- About this Journal
- Abstracting and Indexing
- Aims and Scope
- Annual Issues
- Article Processing Charges
- Articles in Press
- Author Guidelines
- Bibliographic Information
- Citations to this Journal
- Contact Information
- Editorial Board
- Editorial Workflow
- Free eTOC Alerts
- Publication Ethics
- Reviewers Acknowledgment
- Submit a Manuscript
- Subscription Information
- Table of Contents
Discrete Dynamics in Nature and Society
Volume 2012 (2012), Article ID 963864, 20 pages
Stability and Feedback Stabilization of HIV Infection Model with Two Classes of Target Cells
1Department of Mathematics, Faculty of Science, King Abdulaziz University, P.O. Box 80203, Jeddah 21589, Saudi Arabia
2Department of Mathematics, Faculty of Science, Al-Azhar University, Assiut 71511, Egypt
Received 3 April 2012; Accepted 24 June 2012
Academic Editor: Mingshu Peng
Copyright © 2012 A. M. Elaiw and A. M. Shehata. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We study the stability and feedback stabilization of the uninfected steady state of a human immunodeficiency virus (HIV) infection model. The model is a 6-dimensional nonlinear ODEs that describes the interaction of the HIV with two classes of target cells, CD4+ T cells and macrophages, and takes into account the Cytotoxic T Lymphocytes (CTLs) immune response. Lyapunov function is constructed to establish the global asymptotic stability of the uninfected steady state of the model. In a control system framework, the HIV infection model incorporating the effect of Highly Active AntiRetroviral Therapy (HAART) is considered as a nonlinear control system with drug dose as control input. We developed treatment schedules for HIV-infected patients by using Model Predictive Control (MPC-)based method. The MPC is constructed on the basis of an approximate discrete-time model of the HIV infection model. The MPC is applied to the stabilization of the uninfected steady state of the HIV infection model. Besides model inaccuracies that HIV infection model suffers from, some disturbances/uncertainties from different sources may arise in the modelling. In this work the disturbances are modelled in the HIV infection model as additive bounded disturbances. The robustness of the MPC against small model uncertainties or disturbances is also shown.
In the last decade many mathematical models have been proposed to describe the immunological response to infection with the human immunodeficiency virus (HIV). HIV is responsible of acquired immunodeficiency syndrome (AIDS). Some of these models mainly represent the interaction of the HIV with the CD4+ T cells, others take into account the Cytotoxic T Lymphocytes (CTLs) immune response . A tremendous effort has been made in studying the basic properties of these models such as positive invariance properties, boundedness of the model solutions, and stability analysis of the steady states (see e.g., [2–14]). Studying such properties is important for understanding the associated characteristics of the HIV dynamics. The treatment of HIV-infected patients is of major importance in today’s social medicine. Currently, the most important categories of anti-HIV drugs are reverse transcriptase inhibitors (RTIs) drugs and protease inhibitors (PIs) drugs. Reverse transcriptase inhibitors prevent the HIV from infecting cells by blocking the integration of the HIV viral code into the host cell genome. Protease inhibitors prevent already infected host cells from producing infectious virus particles. Recently, Highly Active AntiRetroviral Therapies (HAARTs) which consist of one or more RTI and a PI can suppress viral load below detectable levels and consequently prolong time to the onset of AIDS. Perleson et al., observed that after the rapid first phase of decay during the initial 1-2 weeks of antiretroviral treatment, plasma virus levels declined at a considerably slower rate . This second phase of viral decay was attributed to the turnover of a longer-lived virus reservoir of infected cells. These cells are called macrophages and considered as the second-target cell for the HIV. Therefore, the HIV infection model with two classes of target cells is more accurate than the model with one class of target cells (see [8, 16]). Some HIV infection models exist to describe the interaction process of the HIV not only with the CD4+ T cells but also with the macrophages which are the crucial immune responses and play important roles in phagocytosis (see e.g., [8, 16–19]). The basic properties of the two target cells models are recently studied in [3, 20–22]. In [23, 24] we have studied a class of virus infection models assuming that the virus attacks multiple classes of target cells. However, in [3, 20–24], the CTL immune response is neglected. The role of CTL cells is to attack the infected cells. The first purpose of the present paper is to study the basic properties of an HIV infection model which describes the interaction of the HIV with two target cells, CD4+ T cells and macrophages and takes into consideration the CTL immune response. The global stability of the uninfected steady state of the model is established using a Lyapunov approach.
Optimal treatment scheduling of HIV infection using a control theoretic approach is the subject of substantial research activity. In [17, 18, 25–28], an open-loop optimal controller is designed via Pontryagin’s Maximum Principle. A major drawback of such controller is its lack of robustness against disturbances or model uncertainties. In fact, the interaction of HIV with immune system is not very clear and complicated which leads to model inaccuracies and parameter uncertainties. Moreover, some disturbances may arise from immune system fluctuation or immune system reactions due to coinfections. Therefore, the design of optimal treatment schedules based on an open-loop optimal controller may lead to undesired results. To overcome this drawback, a feedback controller which has inherent robustness property against disturbances has to be designed. Recently, model predictive control (MPC) method is developed for determining optimal treatment schedules for HIV patients ([20, 29–33]). The MPC method obtains the feedback control by solving a finite horizon optimal control problem at each time instant using the current state of the system as the initial state for the optimization and applying “the first part” of the optimal control. The study of stability and robustness properties of MPC schemes has been the subject of intensive research in recent years (see e.g., [34, 35]). In [20, 30, 31, 33], the HIV infection model is discretized and the MPC is constructed on the basis of an approximate discrete-time model. Sufficient conditions are established which guarantee that the MPC designed via approximate discrete-time model stabilizes the original continuous-time model. In [29, 32], the effect of the discretization of the differential equations on the stability analysis was not considered. The importance of approximate discrete-time design is supported by a series of counter examples (see e.g., ), which show that the controller that stabilizes the approximate discrete-time model, may fail to stabilize the original continuous-time system. In [29–33], the MPC is applied to the one target cell (i.e., CD4+ T cell) model. The HIV infection model with two target cells is considered in , but the CTL immune response is neglected.
The second purpose of the present paper is to develop treatment schedules for HIV-infected patients by using MPC. The construction of MPC is based on the approximate discrete-time model of the model. The HIV infection model with additive disturbances is also considered, and the inherent robustness properties of the MPC are established.
The layout of the paper is as follows: in Section 2, we introduce the HIV infection model and study its basic properties. In Section 3, the HIV infection model with additive disturbance is outlined. In Section 4 we outline the MPC design for nonlinear control systems and summarize the main results obtained in [36, 37]. Application of MPC to the HIV infection model is given in Section 5. Section 6 presents the simulation results. The last section is the conclusion.
2. HIV Infection Model
We will study the mathematical model of HIV infection which describes the interaction of HIV with two cocirculation populations of target cells, potentially representing CD4+ T cells and macrophages and takes into consideration the CTL immune response. The model is a modification of the HIV infection model presented in [3, 8, 16, 20], which include an additional state variable for the CTL cells:
Here , , , , , and represent the concentrations of uninfected CD4+ T cells, infected CD4+ T cells, uninfected macrophages, infected macrophages, free virus particles, and CTL cells, respectively, at time . The populations of the uninfected CD4+ T cells and macrophages are described by (2.1) and (2.3), respectively, where and represent, respectively, the rates of which new CD4+ T cell and macrophages are generated from sources within the body, , are the death rate constants, and , are the infection rate constants. Equations (2.2) and (2.4) describe the population dynamics of the infected CD4+ T cells and macrophages and show that they die with rate constant and killed at rate and . The virus particles are produced by the infected CD4+ T cells and infected macrophages with rate constant and are cleared from plasma with rate constant . The CTL cells are produced at a rate and are decayed at a rate . The effect of the RTI and PI drugs is represented by the chemotherapy functions and , where the parameters and are the efficiencies of RTI and PI drugs, respectively. The variables and are the doses from drugs administrated to RTI and PI, respectively (see ). All the parameters of the model are supposed to be positive.
We introduce control functions as , . Then system (2.1)–(2.6) can be considered as a nonlinear control system with as the state vector and is the control input vector. We are now ready to present a study on the basic mathematical properties of the model.
2.1. Positive Invariance
This means that the nonnegative orthant is positively invariant, namely, if a trajectory starts in the nonnegative orthant, it remains there. The boundedness of the solutions of model (2.1)–(2.6) will be given in the following proposition.
Proposition 2.1. There exist positive numbers , , and such that the compact set, is positively invariant.
Proof. Let , then where . Hence for all if . It follows that for all if , where . On the other hand, then , for all if , where .
2.2. Steady States
We will compute the steady states of system (2.1)–(2.6) under constant controllers, that is, for , , . It will be explained in the following that the behavior of model (2.1)–(2.6) crucially depends on the basic reproduction number given by where . We note that can be written as , where are the basic reproduction numbers of each CD4+ T and macrophages dynamics separately. The immune strength of CD4+ T and macrophages are given, respectively, by and (see ).
Proof. The proof can follow the same lines as that of Proposition 2 in  and Theorem 2.2 in .
In the present paper, we are interested in studying the stability of the uninfected steady state of the HIV infection model.
2.3. Global Stability of
In this section, we prove the global stability of the uninfected steady state by using a Lyapunov approach.
Theorem 2.3. If , then is globally asymptotically stable.
Proof. By the method of [3, 5], we consider a Lyapunov function:
We note that is defined, continuous, and positive definite for all . Also, the global minimum occurs at the uninfected steady state . Further, function along the trajectories of (2.1)–(2.6) satisfies
Since , then the last term of (2.14) is negative. Also, since the arithmetical mean is greater than or equal to the geometrical mean, then the first two terms of (2.14) are less than or equal to zero. Therefore, if then for all . The maximal compact invariant set in is the singleton when . The global stability of follows from LaSalle’s Invariance Principle.
Proof. Let and with , where then , therefore the corresponding trajectory will tend to as .
Remark 2.5. We observe that, if , then it is sure that and . But if one considers the three-dimensional model (2.1)-(2.2), and (2.5) and designs a controller such that , then the whole system may be unstable around , because . This shows the importance of considering the effect of the macrophages in the HIV dynamics.
We note that the HIV infection model suffers from model inaccuracies or disturbances that may arise from different sources such as, modeling errors, immune system fluctuation, immune effect of a coinfection, measurement noise, and estimation errors. Therefore, in the next section we will consider the presence of the disturbances in the HIV infection model.
3. HIV Infection Model with Additive Disturbances
3.1. Positive Invariance
We note that the conditions in (3.8) give a lower bound of the disturbances only at the the boundary of .
Proposition 3.1. Suppose that the disturbances satisfy the bound (3.7) then there exist positive numbers , and such that the compact set, is positively invariant.
Proof. The proof can follow the same as Proposition 2.1. It can be easily to show that , , where .
4. MPC for Nonlinear Systems
In this section, we outline the MPC design for nonlinear systems and give a review on the results obtained in [36, 37]. We have shown in the preceding sections that the HIV system states for the nominal model (2.1)–(2.6) and disturbed model (3.1)–(3.6) can be taken from compact sets. Moreover, since the drug dosage of HAART can not arbitrarily increased, thus the controller can also be taken from a compact set.
The sets of real and natural numbers (including zero) are denoted, respectively, by and . The notation denote the set of real numbers in the interval . A continuous function is of class- if , for all and it is strictly increasing. It is of class- if it is of class- and when . A continuous function is of class- if is of class- in for every , it is strictly decreasing in for every and when . The Euclidean norm of a vector is denoted as . Given we define to be a ball of radius centered at the origin. We introduce the following notation for any set .
Consider a continuous-time nonlinear control system given by where , are the state and control input, respectively, is continuous and Lipschitz continuous with respect to in any compact set and , and is compact and .
The control is taken to be a piecewise constant function where is the control sampling period which is fixed.
We will assume that there is a compact set containing the origin that is positively invariant with respect to system (4.1) for any piecewise constant controller . Let denote the solution of (4.1) with given , and . The exact discrete-time model which describes the behavior of the system at the sampling instants , is given by where . We note that, since is typically nonlinear, in (4.3) is not known in most cases, therefore the controller design can be carried out by means of an approximate discrete-time model where is a modelling parameter, which is typically the step size of the underlying numerical method. The applied numerical scheme approximation has to ensure the closeness of the exact models in the following sense. Let be a compact set such that .
Assumption A1. There exists an such that(i), is continuous in both variables uniformly in , and Lipschitz continuous w.r.t in any compact set, uniformly in small ,(ii)there exists a such that
for all , all , and .
We note that, under the conditions on the function , then Assumption A1 can be proven for many one-step numerical methods.
Since we want to find a state-feedback controller, it seems to be reasonable to investigate when it does exist. The next assumption formulates, roughly speaking, a necessary condition for the existence of a stabilizing feedback.
Assumption A2. There exists an such that the exact discrete-time model (4.3) is practically asymptotically controllable from to the origin with piecewise constant controllers for all (See e.g., [36, 40] for the definition).
For the solutions of (4.3) and (4.4) with , and we will use the notations and , respectively.
Let be given and let (4.4) be subject to the cost function where , , , and are given functions, satisfying the following assumptions.
To ensure the existence and the stabilizing property of the proposed controller, the following assumptions are needed.
Assumption A3. Let , (i) is continuous, positive definite radially unbounded, and Lipschitz continuous in any compact set, (ii) is continuous with respect to and , uniformly in small , and Lipschitz continuous in any compact set, (iii) there exist an and two class- functions and such that the inequality,
holds for all and .
The terminal cost function and/or a terminal constraint set given explicitly or implicitly play a crucial role in establishing the desired stabilizing property. We will assume that has to be a local control Lyapunov function.
Assumption A4. There exist and such that for all there exists a such that inequality holds true for all .
Consider the optimization problem:
If this optimization problem has a solution denoted by , then the first element of is applied at the state , that is,
The value function for the optimal control problem is
Let denote the minimum of the values generated by Assumptions A1-A4.
Theorem 4.1 (see ). If Assumptions A1–A4 hold true, then(i)there exist an with , and a constant independent of , such that for all , and ,(ii)there exist constants ,, and and functions such that for all , , and ,
Moreover, for all with for all .
Theorem 4.1 shows that undersuitable conditions; the state feedback MPC renders the origin to be asymptotically stable for the approximate discrete-time model. These conditions concern directly with the data of the problem and the design parameters (the horizon length , the stage cost , the terminal cost , and the terminal constraint set ) of the method but not the results of the design procedure.
Theorem 4.2 (see ). Suppose that Assumptions A1–A4 are valid and is chosen such that , then, there exists , and for any there exists an such that for any and the trajectory of the exact discrete-time system with the MPC, satisfies
Theorem 4.2 lays the foundation for the design of a state feedback MPC via an approximate discrete-time model to achieve practical stability of the exact discrete-time model. Achieving practical stability of the exact model requires that the approximation error can be made sufficiently small. By application of some of one-step numerical approximation formula with possibly variable step size (e.g., a Runge-Kutta formula), the approximation error can be made sufficiently small.
4.1. Robustness Properties of the MPC
In this section we show the inherent robustness properties of the MPC against to bounded disturbances. We consider the continuous-time model (4.1) with additive disturbances where and are the state and disturbance, respectively, is compact, . We assume that there exists such that . For a given we use the following notation . Let us define
The exact discrete-time model for model (4.17) can be given as:
In this paper, the MPC is constructed on the basis of the approximate discrete-time model (4.4) for the nominal model (4.1). The MPC algorithm consists of performing the following steps at certain instants :(1)measure the current state of the system ;(2)compute the open-loop optimal control to the problem ;(3)the control is applied to the system in the interval (the remaining is discarded);(4)the procedure is repeated from (1) for the next sampling instants .
Theorem 4.3 (see ). Suppose that A1–A4 are valid and is chosen such as . Then there exist , and for any there exists an such that for any and , the trajectory of the exact discrete-time model with the MPC, and satisfies
Theorems 4.3 show that under suitable conditions the state feedback MPC practically stabilizes the exact discrete-time model for sufficiently small integration parameter and disturbances. When the full state of the system is not available for feedback, an observer can be designed for estimating the unknown states. In , it is shown that under a set of conditions the output feedback MPC practically stabilize the exact discrete-time model of the plant for sufficiently small approximation and estimation errors.
5. MPC for the HIV Infection Model
In this section we apply the MPC method proposed in Section 4 to the nominal model (2.1)–(2.6) as well as the disturbed model (3.1)–(3.6). We will show that, with a suitable choice of and functions and , the assumptions of Section 4 can be satisfied. We introduce new variables by the definition , , , , , . Let , then in these new variables the model (3.1)–(3.6) takes the form of (4.17) with and . The disturbance vector is assumed to be bounded in a compact set containing the origin. Let the compact set be defined as where , , and are as in Proposition 3.1. For the nominal model, the compact set is defined as by replacing with , where are given in Proposition 2.1.
To verify Assumptions A3 and A4, we linearized the nominal system (5.1) (i.e., ) around the origin in case of constant controllers, that is, with , where is given in Proposition 3.1. Let be the coefficient matrix of the linearized system. Then the discrete-time model for the linearized system is given by:
Let the sampling period be chosen to be and, . The running cost and the terminal cost can be chosen as: where are positive weighting constants, is a positive definite diagonal matrix, and is a positive definite symmetric matrix satisfying the Lyapunov equation for the discrete-time system (5.3):
From (5.4), Assumption A3 is satisfied. Assumption A2 follows from Corollary 2.4 and Assumption A1 holds also true if we choose a suitable numerical integration scheme (e.g., the Runge-Kutta formula). To verify Assumption A4, the weights and the matrix have been chosen through a series of numerical experiments as , , , and . It has been verified numerically by solving a constrained minimization problem with several starting points that Assumption A4 is satisfied over the whole set . Thus all assumptions of the proposed method can be satisfied with suitable choice of the parameters of the MPC method.
6. Numerical Results
We perform simulation studies using the parameter values which are listed in Table 1.
All computations are carried out by MATLAB, in particular, the optimal control sequence is computed by the fmincon code of the Optimization Toolbox. To reduce the computational complexity we chose horizon length to be . Simulations for the continuous-time system are carried out using ode45 program in MATLAB.
In this case no treatment is used (i.e., ). We can see that, for the parameters given in Table 1, , and . Therefore is unstable, exists and stable, while does not exist. To show the simulation results for this case, we assume that the infection occurs with a certain amount of virus particles and CTL cells . Thus the initial conditions for the untreated case are , , , and . From Figures 1, 2, 3, 4, 5, and 6, it can be seen that the concentrations of uninfected CD4+ T cells, macrophages, and CTL cells are decaying, while the concentrations of infected CD4+ T cells, infected macrophages, and free viruses are increasing. Also we note that the trajectory tends to the stable infected steady state .
In this case, the treatment is designed via MPC strategy for the nominal and disturbed HIV infection models. In this case, we assume that the treatment is initiated with time units after the onset of infection. Thus the initial conditions for the MPC are given by , , , , , and . The disturbances are simulated by , where the parameters ’s are uniformly distributed random numbers on and when the system states lie in the interior of the positive orthant . At the boundary of , the lower bound has to be chosen as the following: to guarantee that the positive orthant is positively invariant.
Figures 1, 2, 3, 4, 5, and 6 show also the application of the MPC to the HIV infection models in following cases.
Case (I): nominal HIV infection model that is, .
Case (II): disturbed HIV infection model with , , , , , .
Case (III): disturbed HIV infection model with , , , , , .
Figures 1 and 3 show that when the MPC is applied, the number of uninfected CD4+ T cells is increasing as well as the macrophages. This means that the HAART helps the immune system to recover with some fluctuations due to the presence of disturbances. From Figures 2 and 4 we can see that the number of infected CD4+ T cells and infected macrophages is decaying during the treatment. Figure 5 shows that after initiation of HAART the viral load drops quickly and it can be kept under a suitable level, with a small controller, corresponding to rather mild dosage of HAART. Figure 6 shows that the CTL cells are decaying for untreated as well as treated cases but with a faster rate than untreated case. The model predictive controller as a function of the time for cases (I) and (II) is shown in Figure 7. It is observed that the treatment is initiated with a stronger dosage of HAART and sequentially decreasing over time. Thus we can say that when the MPC strategy is applied in the presence of bounded disturbances the trajectory of the system tends to a ball around the uninfected steady state and remains there (i.e., practical stability). We observe that, for the disturbance-free Case (I), the size of the ball is very small due to small numerical errors. For cases (II) and (III), the size of the ball becomes larger and larger by increasing the bounds of the disturbances.
The basic properties of the 6-dimensional model that describes the interaction of the HIV with two target cells, CD4+ T cells, and macrophages and takes into account the Cytotoxic T Lymphocytes (CTL) immune response were studied. The HIV infection model was incorporated to allow some additive disturbances. The HIV infection model incorporates the effect of HAART is considered as a nonlinear control system, where the control input is defined to be dependent on the drug dose and drug efficiency. The proposed MPC method is applied for determining HAART schedules and stabilizing the HIV infection system around the uninfected steady state. The inherent robustness properties of the MPC were established.
This paper was funded by the Deanship of Scientific Research (DSR), King Abdulaziz University, Jeddah. The authors, therefore, acknowledge with thanks DSR technical and financial support.
- M. A. Nowak and R. M. May, Virus Dynamics: Mathematical Principles of Immunology and Virology, Oxford University Press, Oxford, UK, 2000.
- M. Dehghan, M. Nasri, and M. R. Razvan, “Global stability of a deterministic model for HIV infection in vivo,” Chaos, Solitons & Fractals, vol. 34, no. 4, pp. 1225–1238, 2007.
- A. M. Elaiw, “Global properties of a class of HIV models,” Nonlinear Analysis, vol. 11, no. 4, pp. 2253–2263, 2010.
- P. Georgescu and Y.-H. Hsieh, “Global stability for a virus dynamics model with nonlinear incidence of infection and removal,” SIAM Journal on Applied Mathematics, vol. 67, no. 2, pp. 337–353, 2006.
- A. Korobeinikov, “Global properties of basic virus dynamics models,” Bulletin of Mathematical Biology, vol. 66, no. 4, pp. 879–883, 2004.
- P. D. Leenheer and H. L. Smith, “Virus dynamics: a global analysis,” SIAM Journal on Applied Mathematics, vol. 63, no. 4, pp. 1313–1327, 2003.
- A. S. Perelson, D. Kirschner, and R. De Boer, “Dynamic of HIV infection of CD4+ T cells,” Mathematical Biosciences, vol. 114, pp. 81–125, 1993.
- A. S. Perelson and P. W. Nelson, “Mathematical analysis of HIV-1 dynamics in vivo,” SIAM Review, vol. 41, no. 1, pp. 3–44, 1999.
- J. Prüss, R. Zacher, and R. Schnaubelt, “Global asymptotic stability of equilibria in models for virus dynamics,” Mathematical Modelling of Natural Phenomena, vol. 3, no. 7, pp. 126–142, 2008.
- X. Shi, X. Zhou, and X. Song, “Dynamical behavior of a delay virus dynamics model with CTL immune response,” Nonlinear Analysis, vol. 11, no. 3, pp. 1795–1809, 2010.
- X. Song and A. U. Neumann, “Global stability and periodic solution of the viral dynamics,” Journal of Mathematical Analysis and Applications, vol. 329, no. 1, pp. 281–297, 2007.
- M. O. Souza and J. P. Zubelli, “Global stability for a class of virus models with cytotoxic T-Lymphocyte immune response and antigenic variation,” Bulletin of Mathematical Biology, vol. 73, no. 3, pp. 609–625, 2011.
- L. Wang and M. Y. Li, “Mathematical analysis of the global dynamics of a model for HIV infection of T cells,” Mathematical Biosciences, vol. 200, no. 1, pp. 44–57, 2006.
- R. Xu, “Global stability of an HIV-1 infection model with saturation infection and intracellular delay,” Journal of Mathematical Analysis and Applications, vol. 375, no. 1, pp. 75–81, 2011.
- A. S. Perelson, P. Essunger, Y. Cao et al., “Decay characteristics of HIV-1- infected compartments during combination therapy,” Nature, vol. 387, pp. 188–191, 1997.
- D. S. Callaway and A. S. Perelson, “HIV-1 infection and low steady state viral loads,” Bulletin of Mathematical Biology, vol. 64, pp. 29–64, 2002.
- B. M. Adams, H. T. Banks, M. Davidian et al., “HIV dynamics: modeling, data analysis, and optimal treatment protocols,” Journal of Computational and Applied Mathematics, vol. 184, no. 1, pp. 10–49, 2005.
- B. M. Adams, H. T. Banks, H.-D. Kwon, and H. T. Tran, “Dynamic multidrug therapies for HIV: optimal and STI control approaches,” Mathematical Biosciences and Engineering, vol. 1, no. 2, pp. 223–241, 2004.
- H. T. Banks, H.-D. Kwon, J. A. Toivanen, and H. T. Tran, “A state-dependent Riccati equation-based estimator approach for HIV feedback control,” Optimal Control Applications and Methods, vol. 27, no. 2, pp. 93–121, 2006.
- A. M. Elaiw and X. Xia, “HIV dynamics: analysis and robust multirate MPC-based treatment schedules,” Journal of Mathematical Analysis and Applications, vol. 359, no. 1, pp. 285–301, 2009.
- A. M. Elaiw, I. A. Hassanien, and S. A. Azoz, “Global stability of HIV infection models with intracellular delays,” Journal of the Korean Mathematical Society, vol. 49, no. 4, pp. 779–794, 2012.
- A. M. Elaiw and S. A. Azoz, “Global properties of a class of HIV infection models with Beddington-DeAngelis functional response,” Mathematical Methods in the Applied Sciences. In press.
- A. M. Elaiw, “Global properties of a class of virus infection models with multitarget cells,” Nonlinear Dynamics, vol. 69, pp. 423–435, 2012.
- A. M. Elaiw and M. A. Alghamdi, “Global properties of virus dynamics models with multitarget cells and discrete-time delays,” Discrete Dynamics in Nature and Society, vol. 2011, Article ID 201274, 19 pages, 2011.
- R. V. Culshaw, S. Ruan, and R. J. Spiteri, “Optimal HIV treatment by maximising immune response,” Journal of Mathematical Biology, vol. 48, no. 5, pp. 545–562, 2004.
- K. R. Fister, S. Lenhart, and J. S. McNally, “Optimizing chemotherapy in an HIV model,” Electronic Journal of Differential Equations, vol. 1998, pp. 1–12, 1998.
- H. R. Joshi, “Optimal control of an HIV immunology model,” Optimal Control Applications and Methods, vol. 23, no. 4, pp. 199–213, 2002.
- D. Kirschner, S. Lenhart, and S. Serbin, “Optimal control of the chemotherapy of HIV,” Journal of Mathematical Biology, vol. 35, no. 7, pp. 775–792, 1997.
- H. Shim, S. J. Han, I. S. Jeong, C. C. Chung, S. W. Nam, and J. H. Seo, “Optimal scheduling of drug treatment for HIV infection: continuous dose control and receding horizon control,” International Journal of Control, Automation and Systems, vol. 1, pp. 401–407, 2003.
- A. M. Elaiw, “Receding horizon control method applied to antiviral treatment of AIDS,” Miskolc Mathematical Notes, vol. 5, no. 2, pp. 173–186, 2004.
- A. M. Elaiw, K. Kiss, and M. A. L. Caetano, “Stabilization of HIV/AIDS model by receding horizon control,” Journal of Applied Mathematics and Computing, vol. 18, no. 1-2, pp. 95–112, 2005.
- R. Zurakowski and A. R. Teel, “A model predictive control based scheduling method for HIV therapy,” Journal of Theoretical Biology, vol. 238, no. 2, pp. 368–382, 2006.
- É. Gyurkovics and A. M. Elaiw, “A stabilizing sampled-data -step receding horizon control with application to a HIV/AIDS model,” Differential Equations and Dynamical Systems, vol. 14, no. 3-4, pp. 323–352, 2006.
- D. Q. Mayne, J. B. Rawlings, C. V. Rao, and P. O. M. Scokaert, “Constrained model predictive control: stability and optimality,” Automatica, vol. 36, no. 6, pp. 789–814, 2000.
- F. A. C. C. Fontes, “A general framework to design stabilizing nonlinear model predictive controllers,” Systems & Control Letters, vol. 42, no. 2, pp. 127–143, 2001.
- E. Gyurkovics and A. M. Elaiw, “Stabilization of sampled-data nonlinear systems by receding horizon control via discrete-time approximations,” Automatica, vol. 40, no. 12, pp. 2017–2028, 2004.
- A. M. Elaiw, “Multirate sampling and input-to-state stable receding horizon control for nonlinear systems,” Nonlinear Analysis, vol. 67, no. 5, pp. 1637–1648, 2007.
- M. A. L. Caetano and T. Yoneyama, “Short and long period optimization of drug doses in the treatment of AIDS,” Anais da Academia Brasileira de Ciencias, vol. 74, pp. 589–597, 2002.
- X. Wang, A. M. Elaiw, and X. Song, “Global properties of a delayed HIV infection model with CTL immune response,” Applied Mathematics and Computation, vol. 218, pp. 9405–9414, 2012.
- E. Gyurkovics and A. M. Elaiw, “Conditions for MPC based stabilization of sampled-data nonlinear systems via discrete-time approximations,” in Lecture Notes in Control and Information Sciences, vol. 358, pp. 35–48, 2007.
- A. M. Elaiw, “Stabilizing output feedback receding horizon control of sampled-data nonlinear systems via discrete-time approximations,” Bulletin of Mathematical Analysis and Applications, vol. 2, no. 1, pp. 42–55, 2010.