In this paper we propose the global dynamics of an SIRI epidemic model
with latency and a general nonlinear incidence function. The model is based
on the susceptible-infective-recovered (SIR) compartmental structure with
relapse (SIRI). Sufficient conditions for the global stability of equilibria (the
disease-free equilibrium and the endemic equilibrium) are obtained by means
of Lyapunov-LaSalle theorem. Also some numerical simulations are given to
illustrate this result.
1. Introduction
Epidemic models have long been an important tool for understanding and controlling the spread of infectious diseases. Most of them are described by delayed differential equations. The introduction of time delay is often used to model the latent period, that is, the time from the acquisition of infection to the time when the host becomes infectious [1, 2].
Recently, considerable attention has been paid to model the relapse phenomenon, that is, the return of signs and symptoms of a disease after a remission. Hence, the recovered individual can return to the infectious class (see [3–6]). For the biological explanations of the relapse phenomenon, we cite two examples.
For malaria, Bignami [7] proposed that relapses derived from persistence of small numbers of parasite in the blood. Also, it has been observed that the proportion of patients who have successive relapses is relatively constant (see [8]).
For tuberculosis, relapse can be caused by incomplete treatment or by latent infection, being observed that HIV-positive patients are significantly more likely to relapse than HIV-negative patients, although it is often difficult to differentiate relapse from reinfection (see [9]).
In this paper, we propose the following epidemic model with time delay and relapse (delayed SIRI epidemic model) as follows (see [10, 11]):
(1)dSdt=A-μS-fS,I,dIdt=e-μτfSτ,Iτ-μ+γ+αI+δR,dRdt=γI-μ+δR.
The initial condition for the above system is
(2)Sθ=φ1θ,Iθ=φ2θ,Rθ=φ3θ,iiθ∈-τ,0,
with φ=(φ1,φ2,φ3)∈C+×C+×C+, such that φi(θ)≥0 (-τ≤θ≤0, i=1,2,3). Here C denotes the Banach space C([-τ,0],R) of continuous functions mapping the interval [-τ,0] into R, equipped with the supremum norm. The nonnegative cone of C is defined as C+=C([-τ,0],R+).
Here ψτ=ψ(t-τ) for any given function ψ, A=μN, where N=S+I+R is the total number of population, S is the number of susceptible individuals, I is the number of infectious individuals, R is the number of recovered individuals, A is the recruitment rate of the population, μ is the natural death of the population, α is the death rate due to disease, f is the nonlinear incidence function, γ is the recovery rate of the infective individuals, δ is the rate that recovered individuals relapse and regained infectious class, and τ is the latent period.
In model (1) the incidence function f(S,I) is a locally Lipschitz continuous function on R+×R+ satisfying f(0,I)=f(S,0)=0 for S≥0, I≥0, and the following hold:
f is a strictly monotone increasing function of S≥0, for any fixed I>0, and f is a strictly monotone increasing function of I≥0, for any fixed S≥0.
ϕ(S,I)=f(S,I)/I is a bounded and monotone decreasing function of I>0, for any fixed S≥0, and K(S)=limI→0+ϕ(S,I) is a continuous and monotone increasing function on S≥0.
This incidence function includes different forms presented in literature (see, e.g., [12–23]).
System (1) always has a disease-free equilibrium P0=(A/μ,0,0). On the other hand, under the hypothesis (H2), if
(3)R0≔KA/μe-μτμ+γ+α-γδ/μ+δ>1,
then system (1) also admits a unique endemic equilibrium P*=(S*,I*,R*), where S*, I*, and R* satisfy the following system (see [24]):
(4)iiiiiiiiiiA-μS-fS,I=0,e-μτfS,I-μ+γ+αI+δR=0,iiiiiiiiiiiiγI-μ+δR=0.
Hereafter, we replace μ+γ+α-γδ/(μ+δ) by η.
In [5] (1990), Tudor developed and analyzed qualitatively one of the first SIRI epidemic models for the spread of a herpes-type infection in either human or animal populations. This model consists of a system of nonlinear ordinary differential equations with a bilinear incidence rate (i.e., f(S,I)=βSI) and a constant total population (i.e., N=S+I+R=constant).
In [25] (1997), Moreira and Wang extended a Tudor-model to include nonlinear incidence functions. By using an elementary analysis of Liénard’s equation and Lyapunov’s direct method, they derived sufficient conditions for the global asymptotic stability of the disease-free and endemic equilibria.
In [23] (2000), Castillo-Garsow et al. considered an SIRI model for drug use in a population of adolescents. The authors assumed that N=constant and f(S,I)=βSI/N; they estimated the parameters of the model and they determined a rough approximation of the basic reproductive number. Based on these parameters, they performed some simulations that clearly showed the endemic character of tobacco use among adolescents.
In [26] (2004), Blower developed a compartmental model for genital herpes, assuming standard incidence rate (i.e., f(S,I)=βSI/N) and constant recruitment rate.
In [24] (2006), Korobeinikov proved that the endemic equilibrium of the model (1) with δ=0 and τ=0 is globally asymptotically stable.
In [27] (2007), van den Driessche and Zou proposed an integrodifferential equation to model a general relapse phenomenon in infectious diseases. The resulting model, in particular case, is a delay differential equation with a constant population and standard incidence. The basic reproduction number for this model is identified and some global results are obtained by employing the Lyapunov-Razumikhin technique.
In [10] (2007), Van Den Driessche and co-authors formulated a delay differential SIRI model (System (1) with f(S,I)=βSI). For this system, the endemic equilibrium is locally asymptotically stable if R0>1, and the disease is shown to be uniformly persistent with the infective population size either approaching or oscillating about the endemic level.
In [22] (2011), Liu et al. proposed a mathematical model for a disease with a general exposed distribution, the possibility of relapse and nonlinear incidence rate (f(S,I)=βS(I/N)). By the method of Lyapunov functionals, they showed that the disease dies out if R0=1 and that the disease becomes endemic if R0>1. They also analyzed, as a special case of this model, the system (1) with fS,I=gS·I; the result confirms that the endemic equilibrium is globally asymptotically stable.
In [28] (2012), Abta et al. considered a global asymptotic stability of a delayed SIR model (system (1) with f(S,I)=βSI/(1+α1S+α2I) and δ=0).
In [29] (2013), Vargas-De-Leon presented the global stability conditions of an ordinary SIRI model with bilinear and standard incidence rates, respectively, that includes recruitment rate of susceptible individuals into the community and that the disease produces nonnegligible death in the infectious class. The author presented the construction of Lyapunov functions using suitable combinations of known functions, common quadratic and Volterra-type, and a composite Volterra-type function.
In [21] (2013), Georgescu and Zhang analyzed the dynamics of an ordinary SIRI model under the assumption that the incidence of infection is given by f(S,I)=C(S)g(I). They obtained by means of Lyapunov’s second method sufficient conditions for the local stability of equilibria and they showed that global stability can be attained under suitable monotonicity conditions.
In [30] (2013), Shuai and van den Driessche presented two systematic methods for the construction of Lyapunov functions for general infectious disease models (Ordinary SEIRI, SIS, etc.). Specifically, a matrix-theoretic method using the Perron eigenvector is applied to prove the global stability of the disease-free equilibrium, while a graph-theoretic method based on Kirchhoff’s matrix tree theorem and two new combinatorial identities are used to prove the global stability of the endemic equilibrium.
In [31] (2014), Xu investigated a delayed SIRI model (system (1) with f(S,I)=βSI). The author established the global stability of a disease-free equilibrium and an endemic equilibrium by means of suitable Lyapunov functionals and LaSalle’s invariance principle.
In this paper we extend the global stability results presented in [31] to a delayed SIRI epidemic model (system (1)) with a general nonlinear incidence function. It is shown that global stability can be attained under suitable monotonicity conditions and it is established that the basic reproduction number R0 is a threshold parameter for the stability of a delayed SIRI model. The rest of the paper is organized as follows. In Section 2, the global stability of disease-free and endemic equilibria are established. In Section 3, numerical simulations and concluding remarks are provided. In the appendix, some results on the global stability are stated.
2. Global Stability Analysis of Delayed SIRI Model
In this section, we discuss the global stability of a disease-free equilibrium P0 and an endemic equilibrium P* of system (1). Since (d/dt)(S+I+R)≤A-μ(S+I+R), we have limsup(S+I+R)≤A/μ. Hence we discuss system (1) in the closed set
(5)Ω=∶φ1,φ2,φ3∈C+×C+×C+:φ1+φ2+φ3≤Aμ.
It is easy to show that Ω is positively invariant with respect to system (1). Next we consider the global asymptotic stability of the disease-free equilibrium P0 and the endemic equilibrium P* of (1) by Lyapunov functionals, respectively.
Proposition 1.
If R0≤1, then the disease-free equilibrium P0 is globally asymptotically stable.
Proof.
Define a Lyapunov functional
(6)V0t=e-μτ∫A/μSτ1-KA/μKudu+I+η∫t-τtIudu+δμ+δR.
We will show that dV0(t)/dt≤0 for all t≥0. We have
(7)dV0tdt=e-μτ1-KA/μKSτS˙τ+e-μτfSτ,IτdτW-μ+γ+αI+δR+ηI-IτW+γδμ+δI-δR=e-μτ1-KA/μKSτA-μSτW+KA/μKSτe-μτfSτ,Iτ-ηIτ=μe-μτ1-KA/μKSτAμ-SτW+ηIτϕSτ,IτηKA/μKSτe-μτ-1.
By the hypothesis (H1), we obtain that
(8)1-KA/μKSτAμ-Sτ≤0,
where equality holds if and only if S=A/μ.
Furthermore, It follows from the hypothesis (H2) that
(9)ϕSτ,IτηKA/μe-μτKSτ≤KSτηKA/μe-μτKSτ≤KA/μe-μτη≤R0.
Therefore, R0≤1 ensures that dV0(t)/dt≤0 for all t≥0, where dV0(t)/dt=0 holds if (S,I,R)=(A/μ,0,0). Hence, it follows from system (1) that {P0} is the largest invariant set in S,I,R∣dV0(t)/dt=0. From the Lyapunov-LaSalle asymptotic stability, we obtain that P0 is globally asymptotically stable. This completes the proof.
Proposition 2.
If R0>1, then the endemic equilibrium P1* is globally asymptotically stable.
Proof.
To prove global stability of the endemic equilibrium, we define a Lyapunov functional V(t)=V1(t)+V2(t)+V3(t)+V+, with
(10)V1t=S-S*-∫S*SfS*,I*fu,I*du,V2t=eμτI-I*-I*lnII*,V3t=δeμτμ+δR-R*-R*lnRR*,V+=∫t-τtfS,I-fS*,I*-fS*,I*fSu,IufS*,I*WWWWW·lnfSu,IufS*,I*du.
The time derivative of the function V(t) along the positive solution of system (1) is
(11)dVtdt=1-fS*,I*fS,I*A-μS-fS,IW+eμτ1-I*IW·e-μτfSτ,Iτ+δR-μ+α+γIW+δeμτμ+δ1-R*RγI-μ+δRW+fS,I-fSτ,Iτ+fS*,I*W·lnfSτ,IτfS,I.
Using the relation A=μS*+f(S*,I*), simple calculations give
(12)dVtdt=1-fS*,I*fS,I*-μS-S*+fS*,I*+fS*,I*fS,I*fS,I-eμτμ+α+γI-I*IfSτ,Iτ-δeμτI*IR+eμτμ+α+γI*+δeμτμ+δγI-γIR*R+μ+δR*+fS*,I*lnfSτ,IτfS,I.
Here by using
(13)eμτμ+α+γI*-δeμτμ+δγI*=fS*,I*,μ+δR*=γI*,lnfSτ,IτfS,IW=lnfS*,I*fS,I*+lnI*fSτ,IτIfS*,I*+lnIfS,I*I*fS,I,
straightforward calculations give
(14)dVtdt=-μ1-fS*,I*fS,I*S-S*-fS*,I*fS*,I*fS,I*-1-lnfS*,I*fS,I*-fS*,I*I*fSτ,IτIfS*,I*-1-lnI*fSτ,IτIfS*,I*+fS*,I*fS,IfS,I*-II*+IfS,I*I*fS,I-1-fS*,I*IfS,I*I*fS,I-1-lnIfS,I*I*fS,I+δγI*eμτμ+δ2-I*RIR*-IR*I*R=-μ1-fS*,I*fS,I*S-S*-fS*,I*fS*,I*fS,I*-1-lnfS*,I*fS,I*-fS*,I*I*fSτ,IτIfS,I*-1-lnI*fSτ,IτIfS,I*+II*fS*,I*1-fS,I*fS,IϕS,IϕS,I*-1-fS*,I*IfS,I*I*fS,I-1-lnIfS,I*I*fS,I-δγeμτμ+δIR*RI*RIR*-12.
It follows from (H1) and (H2) that
(15)-μ1-fS*,I*fS,I*S-S*≤0,II*fS*,I*1-fS,I*fS,IϕS,IϕS,I*-1W≤0.
Furthermore, since the function g(x)=1-x+ln(x) is always nonpositive for any x>0, and g(x)=0 if and only if x=1, then dV(t)/dt≤0, for all t≥0, where the equality holds only at the equilibrium point (S*,I*,R*). Hence, the functional V satisfies all the conditions of Theorem A.2. This proves that P1* is globally asymptotically stable.
3. Numerical Simulations and Concluding Remarks
In this section, we give a numerical simulation supporting the theoretical analysis given in Section 2. Let
(16)fS,I=βSI1+α1S+α2I.
We take the parameters of the system (1) as follows:
(17)A=10,α1=0.9,α2=0.9,μ=0.01,γ=0.02,α=0.005,β=0.1,δ=0.001,τ=10.
By Proposition 2, the endemic equilibrium P* is globally asymptotically stable; see Figure 1.
Solutions (S,I,R) of the SIRI epidemic model (1) are globally asymptotically stable and converge to the endemic equilibrium P*.
In this paper, we presented a mathematical analysis and numerical simulations for an SIRI epidemiological model applied to the evolution of the spread of disease with relapse in a given population. We denote R0 the basic reproduction number. It is defined as the average number of contagious persons infected by a typical infectious in a population of susceptible. We prove in this paper that the basic reproduction number, R0, depends on the incubation period and we show that the disease-free equilibrium P0 is globally asymptotically stable if R0≤1 and that a unique endemic equilibrium P* is globally asymptotically stable if R0>1.
AppendixThe Lyapunov-LaSalle Theorem
In the following, we present the method of Lyapunov functionals in the context of a delay differential equations:
(A.1)dxdt=fxt,
where f:C→Rn is completely continuous and solutions of (A.1) are unique and continuously dependent on the initial data. We denote by x(ϕ) the solution of (A.1) through (0,ϕ). For a continuous functional V:C→R, we define
(A.2)V˙=limsuph→0+1hVxhϕ-Vϕ,
the derivative of V along a solution of (A.1). To state the Lyapunov-LaSalle type theorem for (A.1), we need the following definition.
Definition A.1 (see [32, page 30]).
We say V:C→R is a Lyapunov functional on a set G in C for (A.1) if it is continuous on G¯ (the closure of G) and V˙≤0 on G. We also define E={ϕ∈G¯:V˙(ϕ)=0}, and M is the largest set in E which is invariant with respect to (A.1).
The following result is the Lyapunov-LaSalle type theorem for (A.1).
Theorem A.2 (see [32, page 30]).
If V is a Lyapunov functional on G and xt(ϕ) is a bounded solution of (A.1) that stays in G, then ω-limit set ω(ϕ)⊂M; that is, xt(ϕ)→M as t→+∞.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
CookeK. L.Stability analysis for a vector disease model197991314210.1216/RMJ-1979-9-1-31ZBL0423.920292-s2.0-0003100558HethcoteH. W.van den DriesscheP.An SIS epidemic model with variable population size and a delay199534217719410.1007/BF001787722-s2.0-0029445684ChinJ.1999Washington, DC, USAAmerican Public Health AssociationMartinS. W.1994Washington, DC, USANational Academy PressTudorD.A deterministic model for herpes infections in human and animal populations199032113613910.1137/1032003VanLandinghamK. E.MarstellerH. B.RossG. W.HaydenF. G.Relapse of herpes simplex encephalitis after conventional acyclovir therapy198825971051105310.1001/jama.1988.037200700510342-s2.0-0023846305BignamiA.Concerning the pathogenesis of relapses in malarial fevers19136798910.1097/00007611-191302000-00002WhiteN. J.Determinants of relapse periodicity in Plasmodium vivax malaria201110, article 29710.1186/1475-2875-10-2972-s2.0-80053624564CoxH.KebedeY.AllamuratovaS.IsmailovG.DavletmuratovaZ.ByrnesG.StoneC.NiemannS.Rüsch-GerdesS.BlokL.DoshetovD.Tuberculosis recurrence and mortality after successful treatment: impact of drug resistance20063101836184310.1371/journal.pmed.00303842-s2.0-33750894141Van Den DriesscheP.WangL. I. N.ZouX.Modeling diseases with latency and relapse20074220521910.3934/mbe.2007.4.205ZBL1123.920182-s2.0-34547313798KaddarA.On the dynamics of a delayed SIR epidemic model with a modified saturated incidence rate200913317MR2558805GaoS.ChenL.NietoJ. J.TorresA.Analysis of a delayed epidemic model with pulse vaccination and saturation incidence20062435-366037604510.1016/j.vaccine.2006.05.0182-s2.0-33746833233YorkeJ. A.LondonW. P.Recurrent outbreaks of measles, chickenpox and mumps: II19739864694822-s2.0-0015711463GomesM. G. M.WhiteL. J.MedleyG. F.The reinfection threshold2005236111111310.1016/j.jtbi.2005.03.0012-s2.0-20444502433ZhouY.LiuH.Stability of periodic solutions for an SIS model with pulse vaccination2003383-429930810.1016/S0895-7177(03)90088-42-s2.0-0141615743AndersonR. M.MayR. M.Regulation and stability of host-parasite population interactions: I. Regulatory processes1978471219267ZhangF.LiZ.-Z.Global stability of an SIR epidemic model with constant infectious period2008199128529110.1016/j.amc.2007.09.0532-s2.0-41949125685JiangZ.WeiJ.Stability and bifurcation analysis in a delayed SIR model200835360961910.1016/j.chaos.2006.05.0452-s2.0-34548618836CapassoV.SerioG.A generalization of the Kermack-McKendrick deterministic epidemic model1978421-2436110.1016/0025-5564(78)90006-8ZBL0398.920262-s2.0-0018041874XuR.MaZ.Stability of a delayed SIRS epidemic model with a nonlinear incidence rate20094152319232510.1016/j.chaos.2008.09.0072-s2.0-67649304819GeorgescuP.ZhangH.A Lyapunov functional for a SIRI model with nonlinear incidence of infection and relapse2013219168496850710.1016/j.amc.2013.02.044ZBL1288.920192-s2.0-84876139883LiuS.WangS.WangL.Global dynamics of delay epidemic models with nonlinear incidence rate and relapse201112111912710.1016/j.nonrwa.2010.06.0012-s2.0-77958019832Castillo-GarsowC.Jordan-SaliviaG.Rodriguez HerreraA.Mathematical models for the dynamics of tobacco use, recovery, and relapse2000BU-1505-MCornell UniversityKorobeinikovA.Lyapunov functions and global stability for SIR and SIRS epidemiological models with non-linear transmission200668361562610.1007/s11538-005-9037-92-s2.0-33746594608MoreiraH. N.WangY.Global stability in an S→I→R→I model199739349650210.1137/S0036144595295879MR14699472-s2.0-0031234909BlowerS.Modelling the genital herpes epidemic200411, supplement 3138A146Avan den DriesscheP.ZouX.Modeling relapse in infectious diseases200720718910310.1016/j.mbs.2006.09.0172-s2.0-34147109285AbtaA.KaddarA.AlaouiH. T.Global stability for delay SIR and SEIR epidemic models with saturated incidence rates20122012231132-s2.0-84857158981Vargas-De-LeonC.On the global stability of infectious diseases models with relapse201395061ShuaiZ.van den DriesscheP.Global stability of infectious disease models using Lyapunov functions20137341513153210.1137/120876642ZBL062271762-s2.0-84884941404MR3077907XuR.Global dynamics of an SEIRI epidemiological model with time delay201423243644410.1016/j.amc.2014.01.1002-s2.0-84894025380KuangY.1993New York, NY, USAAcademic Press