemodynamic data utilized to evaluate the mathematical analysis technique with respect to human noninvasive peripheral ABP waveforms were obtained from previous experiments designed to address different specific aims and are described in detail elsewhere (10, 27). Here, we briefly describe those aspects of the experiments that are relevant to the present study.
Ten healthy human volunteers [5 men and 5 women, age: 25 ± 4 yr (mean ± SD)] participated in the experiments. Each subject was instrumented for noninvasive measurement of a peripheral ABP waveform, instantaneous CO, and other cardiorespiratory signals. The peripheral ABP waveform was measured with a commercial finger-cuff photoplethysmography system (2300 Finapres continuous blood pressure monitor, Ohmeda; Englewood, CO), and instantaneous CO was measured according to a previously described Doppler ultrasound technique (11) implemented by an expert. Specifically, aortic blood velocity was measured with a bidirectional ultrasound Doppler velocimeter (CFM 750, GE Vingmed; Horten, Norway), which was operated in pulsed mode at 2 MHz with the hand-held transducer placed on the suprasternal notch. The area of the rigid aortic ring was determined in a separate session by parasternal sector-scanner imaging (CFM-750, GE Vingmed). Instantaneous CO was then calculated via the product of the measured instantaneous maximum blood velocity and the area of the aortic valve orifice (see DISCUSSION).
Each instrumented subject was studied on 2 separate days, before and after the administration of atropine (0.04 mg/kg) and/or propranolol (14.6 mg) during 4° head-down tilt, supine, and 30° upright tilt postures, for a total of 10 experimental conditions. Specifically, on 1 day, 6-min intervals of the noninvasive measurements were continuously recorded while the subject was at rest at a sampling frequency of 50 Hz for each of the following experimental conditions: 1) 4° head-down tilt, control (no medications); 2) supine, control; 3) supine, propranolol; 4) supine, double blockade (propranolol+atropine); and 5) 4° head-down tilt, double blockade. On the other day, the noninvasive measurements were likewise collected for each of the following experimental conditions: 1) supine, control; 2) 30° upright tilt, control; 3) 30° upright tilt, atropine; 4) 30° upright tilt, double blockade; and 5) supine, double blockade. At least 5 min were allowed for hemodynamic equilibration between each experimental condition.
On the basis of these noninvasive recordings, we created a data set for technique evaluation as follows. First, we visually examined each noninvasive finger ABP waveform and instantaneous CO waveform and extracted the longest contiguous, artifact-free segment from each waveform. Then we excluded from the study all instantaneous CO waveforms that were less than 1 min in duration and all finger ABP waveforms that were less than 5 min in duration or had unreasonably high-pressure values (see DISCUSSION). Next, the reference CO value corresponding to each of the remaining instantaneous CO waveforms was determined by computing its time average. Then, because the hemodynamic data collected from the supine, control and supine, double blockade conditions hardly differed from 1 experimental day to the next, these data were combined (as the technique estimates changes in CO) by averaging the reference CO values and the (subsequently determined) corresponding proportional CO estimates over the 2 experimental days. Finally, because the 4° head-down tilt posture did not induce a significant hemodynamic change with respect to the supine posture, these data were discarded from the study. (Note that, if the 4° head-down tilt data were instead combined with the corresponding supine data, the overall results reported below would differ by <1%.) A total of 57 pairs of simultaneous measurements of artifact-free, noninvasive finger ABP waveforms and reference Doppler ultrasound CO values from 10 healthy subjects remained for technique evaluation. Table 2 summarizes the hemodynamic data for each of the subjects.
|
After applying the mathematical analysis technique to all of the invasive and noninvasive peripheral ABP waveforms in the two human data sets, we quantitatively compared the resulting, proportional CO estimates with their reference, absolute CO values in each data set as follows. First, we scaled the proportional CO estimates to have the same mean value as the corresponding reference CO within each patient or subject. Then, we pooled the data together from all the patients or subjects in each data set and performed Bland-Altman analysis to comprehensively visualize the CO error (difference between calibrated CO estimate and reference CO value normalized by the reference CO value and given in percent), including its bias (µ) and precision (
) (4). We also computed the CO root-mean-squared normalized error (RMSNE =
) as a scalar metric indicating the size of the overall CO error. [Note that, because the bias component will be small (but not exactly zero) here because of the scaling, the CO RMSNEs reported below are mainly due to its precision component.] These metrics indicate the ability of the technique to measure changes in CO relative to its mean value within an individual. We also computed the correlation coefficients () between the pooled CO error. The corresponding values of CO, MAP, TPR, HR, and CO change with respect to its mean value within each patient or subject (
CO), and the magnitude of these CO changes (|CO|) in each data set to determine the extent to which the hemodynamic conditions affected the performance of the technique.
RESULTS
Figure 4 illustrates a representative example of a measured (actual) and fitted (predicted) human peripheral ABP waveform by the mathematical analysis technique. Note that the figure includes two predicted ABP waveforms. Each sample of the ABP waveform indicated with the dotted line is predicted from the past samples of the actual y(t) and the input x(t) in Fig. 3 (i.e., one-step ABP prediction), whereas each sample of the ABP waveform indicated with the dashed line is predicted from only the input x(t) (i.e., full ABP prediction) (21). Because one-step ABP prediction is what is optimized by the technique (to obtain a closed-form solution; see Eq. 1) and is an easier task than full ABP prediction, the one-step prediction ABP waveform is more accurate and, in fact, virtually superimposable on the actual ABP waveform indicated with the solid line. Nevertheless, the correspondence between the actual and full prediction ABP waveforms is quite good, especially on a beat-to-beat basis. As discussed above, because both predicted ABP waveforms correspond closely to the actual ABP waveform over time scales greater than a cardiac cycle, the subsequent estimation of the windkessel time constant and proportional CO via Ohm's law (see Fig. 3) should, in principle, be accurate.
Figures 5 and 6, respectively, illustrate the overall results of evaluating the technique with respect to the invasive and noninvasive hemodynamic data sets in terms of Bland-Altman plots of the CO error. (Note that the x-axis of these plots is reference CO rather than the average of reference CO and calibrated, estimated CO.) Tables 1 and 2, respectively, summarize the CO RMSNE of the technique with respect to each patient in the invasive hemodynamic data set and each subject in the noninvasive hemodynamic data set. Indeed, these tables and figures indicate that the technique as applied to invasive radial ABP waveforms was in strong agreement with thermodilution measurements in 15 ICU patients with an overall CO RMSNE of 15.3%, whereas the technique as applied to noninvasive finger ABP waveforms was in equally strong agreement with Doppler ultrasound measurements made by an expert in 10 healthy subjects with an overall CO RMSNE of 15.1%. Moreover, the CO error resulting from the invasive hemodynamic data set was essentially uncorrelated with CO ( = –0.14), HR ( = 0.04), CO ( = –0.26), and |CO| ( = 0.03) and only mildly correlated with MAP ( = 0.41) and TPR ( = 0.38), whereas the CO error resulting from the noninvasive hemodynamic data set was essentially uncorrelated with CO ( = –0.17), TPR ( = 0.10), HR ( = 0.19), and CO ( = 0.20) and only mildly correlated with MAP ( = 0.37) and |CO| ( = 0.38). Finally, in the noninvasive hemodynamic data set in which the interventions were known (see DISCUSSION), the CO RMSNE for each intervention (atropine, propranolol, and/or a 30° upright shift in posture) ranged from 8.1 to 20.8%, with the higher errors obtained during the double blockade conditions.
|
|
In summary, we have recently introduced a new technique for continuous [i.e., automatic and without the need for an operator) monitoring of relative changes in CO by long time interval analysis of a peripheral ABP waveform (see Ref. 26 and Figs. 1–4 herein). We have previously demonstrated the validity of the technique with respect to intra-arterial femoral and radial ABP waveforms obtained from open-chest swine instrumented with aortic flow probes over a wide physiological range (26). Here, we present an evaluation of the technique in humans based on previously published invasive and noninvasive hemodynamic data sets (10, 24, 25, 27). Although the evaluation described herein was retrospective, it is noteworthy that neither of these data sets was designed for the evaluation of our technique or any other pulse contour analysis technique. With the former data set, we compared the application of the technique to invasive radial ABP waveforms with reference thermodilution measurements in 15 ICU patients in which CO was changing because of disease progression and therapy. With the latter data set, we compared the application of the technique to noninvasive finger ABP waveforms with reference Doppler ultrasound measurements made by an expert in 10 healthy subjects during pharmacological and postural interventions. We report an overall CO estimation error of about 15% with respect to each of these human data sets (see Tables 1 and 2 and Figs. 5 and 6).
Potential Sources of CO Error
The CO errors reported here could be partly explained by inadequacies in the quality and accuracy of the hemodynamic measurements within the studied invasive and noninvasive data sets. However, as described below, we excluded from the study all data segments of poor quality (e.g., corrupted by significant noise artifact) so as to benchmark technique performance. On the other hand, nothing could be done in this retrospective study to improve on the accuracy of the measurements, which is largely intrinsic to the employed transducers.
The radial ABP waveforms in the invasive data set were measured with generally accurate, intra-arterial catheters. However, in the MIMIC database on which this data set was based, ABP waveform artifact was sometimes present. The artifact may have been due, for example, to patient movement, arterial line flushing, catheter obstruction, loss of signal, and proximal ABP cuff inflation. We excluded from the study all radial ABP waveforms that were significantly corrupted by such artifact (11% of the available, simultaneous pairs of radial ABP waveforms and thermodilution measurements). Although we identified ABP artifact here by visual means, it may be possible to automatically and reliably detect ABP artifact in real time [e.g., with a simultaneous surface ECG measurement based on an algorithm recently introduced by Zong et al. (39)] so as to warn the clinician that the CO estimate derived from the ABP analysis may not be valid or preclude the output of such a CO estimate. Note that, for reasons described above, we also excluded from the study all radial ABP waveforms with linear trends of 20 mmHg (13% of the available, simultaneous pairs of radial ABP waveforms and thermodilution measurements).
In contrast to the radial ABP waveforms in the invasive data set, it was not possible to assess the quality of the corresponding thermodilution measurements. However, the error in clinical thermodilution measurements is known to be in the 15–20% range (18, 36). Assuming such an error, Critchley and Critchley (8) argued that a new CO measurement method should be accepted as an alternative to thermodilution, provided that their limits of agreements are within ±30%. We note that the level of agreement between our technique and thermodilution measurements from 15 ICU patients is essentially in this range (see Table 1 and Fig. 5).
The noninvasive hemodynamic data utilized in this study were obtained from controlled experiments, and none of the noninvasive finger ABP waveforms were excluded from the study because of noise artifact. However, two noninvasive finger ABP waveforms were excluded from the study because of unreasonably high pressure levels. This artifact may have been due to an upward drift in finger ABP caused by prolonged application of cuff pressure (34). Note that this finger-cuff photoplethysmography artifact may be attenuated or eliminated by intermittent finger exercise (34) or with the currently available Portapres system, which alternates the application of cuff pressure between different fingers.
Although artifact in the noninvasive finger ABP waveforms was not a significant factor here, the intrinsic level of accuracy of the employed Finapres system may have affected our results. In a review of Finapres technology, Imholz et al. (16) gathered 43 previous studies comparing the Finapres system with intra-arterial or noninvasive, discrete ABP measurements and reported that the systolic and mean pressure levels of the Finapres system were not within the limits of accuracy suggested by the American Association for the Advancement of Medical Instruments. On the other hand, Omboni et al. (29) showed that the Finapres system and radial artery catheterization produce similar beat-to-beat (mean) ABP fluctuations. That is, although the Finapres system distorts the peripheral ABP waveform over short time scales, it may have little effect on the waveform over longer time scales. Thus, because the technique analyzes the ABP waveform over long time intervals, the inherent inaccuracies of the Finapres system may not have been a major contributor to the CO error reported herein. However, we cannot confirm this possibility, because simultaneous intra-arterial ABP recordings were not available.
The reference Doppler ultrasound measurements in the noninvasive data set were made by an expert, and only four of these CO measurements were excluded from the study because of poor signal quality. Despite the generally excellent signal quality, the assumptions underlying the Doppler ultrasound technique could have been partially violated, thereby affecting our results. These assumptions include the following: 1) the velocity profile in the aortic valve orifice is rectangular and this velocity is conserved as the central maximum velocity of a jet 3–4 cm downstream; 2) the insonication angle is 20°; and 3) the aortic valve is circular (11). However, assumptions 2 and 3 are likely to have little effect on the accuracy of the measured relative changes in CO, whereas the same argument has been made for assumption 1 (37). Because our technique is designed to measure relative changes in CO and we evaluated it here as such (see Statistical Analysis), we believe that the Doppler ultrasound measurements serve as a reasonably accurate reference here (i.e., at least as accurate as standard thermodilution measurements).
Other potential sources of the CO errors reported here are any violations to the assumptions on which the technique is based. These assumptions include the following: 1) AC is constant within each individual; 2) peripheral venous pressure is negligible with respect to ABP; 3) ABP exceeds the critical closing pressure; and 4) the time constant governing arterial viscoelastic effects is negligible with respect to the windkessel time constant (26). The first of these assumptions is perhaps the most controversial, because there is currently no generally accepted, gold-standard method for measuring in vivo AC. Nevertheless, we believe that in vivo AC must be nearly constant over a wide hemodynamic range in at least some animals on the basis of the success of the pulse contour analysis of Bourgeois et al. (5, 6) and Osborn et al. (31) with respect to canine central ABP waveforms and the present mathematical analysis technique with respect to swine peripheral ABP waveforms (26). We are unaware of any existing in vivo data likewise demonstrating constancy of the human AC. However, Hallock and Benson (15) did show that, although the compliance of excised human aortas of various ages at autopsy (ranging from young adults to the elderly) decreased with increasing pressure, in vitro aortic compliance could be approximated as constant over a wide pressure range. If in vivo AC sharply changed in the opposite direction of MAP within each individual record of our study, then our technique would have grossly overestimated CO at high MAP levels and underestimated CO at low MAP levels (i.e., a strong, positive correlation between CO error and MAP). Although the correlation between CO error and MAP is positive, the degree of correlation is mild (see above), suggesting that in vivo AC within each of the 15 ICU patient and 10 healthy young adult records may have been approximately constant.
Comparison to Intrabeat Pulse Contour Analysis Techniques
In a previous paper introducing our mathematical analysis technique (26), we used signal-to-noise theory to argue that estimating the average windkessel time constant (and thus average, proportional CO via Ohm's law) by analyzing a peripheral ABP waveform over time intervals greater than a cardiac cycle should be more accurate than analyzing the ABP waveform over individual cardiac cycles and then averaging the beat-to-beat results. To support this theoretical argument, we fitted complex exponentials function(s) to individual diastolic decay intervals of swine peripheral ABP waveforms to estimate on a beat-to-beat basis, averaged the resulting individual estimates, and then computed average, proportional CO via Ohm's law. The best result we were able to achieve with this intrabeat analysis was an overall CO RMSNE that was 52% larger than that obtained by our technique. We repeated this intrabeat analysis with respect to the human invasive and noninvasive hemodynamic data sets here and obtained overall CO RMSNEs that were, respectively, 23 and 81% higher than those obtained by our technique. (Note that one possible reason that this intrabeat analysis is much less effective with respect to the noninvasive hemodynamic data set is that, as described above, the noninvasive ABP waveforms may suffer from high-frequency distortion due to the employed Finapres system). We believe that these comparative studies confirm the theory that important information is indeed present in beat-to-beat ABP variations and that analysis of these subtle variations leads to improved average, proportional CO estimation in practice. However, we note that future studies should also be conducted to compare our technique with the recent intrabeat techniques of Wesseling et al. (38) and Linton and Linton (20), which also require a single peripheral ABP waveform for analysis.
Limitations of the Long Time Interval Analysis Technique
Two limitations of the current form of the long time interval analysis technique are: 1) beat-to-beat CO monitoring is not feasible and 2) artifact is a more significant problem (compared with beat-to-beat pulse contour analysis techniques such as the aforementioned). With respect to the former limitation, we feel that attempts to improve the accuracy of average, proportional CO estimation, even at the cost of temporal resolution, are worthwhile from a clinical point of view. For example, although many previous pulse contour analysis techniques can offer beat-to-beat proportional CO monitoring, they have still not been widely adopted in clinical practice, presumably because of accuracy concerns. Moreover, automatic estimation of proportional CO at intervals on the order of seconds but representing the last 6 min (i.e., boxcar moving average) would represent a significant improvement on discrete, operator-required determinations of CO by the clinical thermodilution method (assuming similar accuracy). With respect to the second limitation, the requirement of 6-min intervals of relatively artifact-free ABP waveforms does not substantially limit the practical applicability of the technique. For example, only 11% of the invasive radial ABP waveforms from the real-world MIMIC database were discarded in our study because of artifact. Moreover, the 6-min intervals of analysis may be reduced to smaller intervals (e.g., 1 min) without materially affecting the accuracy of the estimates (e.g., CO RMSNE of 16.1% in the invasive hemodynamic data set and 15.7% in the noninvasive hemodynamic data set). Future formal studies are needed to determine the minimum interval for analysis that does not significantly compromise the accuracy of the technique.
Limitations of the Human Evaluation Study
In the invasive hemodynamic data set, CO was naturally changing within each ICU patient record because of disease progression and therapy. Typical ICU therapy is known to include medications such as dobutamine, dopamine, intravenous fluids, and nitroprusside (i.e., both cardiac and vascular interventions) (22). However, because time-stamped annotations were not available here, we were not able to evaluate the technique in the ICU patients with respect to each of these common therapeutic interventions. In contrast, in the noninvasive hemodynamic data set, CO was changing in each healthy subject because of precisely known interventions of atropine, propranolol, and/or a 30° upright shift in posture. As described above, the CO RMSNEs were largest during the double blockade conditions, presumably because beat-to-beat HR variability was totally abolished. Although vascular changes (TPR and fluid shifts) occurred reflexively on administration of atropine and propranolol as well as via the postural shift (see Table 2), we were not able to test the technique with respect to noninvasive ABP waveforms during interventions that directly act on the vasculature (e.g., phenylephrine, nitroprusside). Finally, because the reference thermodilution CO in the invasive hemodynamic data set could not be assumed to be valid during unsteady conditions (e.g., ABP waveform segments with significant trends) and the noninvasive ABP waveforms were only recorded during steady conditions (Invasive Human Hemodynamic Data Set), we were not able to evaluate the technique in humans during unsteady conditions (i.e., rapid changes in CO). However, we have previously shown that the technique performs quite accurately during unsteady conditions in swine instrumented with aortic flow probes measuring instantaneous flow (26). Moreover, in the present human study, we were at least able to show that the technique performed approximately the same regardless of the size or direction of the CO change, as the correlations between the CO error and CO and |CO| were only mild (Noninvasive Human Hemodynamic Data Set).
Potential Applications of the Mathematical Analysis Technique
Our technique mathematically analyzes a single peripheral ABP waveform over long time intervals to continuously (i.e., automatically and without the need for an operator) measure CO to within a constant scale factor. The technique may therefore be utilized to quantitatively monitor relative changes in CO. The proportional CO may be calibrated, if desired, with a single, absolute CO measurement (e.g., thermodilution). For normal individuals, it may be possible to determine the proportionality constant from a nomogram. However, we believe determination of the proportionality constant is unnecessary in the context of continuous monitoring in the acute setting in which only CO changes are clinically relevant.
The results of this retrospective human evaluation study indicate that the technique may be sufficiently accurate in terms of estimating relative changes in CO with respect to invasive radial ABP waveforms from critically ill patients and noninvasive finger ABP waveforms from healthy subjects. With further mathematical analysis development (including the incorporation of an automated artifact detector) and successful prospective testing, the technique may potentially be applied to continuously monitor CO in the acute setting. The most prominent such application is in critically ill patients in the ICU and operating and recovery rooms. In critically ill patients instrumented with both pulmonary and radial artery catheters, the technique could be calibrated with a single thermodilution measurement to permit subsequent continuous monitoring of absolute CO. In the numerous critically ill patients with only radial artery catheters installed (see Introduction), the technique could provide continuous, quantitative monitoring of relative changes in CO. Other such applications in which noninvasive peripheral ABP transducers would be most appropriate include patients in the emergency room and the hospital ward, trauma patients in transport, as well as soldiers in combat. The human evaluation study described herein represents an initial step toward the realization of such applications.
GRANTS
This work was supported by an award from the American Heart Association and Michigan State University.
ACKNOWLEDGMENTS
The authors thank Prof. Richard J. Cohen for providing the noninvasive human hemodynamic data set utilized in this study.
FOOTNOTES
Address for reprint requests and other correspondence: R. Mukkamala, Dept. of Electrical and Computer Engineering, Michigan State Univ., 2120 Engineering Bldg., East Lansing, MI 48824 (e-mail: rama{at}egr.msu.edu)
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  R. Mukkamala and A. T. Reisner Reply to van Lieshout and Jansen J Appl Physiol, February 1, 2007; 102(2): 827 - 827. [Full Text] [PDF]
|
|
|
|
|
|
J. J. van Lieshout and J. R. C. Jansen Continuous cardiac output monitoring by blood pressure analysis J Appl Physiol, February 1, 2007; 102(2): 826 - 826. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
(t)] with 
