946
Views
4
CrossRef citations to date
0
Altmetric
Articles

Theoretical and numerical results for an age-structured SIVS model with a general nonlinear incidence rate

&
Pages 789-816 | Received 14 Nov 2017, Accepted 20 Sep 2018, Published online: 14 Oct 2018

ABSTRACT

In this paper, we propose an SIVS epidemic model with continuous age structures in both infected and vaccinated classes and with a general nonlinear incidence. Firstly, we provide some basic properties of the system including the existence, uniqueness and positivity of solutions. Furthermore, we show that the solution semiflow is asymptotic smooth. Secondly, we calculate the basic reproduction number R0 by employing the classical renewal process, which determines whether the disease persists or not. In the main part, we investigate the global stability of the equilibria by the approach of Lyanpunov functionals. Some numerical simulations are conducted to illustrate the theoretical results and to show the effect of the transmission rate and immunity waning rate on the disease prevalence.

1. Introduction

Vaccination is one of the most effective methods of preventing infectious diseases. Indeed, for some children diseases like Measles, Rubella, and Chicken pox, preventive vaccines may provide a permanent immunity against the diseases. However, life-long immunity cannot be offered by preventive vaccines for some diseases such as Hepatitis B, Influenza, and Mumps. The immunity of vaccinated persons will wane and they will become vulnerable to the diseases again. It is necessary to design a framework to study the effect of waning immunity on the spread of an epidemic. Compartmental models have been developed to provide deep insights on the dynamics of an epidemic with non-permanent immunity (see, for example, [Citation5,Citation19,Citation29]).

It is strongly supported by data that a vaccine usually wanes with respect to the vaccinated time. Many scholars have successfully addressed this by adding a vaccinated compartment to classic epidemic models [Citation6,Citation10,Citation15,Citation18,Citation30,Citation31]. Obtained results include threshold dynamics [Citation6,Citation30,Citation31] and backward bifurcation [Citation18]. In this regard, the immunity duration has been becoming an important issue for the evolution and efficacy of the vaccine. Recently, we proposed an SIVS-type epidemic model with vaccination age in [Citation31] to explore the non-fixed immunity duration, (1) dS(t)dt=ΛμS(t)S(t)f(I(t))φS(t)+0ε(a)v(t,a)da,dI(t)dt=S(t)f(I(t))(μ+γ+δ)I(t),v(t,a)t+v(t,a)a=(μ+ε(a))v(t,a),v(t,0)=φS(t),S(0)=S00,I(0)=I0>0,v(0,)=v0()L+1L1(R+;R+),(1) where S(t) and I(t) denote the population sizes of the susceptible and infected at time t, respectively; v(t,a) denotes the population density of the vaccinated at time t with vaccination age a. The parameters have the following meanings: Λ is the input rate of the new members, μ denotes the natural death rate, φ represents the vaccinated rate for the susceptible, γ denotes the cure rate for infected individuals, δ represents the disease-caused death rate of infected individuals, ε(a) denotes the immunity waning rate at age a, f(I) represents the incidence rate and satisfies the following property: f(0)=0,f(I)>0,f(I)<0. We showed that system (Equation1) exhibits a threshold dynamics by constructing appropriate Lyapunov functionals. In this sense, the basic reproduction number R0 is a key value determining whether the disease dies out or persists.

As we know, the incidence rate is an important factor affecting the disease dynamics. In the above mentioned works, the incidences used include the bilinear-type (βSI) [Citation5,Citation6,Citation18,Citation19], the standard-type (βSI/(S+I)) [Citation15], the saturated-type Sf(I) [Citation30,Citation31], and the nonlinear-type SpIq [Citation29]. In the literature, Feng and Thieme firstly proposed a nonlinear general incidence of the form f(S,I) [Citation8]. Thereafter, Huang et al. [Citation13] and Korobeinikov [Citation17] studied some epidemic models with incidence rates of the form f(S)g(I). Furthermore, Korobeinikov [Citation16] obtained the global stability of basic SIR and SIRS epidemic models with the incidence rate of the form f(S,I).

Most classical epidemic models are compartmental models described by ordinary differential equations, where all infectious individuals are assumed to be homogeneous during their infectious period. This assumption has been proved to be reasonable in the study of the dynamics of communicable diseases such as influenza as well as in the study of sexually transmitted diseases. However, infectivity experiments on HIV/AIDS indicate that the transmission style follows an early infectivity peak (a few weeks after exposure) and a late infectivity plateau [Citation9]. To describe such a phenomenon, the concept of infection age (the time that has passed since infection) has been introduced into classical models. Therefore, epidemic models with infection age have been extensively studied in the literature. To name a few, see [Citation3,Citation4,Citation22,Citation25,Citation28,Citation32], where the incidence is bilinear in most of the models. Recently, epidemic models with infection age and nonlinear incidence have been extensively studied (see, for instance, [Citation4,Citation25,Citation28]).

To the best of our knowledge, not much has been done for epidemic models with two age structures [Citation7,Citation23]. In [Citation23], McCluskey considered an SEI model with continuous age structures in both the exposed and infectious classes and a threshold dynamics was established by using the approach of Lyapunov functionals while in [Citation7], Duan et al. studied an SVEIR epidemic model with ages of vaccination and latency and also obtained a threshold dynamics. Magal and McCluskey in [Citation21] proposed a two group SI epidemic model with age of infection and discussed the global stability of steady states.

Motivated by the above discussion, the purpose of this paper is to make further contribution to the study of epidemic models with two age structures and nonlinear incidence. Precisely, we introduce age structure into the infected individuals in (Equation1) and use a general incidence. Furthermore, let i(t,a) denote the density of infected individuals at time t with infection age a. The model to be studied is as follows, (2) dS(t)dt=ΛμS(t)f(S(t),J(t))φS(t)+0ε(a)v(t,a)da,i(t,a)t+i(t,a)a=(μ+γ(a)+δ(a))i(t,a),i(t,0)=f(S(t),J(t)),J(t)=0β(a)i(t,a)da,v(t,a)t+v(t,a)a=(μ+ε(a))v(t,a),v(t,0)=φS(t),(2) with initial condition S(0)=S00,i(0,)=i0()L+1,v(0,)=v0()L+1. Here Λ, μ, φ, and ε(a) have the same biological meanings as in those (Equation1). For the other parameters, γ(a) is the recovery rate of the infected with infection age a, δ(a) is the disease-induced death rate with infection age a, and β(a) is the transmission coefficient with infection age a. In epidemiology, J(t) is called the force of infection, which justifies the form of incidence f(S(t),J(t)). Note that the models in [Citation4,Citation6,Citation7,Citation22,Citation31] are just special cases of (Equation2) and hence over results will cover those in the above-mentioned references.

Throughout this paper, we make the following assumptions on the parameter functions.

(A1)

The functions ϵ, βCBU(R+,R+), where CBU(R+,R+) is the set of all bounded and uniformly continuous functions from R+ to R+.

(A2)

The functions γ, δL+(R+), the nonnegative cone of L(R+).

Moreover, we suppose that the nonlinear incidence f satisfies:

(B1)

For S, JR+, f(0,J)=f(S,0)=0; f(S,J)/S>0 for J>0; f(S,J)/J>0 for S>0; and f2(S,J)/J20.

Assumption (B1) is a combination of those in [Citation17,Citation25]. It is easy to see that f is locally Lipschitz continuous on S and J, that is, for every C>0, there exists some L:=LC>0 such that (3) f(S1,J1)f(S2,J2)L(|S1S2|+|J1J2|),(3) whenever 0S1, S2, J1, J2C.

The pahse space of (Equation2) is X=R+×L+1×L+1. For (S0,i0,v0)X, we denote (S0,i0,v0)X=S0+i01+v01. In order to study the existence of solutions to (Equation2), we will extend X.

Let Y=R×L1(R+;R), Y0={0}×L1(R+;R), Y+=R+×L+1, and Y+0=Y+Y0. Define two linear operators Aj:D(Aj)YY (j=1, 2) as follows. (4) A10φ1=φ1(0)φ1(μ+γ(a)+δ(a))φ1,D(A1)=0φ1Y|φ is absolutely continuous andφ1L1(R+;R),(4) and (5) A20φ2=φ2(0)φ2(μ+ε(a))φ2,D(A2)=0φ2Y|φ2 is absolutely continuous andφ2L1(R+;R).(5) For any λρ(A1) (ρ(A1) denotes the resolvent set of A1) and 0φ1D(A1), if (λIA1)1α1θ1=0φ1,α1θ1Y, then by simple calculation we obtain (6) φ1(a)=α1e(λ+μ)a0a(γ(ξ)+δ(ξ))dξ+0aθ1(s)e(λ+μ)(as)esa(γ(ξ)+δ(ξ))dξds.(6) Similarly, for any λρ(A2) and 0φ2D(A2), if (λA2)1α2θ2=0φ2 with α2θ2Y, then (7) φ2(a)=α2e(λ+μ)a0aε(ξ)dξ+0aθ2(s)e(λ+μ)(as)esaε(ξ)dξds.(7) Define a linear operator A=diag{μ,A1,A2} with D(A)=R×D(A1)×D(A2). It is easy to see that D(A)¯=R×{0}×L1(R+;R)×{0}×L1(R+;R). For convenience, denote Ω:={λC:Re(λ)>μ}. It follows from the definition of A that it has the following properties.

Lemma 1.1

If λΩ then λρ(A). More precisely, for any λC with Re(λ)>μ, any ψ0,α1ψ1, α2ψ2X,andφ0,0φ1, 0φ2D(A), we have (λIA)1ψ0,α1ψ1, α2ψ2=φ0,0φ1, 0φ2 if and only if (8) φ1(a)=e0a(λ+μ+γ(l)+δ(l))dlα1+0aesa(λ+μ+γ(l)+δ(l))dlψ1(s)ds,(8) (9) φ2(a)=e0a(λ+μ+ε(l))dlα2+0aesa(λ+μ+ε(l))dlψ2(s)ds,(9) (10) φ0=ψ0λ+μ.(10) Moreover, A is a Hille-Yosida operator and (λIA)nM(Re(λ)+μ)nfor\ Re(λ)>μ and\ n1.

Proof.

By Equation (Equation6) and (Equation7), we immediately obtain (Equation8) and (Equation9). It follows from the definition of A that (λ+μ)φ0=ψ0 and hence φ0=ψ0λ+μ. For any λC with Re(λ)>μ, we have (λA)1ψ0,α1ψ1, α2ψ2X=φ0,0φ1(a), 0φ2(a)X|ψ0||λ+μ|+0e0a(λ+μ+γ(l)+δ(l))dlda|α1|+00aesa(λ+μ+γ(l)+δ(l))dlψ1(s)dsda+0e0a(λ+μ+ε(l))dlda|α2|+00aesa(λ+μ+ε(l))dlψ2(s)dsda|ψ0|+|α1|+|α1||λ+μ|+0e(λ+μ)s(|ψ1(s)|+|ψ2(s)|)s|e(λ+μ)a|dads1|Re(λ)+μ|(|ψ0|+|α1|+ψ1L1+|α2|+ψ2L1). The results immediately follow.

Let Xˆ=R+×Y+×Y+. Define a nonlinear operator F:D(A)¯XX by (11) F(φ):=Λpφ0+0ε(a)φ2(a)daB(φ0,φ1)B(φ0,φ1)0pφ00,(11) where B(φ0,φ1)=f(φ0,0β(a)φ1(a)da). Then, setting u=S,0i,0vXˆ, we can rewrite (Equation2) as an abstract Cauchy problem in Xˆ (12) du(t)dt=Au(t)+F(u(t)),u(0)=u0Xˆ.(12) It follows from Lemma 1.1 and (Equation3), together with Lemma 3.1 [Citation20] that (Equation12) has a unique continuous solution if the initial condition u0=S0,0i0,0v0Xˆ satisfies the compatibility condition i0(0)=f(S0,J0),J0=0β(a)i0(a)da,v0(0)=φS0. Hence, for each (S0,i0,v0)Xˆ satisfying the above coupling condition, (Equation2) has a unique solution in X. Then we can define a solution semiflow Φ:R+×XX of (Equation2) by Φ(t,u0)=u(t)fortR+, where u(t) is the unique solution of (Equation2) with the initial condition u0X.

Let N(t)=S(t)+i(t,)1+v(t,)1. Then it follows from (Equation2) that (13) dN(t)dtΛμN(t).(13) From this, we can easily deduce that Γ is a positively invariant and attracting set of the semiflow Φ, where Γ=(S0,i0,v0)Xi0(0)=f(S0,J0),J0=0β(a)i0(a)da,v0(0)=φS0,(S0,i0,v0)XΛμ. The aim of this paper is to establish the global dynamics of (Equation2). As a result, we only need to consider (Equation2) with initial conditions in Γ. The following estimates are easy to obtain.

Theorem 1.2

Let Assumption (B1) hold. For all u0Γ, the following statements are true.

  1. S(t)+0i(t,a)da+0v(t,a)daΛ/μ, and J(t)β(Λ/μ), for all tR+,

  2. liminftS(t)Λ/(μ+φ+L), and φΛε/(μ+φ+L)0ε(a)v(t,a)daε(Λ/μ), where L is the Lipschitz coefficient.

Proof.

We readily obtain (i) by the positivity of the solution and Equation (Equation13). By Assumption (B1), we conclude that f(S(t),J(t))LS(t). Substituting this inequality into S equation of (Equation2), we have S(t)Λ(μ+φ+L)S(t). From Fluctuate Lemma, it follows that there exists a sequence tn such that S(tn)0 and S(tn)S:=liminftS(t). Then SΛ/(μ+φ+L). Integrating the third equation of (Equation2) along the characteristic line yields (14) v(t,a)=φS(ta)π(a),ta,v0(at)π(a)π(at),t<a,(14) where π(a)=e0a(μ+ε(s))ds denotes the probability of a vaccinated individual having immunity until age a. Therefore, (15) 0ε(a)v(t,a)da=φ0tε(a)S(ta)da+tε(a)v0(at)π1(a)π1(at).(15) Taking limit inferior on both sides of (Equation15) (16) lim inft0ε(a)v(t,a)daφεS.(16) On the other hand, 0ε(a)v(t,a)daε(Λ/μ). This completes the proof.

Corollary 1.3

Suppose Assumptions (A1) (A2) and B(1) hold. Then the semiflow Φ is point dissipative. In fact, there is a bounded set that attracts all points in X.

The rest of this paper is organized as follows. In Section 2, we establish the asymptotic smoothness of the semiflow Φ. Then we study the existence and local stability of equilibria in Section 3. Before obtaining the main result, a threshold dynamics of (Equation2), in Section 5, we show the uniform persistence in Section 4. In Section 6, we provide some numerical simulations to demonstrate the main results and to analyze the effect of the transmission rate and immunity waning rate on the disease prevalence. The paper concludes with a brief discussion.

2. Asymptotic smoothness

In this section, we establish the asymptotic smoothness of the solution semiflow Φ. For any closed, bounded, and positively invariant set BΓ, we need to show that there exists a compact set KΓ such that dH(Φ(t,B),K)0 as t, where dH is the Hausdorff semi-distance (see, for example, [Citation11]).

Before proceeding, we get the expressions of i and v as follows by integrating along the characteristic lines, i(t,a)=f(S(ta),J(ta))π1(a),ta0i0(at)π1(a)π1(at),a>t0 where π1(a)=e0a(μ+γ(s)+δ(s))ds denotes the probability of an infected individual surviving to infection age a time units later.

Proposition 2.1

The function J:R+R+ is uniformly continuous, that is, for any η>0 there exists h>0 such that |J(t+h)J(t)|<ηfor allt0R+andu0Γ

Proof.

For tR+ and h>0, we have |J(t+h)J(t)|=0β(a)i(t+h,a)0β(a)i(t,a)dahβ(a)i(t+h,a)da0β(a)i(t,a)da+0hβ(a)i(t+h,a)da. The last integral is estimated by 0hβ(a)i(t+h,a)da=0hβ(a)f(S(t+ha),J(t+ha))π1(a)daβfΛμ,βΛμh. Note that i(t+h,a+h)=i(t,a)(π1(a+h)/π1(a)) for (t,a,h)R+3. Now we are in position to estimate |J(t+h)J(t)| as follows |J(t+h)J(t)|0β(a+h)i(t,a)π1(a+h)π1(a)da0β(a)i(t,a)da+βfΛμ,βΛμh0|β(a+h)β(a)|i(t,a)π1(a+h)π1(a)da+0β(a)i(t,a)1π1(a+h)π1(a)da+βfΛμ,βΛμh0|β(a+h)β(a)|i(t,a)π1(a+h)π1(a)da+βhΛμ(μ+δ+γ)+fΛμ,βΛμ. Now the result follows immediately from Assumption (A1).

Proposition 2.2

The semiflow Φ is asymptotically smooth.

Proof.

Let BX be a bounded set with Φ(t,u0)M for u0B. For tR+ and u0B, define Φˆ(t,u0)=(0,iˆ(t,),vˆ(t,)),Φ~(t,u0)=(S(t),i~(t,),v~(t,)), where (17) i~(t,a)=i(t,a)if0at0if t < a=f(S(tτ),J(ta))π1(a)for0at,0for t < a,(17) (18) iˆ(t,a)=i(t,a)i~(t,a)=0for0at,i0(at)π1(a)π1(at)for t < a,(18) (19) v~(t,a)=v(t,a)for0at0for t < a=φS(ta)π(a)for0at,0for t < a,(19) (20) vˆ(t,a)=v(t,a)v~(t,a)=0for0at,v0(at)π(a)π(at)for t < a.(20) Then Φ=Φˆ+Φ~. It is easy to see that iˆ, vˆ, i~, and v~ are nonnegative. It follows from (Equation18) and (Equation20) that Φˆ(t,u0)X=iˆ(t,)1+vˆ(t,)1=ti0(at)π1(a)π1(at)da+tv0(at)π(a)π(at)da=0i0(a)π1(a+t)π1(a)da+tv0(a)π(a+t)π(a)daeμt0(i0(a)+v0(a))da=eμt(i01+v01)eμtu0B and thus Assumption (1) in Lemma 3.2.3 [Citation11] holds.

Next, we establish that Φ~ is completely continuous. This means that for any fixed tR+ and any bounded set BΓ0, the set Bt{Φ~(t,(S0,i0,v0))|(S0,i0,v0)B} is precompact. It is enough to show that Bt,i,v={(i~(t,),v~(t,))L+1×L+1|(S(t),i~(t,),v~(t,))Bt} is precompact. This can be obtained by Fréchet-Kolmogrov Theorem [Citation24]. Firstly, it follows from the definitions of Φ~ and Γ0 that Bt,i,v is bounded. This implies that the first condition of the Fréchet-Kolmogrov Theorem holds. Secondly, it is easy to see that ti~(t,a)da+tv~(t,a)da=0 by (Equation19) and this indicates that the third condition of the Fréchet-Kolmogrov Theorem is satisfied. Finally, to verify the second condition of the Fréchet-Kolmogrov Theorem, we need to show that Bt,i,v is uniformly continuous under Φ~ that is, (21) limh0+i~(t,)i~(t,+h)1=0uniformly inBt,i,v(21) and (22) limh0+v~(t,)v~(t,+h)1=0uniformly inBt,i,v(22) Equation (Equation22) has been proved in Yang et al. [Citation31, Proposition 3.7] and hence we only need to prove (Equation21). Obviously (Equation21) holds when t=0 since i~(0,)=0 by (Equation19). Now let t>0. Since we are concerned with the limit as h tends to 0+, we assume that h(0,t). Then i~(t,)i~(t,+h)1=0|i~(t,a)i~(t,a+h)|da=0th|f(S(tah),J(tah))π1(a+h)f(S(ta),J(ta))π1(a)|da+thtf(S(ta),J(ta))π1(a)da0thf(S(tah),J(tah))|π1(a+h)π1(a)|da+0th|f(S(tah),J(tah))f(S(ta),J(ta))|π1(a)da+fΛμ,βΛμhfΛμ,βΛμt(μ+δ+γ)h+fΛμ,βΛμh+L0th(|S(tah)S(ta)|+|J(tah)J(ta)|)π1(a)da. Here we have used S(t)Λ/μ and J(t)βΛ/μ for tR+ and |π1(a+h)π1(a)|1eaa+h(μ+δ(s)+γ(s))ds1π1(a+h)/π1(a)(μ+δ+γ)h. By the first equation of (Equation2), dS(t)dtΛ+(μ+φ+ε)Λμ+fΛμ,βΛμL1. It follows that i~(t,)i~(t,+h)1fΛμ,βΛμh[1+t(μ+δ+γ)]+LL1th+L0th|J(tah)J(ta)|π1(a)da. Then with the help of Proposition 2.1, we easily see that (Equation21) holds.

The following result follows immediately from Proposition 1.2, Proposition 2.2, and Theorem 2.33 of [Citation24].

Theorem 2.3

Suppose that Assumptions (A1), (A2), and (B1) hold. Then the semiflow Φ has a compact attractor A in Γ.

3. The existence and local stability of equilibria

In this section, we mainly focus on the calculation of the basic reproduction number R0 and investigate the existence and local stability of equilibria of (Equation2). Denote K=0ε(a)π(a)daandK1=0β(a)π1(a)da. Clearly, K1. Note that K1 is the total transmission rate of an infectious individual in its infectious period.

Let E¯=(S¯,i¯,v¯) be an equilibrium of (Equation2). Then (23) 0=ΛμS¯f(S¯,J¯)φS¯+0ε(a)v¯(a)da,di¯(a)da=(μ+γ(a)+δ(a))i¯(a),i¯(0)=f(S¯,J¯),J¯=0β(a)i¯(a)da,dv¯(a)da=(μ+ε(a))v¯(a),v¯(0)=φS¯.(23) Obviously, v¯(a)=φS¯π(a) and i¯(a)=f(S¯,J¯)π1(a). It follows that (24) 0=Λ(μ+φ(1K))S¯f(S¯,J¯),J¯=f(S¯,J¯)K1.(24) Therefore, S¯=ΛJ¯K1μ+φ(1K), where J¯ is a nonnegative zero of g with (25) g(x)=xfΛxK1μ+φ(1K),xK1.(25) Clearly, J¯ΛK1 as S¯0. Note that g(0)=0, which implies that (Equation2) always has a disease-free equilibrium E0=(S0,0,v0)=Λμ+φ(1K),0,φΛμ+φ(1K)π(). Next, we calculate the basic reproduction number R0. Linearizing (Equation2) at the disease-free equilibrium E0, we obtain the following linear system in the disease invasion phase: (26) i(t,a)t+i(t,a)a=(μ+γ(a)+δ(a))i(t,a),i(t,0)=f(S0,0)JJ(t),J(t)=0β(a)i(t,a)da,i(0,a)=i0(a).(26) Define I(t)=0i and Bφ=(f(S0,0)/J)0β(a)φ(a)da0. Borrowing the definition of A1 in (Equation4), we obtain the following abstract Cauchy problem: (27) dI(t)dt=A1I(t)+BI(t),I(0)=I0.(27) Let M(t)=etA1 be the C0-semigroup generated by A1. From the variation of constant formula, we obtain (28) I(t)=M(t)I0+0tM(ts)BI(s)ds.(28) Applying B on both sides of (Equation28) yields (29) m(t)=g(t)+0tΨ(s)m(ts)ds,(29) where m(t)=BI(t) denotes the density of newly infected, g(t)=BM(t)I0 and Ψ(s)=BM(s). Then the next generator is defined by L=0Ψ(s)ds=B(A1)1, where (zA1)1=0ezsM(s)ds. So the basic reproduction number R0 can be defined as follows: (30) R0=r(L)=f(S0,0)JK1.(30) In epidemiology, R0 is the average number of cases produced by an infectious individual in the whole infectious period when introduced into a wholly susceptible population.

It is easy to see that an equilibrium must be endemic if it is not disease free. In the following, we discuss the existence of endemic equilibria. We firstly derive a necessary condition on the existence of endemic equilibria.

Suppose that there is an endemic equilibrium (S¯,i¯,v¯). Then from g(J¯)=0=g(0), we know that there exists a Jˆ(0,J¯) such that (31) 0=g(Jˆ)=1+fSΛJˆK1μ+φ(1K),JˆK1fJΛJˆK1μ+φ(1K),Jˆ.(31) This, combined with (B1), implies that K1fJ(S0,0)K1fJΛJˆK1μ+φ(1K),0K1fJΛJˆK1μ+φ(1K),Jˆ>1. Therefore, a necessary condition on the existence of endemic equilibria is R0>1.

Now we show that R0>1 is also a sufficient condition on the existence of endemic equilibria. In fact, suppose that R0>1. Note that g(0)=0 and g(0)=1R0<0. It follows that g(J)<0 for J>0 and sufficiently small. This, combined with the Intermediate Value Theorem and g(ΛK1)ΛK1>0, implies that g(x)=0 has a positive solution in (0,ΛK1). Hence there exists at least one endemic equilibrium. Actually, there is only one endemic equilibrium. Otherwise, let (S¯,i¯,v¯) and (Sˆ,iˆ,vˆ) be two distinct endemic equilibria. Without loss of generality, we assume that i¯(0)>iˆ(0). Denote l=i¯(0)iˆ(0) (>1). Then J¯=lJˆ>Jˆ, which implies that S¯<Sˆ. With the help of (B1), we get iˆ(0)=f(Sˆ,Jˆ)>f(S¯,Jˆ)=fS¯,J¯lf(S¯,J¯)J¯J¯l=i¯(0)l, a contradiction.

To summarize, we have the following result on the existence of equilibria.

Theorem 3.1

Let R0 be defined as in (Equation30).

  1. If R01, then (Equation2) has a unique equilibrium, which is the disease-free equilibrium E0.

  2. If R0>1, then, besides E0, (Equation2) also has a unique endemic equilibrium, E=(S,i,v), where S=(ΛJ/K1)/(μ+φ(1K)), i(a)=f(S,J)π1(a), v(a)=φSπ(a), with J being the unique positive zero of g (defined by (Equation31)) on (0,ΛK1).

In the remaining of this section, we study the local stability of equilibria by linearization. Linearizing (Equation2) at an equilibrium E¯=(S¯,i¯,v¯) will produce the associated characteristic equation 0=λ+μ+f(S¯,J¯)S+φ(1Kˆ(λ))f(S¯,J¯)JKˆ1(λ)f(S¯,J¯)S1f(S¯,J¯)JKˆ1(λ), where Kˆ(λ)=0ε(a)π(a)eλadaandK1ˆ(λ)=0β(a)π1(a)eλada are the Laplace transforms of επ and βπ1, respectively. The equilibrium E¯ is locally (asymptotically) stable if all eigenvalues of the characteristic equation have negative real parts and it is unstable if at least one eigenvalue has a positive real part.

Theorem 3.2

Let R0 be defined in (Equation30).

  1. The disease-free equilibrium E0 is locally asymptotically stable if R0<1 and it is unstable if R0>1.

  2. If R0>1, then the endemic equilibrium E is locally asymptotically stable.

Proof.

(i) The characteristic equation at E0 is 0=C1(λ)C2(λ), where C1(λ)=λ+μ+φ(1Kˆ(λ))andC2(λ)=1f(S0,0)JKˆ1(λ). We claim that all roots of C1(λ)=0 have negative real parts. In fact, if λ0 is a root with nonnegative real part, then φ<|λ0+μ+φ|=φ|Kˆ(λ0)|φKφ, a contradiction. Thus we have proved the claim.

First, suppose R0>1. Then C2(0)=1R0<0. This, combined with limλC2(λ)=1 and the Intermediate Value Theorem, tells us that C2(λ)=0 has a positive root and hence E0 is unstable if R0>1.

Now, suppose R0<1. We claim that all roots of C2(λ)=0 have negative real parts. Otherwise, let λ0 be a root of C2(λ)=0 with Re(λ0)0. Then 1(f(S0,0)/J)Kˆ1(λ0)=0 implies that 1=f(S0,0)JKˆ1(λ0)R0, a contradiction. This proves the claim and hence E0 is locally asymptotically stable ifR0<1.

(ii) The characteristic equation at E is (32) (C1(λ)+f(S,J)S)C1(λ)Kˆ1(λ)f(S,J)J=0.(32) We claim that (Equation32) has no root with a nonnegative real part. Otherwise, suppose (Equation32) has a root λ0 with Re(λ0)0. Since |Kˆ(λ0)|K1, we know Re(C1(λ0))>0 and hence C1(λ0)+f(S,J)SC1(λ0)>1 as f(S,J)/S>0. On the other hand, from (B1) and the second equation of (Equation24), we have Kˆ1(λ0)f(S,J)Jf(S,J)K1J=1. It follows that Kˆ1(λ0)f(S,J)J<C1(λ0)+f(S,J)SC1(λ0), a contradiction to the assumption that λ0 is a root of (Equation32). This completes the proof.

4. Uniform persistence

We start with the uniformly weak ρ-persistence.

Define ρ:ΓR+ by ρ(S,i,v)=0β(a)i(t,a)dafor(S,i,v)Γ. Let Γ0={(S0,i0,v0)Γ:ρ(Φ(t0,(S0,i0,v0)))>0forsomet0R+}. Obviously, if (S0,i0,v0)ΓΓ0, then (S(t),i(t,),v(t,))E0 as t.

Definition 4.1

[Citation24, pp. 61]

System (Equation2) is said to be uniformly weakly ρ-persistent (respectively, uniformly strongly ρ-persistent) if there exists an ϵ>0, independent of the initial conditions, such that lim suptρ(Φ(t,(S0,i0,v0)))>ϵ(respectively,lim inftρ(Φ(t,(S0,i0,v0)))>ϵ) for (S0,i0,v0)Γ0.

To show that (Equation2) is uniformly weakly ρ-persistent, we need the following Fluctuation Lemma. For a function h:R+R, we denote h=lim infth(t)andh=lim supnh(t).

Lemma 4.2

Fluctuation Lemma [Citation12]

Let h:R+R be a bounded and continuously differentiable function. Then there exist sequences {sn} and {tn} such that sn, tn, h(sn)h, h(tn)h, h(sn)0, and h(tn)0 as n.

The next result will be helpful in the coming discussion.

Lemma 4.3

[Citation14]

Suppose h:R+R is a bounded function and kL+1. Then lim supt0tk(θ)h(tθ)dθhk1.

Lemma 4.4

Let (S,i,v) be a solution of (Equation2). Then SS0.

Proof.

By Lemma 4.2, there exists {tn} such that tn, S(tn)S, and dS(tn)/dt0 as n. Then dS(tn)dt=Λ(μ+φ)S(tn)f(S(tn),J(tn))+0ε(a)v(tn,a)daΛ(μ+φ)S(tn)+0tnε(a)φS(tna)π(a)da+tnε(a)v0(atn)π(a)π(atn)daΛ(μ+φ)S(tn)+0snε(a)φS(tna)π(a)da+eμtn0ε(a+tn)v0(a)da. Letting n and using Lemma 4.3, we have 0Λ(μ+φ)S+φSK or SΛ/(μ+φ(1K))=S0 as required.

Proposition 4.5

If R0>1, then (Equation2) is uniformly weakly ρ-persistent.

Proof.

By way of contradiction, for any ε>0, there exists an (Sε,iε,vε)Γ0 such that lim suptρ(Φ(t,(Sε,iε,vε))ε. Since R0>1, there exists an ε0>0 such that (33) f(S(ϵ0),ϵ0)JKˆ1(ε0)>1,(33) where S(ε0)=(Λf(S0+ε0,ε0))/(μ+(1K)φ)ε0(>0) and Kˆ1 is the Laplace transform of βπ as before. In particular, for this ε0, there exists an(S0,i0,v0)Γ0 such that lim suptρ(Φ(t,(S0,i0,v0)))ε02. We will get a contradiction as follows.

Firstly, there exists t0R+ such that ρ(Φ(t,(S0,i0,v0)))ε0 for tt0. Without loss of generality, we assume that t0=0 as we can replace (S0,i0,v0) with Φ(t0,(S0,i0,v0)). Then J(t)ε0 for tt0=0.

Secondly, we show S(Λf(S0+ε0,ε0))/(μ+(1K)φ). Using the Fluctuation Lemma, there exists a sequence {tn} such that tn, S(tn)S, dS(tn)/dt0 as n. By Lemma 4.4, without loss of generality, we can assume that S(t)S0+ε0 for tR+. Then by (B1), we have dS(tn)dtΛ(μ+φ)S(tn)f(S0+ε0,ε0)+0ε(a)v(tn,a)da. This, combined with (Equation14), gives dS(tn)dtΛ(μ+φ)S(tn)f(S0+ε0,ε0)+0tnε(a)φS(tna)π(a)da. Now, for any ξ>0, there exists TR+ such that S(t)Sξ for tT. Then, for tnT, dS(tn)dtΛ(μ+φ)S(tn)f(S0+ε0,ε0)+0tnTε(a)φ(Sξ)π(a)da. Letting n gives 0Λ(μ+φ)Sf(S0+ε0,ε0)+φ(Sξ)K. This implies that S(Λf(S0+ε0,ε0)φξK)/(μ+(1φ)K). Since ξ is arbitrary, we immediately get S(Λf(S0+ε0,ε0))/(μ+(1φ)K).

Finally, since S(Λf(S0+ε0,ε0))/(μ+(1φ)K), there exists t1R+ such that S(t)S(ε0) for tt1. Again we can assume t1=0. Then J(t)=0β(a)i(t,a)da0tβ(a)f(S(ta),J(ta))π1(a)da0tβ(a)f(S(ϵ0),J(ta))π1(a)da0tβ(a)f(S(ε0),ε0)JJ(ta)π1(a)da. Here we have used the Mean Value Theorem for f(S,J) with respect to J, J(t)ε0, and assumption (B1). Taking Laplace transforms on both sides of the above inequality yields Jˆ(λ)f(S(ε0),ε0)JJˆ(λ)Kˆ1(λ). Then 1(f(S(ε0),ε0)/J)Kˆ1(λ) for all λ>0 since Jˆ(λ)>0 for λ>0. In particular, (f(S(ε0),ε0)/J)Kˆ1(ε0)1, a contradiction. This completes the proof.

Any global attractor of Φ only contains points where a total trajectory passes through it. A total trajectory of Φ is a function h:RX such that Φ(η,h(t))=h(t+η) for all tR and all ηR+. For a total trajectory, for tR and aR+, (34) dS(t)dt=Λf(S(t),J(t))(μ+φ)S(t)+0ε(a)v(t,a)da,J(t)=0β(a)π1(a)f(S(ta),J(ta))da,i(t,a)=f(S(ta),J(ta))π1(a),v(t,a)=φS(ta)π(a).(34) In what follows, we will show the semiflow Φ(t,u0) is uniformly strongly ρpersistent. For u0Γ, we have (35) J(t)=0β(a)i(t,a)da=0tβ(a)π1(a)f(S(ta),J(ta))da+J~(t),(35) where J~(t)=tβ(a)i0(at)(π1(a)/π1(at))da. Following the approach in [Citation24], we have the following proposition.

Proposition 4.6

Let h(t) be a total trajectory in Γ for all tR. Then S(t) is positive and either J is identically zero or it is strictly positive.

Proof.

Define ht(η)=h(t+η) is a semi-trajectory of system (Equation2) with initial condition ht(0)=h(t)Γ for tR and ηR+. First, we show S(t)>0 for any sR. Suppose that it doesn't hold. Then for some tR S(s)=0 and dS(t)/dt=Λ>0. By the continuity of S(t), there exists a sufficiently small η>0 such that S(tη)<0, which is a contradiction with S(t)Γ. Therefore, S(t) is strictly positive for each tR.

Secondly, we claim that J(t) is identically zero for all tR if i(t,)=0. For tη, we have J(t)=0tβ(a)π1(a)f(S(ta),J(ta))da0tβ(a)π1(a)fJ(S(ta),0)J(ta)da. It follows from Gronwall inequality that J(t)=0. Besides, for t<η, i(η,ηt)=i(η,0)π1(ηt)=f(S(t),J(t))π1(ηt)=0. Thus J(η)=0. So that for all tR J(t) is identically zero.

Now we are going to assume that J(t) is non-zero for each tR. If there exists a t0 such that J(t)=0 for alltt0, then J(t)=0β(a)i(t,a)da=0tβ(a)f(S(ta),J(ta))π1(a)da+J~(t)=t0tβ(a)f(S(ta),J(ta))π1(a)da for aR+. Gronwall inequality ensures that J(t) is identically zero, giving a contradiction. Thus, there exists a sequence {tn} toward as n goes to infinity such that J(tn)>0. For each nN, let Jn(t)=J(tn+η). Since S(η)>0 for each ηR. Hence, there exists a positive value ξ such that S(t)>ξ for each tR. Recalling Equation (Equation35), we have Jn(t)0tβ(a)f(ξ,J(ta))π1(a)da+J~n(t), where J~n(t)=tβ(a)in0(at)(π(a)/π(at))da. Hence, J~n(0)=0β(a)i(tn,a)da=J(tn)>0 for each nN. From Corollary B.6 in [Citation24], we conclude that there exists a constant b>0 such that Jn(t)>0 for all t>b. Letting n, we have tn. Therefore, for all tR, J(t) is strictly positive.

Proposition 4.5 implies that Γ0 is invariant under Φ. By Theorem 2.3, Φ has a compact attractor A. Then A0=AΓ0 is a compact attractor of the restricted semiflow Φ|Γ0. This, combined Propositions 4.5 and 4.6 with Theorem 3.2 in [Citation27], immediately yields the uniformly strong ρ-persistence.

Theorem 4.7

If R0>1, system (Equation2) is uniformly strongly ρ-persistent.

Corollary 4.8

Suppose R0>1. Let h(t)=(S(t),i(t,),v(t,)) be a total trajectory in A0. Then there exists an η0>0 such that S(t)η0, i(t,a)η0π1(a), and v(t,a)>η0π(a) for all tR and aR+.

Proof.

Theorem 1.2 provides a positive lower bound η1 such that S(t)>η1 for all tR. It follows from Theorem 4.7 that there exists a positive number η2 such that ρ(h(t))=J(t)>η2 for all tR. Thus, f(S(t),J(t))f(η1,η2). Hence,i(t,a)=f(S(t),J(t))π1(a)f(η1,η2)π1(a)=η3π1(a) for all tR and aR+. By the v equation in (Equation34), we have v(t,a)=φS(ta)π(a)φη1π(a):=η4π(a). Letting η0=min{η1,η3,η4} completes the proof.

5. A threshold dynamics

The main result of this paper is a threshold dynamics determined by R0. We start with the global stability of the disease-free equilibrium E0.

Theorem 5.1

If R0<1, then the disease-free equilibrium E0 is globally asymptotically stable in X.

Proof.

By Theorem 3.2, it suffices to show that limtΦ(t,u0)=E0 for u0A. Let u0A and h(t) be a total trajectory in A. Then Φ(t,u0)=h(t) for tR+.

Firstly, J(t)=0β(a)f(S(ta),J(ta))π1(a)da0β(a)fJ(S0,0)J(ta)π(a)da. Taking superior limit on both sides of the above inequality and applying Lemma 4.3, we have JR0J. Therefore, J=0 as R0<1.

Secondly, we show that (f(S(),J())=0. For any ξ>0, there exists a TR+ such that S(t)S0+ξ for tT by Lemma 4.4. Then for tT, it follows from (B1) and (Equation34) that f(S(t),J(t))f(S0+ξ,J(t))f(S0+ξ,0)JJ(t)=f(S0+ξ,0)J0β(a)f(S(ta),J(ta))π1(a)da. Taking limit suprema and using Lemma 4.3 again, we obtain (f(S(),J())f(S0+ξ,0)JK1(f(S(),J()). As ξ is arbitrary, this immediately gives (f(S(),J())R0(f(S(),J()). Since R0<1, we have (f(S(),J())=0.

Thirdly, we show limti(t,)1=0. In fact, we use (Equation34) again to get i(t,)1=0f(S(ta),J(ta))π1(a)da. With the help of Lemma 4.3, one has lim supti(t,)1(f(S(),J()))π11=0, which implies that limti(t,)1=0.

Fourthly, we show limtS(t)=S0. It suffices to show SS0. We achieve this by using Lemma 4.2. There exists {tn} such that S(tn)S and dS(tn)/dt0 as n. With (B1) and (Equation34), we have dS(tn)dtΛ(μ+φ)S(tn)f(S(tn),0)JJ(tn)+0tnε(a)φS(tna)π(a)da. Letting n immediately yields 0Λ(μ+φ(1K))S. Here we have used J(tn)=0β(a)i(tn,a)daβi(tn,)10 as n. So we have SS0 as sought.

Finally, we show limtv(t,)v0()1=0. With (Equation14), we have v(t,)v0()10tφ|S(ta)S0|π(a)da+tv0(at)π(a)π(at)da+tφS0π(a)da=0tφ|S(ta)S0|π(a)da+0v0(a)π(a+t)π(a)da+tφS0π(a)da0tφ|S(ta)S0|π(a)da+eμtv01+tφS0π(a)da. Applying Lemma 4.3, we get lim suptv(t,)v0()1=0. Therefore, limtv(t,)v0()1=0 and this completes the proof.

The following corollary guarantees the well-definition of the constructive Lyapunov functional.

Corollary 5.2

If R0>1, the following statements hold. For all tR and aR+ i(t,a)i(a)η0f(S,J),v(t,a)v(a)Λ(μ+L+φ)S, where L is the Lipschitz coefficient and η0 is defined in Corollary 4.8.

To establish the global stability of the endemic equilibrium E by constructing a suitable Lyapunov functional, we need an additional assumption on the incidence rate, that is,

(B2)

For S>0, xJSf(S,x)Sf(S,J)<1for 0<x<J,1<Sf(S,x)Sf(S,J)xJfor x>J.

This condition holds automatically for most nonlinear incidences. For example, f(S,J)=SG(J) is one of such. Hence our results cover some existing ones. To build the Lyapunov functional, we introduce the function ϕ:(0,)R defined by ϕ(x)=x1lnxforx(0,). It is well-known that ϕ(x)0 for x(0,) and it attains the global minimum 0 only at x=1.

Theorem 5.3

Suppose that R0>1 and (B2) holds. Then the endemic equilibrium E is globally asymptotically stable in A0.

Proof.

By Theorem 3.2, it suffices to show A0={E}. Let h(t)=(S(t),i(t,),v(t,)) be a total trajectory in A0. By Corollary 4.8, there exists ϵ0>0 such that 0ϕ(z)<ϵ0 with z=S(t)/S, i(t,a)/i(a), and v(t,a)/v(a) for any tR and aR+.

Let α(a)=aε(s)v(s)ds,α1(a)=f(S,J)aβ(s)π1(s)dsK1. Then dα(a)da=ε(a)v(a),dα1(a)da=f(S,J)β(a)π1(a)K1. Define V(t)=VS(t)+Vi(t)+Vv(t), where VS(t)=SϕS(t)S,Vi(t)=0α1(a)ϕi(t,a)i(a)da,Vv(t)=0α(a)ϕ(v(t,a)v(a))da. Then V is well-defined because of Corollary 5.2.

Now we show that the upper-right derivative dV(t)/dt along the solution is non-positive. We first have dVS(t)dt=1SS(t)Λ(μ+φ)S(t)f(S(t),J(t))+0ε(a)v(t,a)da=1SS(t)(μ+φ)S+f(S,J)0ε(a)v(a)da(μ+φ)S(t)f(S(t),J(t))+0ε(a)v(t,a)da=(μ+φ(1K))SSS(t)+S(t)S2+f(S,J)1+Sf(S(t),J(t))S(t)f(S,J)SS(t)f(S(t),J(t))f(S,J)+0ε(a)v(a)v(t,a)v(a)Sv(t,a)S(t)v(a)S(t)S+1da=(μ+φ(1K))SSS(t)+S(t)S2+f(S,J)ϕSf(S(t),J(t))S(t)f(S,J)ϕSS(t)ϕf(S(t),J(t))f(S,J)+0ε(a)v(a)ϕv(t,a)v(a)ϕSv(t,a)S(t)v(a)ϕS(t)Sda. Here we have used 0ε(a)v(a)da=φKS. Noting i(t,a)=f(S(ta),J(ta))π1(a) and i(a)=f(S,J)π1(a), we obtain dVi(t)dt=0α1(a)aϕi(t,a)i(a)da=α1(a)ϕi(t,a)i(a)|a=0+0α1(a)ϕi(t,a)i(a)da=f(S,J)K10β(a)π1(a)ϕi(t,0)i(0)ϕi(t,a)i(a)da=f(S,J)ϕf(S(t),J(t))f(S,J)sf(S,J)K10β(a)π1(a)ϕf(S(ta),J(ta))f(S,J)da. Similarly, noting v(t,0)=φS(t) and v(0)=φS, we have dVv(t)dt=0ε(a)v(a)ϕv(t,0)v(0)ϕv(t,a)v(a)da=0ε(a)v(a)ϕS(t)Sϕv(t,a)v(a)da. Therefore, dV(t)dt=(μ+φ(1K))SSS(t)+S(t)S2f(S,J)ϕSS(t)0ε(a)v(a)ϕSv(t,a)S(t)v(a)da+f(S,J)ϕSf(S(t),J(t))S(t)f(S,J)f(S,J)K0β(a)π1(a)ϕf(S(ta),J(ta))f(S,J)da. By Jensen's inequality and the concavity of ϕ, f(S,J)K0β(a)π1(a)ϕf(S(ta),J(ta))f(S,J)daf(S,J)ϕ0β(a)π1(a)f(S(ta),J(ta))Kf(S,J)da=f(S,J)ϕ0β(a)i(t,a)Kf(S,J)da=f(S,J)ϕJ(t)J. Because of (B2) and the monotonicity of ϕ and f, we have dV(t)/dt0, that is, V is nonincreasing. Since V is bounded on h(), the α-limit set of h() must be contained in M, the largest invariant subset of {dV/dt=0}. It follows from dV(t)/dt=0 that S(t)=S, Sv(t,a)/S(t)v(a)=1, and J(t)=J. Consequently, M={E} since i(t,a)=f(S,J)π1(a)=i(a).

The above analysis indicates that the α-limit set of h() consists of just the endemic equilibrium E and hence V(h(t))V(E) for all tR. It follows that hE for tR. Thus A0={E} and the proof is complete.

6. Numerical simulations

In this section, we first perform numerical experiments to illustrate the theoretical results. It follows from Theorem 5.1 and Theorem 5.3 that the basic reproduction number R0 is a key threshold to determine whether or not the disease persists. For convenience, we choose F(S,J)=SJ1+αJ. We take the values of some parameters as in Table . Besides, we take the transmission rate β(a) in the form of β(a)=0,0<a<5,β(a5)2e0.6(a0.5),a5, to illustrate the theoretical results by changing β.

Table 1. List of parameter values.

First, we fix ψ=0.006. If we take β=1.527, then the basic reproduction number R0=(Λ/(μ+p(1K)))K10.9927<1. It follows from Theorem 5.1 that the disease-free equilibrium E0 is globally asymptotically stable. (a) and (b) of Figure  show this fact. Then we enlarge the transmission rate to β=6.667 and have R04.3361>1. (c) and (d) of Figure  indicates that the endemic equilibrium E is asymptotically stable, which supports Theorem 5.3.

Figure 1. Time evolution of the infective population i(t,a), 0t100, 0a20 for system (Equation2) with initial value i0(a)=1+cos(2a) and v0(a)=ea. (a) and (b) with β=1.527, (c) and (d) with β=6.667.

Figure 1. Time evolution of the infective population i(t,a), 0≤t≤100, 0≤a≤20 for system (Equation2(2) dS(t)dt=Λ−μS(t)−f(S(t),J(t))−φS(t)+∫0∞ε(a)v(t,a)da,∂i(t,a)∂t+∂i(t,a)∂a=−(μ+γ(a)+δ(a))i(t,a),i(t,0)=f(S(t),J(t)),J(t)=∫0∞β(a)i(t,a)da,∂v(t,a)∂t+∂v(t,a)∂a=−(μ+ε(a))v(t,a),v(t,0)=φS(t),(2) ) with initial value i0(a)=1+cos⁡(−2a) and v0(a)=e−a. (a) and (b) with β∗=1.527, (c) and (d) with β∗=6.667.

Next we study the effect of some parameters on the disease spread.

Vaccination plays an important role in controlling the disease prevalence. (a) of Figure  indicates that improving the vaccine coverage for the susceptible decreases the final size of the disease prevalence. In fact, improving the vaccine rate can reduce the basic reproduction number. Theorem 5.1 implies that taking suitable vaccine measures can slow down the disease prevalence and even make the disease die out. (b) of Figure  shows that taking vaccine on newborns has the familiar effect as increasing φ.

Figure 2. Time evolution of the infective population i(t,a), 0t100, 0a20 for system (Equation2) with different vaccinated rates and input rate. (a) with φ=0.3,0.4,0.5. (b) with Λ=10,20,40.

Figure 2. Time evolution of the infective population i(t,a), 0≤t≤100, 0≤a≤20 for system (Equation2(2) dS(t)dt=Λ−μS(t)−f(S(t),J(t))−φS(t)+∫0∞ε(a)v(t,a)da,∂i(t,a)∂t+∂i(t,a)∂a=−(μ+γ(a)+δ(a))i(t,a),i(t,0)=f(S(t),J(t)),J(t)=∫0∞β(a)i(t,a)da,∂v(t,a)∂t+∂v(t,a)∂a=−(μ+ε(a))v(t,a),v(t,0)=φS(t),(2) ) with different vaccinated rates and input rate. (a) with φ=0.3,0.4,0.5. (b) with Λ=10,20,40.

As for the transmission rate, we take α to be 0 and 0.5, respectively. Whenever we take any value for α, the value of the basic reproduction number doesn't change any more. Since R0>1, Theorem 5.3 implies that the disease must break out. From (a) of Figure , we see that enlarging α can lower the final size of the disease (the value of the component I of the endemic equilibrium). Furthermore, we readily see that enlarging the immunity waning rate can raise the value of the basic reproduction number. Hence this move increases the transmission risk. (b) of Figure  shows that the final size of the disease increases as the immunity waning rate increases.

Figure 3. Time evolution of the infective population i(t,a), 0t100, 0a20 for system (Equation2) with different transmission rates and different immunity waning rates. (a) with different values α=0,0.5, (b) with different immunity waning rates ψ=0.06,0.09,0.12.

Figure 3. Time evolution of the infective population i(t,a), 0≤t≤100, 0≤a≤20 for system (Equation2(2) dS(t)dt=Λ−μS(t)−f(S(t),J(t))−φS(t)+∫0∞ε(a)v(t,a)da,∂i(t,a)∂t+∂i(t,a)∂a=−(μ+γ(a)+δ(a))i(t,a),i(t,0)=f(S(t),J(t)),J(t)=∫0∞β(a)i(t,a)da,∂v(t,a)∂t+∂v(t,a)∂a=−(μ+ε(a))v(t,a),v(t,0)=φS(t),(2) ) with different transmission rates and different immunity waning rates. (a) with different values α=0,0.5, (b) with different immunity waning rates ψ=0.06,0.09,0.12.

7. Conclusion and discussion

We proposed a general SIVS model with infection age and vaccinated age. It turned out that the dynamics of the model is determined by the basic reproduction number R0, which determines whether the disease dies out or persists. Our model generalizes many existing ones with some being listed in Table . Here f, F, G have familiar features as Assumption of (B1) in our paper. Moreover, the idea of this paper can be applied to some other structured epidemic models such as SVIR [Citation6], SEIR [Citation13,Citation19] models, and even if two-group model [Citation21].

Table 2. Incidence rates.

In this paper, we mainly focus on the global stability of equilibria. However, some complex phenomena such as multiple endemic equilibria (backward bifurcations) [Citation18,Citation29] and unstable endemic equilibrium [Citation1,Citation2,Citation26] have appeared in some epidemic models with vaccination. In the future, we shall address them with more age structures.

Disclosure statement

There is no any potential conflict of interest in this work.

Additional information

Funding

Research is partially supported by the National Natural Science Foundation of China [Nos. 61573016,61203228,11371313], Shanxi Scholarship Council of China (2015-094), Shanxi Scientific Data Sharing Platform for Animal Diseases, and the Natural Sciences and Engineering Research Council (NSERC) of Canada.

References

  • V. Andreasen, Instability in an SIR-model with age-dependent susceptibility, in Mathematical Population Dynamics: Analysis of Heterogeneity, O. Arino, M. Kimmel and M. Langlais, eds., Wuerz Publ., Winnipeg, 1995, pp. 3–14.
  • Y. Cha, M. Iannelli, and F.A. Milner, Stability change of an epidemic model, Dynam. Syst. Appl. 9 (2000), pp. 361–376.
  • Y. Chen, J. Yang and F. Zhang, The global stability of an SIRS model with infection age, Math. Biosci. Eng. 11 (2014), pp. 449–469. doi: 10.3934/mbe.2014.11.449
  • Y. Chen, S. Zou, and J. Yang, Global analysis of an SIR epidemic model with infection age and saturated incidence, Nonlinear Anal. Real World Appl. 30 (2016), pp. 16–31. doi: 10.1016/j.nonrwa.2015.11.001
  • D. Ding and X. Ding, Global stability of multi-group vaccination epidemic models with delays, Nonlinear Anal. Real World Appl. 12 (2011), pp. 1991–1997. doi: 10.1016/j.nonrwa.2010.12.015
  • X. Duan, S. Yuan, and X. Li, Global stability of an SVIR model with age of vaccination, Appl. Math. Comput. 226 (2014), pp. 528–540.
  • X. Duan, S. Yuan, Z. Qiu, and J. Ma, Global stability of an SVEIR epidemic model with ages of vaccination and latency, Comput. Math. Appl. 68 (2014), pp. 288–308. doi: 10.1016/j.camwa.2014.06.002
  • Z. Feng and H.R. Thieme, Endemic models with arbitrarily distributed periods of infection I: Fundamental properties of the model, SIAM J. Appl. Math. 61 (2000), pp. 803–833. doi: 10.1137/S0036139998347834
  • D.P. Francis, P.M. Feorino, J.R. Broderson, H.M. Mcclure, J.P. Getchell, C.R. Mcgrath, B. Swenson, J.S. Mcdougal, E.L. Palmer, A.K. Harrison, F. Barre-Sinoussi, J.-C. Chermann, L.Montagnier, J.W. Curran, C.D. Cabradilla, and V.S. Kalyanaraman, Infection of chimpanzees with lymphadenopathy-associated virus, Lancet 2 (1984), pp. 1276–1277. doi: 10.1016/S0140-6736(84)92824-1
  • H. Gulbudak and M. Martcheva, A Structured avian influenza model with imperfect vaccination and vaccine induced asymptomatic infection, Bull. Math. Biol. 76(10) (2014), pp. 2389–2425. doi: 10.1007/s11538-014-0012-1
  • J.K. Hale, Asymptotic Behavior of Dissipative Systems, AMS, Providence, 1988.
  • W.M. Hirsch, H. Hanisch, and J.P. Gabriel, Differential equation models of some parasitic infections: Methods for the study of asymptotic behavior, Comm. Pure Appl. Math. 38 (1985), pp. 733–753. doi: 10.1002/cpa.3160380607
  • G. Huang, Y. Takeuchi, W. Ma, and D. Wei, Global stability for delay SIR and SEIR epidemic models with nonlinear incidence rate, Bull. Math. Biol. 72 (2010), pp. 1192–1207. doi: 10.1007/s11538-009-9487-6
  • M. Iannelli, Mathematical Theory of Age-Structured Population Dynamics, Applied Mathematics Monographs, vol. 7, Giardini Editorie Stampatori, Pisa, 1995.
  • M. Iannelli, M. Martcheva, and X.-Z. Li, Strain replacement in an epidemic model with super-infection and perfect vaccination, Math. Biosci. 195 (2005), pp. 23–46. doi: 10.1016/j.mbs.2005.01.004
  • A. Korobeinikov, Lyapunov functions and global stability for SIR and SIRS epidemiological models with non-linear transmission, Bull. Math. Biol. 68 (2006), pp. 615–626. doi: 10.1007/s11538-005-9037-9
  • A. Korobeinikov and P.K. Maini, Nonlinear incidence and stability of infectious disease models, Math. Med. Biol. 22 (2005), pp. 113–128. doi: 10.1093/imammb/dqi001
  • X. Li, J. Wang, and M. Ghosh, Stability and bifurcation of an SIVS epidemic model with treatment and age of vaccination, Appl. Math. Model. 34 (2010), pp. 437–450. doi: 10.1016/j.apm.2009.06.002
  • J. Li, Y. Yang, and Y. Zhou, Global stability of an epidemic model with latent stage and vaccination, Nonlinear Anal. Real World Appl. 12 (2011), pp. 2163–2173. doi: 10.1016/j.nonrwa.2010.12.030
  • P. Magal, Compact attractors for time periodic age-structured population models, Electron. J. Differential Equations 65 (2001), pp. 1–35.
  • P. Magal and C.C. McCluskey, Two group infection age model: An application to nosocomial infection, SIAM J. Appl. Math. 73(2) (2013), pp. 1058–1095. doi: 10.1137/120882056
  • P. Magal, C.C. McCluskey, and G.F. Webb, Lyapunov functional and global asymptotic stability for an infection-age model, Appl. Anal. 89 (2010), pp. 1109–1140. doi: 10.1080/00036810903208122
  • C.C. McCluskey, Global stability for an SEI epidemiological model with continuous age-structure in the exposed and infectious classes, Math. Biosci. Eng. 9 (2012), pp. 819–841. doi: 10.3934/mbe.2012.9.819
  • H.L. Smith and H.R. Thieme, Dynamical Systems and Population Persistence, Grad. Stud. Math., vol. 118, AMS, Providence, RI, 2011.
  • B. Soufiane and T.M. Touaoula, Global analysis of an infection age model with a class of nonlinear incidence rate, J. Math. Anal. Appl. 434 (2016), pp. 1211–1239. doi: 10.1016/j.jmaa.2015.09.066
  • H.R. Thieme, Stability change of the endemic equilibrium in age-structured models for the spread of S-I-R type infectious diseases, in Differential Equations Models in Biology, Epidemiology and Ecology, S. Busenberg and M. Martelli, eds., Springer, Berlin, 92, 1991, 139–158.
  • H.R. Thieme, Uniform persistence and permanence for non-autonomous semiflows in population biology, Math. Biosci. 166 (2000), pp. 173–201. doi: 10.1016/S0025-5564(00)00018-3
  • J. Wang, R. Zhang, and T. Kuniya, Global dynamics for a class of age-infection HIV models with nonlinear infection rate, J. Math. Anal. Appl. 432 (2015), pp. 289–313. doi: 10.1016/j.jmaa.2015.06.040
  • Y. Xiao and S. Tang, Dynamics of infection with nonlinear incidence in a simple vaccination model, Nonlinear Anal. Real World Appl. 11 (2010), pp. 4154–4163. doi: 10.1016/j.nonrwa.2010.05.002
  • J. Xu and Y. Zhou, Global stability of a multi-group model with vaccination age, distributed delay and random perturbation, Math. Biosci. Eng. 12 (2015), pp. 1083–1106. doi: 10.3934/mbe.2015.12.1083
  • J. Yang, M. Maia, and L. Wang, Global threshold dynamics of an SIVS model with waning vaccine-induced immunity and nonlinear incidence, Math. Biosci. 268 (2015), pp. 1–8. doi: 10.1016/j.mbs.2015.07.003
  • J. Yang, X. Li, and F. Zhang, Global dynamics of a heroin epidemic model with age structure and nonlinear incidence, Int. J. Biomath. 9(3) (2016), p. 1650033. doi: 10.1142/S1793524516500339