Professional Documents
Culture Documents
Bondgraph Published CardioVascular 5
Bondgraph Published CardioVascular 5
The role of the autonomic nervous system in hypertension: a bond graph model study
This content has been downloaded from IOPscience. Please scroll down to see the full text.
(http://iopscience.iop.org/0967-3334/29/4/005)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 61.129.42.15
This content was downloaded on 10/05/2015 at 13:03
E-mail: gmning@zju.edu.cn
Abstract
A bond graph model of the cardiovascular system with embedded autonomic
nervous regulation was developed for a better understanding of the role of the
autonomic nervous system (ANS) in hypertension. The model is described
by a pump model of the heart and a detailed representation of the head and
neck, pulmonary, coronary, abdomen and extremity circulation. It responds to
sympathetic and parasympathetic activities by modifying systemic peripheral
vascular resistance, heart rate, ventricular end-systolic elastance and venous
unstressed volumes. The impairment of ANS is represented by an elevation
of the baroreflex set point. The simulation results show that, compared
with normotensive, in hypertension the systolic and diastolic blood pressure
(SBP/DBP) rose from 112/77 mmHg to 144/94 mmHg and the left ventricular
wall thickness (LVWT) increased from 10 mm to 12.74 mm. In the case
that ANS regulation was absent, both the SBP and DBP further increased by
8 mmHg and the LVWT increased to 13.22 mm. The results also demonstrate
that when ANS regulation is not severely damaged, e.g. the baroreflex set point
is 97 mmHg, it still has an effect in preventing the rapid rise of blood pressure
in hypertension; however, with the worsening of ANS regulation, its protective
role weakens. The results agree with human physiological and pathological
features in hemodynamic parameters and carotid baroreflex function curves,
and indicate the role of ANS in blood pressure regulation and heart protection.
In conclusion, the present model may provide a valid approach to study the
pathophysiological conditions of the cardiovascular system and the mechanism
of ANS regulation.
1. Introduction
2. Methods
In this section, the blood circulation system and autonomic nervous regulation submodels are
introduced in detail and the methods of calculating oxygen consumption per heart beat and
left ventricular wall thickness are presented.
ith segment
Qi-1 Qi
(i-1)th segment (i+1)th segment
Pi Pi+1
compliance and unstressed volume, i.e. the contractile activity of the atria is neglected (Ursino
1998), while the left and right ventricles are simulated by an improved pressure–volume
relationship model (Guarini et al 1998, Hunter 1989, Shroff et al 1989, Suga and Sagawa
1974, Ursino 1998). More details about the heart model are given in appendix A.
The vascular system is described by a detailed tree structure accounting for the head,
neck, pulmonary, coronary, abdomen and extremity circulation. Each circulation structure is
further divided into several vessel segments representing arteries, arterioles, capillaries and
veins, thus in total 30 vessel segments (see appendix C.1 for the notation of the segments) are
constructed in the vascular model. Each segment is regarded as a modified Windkessel model,
which is composed of capacitance (C), resistance (R) and inertial (I) elements.
To represent the changes in peripheral vascular resistance (PVR) and capacitance of large
arteries, two gain factors Kr and Kc are introduced. For the ith segment of peripheral vessels,
its resistance is described as
Ri = Kr Ri0 , (1)
where Ri0 is the initial resistance, Kr is an index (Kr 1) representing the change in PVR,
and it is set to be 1.0 for normal state.
Similarly, for the ith segment of large arteries, its capacitance is described as
Ci = Kc Ci0 , (2)
where Ci0 is the initial capacitance, Kc is an index (0 < Kc 1) representing the change in
vessel compliance, and it is set to be 1.0 for normal state.
Using the bond graph technique, a vessel segment is expressed in figure 1. The coupling
of vessel segments is accomplished by the pressure–volume relationship of adjacent segments.
For the ith segment, the flow is provided by the (i − 1)th segment, which is denoted as Qi-1,
and the pressure is provided by the (i + 1)th segment, which is denoted as Pi+1. Meanwhile,
the output of the ith segment is the flow to the (i + 1)th segment and the pressure for the
(i − 1)th segment, which are denoted as Qi and Pi, respectively. Concretely, the mathematical
equations of the ith segment are deduced by
dVi
= Qi−1 − Qi (3)
dt
dλi
= Pi − Pi+1 − Ri × Qi , (4)
dt
476 S Chen et al
where
Qi = λi /Ii (5)
Pi = (Vi − Vi0 )/Ci , (6)
dλi
Vi is the volume on Ci, λi is the pressure momentum of Ii, dt
denotes the pressure difference
on Ii, and Vi0 is the unstressed volume of the ith segment.
Integration of the heart and vascular system model. A complete bond graph model for the
blood circulation system is illustrated in figure 2. For the integration of the heart and vascular
system, the cardiac cycle is divided into three phases, i.e. the ventricular ejection period,
isovolumic contraction–relaxation period and the filling period. In each phase the coupling
of the heart and vascular system is performed by their inner pressure–flow relationship.
Equations (7) are the mathematical expressions for the ventricular ejection period. In this
period, the flows through the left ventricle and aortic valve are assigned to be equal, since the
aortic valve is opened, and the flow into the left ventricle through the mitral valve is assigned
to be zero: ⎧
⎪
⎪
dVlat
= Qpvn
⎪
⎪
⎪
⎪ dt
⎪
⎪
⎪
⎪ dλmvv
⎨ =0
dt (7)
⎪
⎪ dVtat
⎪
⎪ = Q − Q − Q − Q − Q − Q
⎪
⎪ dt
avv lsa rsa eca cor aat
⎪
⎪
⎪
⎪ dλavv
⎩ = Plvt − Ptat − Pavv
dt
Likewise, for the isovolumic contraction relaxation and filling period, the corresponding
model expression can also be defined. Consequently, an extendable circulation system model
is constructed of a total of 54 ordinary differential equations (appendix B).
Carotid
Baroreceptor
Afferent
Pathway
Efferent
Pathway
Peripheral
Resistance
Heart
Ventricular Elastance
Effectors
Heart
Rate
Venous
Volumes
2.4. Oxygen consumption per heart beat and left ventricular wall thickness calculation
Calculation of oxygen consumption per heart
beat.
In order to investigate the function of the
heart, oxygen consumption per heart beat VO2 is computed and presented in the simulation
results. According to Suga (Suga et al 1981)
VO2 = A × PVA + B, (8)
where
T
1
PVA = P (t) × Q(t) dt + (LVESV − Vd ) × SBP, (9)
0 2
where T is the cardiac period, LVESV is the end-systolic volume of the left ventricle, SBP
is the systolic blood pressure, Vd is the unstressed volume of the ventricle, A = 1.6 ×
10−5 ml · O2 /(mmHg ml) and B = 0.015 ml · O2 /beat.
Calculation of the left ventricular wall thickness. Sustained pressure overload induced by
hypertension may lead to self-adaptation of the left ventricle, which causes left ventricle
hypertrophy (LVH). Thus, the left ventricular wall thickness (LVWT) under various states was
computed for a better understanding of the effect of ANS on LVH.
According to the literature (Segers et al 2000, 2001), the maximal myocardial tension of
the left ventricle during systolic phase is maintained at normal levels despite chronic pressure
overload such as hypertension. The left ventricle is assumed to be a thick-walled prolate
spheroid (Mirsky 1969); thus the myocardial tension in the ventricle can be represented by
Laplace’s law:
σs = P · r 2 / h · (2r + h), (10)
where σs is the maximal myocardial tension of the left ventricle in the systolic period, P is
the systolic blood pressure, h is the wall thickness ranging from 9 to 12 mm in a normal case,
r is the radius of the left ventricle and is obtained by the hemisphere volume formula:
1
3Vs 3
r= , (11)
2π
where Vs is the volume of the left ventricle when the myocardial tension of the left ventricle
reaches its maximum in the systolic period. It is affected by both systolic elastance and
diastolic stiffness.
Thus, the left ventricle wall thickness h is calculated as
1
1
3Vs 3 P + σs 2
h= −1 . (12)
2π σs
A bond graph model for ANS in hypertension 479
180 600
(a) (b)
160
500
140
Blood pressure of aorta/mmHg
100
300
80
60 200
40
100
20
0 0
175 176 177 178 179 180 181 182 183 175 176 177 178 179 180 181 182 183
t/s t/s
140 180
(c) (d )
130 160
120
140
110
120
100
100
90
80
80
60
70
40
60
50 20
40 0
175 176 177 178 179 180 181 182 183 175 176 177 178 179 180 181 182 183
t/s t/s
Figure 3. The simulation results under normal physiological condition: (a) blood pressure in aorta;
(b) outflow wave of the left ventricle; (c) volume of the left ventricle; (d) blood pressure in the left
arm.
3. Results
Baroreflex function. To validate the appropriateness of the ANS regulation model, two kinds
of simulation addressed in the literature (Ursino 1998) were respectively repeated.
(i) Carotid baroreflex in open-loop nonpulsatile conditions. To analyze the function of
carotid baroreflex, the close-loop baroreflex regulation was removed, i.e. the input of the
480 S Chen et al
160
140
Arterial pressure/mmHg
120
100
80
60
40
40 60 80 100 120 140 160
Intrasinus pressure/mmHg
Figure 4. Baroreflex function curve under the open-loop condition with static loading of carotid
sinuses. The abscissa is the intrasinus pressure and the ordinate is the change in the systemic
arterial pressure.
SBP: systolic blood pressure, DBP: diastolic blood pressure, CO: cardiac output, VO2 : oxygen consumption per heart
beat, EF: eject fraction, LVESV: end-systolic volume of left ventricle, LVEDV: end-diastolic volume of left ventricle.
afferent pathway was not obtained from the carotid artery pressure, but was given by
adjusting the pressure from 50 mmHg to 150 mmHg with a step of 10 mmHg. The mean
systemic arterial pressure (MAP) was computed and the function curve (MAP versus
mean intrasinus pressure) is illustrated in figure 4. As shown in the figure, MAP changes
from 146 mmHg to 55 mmHg with the alteration of the mean intrasinus pressure.
(ii) Effect of intrasinus pressure pulsatility. The effect of intrasinus pressure pulsatility on
the carotid baroreflex control was investigated. The mean intrasinus pressure varies from
50 mmHg to 150 mmHg with a step of 10 mmHg. And a 2 Hz pulsatile fluctuation
with varying peak-to-peak amplitudes (0 mmHg, 20 mmHg, 30 mmHg and 40 mmHg,
respectively) is superposed on the mean intrasinus pressure. Figure 5 demonstrates
that the pulsating pressure at the carotid sinuses decreases the sensitivity of the carotid
baroreflex.
160
0mmHg
20mmHg
30mmHg
140
40mmHg
Arterial pressure/mmHg
120
100
80
60
40
40 60 80 100 120 140 160
Intrasinus pressure/mmHg
Table 2. Blood pressures under different set points under hypertensive condition.
Resetting of the baroreflex set point in hypertension. Simulation results under different
carotid baroreflex set points are given in table 2. As indicated in the results, the blood pressure
increased as the baroreflex set point rose. Referring to the classification of hypertension
(JNC7 2003), it developed from prehypertension (set points are 87 and 97 mmHg) to stage 1
hypertension (set points are 102, 107, 112 and 117 mmHg).
The following simulation results were under stage 1 hypertension condition in which the
baroreflex set point was reset to 107 mmHg (SBP/DBP were assumed as 140/90 mmHg).
ANS regulation on hemodynamic parameters. Figure 6 illustrates the aorta blood pressure
and the left ventricle pressure–volume loop for normal and hypertensive states. It is shown that
under hypertension the blood pressure is higher than that in normal (figure 6(a)) and the left
ventricle pressure–volume loop shifts toward right, i.e. both LVESV and LVEDV are increased
(figure 6(b)). Additionally, it is observed that if the ANS function is damaged completely,
SBP/DBP under hypertension increase more and the left ventricle pressure–volume loop shifts
482 S Chen et al
170
(a) normal with baroreflex 200 (b)
160 normal with baroreflex
hypertension with baroreflex
180 hypertension with baroreflex
hypertension absent of baroreflex
150 hypertension absent of baroreflex
160
140
140
130
Pressure/mmHg
Pressure/mmHg
120
120
100
110
80
100
90 60
80 40
70 20
60 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 20 40 60 80 100 120 140 160 180 200
t/s Volume/mL
50
hypertension with
intact regulation
40 of all effectors
no SPVR
30
Percent of changes(%)
20 no elastance
10
no venous volume
VO2
0
no HR
-10
SBP DBP CO EF VO2 LVESV LVEDV
-20
Figure 7. Sensitivity analysis of SPVR, elastance, venous volume and HR. The abscissa indicates
parameters and the ordinate is their percentage change.
further right. As shown in table 1, when hypertension occurs, the blood pressure rises from
112/77 mmHg to 144/94 mmHg. If autonomic nervous regulation is absent in hypertension,
SBP and DBP both increase further by 8 mmHg, while EF decreases by 0.19% and VO2
increases by 20 × 10−3 ml · O2 /beat.
SPVR, ventricular end-systolic elastance, venous unstressed volumes and HR are four
different categories of effectors in ANS regulation. To clarify their roles in hemodynamic
regulation, a sensitivity analysis was carried out in the hypertension simulation. This was
done by isolating SPVR, ventricular end-systolic elastance, venous unstressed volumes and
HR one by one from the efferent pathway, and thus a total of four tests were processed.
The corresponding changes of the main hemodynamic parameters in the tests were obtained
and are illustrated in figure 7. It shows that when SPVR is not modified by the efferent
pathway, the changes in blood pressure are the largest (SBP: 33.04%, DBP: 29.87%). The
effect of the venous unstressed volume is the second largest; when it is absent the changes in
SBP and DBP are 31.25% and 24.68%, respectively. It is also observed that CO decreases
the least (−0.356%) when the venous unstressed volume is neglected. This suggests that
A bond graph model for ANS in hypertension 483
the venous unstressed volume contributes much to the modulation of the cardiac output and
then indirectly affects the blood pressure. Thus evidently, it indicates that SPVR and venous
unstressed volume play critical roles in blood pressure regulation, while ventricle elastance
and HR contribute less.
ANS regulation on LVH. To elucidate the effect of autonomic nervous system regulation on
LVH in hypertension, LVWT was calculated in the hypertensive state. In the simulation, the
initial ventricle wall thickness is assigned to be 10 mm under normal physiological condition.
The left ventricle end-systolic volume and systolic pressure were obtained by the simulation
and LVWT under hypertension was computed by equation (12).
Table 3 gives the simulation results for LVWT. It reveals that the occurrence of
hypertension induces the thickening of LVWT from 10 mm to 12.74 mm. If autonomic
regulation is absent in hypertension, the LVWT visibly increases to 13.22 mm.
4. Discussion
Various mathematical models have been recently proposed for physiological study (Heldt
et al 2001, Li et al 2002, Segers et al 2000, 2001, Ursino 1998, Ursino and Magosso 2003,
Van Roon et al 2004). Most of them were constructed based on electrical analog theory
and the model equations are usually difficult to deduce. In the present model, the bond
graph technique is applied. It describes different energy domains by unified elements, and
has a single form in deducing model equations. Meanwhile, a number of software tools are
available for bond graph modeling, such as 20-sim (www.20sim.com), CAMP-G (Granda and
Reus 1997) etc. Thus the complexity in model solving is not so huge as it looks like. The
graphical characteristic of the bond graph makes the interpretation of the model easy as its
design approaches the anatomy structure (Le Rolle et al 2005). In this sense, the bond graph
technique is more convenient for understanding and extending the model in contrast with the
electrical analog method.
Another benefit of the current model is the detailed description of the vascular system.
Each part of the vascular system is lumped as a modified Windkessel model and can be further
divided into much more detailed segments to satisfy intensive research purposes. For instance,
the cerebral arteries which are represented by a lumped segment in the present model can be
further depicted by a Willis circle. Such a clear distinction of the vascular network may
improve the understanding of the role of a specific vessel segment under physiological and
pathological conditions.
blood pressure in the left arm were obtained (figure 3). It also shows that the systolic blood
pressure in the left arm is higher than that in the aorta, while the diastolic blood pressure is
lower. These simulation results are in good agreement with human physiological features
(Fung 1996, Guyton and Hall 2000) where the normal ranges of the main parameters are
SBP < 120 mmHg, DBP < 80 mmHg, EF: 55%–65%. Thus, our model provides the
availability of a tool for the study of main cardiovascular characteristics. Second, the baroreflex
function simulation results from the present model are accordant with the results obtained by
Ursino and other observations (Angell James and Daly 1971, Chen and Chang 1991, Guyton
and Hall 2000, Schmidt et al 1972, Ursino 1998). The carotid baroreflex function curve
illustrated in figure 4 demonstrates a sigmoid shape. A mean intrasinus pressure alteration from
50 mmHg to 150 mmHg results in a MAP change from 146 mmHg to 55 mmHg. It agrees
well with the autonomic nervous regulation mechanism in normal physiological state (Chen
and Chang 1991, Guyton and Hall 2000). The curves in figure 5 indicate that the sensitivity
of the carotid baroreflex decreases when the intrasinus pressure shifts from nonpulsatile to
pulsatile conditions. These results are consistent with other simulation studies (Angell James
and Daly 1971, Ursino 1998, Schmidt et al 1972); thus, the ANS regulation model in our work
is appropriate. Furthermore, a small fluctuation on the parameter curves in figure 3 indicates
the tone of baroreflex, which disappears in the models without autonomic regulation. These
facts confirm the validity of the autonomic nervous regulation model.
LVH in hypertensive patients (Milan et al 2007, Schlaich et al 2003). The complex interaction
mechanisms of LVH, arterial system and ANS are not well understood (Guyenet 2006). LVH
is considered to compensate for the increased systolic wall stress induced by the chronic
elevation of blood pressure (Segers et al 2001). It was investigated by a simple model which
comprises a heart model and a 4-element lumped parameter Windkessel model (Segers et al
2000, 2001). However, the role of ANS in LVH was not mentioned.
The present work aimed at a better understanding of the role of ANS in hypertension,
in particular its regulation on both hemodynamics and LVH. The results demonstrate that
autonomic regulation is crucial in the protection of hypertension. It is found that despite
ANS being partially impaired in the hypertensive situation, the status of both hemodynamics
and LVH is less serious than the case without baroreflex regulation. If the ANS regulation
is lacking, the blood pressure increases notably. Consequently, the elevated pressure results
in an increase of the left ventricle wall stress (Segers et al 2000); thus LVWT is increased
to maintain the wall stress at normal levels and correspondingly more oxygen is needed.
Such straightforward adaptive compensation preserves the cardiac output at the cost of
fundamentally impairing the heart.
The effectors of ANS regulation. The sensitivity analysis helps to clarify the effectors of
ANS regulation. The results demonstrate that the most important effector of ANS regulation
on the hypertension is SPVR, followed by venous unstressed volume. From the results,
one can deduce that ANS regulation on the hypertension performs mainly by a remarkable
reduction of SPVR and increase of venous unstressed volume. The reduction of SPVR
decreases the overloading of the heart. This reduction in the load of the heart may prevent
it from being damaged. The increase of venous unstressed volume reduces the end-diastolic
volume and causes the decrease of cardiac output, which leads to the reduction of blood
pressure. Therefore, the present work gives more insight into the protective effect of ANS in
hypertension.
Being a mathematical model of the intricate human system, the present study inevitably
has some limitations. Other regulations such as humoral factors, chemoreflex and pulmonary
stretch reflex are not taken into account in the model. The humoral factor is important in
the regulation of blood pressure. A long-term regulation of the blood pressure is achieved
through the combined actions of a number of hormones, which influence the kidney and the
regulation of extracellular fluid volume. The maintenance of blood volume is achieved by the
renin-angiotensin system, aldosterone and vasopressin (antidiuretic hormone) which act on
the kidney to promote the retention of sodium and water. Also, atrial natriuretic peptide causes
diuresis and natriuresis to reduce the blood volume (Ashton 2007, Schroeder et al 1996). In
addition to direct regulation on blood pressure, some receptor-mediated humoral factors affect
the blood pressure indirectly by modulating the reflex sensitivity of the receptors distributing
on the baroreflex arc (Schlaich et al 2003, 2004). Further, the simplification of the heart model
may be another shortcoming. In a simplified pump model, specific details related to myocardial
contraction, such as the influence of calcium concentration during the contraction process, are
not accounted for. Also, the Windkessel model components R, C, and I are simplified to be
constant during simulation; thus the model is unable to represent some nonlinear characteristics
such as nonlinear pressure–volume relationship in some vessel segments. Besides, neglecting
atrial contraction may impact some of the hemodynamic parameters. Early studies show
that atrial contraction results in about 10% increase in aortic pressure and left ventricular
systolic pressure; it also contributes to stroke volume (Korakianitis and Shi 2006). Lastly, the
whole model is a zero-dimensional lumped-parameter model without geometrical information.
Lumped-parameter methods can be used to represent a single-vessel segment or a collection
486 S Chen et al
of vessels and have been applied to model blood flow in large portions of the human arterial
system. However, since the geometry is not explicitly defined, the model is not able to simulate
wave propagation and reflection effects. Based upon the considerations above, more effort
should be taken in future work to improve these aspects.
5. Conclusion
Acknowledgments
This work is supported by the National Nature Science Foundation of China (Grant 30570483),
the Science & Technology Department of Zhejiang Province, China (Grant 2006C13018). The
authors would like to thank Professor Mauro Ursino (University of Bologna, Italy) for the
helpful discussions and suggestions.
Appendix A
Afferent pathway. It is characterized byfcs , the frequency of spikes in the afferent fibers:
P̃ − Pn P̃ − Pn
fcs = fmin + fmax exp 1 + exp . (A.5)
ka ka
Efferent vagal pathway. It is characterized by fev , the frequency of spikes in the efferent
vagal fibers:
fcs − fcs,0 fcs − fcs,0
fev = fev,0 + fev,∞ exp 1 + exp . (A.7)
kev kev
T is regulated by
GT ,s · ln[fes (t − DT ,s ) − fes,min + 1] if fes fes,min
σT ,s (t) = (A.11)
0 if fes < fes,min
dTs (t) 1
= · (−Ts (t) + σT ,s (t)) (A.12)
dt τT ,s
σT ,v (t) = GT ,v · fev (t − DT ,v ) (A.13)
dTv (t) 1
= · (−Tv (t) + σT ,v (t)) (A.14)
dt τT ,v
T = Ts + Tv + T0 . (A.15)
488 S Chen et al
Appendix B
where PLV and PRV are the left and right ventricular pressures which are calculated by the
heart model, Ctvc+rat is the sum of the capacitances of the thoracic vena cava and left atrium.
Appendix C
Internal carotid artery and arteries in brain ici 31.23 0.15 15.51 0 a
491
492
(Continued.)
Circulation Vascular segment Notation R (mmHg s mL−1 × 10−3) C (mmHg−1 mL) I (mmHg s2 mL−1× 10−3) V0 (ml) Source
Extremity Left subclavian artery lsa 12.45 0.14 7.76 0
Arteries in the left arm laa 31.23 0.11 15.51 0
Left radial artery lra 37.00 0.08 20.00 0 a,b
S Chen et al
A bond graph model for ANS in hypertension 493
Notation R (mmHg s mL−1 × 10−3) C (mmHg−1 mL) I (mmHg s2 mL−1 × 10−3) Source
References
Angell James J E and Daly M de B 1971 Effects of graded pulsatile pressure on the reflex vasomotor responses elicited
by changes of mean pressure in the perfused carotid sinus-aortic arch regions of the dog J. Physiol. (Lond.)
214 51–64
Ashton N 2007 Neurological and humoral control of blood pressure Anaesth. Intensive Care Med. 8 221–6
Chen H I and Chang K C 1991 Assessment of threshold and saturation pressure in the baroreflex function curve: a
new mathematical analysis Japan. J. Physiol. 41 861–77
Diaz-insua M and Delgado M 1996 Modeling and simulation of the human cardiovascular system with bond graph:
a basic development IEEE Proc. 23rd Annual Meeting on Computers in Cardiology pp 393–6
Freeman R 2006 Assessment of cardiovascular autonomic function Clin. Neurophysiol. 117 716–30
494 S Chen et al
Ursino M and Magosso E 2003 Role of short-term cardiovascular regulation in heart period variability: a modeling
study Am. J. Physiol. Heart Circ. Physiol. 284 1479–93
US Department of Health and Human Services, National Institutes of Health, National Heart, Lung and Blood Institute
2003 The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment
of High Blood Pressure NIH Publication No 03-5233
Van Roon A M, Mulder L J, Althaus M and Mulder G 2004 Introducing a baroreflex model for studying cardiovascular
effects of mental workload Soc. Psychophysiol. Res. 41 961–81
Wu W Y, Dai G H and Wen G B 1999 The dynamic coupling of left ventricle and systemic arteries Appl. Math. Mech.
20 661–73 (in Chinese)
Zollei E, Paprika D and Rudas L 2003 Measures of cardiovascular autonomic regulation derived from spontaneous
methods and the Valsalva maneuver Auton. Neurosci.: Basic Clin. 103 100–5