16th IFAC workshop on Control Applications of Optimization

October
16th IFAC6-9, 2015. Garmisch-Partenkirchen,
workshop on Control ApplicationsGermany
of Optimization
16th IFAC6-9,
October workshop on Control Applications of online
Available
2015. Garmisch-Partenkirchen, Optimization
Germany at www.sciencedirect.com
October 6-9, 2015. Garmisch-Partenkirchen, Germany

ScienceDirect
IFAC-PapersOnLine 48-25 (2015) 206–209
HIV-infection modeling
HIV-infection
HIV-infection modeling
modeling
A. Kim*, V. Kormyshev*, H. Kwon**, M. Safronov***, A. Tarasyev***
A.
A. Kim*,
Kim*, V.V. Kormyshev*,
Kormyshev*, H. H. Kwon**,
Kwon**,  M. Safronov***, A. Tarasyev***
M. Safronov***, A. Tarasyev***

*Ural Federal University named after the  first President of Russia B.N. Yeltsin,
*Ural Federal University
Yekaterinburg, named
Russian after the first
Federation President
(e-mail: of Russia B.N. Yeltsin,
avkim@imm.uran.ru,
*Ural Federal University named after the first President of Russia B.N. Yeltsin,
Yekaterinburg, Russian Federation (e-mail: avkim@imm.uran.ru,
Kormyshev@gmail.com).
Yekaterinburg, Russian Federation (e-mail: avkim@imm.uran.ru,
** Inha university, Inchon,Kormyshev@gmail.com).
Republic of Korea (e-mail: hdkwon@inha.ac.kr)
Kormyshev@gmail.com).
** Inha
*** The university,
Institute of Inchon, Republic
Mathematics of Korea(IMM)
and Mechanics (e-mail: hdkwon@inha.ac.kr)
named after N.N. Krasovskii
** Inha university, Inchon, Republic of Korea (e-mail: hdkwon@inha.ac.kr)
*** The Institute
of the of Mathematics
Ural Branch of and
the Mechanics
Russian Academy(IMM)of named after
Sciences (UrBN.N.
RAS)Krasovskii
*** The Institute of Mathematics and Mechanics (IMM) named after N.N. Krasovskii
of the Ural Branch
(e-mail: of the Russian Academy
samaxan@gmail.com, of Sciences (UrB RAS)
tam@imm.uran.ru)
of the Ural Branch of the Russian Academy of Sciences (UrB RAS)
(e-mail: samaxan@gmail.com, tam@imm.uran.ru)
(e-mail: samaxan@gmail.com, tam@imm.uran.ru)
Abstract: In the paper a problem of stabilizing a HIV infection dynamics mathematical model is
considered.In
Abstract:
Abstract: The
In
themodel
paper a problembyofa stabilizing
the paperis adescribed
problem of stabilizing
a HIV infection
system of functional dynamics
differential
a HIV infection
mathematical
equations.
dynamics A stabilizing
mathematical
model is
control
model is
considered.
is constructed Thebasing
modelonisthe
described
method by
of aexplicit
system solutions
of functional
of differential Riccati
Generalized equations. A stabilizing
Equations of the control
theory
considered. The model is described by a system of functional differential equations. A stabilizing control
is
of constructed basing on the
analytical constructing methodfor
regulator of explicit solutions of Generalized
For constructRiccati Equations of we
the theory
is constructed basing on the method of systems with delays.
explicit solutions of Generalized a feedback
Riccati control
Equations of theuse the
theory
of analytical
first variant ofconstructing
explicit regulator
solutions of for
the systems with
generalized delays.
Riccati For construct
equations (thea feedback
study of control
control we use the
stabilizing
of analytical constructing regulator for systems with delays. For construct a feedback control we use the
first
first variant
properties of
of explicit
variantbased on other
explicit solutions
variantsof
solutions the
the generalized
ofdiscussed in otherRiccati
generalized authorsequations
Riccati (the
(the study
study of
articles). Stabilizing
equations control
control
of stabilizing
for the
control system
stabilizing
properties
of based
differential on otherwith
equations variants
delaydiscussed
supports inHIV-infection
other authorsmodel
articles). Stabilizing
spread at a control
certain for the system
sufficiently small
properties based on other variants discussed in other authors articles). Stabilizing control for the system
of differential
non-zero level. equations
Results ofwith
the delay supports
research can be HIV-infection
applied to model
analysis of spreadaspects
some at a certain
of HIV sufficiently
dynamics. small
of differential equations with delay supports HIV-infection model spread at a certain sufficiently small
non-zero level. Results of the research can be applied to analysis of some aspects of HIV dynamics.
© 2015, IFAC
non-zero
Keywords: level. ResultsHIV;
modelling; of the
(International research can
Federation
differential be applied
ofequations
Automatic todelay;
analysis
Control)
with of some
Hosting aspects
by Elsevier
generalized Riccati of
Ltd. HIV dynamics.
All rights
equations. reserved.
Keywords: modelling; HIV; differential equations with delay; generalized Riccati equations.
Keywords: modelling; HIV; differential equations with delay; generalized Riccati equations.
 In this paper we consider the problem of constructing the
1. INTRODUCTION 
1. INTRODUCTION In
feedback this paper we consider
control, the problem
stabilizing the HIV of constructing
model, the
1. INTRODUCTION In this paper we consider the problem of constructing the
It has been over 30 years since the discovery of human feedback
mathematical control,
model stabilizing
described the a HIV
by system model,
of linearthe
feedback control, stabilizing the HIV model, the
It has been over 30 years sinceasthe thediscovery
causative of human mathematical
functional model
differential described by a system of linear
immunodeficiency
It has been over 30 years virus (HIV)since the discovery ofagent
human of mathematical model equations
describedthat byallows
a system you to ofapply for
linear
immunodeficiency
AIDS. Today, the virus
HIV (HIV)
pandemic as thehas causative
affected agent 35
about of functional
the
functional constructiondifferential
differential equations
of control theory
equations that allows
allows you
thatanalytical to
to apply
apply for
construction
you of
for
immunodeficiency virus (HIV) as the causative agent of
AIDS.
million Today, the HIV
peopletheworldwide. pandemic has affected
Evenhasif affectedsuccessfulabout 35 the
the construction
regulators for systems of
of control
with theory
(Kwon,analytical construction
2010; Krasovskiy, of
1962;
AIDS. Today, HIV pandemic aboutviral
35 construction control theory analytical construction of
million regulators forInsystems with (Kwon, 2010; Krasovskiy, 1962;
million people
suppression worldwide.
with highly
people worldwide.active Even if
if successful
antiretroviral
Even viral
therapy (Arts,
successful viral Kim, 1998). Kwon (2010)
regulators for systems with (Kwon, 2010; Krasovskiy, 1962; and Kim (1998) based on the
suppression with highly
2012), the infection can notactive antiretroviral
be completely therapy
cured. (Arts,
Moreover, Kim,
explicit 1998). In Kwon
solution of the (2010) and Kim
generalized Riccati(1998) basedobtained
equation on the
suppression with highly active antiretroviral therapy (Arts, Kim, 1998). In Kwon (2010) and Kim (1998) based on the
2012),
so far the
not infection
developed canannoteffective
be completelyvaccine cured.
(Kwong,Moreover,
2011; explicit
several solutionfor
options of the generalized
feedback control Riccati
for equation
linear obtained
systems with
2012), the infection can not be completely cured. Moreover, explicit solution of the generalized Riccati equation obtained
so far2011).
Jang, not developed
The reasons an for
effective
this are vaccine
directly (Kwong,
related to2011;
two several
delays. options
This for feedback
approach is control
used to for
studylinearthe systems
problem with
of
so far not developed an effective vaccine (Kwong, 2011; several options for feedback control for linear systems with
Jang, 2011).
important The reasons
features of HIV. forFirstly,
this are the directly
HIV related can
provirus to twobe delays.
stabilization Thisofapproach
the is used
considered modelto ofstudy
HIV. the
In problem
the first of
part
Jang, 2011). The reasons for this are directly related to two delays. This approach is used to study the problem of
important
hidden in features
the of HIV.
target cell Firstly,
(T-cells) theandHIVtherefore
provirus remain
can be stabilization
of the paper ofexamines
the considered model ofHIV
stabilizability HIV.infection
In the first part
model,
important features of HIV. Firstly, the HIV provirus can be stabilization of the considered model of HIV. In the first part
hidden in bythe target cell system
(T-cells) andand drugtherefore remain of
built the paper examines stabilizability
embodimentHIV infectionsolutions
model,
undetected
hidden in thethe immune
target cell (T-cells) and used. Secondly,
therefore remain of theonpaper the basisexaminesof a firststabilizability HIVof explicit
infection model,
undetected
the by
variability the immune
of HIV system
maysystem and
lead toand drug
thedrug used.
emergence Secondly,
of drug built
GRE. onIn the basis
second of a
partfirst
of embodiment
the
built on the basis of a first embodiment of explicit solutionsresearch of
appliedexplicit solutions
management,
undetected by the immune used. Secondly,
the GRE. In the
the second ofpart of the research applied management,
the variability
resistance of
of HIV
in immune
variability may
may lead
HIVrecognition lead to andthe
to emergence
thereduced (Arts,of
emergence drug
of2012).
drug built
GRE.on In the basis
second a third
part embodiment
of the of explicit
research applied solutions
management,
resistance
The adventin ofimmune
modern recognition
methods andinvestigation
of reduced (Arts, led 2012).
to an built
GRE. onComputer
the basis ofsimulations
a third embodiment
showed ofthat
explicit
both solutions
control
resistance in immune recognition and reduced (Arts, 2012). built on the basis of a third embodiment of explicit solutions
The advent of ofmodern
understanding the methods of
importance of investigation
specific led ofto the
factors an GRE.
stabilize Computer
consider HIVsimulations
model. showed that both control
The advent of modern methods of investigation led to an GRE. Computer simulations showed that both control
understanding
human body to of the importance
support or limit HIV ofreplication.
specific factors
In of the
particular, stabilize consider HIV model.
understanding of the importance of specific factors of the stabilize consider HIV model.
human
the bodyresearch
current to support or limit
is aimed at HIV replication.
the source of theIn particular,
human body to support or limit HIV replication. Indepletion of
particular, 1.1 HIV-infection mathematical model
the current
latent HIV research2011).
(Cohen, is aimed at the source of the depletion of 1.1 HIV-infection mathematical model
the current research is aimed at the source of the depletion of 1.1 HIV-infection mathematical model
latent HIV (Cohen, 2011).
latent HIV (Cohen,
Beginning with the 2011).work of G.I. Marchuk and R.V. Classical mathematical model HIV-infection spread in the
Classical
human bodymathematical
is a systemmodel HIV-infection
of differential equations spread
within the
Beginning
Khokhlova, with
Beginning with the work
the
mathematical work of G.I.
modeling
of G.I. is anMarchuk
importantand
Marchuk part R.V.
and of a
R.V. Classical mathematical model HIV-infection spread indelay
the
human
(Bocharov, body is a system of differential equations with delay
Khokhlova, mathematical
serious immunological
Khokhlova, modeling
mathematical modeling research.is anis an important
Broad
important part
part of
classes ofofaa human body2012; is a Ciupe,
system 2006):of differential equations with delay
serious immunological research. describe Broad the classes of (Bocharov,
(Bocharov, 2012;
2012; Ciupe,Ciupe, 2006):
2006):
mathematical models that adequately
serious immunological research. Broad classes of dynamics T (t )  s  dT  kVT ,
mathematical
of HIV allow models that adequately
mathematical tools to describe
analyze the
the dynamics
immune T *(t )  s  dT  kVT
mathematical models that adequately describe the dynamics T * ,, d ET * ,
of HIV allow
processes mathematical
and develop appropriatetoolsmethods.
to analyze the immune T *(t()t )s kVT  dTkVT
of HIV allow mathematical tools to analyze the immune
T (t )  kVT*  T *  d x ET * ,,
processes and develop appropriate methods.
processes and most
develop appropriate V T*(t()t )NkVTT cV T *,  d x ET * ,,
One of the interesting andmethods.
important problems in V (t )  NT* *
x
*(t  cV
 , d E, ,

One of the most
mathematical interesting
immunology is to
One of the most interesting and important
and
developimportant
methods problems
of control
problems
in
in V
 E ( t )  N 
pT T
*  cV ) , E ,
mathematical
immune models,immunology
as progressis toindevelop
this
mathematical immunology is to develop methods of control
methods of
direction, control
given the E ((tt )) 
 E  pT   )) 
pT * ((tt  d dE E E,
immune models, as  E
current
immunelevel as progress
of experimental
models, progress and in
in this
clinical
this direction,
studies given
direction, the
can, with
given the
current level
sufficient
current level of
funding experimental
and proper
of experimental and clinical
andmanagement, studies
clinical studies can,
to can, with
translate
with
sufficient
research funding
practicaland proper management, to translate
sufficientinto
funding management
and of HIV dynamics.
proper management, to translate
research into practical management of HIV dynamics.
research into practical management of HIV dynamics.
Copyright IFAC 2015
2405-8963 © 2015, 206Hosting by Elsevier Ltd. All rights reserved.
IFAC (International Federation of Automatic Control) 206
Copyright
Peer review©under
IFAC responsibility
2015 206Control.
of International Federation of Automatic 206
Copyright © IFAC 2015 206 206
10.1016/j.ifacol.2015.11.086
 IFAC CAO 2015
October 6-9, 2015. Garmisch-Partenkirchen, Germany
A. Kim et al. / IFAC-PapersOnLine 48-25 (2015) 206–209 207

T (t )  s  dt  kVT 0 0 0 0 
T * (t )  kVT  T *  d ET *  0 - 3 0 3 ,
V (t )  NT *  cV A
X (1) 0 1285 - 6 0 
 0 0 0 - 16
E (t )  pT * (t   )  d EU
where T (t ) - number of uninfected T-cells, 0 0 0 0

A   0 0 0 0 ,
T * (t ) - number of infected T-cells, 0 0 0 0
0 1.473 0 0
V (t ) - number of free viral cells,
0 
0 
E (t ) - immune response, the number of effector cells B   .
0
generating by human body after drug stimulation, 1
 
s - source of healthy cells,
The system (2) has a root with a positive real part (Ciupe,
d - healthy cells mortality, 2006) and is unstable (El'sgol'ts, 1971). The considered task
is to construct a feedback control
p - effector cells activation rate, 0
u[t , x(t ), x(t  s)]  Cx(t )   ( s) x(t  s)ds, (3)
k - the rate of infection, 

stabilizing system (3). As mentioned above is not always

b - infected cells mortality, HIV-infected cells may be derived from the human body,
however, in contrast to the technical concept of the
d x - immune response effectiveness,
stabilization, when the solution tends to zero, we understand
the following stabilization.
N - number of virus particles obtained from one infected
cell,
Control (3)  -stabilizes (   R n ) system (2), if system (2)
c - viral cells clearance, is  -asymptotically stable, i.e. every decision of this closed
system tends to  .
d E - effector cell mortality,

 - (delay) - time required effector cell for infection 1.2 The purpose and methodology of research
recognition,
The purpose of this paper is to simulate the process of HIV
U - drug immune response stimulation that help human
replication in the human body and research the properties of
body generate effector cells and kill free viral cells in human
stabilizing feedback control, that was built on the basis of a
blood.
first variant of explicit solutions GRE.
The process regulation can be carried out at the expense of
In Krasovskiy (1962), Kim (1996, 1998, 2004, 2011) and
drug enhancing the immune response: in this case U is a Kwon (2010) discuss how stabilizing control on the basis of
control parameter of u (t ) . generalized Riccati equation explicit solutions.
The parameters of the system (1) are presented in (Ciupe, A first variant of explicit solutions GRE (Kwon, 2010)
2006) corresponds to the feedback control
 0

T(0) = T0 , T* ( ) = 0, V(0) = V0 , E(0) = 0,  [- , 0] , u * ( x, y())   N 1 B'  Px   e [ PK  A']( s  ) PA y( s)ds , (4)
  
T0  10 , T *0  0 , V 0 109 , E0  0.1 , and closed system
s  0.103 , d  0.01025 , p  1.473 , c  0.557 ,
 0
 (5)
x (t )  Ax(t )  A x(t  1)  K *  Px   e [ PK  A']( s  ) PA y( s)ds ,
k  0.0000065 , b  0.514 , d x  0.812 , N  2500 ,   

d E  1.618 ,   16.05 , This control is obtained by substituting (4) into the original
system (2). There is
V  -729.2579, T  -0.3161, E  -0.2878, T  18.6939
*

K  BN 1 B'.
System (1) can be conveniently represented in matrix form as
(Ciupe, 2006) In control (4), the 4 4 matrix P is a solution of the matrix
x(t )  Ax(t )  A x(t   )  Bu(t ), (2) equation (Kim, 2004)
PA  A' P  M  PKP, (6)
where M - the identity matrix.

207
IFAC CAO 2015
208 A. Kim et al. / IFAC-PapersOnLine 48-25 (2015) 206–209
October 6-9, 2015. Garmisch-Partenkirchen, Germany

To verify that the control (4)  -stabilizes system (2) is application control, built on the first variant explicit solutions
of generalized Riccati equation.
sufficient to show  -asymptotically stability of the closed
system The negative consequences using control (4) for stabilizing
1
the system (5) is reducing the number of effectors and
x (t )  ( A  BN B' P) x(t )  A x(t   )  uninfected T-cells ( E (t ) and T (t ) parameters in the system
(7)
0
1 [ PK  A']'( s  )
of functional differential equations).
  (G( s)  BN B' e PA )x(t  s)ds.
 This inevitably occurs reduction in the immune functions of
2. COMPUTER SIMULATION RESULTS the human organism as a whole. With long-term using of
stabilization may occur development immunopathological
processes that ultimately can lead to autoimmune diseases
2.1 Solution of the matrix equation and hypersensitivity to various infections.

Solution of equation (6) can be found analytically and equal In order to reduce the negative impacts of the considered
control is proposed to apply it is not continuous, but discrete.
0 41 0 1  Period of application of immune drugs offered alternate with
P1  41 11767 41 315. (8) recover periods.
0 41 0 1
 1 315 1 11  This approach will significantly increase the amount of time
needed to lead the system to a steady state, but will reduce
2.2 System trajectory construction the negative effects of therapy.

The trajectory of the system (4) (see Figure 1) converges to 3. CONCLUSION

 , and hence control (4) is  -stabilizes.
When   16.05 the system (5) tends to
 2.5930 
0.001 .
  - -2.3601
 value. That is, the control maintained replication of HIV-
 0.0479 
Fig. 1. The trajectory of the system (5).
In the case of using the control (4) for stabilizing the system
(5) can be observed a significant (to a negative value)
reduction count of free viral cells in the human body ( V (t )
parameter in the (1) system of functional differential
equations).
The number of infected T-cell ( T * (t ) parameter in the (1)
system of functional differential equations) is stabilized at a
low level negative value.
Control leads system (5) to a steady state with low viral load
and a small amount of infected T-cells and can to stabilize
system (5) in state with negative values of these parameters.
These factors can be attributed to the positive results of
Thus, the feedback control is constructed on the basis of a
first variant of generalized Riccati equation explicit solutions
is  -stabilizes the spread of HIV-infection in humans.
At the same control characteristics tend to some non-zero
infection in the human body in a certain steady state.
This steady state is characterized by low level of infected T-
cells and viral loads, but also human body immune function
is in depressed state.
To reduce the negative impact of therapy on the human body
are encouraged to apply discrete control, although this
approach will increase the total time to lead the system in a
steady state.
Comparing our results with the results of other authors Ciupe
(2006), Jang (2011), Kwong (2011), Arts (2012), it should be
noted that the considered control, built on the basis of a first
variant of generalized Riccati equations explicit solutions not
only leads the (1) system of functional differential equations
describing the HIV-infection model dynamics in the human
body, in a steady state with low viral load and allows to
achieve remission of the disease at a level not dangerous to
human life, but also can contribute to the complete
disappearance of free viral cells and infected T-cells.
Remains an open the question about stabilizing properties of
controls, built on the basis of the second and third variant of
explicit solutions of generalized Riccati equations.
Subsequent articles will be considered building a stabilizing
control on the basis of the remaining variants of explicit
solutions GRE and discuss the results of their using for
stabilizing the HIV-infection dynamics model that was built
in the current article.

208
 IFAC CAO 2015
October 6-9, 2015. Garmisch-Partenkirchen, Germany
A. Kim et al. / IFAC-PapersOnLine 48-25 (2015) 206–209 209

REFERENCES Conjugate Gradient Method \\ Bull Math Biol (2011),

73:2408-2429.
Kim, A.V., Volokhova, L.E., Zavodnikov, D.E. (2011).
Vestnik Nizhegorodskogo universiteta im N.I.
Lobachevskogo [Bulletin of the Nizhny Novgorod
University after the name of Lobachevsky], no. 4, pp.
172-173.
Kwon, W.H., Kim, A.V., Kormyshev, V.M., Pimenov, V. G.,
Solodushkin, S.I. (2010). Analiticheskoe
konstruirovanie i sintez regulyatorov dlya sistem s
posledeystviem [Analytical design and synthesis of
controllers for systems with delay]. Ekaterinburg, Urfu.
Kim, A.V. (1996). i-Gladkiy analiz i funktsional'no-
differentsial'nye uravneniya [i-smooth analysis and
functional-differential equations], p 236. Ekaterinburg,
IMM UrO RAN.
Kim A.V., Pimenov, V.G. (2004). i-Gladkiy analiz i
chislennye metody resheniya funktsional'no-
differentsial'nykh uravneniy [i-smooth analysis and
numerical methods of solution of functional differential
equations]. M.-Izhevsk, Regulyarnaya i khaoticheskaya
dinamika.
Krasovskiy, N.N. (1962). Prikladnaya matematika i
mekhanika [Applied Mathematics and Mechanics], no.
26, pp. 39-51.
El'sgol'ts, L.E., Norkin, S.B. (1971). Vvedenie v teoriyu
differentsial'nykh uravneniy s otklonyayushchimsya
argumentum [Introduction to the theory of differential
equations with deviating argument], p. 296. M. Nauka.
Kim, A.V., Han, S.H., Kwon, W.H., Pimenov, V.G. (1998).
Explicit numerical methods and LQR control algorithms
for time-delay systems. Proceedings of the international
Conference on Electrical Engineering. Kyungju, Korea,
July 21-25, 1998.
Arts, E. J., and Hazuda, D. J. (2012). HIV-1 Antiretroviral
Drug Therapy. Cold Spring Harbor perspectives in
medicine, 2(4) (2012), p. a007161.
Bocharov, G., Chereshnev, V., Gainova, I., Bazhan, S.,
Bachmetyev, B., Argilaguet, J., Martinez, J., Meyerhans,
A. (2012). Human Immunodeciency Virus Infection:
from Biological Observations to Mechanistic
Mathematical Modelling. EDP Math. Model. Nat.
Phenom, pp. 1-29. Vol. 7, No. 2, 2012.
Ciupe, M. S., Bivort, B. L., Bortz, D. M. and Nelson, P. W.
(2006). Estimating kinetic parameters from HIV primary
infection data through the eyes of three different
mathematical models. Mathematical biosciences, 200(1)
(2006), pp. 1-27.
Cohen, J. (2011). Understanding HIV latency to undo it.
Science, 332(6031) (2011), p. 786.
Kim, A.V., Kwon, W.H., Pimenov, V.G. (1999). Numerical
methods and a software package for delay differential
equations // The Third International Conference on
Dynamical Systems and Applications. Atlanta. USA.
May 26-29. 1999.
Kwong, P. D., Mascola, J. R. and Nabel, G. J. (2011).
Rational Design of Vaccines to Elicit Broadly
Neutralizing Antibodies to HIV-1. Cold Spring Harbor
perspectives in medicine, 1(1) (2011), p. a007278.
Jang, T., Kwon, H.-D., Lee, J. (2011). Free Terminal Time
optimal Control Problem of an HIV Model Based on a

209