Research article

Optimal control and analysis of a stochastic SEIR epidemic model with nonlinear incidence and treatment

  • Received: 25 September 2024 Revised: 12 November 2024 Accepted: 19 November 2024 Published: 26 November 2024
  • MSC : 34D05, 34K20, 92B05, 92D25

  • In this paper, we represented the optimal control and dynamics of a stochastic SEIR epidemic model with nonlinear incidence and treatment rate. By using the Lyapunov function method, the existence and uniqueness of the global positive solution of the model were proved. The dynamic analysis of the stochastic model was studied and we found that the model has an ergodic stationary distribution when Rs0>1. The disease was extinct when Re0<1. The optimal solution of the disease was obtained by using the stochastic control theory. The numerical simulation of our conclusion was carried out. The results showed that the disease decreased with the increase of the control variables.

    Citation: Jinji Du, Chuangliang Qin, Yuanxian Hui. Optimal control and analysis of a stochastic SEIR epidemic model with nonlinear incidence and treatment[J]. AIMS Mathematics, 2024, 9(12): 33532-33550. doi: 10.3934/math.20241600

    Related Papers:

    [1] Xiaodong Wang, Kai Wang, Zhidong Teng . Global dynamics and density function in a class of stochastic SVI epidemic models with Lévy jumps and nonlinear incidence. AIMS Mathematics, 2023, 8(2): 2829-2855. doi: 10.3934/math.2023148
    [2] Tingting Wang, Shulin Sun . Dynamics of a stochastic epidemic model with quarantine and non-monotone incidence. AIMS Mathematics, 2023, 8(6): 13241-13256. doi: 10.3934/math.2023669
    [3] Yuhuai Zhang, Xinsheng Ma, Anwarud Din . Stationary distribution and extinction of a stochastic SEIQ epidemic model with a general incidence function and temporary immunity. AIMS Mathematics, 2021, 6(11): 12359-12378. doi: 10.3934/math.2021715
    [4] Xiangming Zhao, Jianping Shi . Dynamic behavior of a stochastic SIR model with nonlinear incidence and recovery rates. AIMS Mathematics, 2023, 8(10): 25037-25059. doi: 10.3934/math.20231278
    [5] Yue Wu, Shenglong Chen, Ge Zhang, Zhiming Li . Dynamic analysis of a stochastic vector-borne model with direct transmission and media coverage. AIMS Mathematics, 2024, 9(4): 9128-9151. doi: 10.3934/math.2024444
    [6] Ruoyun Lang, Yuanshun Tan, Yu Mu . Stationary distribution and extinction of a stochastic Alzheimer's disease model. AIMS Mathematics, 2023, 8(10): 23313-23335. doi: 10.3934/math.20231185
    [7] Yue Dong, Xinzhu Meng . Stochastic dynamic analysis of a chemostat model of intestinal microbes with migratory effect. AIMS Mathematics, 2023, 8(3): 6356-6374. doi: 10.3934/math.2023321
    [8] Shah Hussain, Naveed Iqbal, Elissa Nadia Madi, Thoraya N. Alharthi, Ilyas Khan . Vaccination strategies in a stochastic SIVR epidemic model. AIMS Mathematics, 2025, 10(2): 4441-4456. doi: 10.3934/math.2025204
    [9] Rukhsar Ikram, Ghulam Hussain, Inayat Khan, Amir Khan, Gul Zaman, Aeshah A. Raezah . Stationary distribution of stochastic COVID-19 epidemic model with control strategies. AIMS Mathematics, 2024, 9(11): 30413-30442. doi: 10.3934/math.20241468
    [10] Yassine Sabbar, Aeshah A. Raezah, Mohammed Moumni . Enhancing epidemic modeling: exploring heavy-tailed dynamics with the generalized tempered stable distribution. AIMS Mathematics, 2024, 9(10): 29496-29528. doi: 10.3934/math.20241429
  • In this paper, we represented the optimal control and dynamics of a stochastic SEIR epidemic model with nonlinear incidence and treatment rate. By using the Lyapunov function method, the existence and uniqueness of the global positive solution of the model were proved. The dynamic analysis of the stochastic model was studied and we found that the model has an ergodic stationary distribution when Rs0>1. The disease was extinct when Re0<1. The optimal solution of the disease was obtained by using the stochastic control theory. The numerical simulation of our conclusion was carried out. The results showed that the disease decreased with the increase of the control variables.



    Mathematical models are very effective tools in studying the dynamical behavior of infectious diseases [1,2,3,4,5]. When there is no way to eradicate diseases completely, researchers are always looking for and developing the best methods to control the spread of diseases. A lot of mathematical models have been presented for control effects of infectious diseases[6,7,8]. Disease control is mainly considered from two aspects: vaccine and treatment. Some researchers have considered vaccine control, such as [9,10,11], while others considered treatment control. Among them, in order to measure the effect of delayed treatment of the infected, Zhang and Liu [12] proposed the form of saturated treatment function T(I),T(I)=γI1+αI, where γ>0, α>0. A treatment function containing both the control and the infected is proposed and is defined as T(u,I)=ϕuI1+αuI in [13], which better reflects the characteristics of natural epidemics.

    Many human epidemics, such as measles, smallpox, epidemics, and dengue fever, are represented by the SEIR model [14,15]. In particular, the literature [15] takes into account the Crowley-Martin-type incidence rate and Holling type Ⅱ treatment rate, and proposes the following model:

    {dS=[ΛμSβ1SI(1+α1S)(1+γI)]dt,dE=[β1SI(1+α1S)(1+γI)(μ+φ)E]dt,dI=[φEβ2I1+α2IauI1+buI(μ+ν)I]dt,dR=[β2I1+α2I+auI1+buIμR]dt, (1.1)

    where the total population N is divided into four parts: the susceptible(S), exposed(E), infected(I), and recovered(R), β1SI(1+α1S)(1+γI) represents the transmission population of disease from S to I by the Crowley-Martin incidence rate, β2I1+α2I represents the treatment rate of the infected population, and auI1+buI is the saturated treatment function of infected population where u is treatment control. Other parameters and their definition are shown in Table 1. b is a nonnegative quantity, and other parameter are all positive. Neglecting the fourth equation, they considered an equivalent model where the basic reproduction number was described as

    R0=Λβ1φ(Λα1+μ)(μ+φ)(β2+μ+au+υ).

    They have performed the stability and bifurcation analysis of the model system. If R0<1, then symtem (1.1) has a unique disease-free equilibrium P0(Λμ,0,0), which is locally asymptotically stable. Conversely, if R0>1, then system (1.1) has two equilibrium points: one disease-free equilibrium P0 that is unstable, and another endemic equilibrium P(S,E,I) that is locally asymptotically stable, where (S,I) is presented numerically, E=β1SI(μ+φ)(1+α1S)(1+γI).

    Table 1.  Parameters and their definition.
    Symbol Definition
    Λ Total recruitment
    β1 Disease transmission rate
    φ Transition rate from E to I
    α1 Inhibition effect due to susceptible population
    γ Inhibition effect due to infected population
    μ Natural death rate
    ν Death rate due to disease

     | Show Table
    DownLoad: CSV

    In real life, the spread of diseases is inevitably affected by environmental white noise, as it is an integral part of nature, therefore, considering deterministic models no longer fits the actual needs. Some scholars have studied the dynamical behaviors of epidemic models affected by white noise, such as [16,17,18,19]. Hence, we incorporate white noise perturbations into model (1.1). We propose the following stochastic SEIR epidemic model with nonlinear incidence and treatment.

    {dS=[ΛμSβ1SI(1+α1S)(1+γI)]dt+σ1SdB1(t),dE=[β1SI(1+α1S)(1+γI)(μ+φ)E]dt+σ2EdB2(t),dI=[φEβ2I1+α2IauI1+buI(μ+ν)I]dt+σ3IdB3(t),dR=[β2I1+α2I+auI1+buIμR]dt+σ4RdB4(t), (1.2)

    where independent standard Brownian motions are expressed as Bi(t)(i=1,2,3,4), and σi(i=1,2,3,4) are positive constants that represent the intensity of the environment white noise, respectively.

    This paper is organized as follows: In Section 2, we prove that the proposed model has a unique positive solution and the solution is global. In Sections 3 and 4, we study the dynamical behaviors of the proposed model in terms of the existence of stationary distribution and the extinction of the disease, respectively. In Section 5, we discuss the optimal control problem of the proposed model. In Section 6, we give a series of numerical simulations. Finally, in Section 7, conclusions are given.

    Let (Ω,F,{Ft}t0,P) be a complete probability space with a filtration {Ft}t0 that satisfies the usual conditions (i.e., it is increasing and right continuous while F0 contains all P-null sets).

    Consider the stochastic differential equation (SDE) of n-dimensional of the form

    dX(t)=F(t,X(t))dt+G(t,X(t))dB(t), (1.3)

    where F(t,X):R+×RnRn and G(t,X):R+×RnRn×m are measurable functions and B(t) is Rm-valued standard Brownian motion. Given V(X,t)C2,1(Rn×R+,R+), we define the operator LV corresponding to the SDE (1.3) by

    LV=Vt(X,t)+Vx(X,t)F(X,t)+12trace[GT(X,t)Vxx(X,t)G(X,t)], (1.4)

    where

    Vt(X,t)=V(X,t)t,Vx(X,t)=(Vx1,Vx2,,Vxn),Vxx(X,t)=(2Vxixj)n×n.

    Then, the Itô formula is obtained:

    dV(X,t)=LV(X,t)dt+Vx(X,t)G(X,t)dB(t).

    In this section, using the Lyapunov analysis method[20], we first show that the system (1.2) has a unique local positive solution, then we show that this solution is global.

    Theorem 1. If (S(0),E(0),I(0),R(0))R4+ is any initial value to (1.2), then (S(t),E(t),I(t),R(t)) is a unique existing positive solution to (1.2) for t0 and the solution remains in R4+ with probability 1.

    Proof. Since the local Lipschitz condition is satisfied by system (1.2), for any initial value (S(0),E(0),I(0),R(0))R4+, there exists a unique local solution (S(t),E(t),I(t),R(t)) for t[0,τe), where τe denotes the explosion time [21]. To prove that the solution is global, we only need to prove τe=+ a.s. To this end, let k01 be a sufficiently large constant such that S(0),E(0),I(0), and R(0) lie within the interval [1k0,k0]. For kk0, we define the stopping time as follows:

    τk=inf{t[0,τe):min{S(t),E(t),I(t),R(t)}1k,ormax{S(t),E(t),I(t),R(t)}k}.

    Clearly, ττk a.s. If τ=+ a.s., then we have τe=+ a.s., and (S(t),E(t),I(t),R(t))R4+ a.s. If this is false, then there exists a pair of constants T>0 and ε(0,1) such that P{τT}>ε. Therefore, there is an integer k1k0 satisfying

    P{τT}ε,kk1. (2.1)

    Define the C2-function V1:R4+R4+:

    V1(S,E,I,R)=(S1lnS)+(E1lnE)+(I1lnI)+(R1lnR). (2.2)

    Applying the Itô formula, we obtain

    LV1=S1S[ΛμSβ1SI(1+α1S)(1+γI)]+E1E[β1SI(1+α1S)(1+γI)(μ+φ)E]+I1I[φEβ2I1+α2IauI1+buI(μ+ν)I]+R1R(β2I1+α2I+auI1+buIμR)+σ21+σ22+σ23+σ242=Λ+4μ+φ+ν+β1I(1+α1S)(1+γI)+β21+α2I+au1+buIμ(S+E+I+R)ΛSνIβ1SIE(1+α1S)(1+γI)φEIβ2IR(1+α2I)auIR(1+buI)+σ21+σ22+σ23+σ242Λ+4μ+φ+ν+β1γ+au+β2+σ21+σ22+σ23+σ242=K, (2.3)

    where K is a positive constant.

    The following proofs are similar to references[22].

    The unique stationary distribution of the stochastic SEIR model indicates that the persistence of the disease in the future under certain conditions, that is, the stochastic model fluctuates around the endemic equilibrium of the corresponding deterministic model.

    Let X(t) be a regular time-homogeneous Markov process described by the following stochastic differential equation in Rd:

    dX(t)=b(X)+kr=1hr(X)dBr(t).

    The diffusion matrix of the process X(t) is defined as follows:

    A(X)=(aij(x)),aij(x)=kr=1hirhjr.

    Lemma 1. [23] If there is a bounded open domain DEd with regular boundary Γ, it has the following properties:

    (i) The diffusion matrix A(x) is strictly positive definite for all xD;

    (ii) For any xEdD, it has a nonnegative C2 function V such that LV is negative.

    Then there exists a unique ergodic stationary distribution π() for the Markov process X(t).

    Theorem 2. For any initial value (S(0),E(0),I(0),R(0))R4+, the system (1.2) admits a unique ergodic stationary distribution π(), if

    Rs0=4Λβ1φμ(μ+ν)α1γ(μ+σ212)(μ+φ+σ222)(μ+ν+au+β2+σ232)(Λ+μα1+μ+νγ)2>1.

    Proof. In order to prove the theorem, we first verify that condition (i) in Lemma 1 holds. From (1.2), we obtain that the diffusion matrix of system (1.2) is

    A=(σ21S20000σ22E20000σ23I20000σ24R2).

    It is easy to see that the matrix A is positive definite for any compact subset of R4+. Therefore, condition (ⅰ) of Lemma 1 is satisfied.

    Next, we verify that the condition (ⅱ) in Theorem 2 also holds. Define C2 functions V1:R4+R:

    V1=c1lnSc2lnEc3lnI+c4(S+E+I+R),

    where

    c1=Λβ1φμμ+να1γ(μ+σ212)2(μ+φ+σ222)(μ+ν+au+β2+σ232),c2=Λβ1φμμ+να1γ(μ+σ212)(μ+φ+σ222)2(μ+ν+au+β2+σ232),c3=Λβ1φμμ+να1γ(μ+σ212)(μ+φ+σ222)(μ+ν+au+β2+σ232)2,c24=Λβ1φμμ+να1γ(μ+σ212)(μ+φ+σ222)(μ+ν+au+β2+σ232).

    Making use of the Itô formula, we obtain

    LV1=c1ΛSc2β1SIE(1+α1S)(1+γI)c3φEI+c1β1I(1+α1S)(1+γI)+c3β21+α2I+c3au1+buI+c1(μ+σ212)+c2(μ+φ+σ222)+c3(μ+ν+σ232)+c4[Λμ(S+E+I+R)νI]=c1ΛSc2β1SIE(1+α1S)(1+γI)c3φEIc4(μ+ν)(1+γI)γc4μ(1+α1S)α1c4μ(E+R)+c1β1I(1+α1S)(1+γI)+c3β21+α2I+c3au1+buI+c1(μ+σ212)+c2(μ+φ+σ222)+c3(μ+ν+σ232)+c4(Λ+μα1+μ+νγ)5(c1c2c3c24Λμβ1φμ+να1γ)15+c1(μ+σ212)+c2(μ+φ+σ222)+c3(μ+ν+au+β2+σ232)+c4(Λ+μα1+μ+νγ)+c1β1I(1+α1S)(1+γI)=2Λβ1φμ(μ+ν)α1γ(μ+σ212)(μ+φ+σ222)(μ+ν+au+β2+σ232)+c4(Λ+μα1+μ+νγ)+c1β1I(1+α1S)(1+γI)=c4(Λ+μα1+μ+νγ)(Rs01)+c1β1I(1+α1S)(1+γI). (3.1)

    Set V2(S,E,R)=lnSlnElnR. Then, we have

    LV2=1S[ΛμSβ1SI(1+α1S)(1+γI)]1E[β1SI(1+α1S)(1+γI)(μ+φ)E]1R(β2I1+α2I+auI1+buIμR)+σ21+σ22+σ242=ΛS+β1I(1+α1S)(1+γI)β1SIE(1+α1S)(1+γI)β2IR(1+α2I)auIR(1+buI)+3μ+φ+σ21+σ22+σ242. (3.2)

    Define

    V3(S,E,I,R)=S+E+I+R,
    V4(S,E,I,R)=1θ+1(S+E+I+R)θ+1.

    Then, we have

    LV3=Λμ(S+E+I+R)νI, (3.3)
    LV4(S+E+I+R)θ[Λμ(S+E+I+R)]+θ2(S+E+I+R)θ+1(σ21σ22σ23σ24)=[μθ2(σ21σ22σ23σ24)](S+E+I+R)θ+1+Λ(S+E+I+R)θG12[μθ2(σ21σ22σ23σ24)](Sθ+1+Eθ+1+Iθ+1+Rθ+1), (3.4)

    where

    G=sup(S,E,I,R)R4+{Λ(S+E+I+R)θ12[μθ2(σ21σ22σ23σ24)](S+E+I+R)θ+1}<.

    Define a C2 function Q:R4+R in the following form:

    Q(S,E,I,R)=MV1(S,E,I,R)+V2(S,E,R)+V3(S,E,I,R)+V4(S,E,I,R),

    where M>0 is sufficiently large and satisfies the condition

    Mc4(Λ+μα1+μ+νγ)(Rs01)+B2, (3.5)

    where

    B=sup(S,E,I,R)R4+{3μ+Λ+φ+μ(α1+γ)α1γ+σ21+σ22+σ242+G12(Sθ+1+Eθ+1+Iθ+1+Rθ+1)[μθ2(σ21σ22σ23σ24)]}<.

    In addition, Q(S,E,I,R) is continuous, and (S0,E0,I0,R0) is a minimum value point of Q(S,E,I,R) in R4+. Therefore, define a C2 function

    V(S,E,I,R)=Q(S,E,I,R)Q(S0,E0,I0,R0).

    Clearly, V is nonnegative. By the Itô formula and combining (3.2)(3.4), we get

    LVMc4(Λ+μα1+μ+νγ)(Rs01)+(Mc1+1)β1I(1+α1S)(1+γI)μ(S+E+I+R)νIβ1SIE(1+α1S)(1+γI)ΛSβ2IR(1+α2I)auIR(1+buI)+σ21+σ22+σ242+3μ+Λ+φ+G12[μθ2(σ21σ22σ23σ24)](Sθ+1+Eθ+1+Iθ+1+Rθ+1)Mc4(Λ+μα1+μ+νγ)(Rs01)+(Mc1+1)β1I(1+α1S)(1+γI)3(β1μ2SIEα1γ)13ΛSμ(E+R)νIβ2IR(1+α2I)auIR(1+buI)+μ(α1+γ)α1γ+σ21+σ22+σ242+3μ+Λ+φ+G12[μθ2(σ21σ22σ23σ24)](Sθ+1+Eθ+1+Iθ+1+Rθ+1). (3.6)

    The tectonic compact set is

    D={(S,E,I,R)R4+:ε1S1ε1,ε2E1ε2,ε3I1ε3,ε4R1ε4}.

    For the sake of discussion, let's divide R4+D into eight regions:

    D1={(S,E,I,R)R4+:0<S<ε1},D2={(S,E,I,R)R4+:ε1S,0<E<ε2,ε3I},D3={(S,E,I,R)R4+:ε1S,0<I<ε3},D4={(S,E,I,R)R4+:ε3I,0<R<ε4},D5={(S,E,I,R)R4+:S>1ε1},D6={(S,E,I,R)R4+:I>1ε3},D7={(S,E,I,R)R4+:E>1ε2},D8={(S,E,I,R)R4+:R>1ε4},

    where εi(0<εi<1,i=1,2,3,4) are positive numbers small enough to satisfy that the following conditions hold

    ε2=ε41,ε3=ε21,ε4=ε31, (3.7)
    Λε1+F<1, (3.8)
    3(β1μ2α1γε1)13+F<1, (3.9)
    Mc4(Λ+μα1+μ+νγ)(Rs01)+(Mc1+1)β1ε1α1+B<1, (3.10)
    β1ε1(1+α2ε21)auε1(1+buε21)+F<1, (3.11)
    12εθ+11[μθ2(σ21σ22σ23σ24)]+K<1, (3.12)
    12ε2(θ+1)1[μθ2(σ21σ22σ23σ24)]+K<1, (3.13)
    12ε4(θ+1)1[μθ2(σ21σ22σ23σ24)]+K<1, (3.14)
    12ε3(θ+1)1[μθ2(σ21σ22σ23σ24)]+K<1, (3.15)

    where

    F=sup(S,E,I,R)R4+{(Mc1+1)β1I(1+α1S)(1+γI)+3μ+Λ+φ+μ(α1+γ)α1γ+σ21+σ22+σ242+G12[μθ2(σ21σ22σ23σ24)](Sθ+1+Eθ+1+Iθ+1+Rθ+1)}<,
    K=sup(S,E,I,R)R4+{(Mc1+1)β1I(1+α1S)(1+γI)+3μ+Λ+φ+μ(α1+γ)α1γ+σ21+σ22+σ242+G}<.

    In the following, we prove that the eight regions have LV(S,E,I,R)1 for any (S,E,I,R)Dc.

    Case 1. For any (S,E,I,R)D1, by (3.8), we have

    LVΛS+(Mc1+1)β1I(1+α1S)(1+γI)+μ(α1+γ)α1γ+σ21+σ22+σ242+3μ+Λ+φ+G12[μθ2(σ21σ22σ23σ24)](Sθ+1+Eθ+1+Iθ+1+Rθ+1)ΛS+FΛε1+F<1.

    Case 2. On D2, by (3.7) and (3.9), we have

    LV3(β1μ2SIEα1γ)13+F3(β1μ2ε1ε3ε2α1γ)13+F=3(β1μ2α1γε1)13+F<1.

    Case 3. When (S,E,I,R)D3, by (3.7) and (3.10), we obtain

    LVMc4(Λ+μα1+μ+νγ)(Rs01)+(Mc1+1)β1Iα1S+BMc4(Λ+μα1+μ+νγ)(Rs01)+(Mc1+1)β1ε1α1+B<1.

    Case 4. On D4, by (3.7) and (3.11), we get

    LVβ2IR(1+α2I)auIR(1+buI)+Fβ2ε3ε4(1+α2ε3)auε3ε4(1+buε3)+F<1.

    Case 5. For any (S,E,I,R)D5, by (3.7) and (3.12), we have

    LV(Mc1+1)β1I(1+α1S)(1+γI)+μ(α1+γ)α1γ+σ21+σ22+σ242+3μ+Λ+φ+G12[μθ2(σ21σ22σ23σ24)](Sθ+1+Eθ+1+Iθ+1+Rθ+1)12[μθ2(σ21σ22σ23σ24)]1Sθ+1+K12[μθ2(σ21σ22σ23σ24)]1εθ+11+K<1.

    Case 6. On D6, by (3.7) and (3.13), we have

    LV12[μθ2(σ21σ22σ23σ24)]Iθ+1+K12[μθ2(σ21σ22σ23σ24)]1ε2(θ+1)1+K<1.

    Case 7. When (S,E,I,R)D7, by (3.7) and (3.14), we obtain

    LV12[μθ2(σ21σ22σ23σ24)]Eθ+1+K12[μθ2(σ21σ22σ23σ24)]1ε4(θ+1)1+K<1.

    Case 8. On D8, by (3.7) and (3.15), we get

    LV12[μθ2(σ21σ22σ23σ24)]Rθ+1+K12[μθ2(σ21σ22σ23σ24)]1ε3(θ+1)1+K<1.

    Thus, for sufficiently small positive numbers εi(i=1,2,3,4), we obtain

    LV1,(S,E,I,R)R4+D.

    Consequently, Theorem 2 holds.

    In this section, we will demonstrate that under certain assumptions, the disease will become extinct.

    Define a parameter

    Re0=φβ1+α1(μ+φ)(β2+au)α1(μ+φ)(μ+ν+β2+au)σ232(σ222+σ222σ232+σ224)μ+υ+β2+au.

    Theorem 3. Let (S(t),E(t),I(t),R(t)) be the solution of system (1.1) with any given initial value (S(0),E(0),I(0),R(0))R4+. If Re0<1, then

    limt+supln[φE(t)+(μ+φ)I(t)]t<0,a.s.,
    limt+R(t)=0,a.s.,

    that is to say, (E(t),I(t),R(t)) exponentially converges to (0,0,0) a.s.

    Proof. Let p(t)=φE(t)+(μ+φ)I(t). By the Itô formula, we obtain

    dlnp(t)=Llnp(t)+1φE+(μ+φ)I[φσ2EdB2(t)+(μ+φ)σ3IdB3(t)], (4.1)

    where

    Llnp(t)=1φE+(μ+φ)I{φβ1SI(1+α1S)(1+γI)(μ+φ)[β21+α2I+au1+buI+(μ+ν)]I}φ2σ222E2(φE+(μ+φ)I)2(μ+φ)2σ232I2(φE+(μ+φ)I)21φE+(μ+φ)I{[φβ1α1(μ+φ)+β2+au][(μ+φ)I+φE][φβ1α(μ+φ)+β2+au]φE(μ+φ)(μ+υ+β2+au)I}φ2σ222E2(φE+(μ+φ)I)2(μ+φ)2σ232I2(φE+(μ+φ)I)2=φβ1α1(μ+φ)+β2+auφEφE+(μ+φ)I[φβ1α1(μ+φ)+β2+au](μ+φ)(μ+ν+β2+au)IφE+(μ+φ)Iφ2σ222E2(φE+(μ+φ)I)2(μ+φ)2σ232I2(φE+(μ+φ)I)2=φβ1α1(μ+φ)+β2+au[φβ1α1(μ+φ)+β2+au+σ222η]φ2E2(φE+(μ+φ)I)2φ(μ+φ)[φβ1α1(μ+φ)+β2+au+μ+ν+β2+au]EI(φE+(μ+φ)I)2(μ+φ)2(μ+ν+β2+au+σ232η)I2(φE+(μ+φ)I)2η[φ2E2+(μ+φ)2I2](φE+(μ+φ)I)2. (4.2)

    By Re0<1, we get

    φβ1+α1(μ+φ)(β2+au)α1(μ+φ)<σ232(σ222+σ222σ232+σ224).

    There is 0<η<min{σ222,σ232}. Setting ˉσ222=σ222η, ˉσ232=σ232η, we can get

    φβ1α1(μ+φ)ˉσ232+μ+ν, (4.3)
    φβ1α1(μ+φ)ˉσ22+ˉσ22ˉσ23+ˉσ42+μ+ν. (4.4)

    Combining (4.3) and (4.4), we obtain

    ˉσ222φ2E2+(μ+φ)2[ˉσ232(φβ1α1(μ+φ)+β2+au)+μ+ν+β2+au]I22φ(μ+φ)ˉσ222[ˉσ232(φβ1α1(μ+φ)+β2+au)+μ+ν+β2+au]EIφ(μ+φ)[φβ1α1(μ+φ)+β2+au(μ+ν+β2+au)]EI. (4.5)

    By (4.1), (4.2), and (4.5), we get

    dlnp(t)η[φ2E2+(μ+φ)2I2](φE+(μ+φ)I)2dt+φσ2EdB2(t)+(μ+φ)σ3IdB3(t)φE(t)+(μ+φ)I(t). (4.6)

    Integrating both sides of (4.6) from 0 to t and dividing by t, we get

    lnp(t)lnP(0)tη2+1tt0φσ2E(s)dB2(s)φE(s)+(μ+φ)I(s)+1tt0(μ+φ)σ3I(s)dB3(s)φE(s)+(μ+φ)I(s). (4.7)

    According to (4.7), we have

    limt+supln[E(t)+μ+φφ(I(t)+R(t))]tη2<0,a.s.

    The upper formula indicates that

    limtE(t)=0, a.s. limtI(t)=0, a.s.

    According to (1.2), we get limtR(t)=0 a.s. That shows that (E(t),I(t),R(t)) exponentially converges to (0,0,0) a.s. We complete the proof of Theorem 3.

    If sustained control is implemented, the processing level will remain at a relatively high level over time. From the previous sections, we conclude that the cost eradicating the disease may be too high. In order to eliminate the disease within a limited time, time-dependent control should be considered.

    As in previous studies[24], using the stochastic maximum principle as in [25], we find the characteristics of optimal control problem of model (1.3). Our objective is to minimize both the number of infectious individuals and the cost of treatment control; thus, we establish the following objective function.

    J(U)=minuΓt10(AE(t)+BI(t)+Cu(t))dt,

    where A,B, and C, respectively, represent the weights of the relationship between the state variables E,I, and u. The control set is given by Γ={u is measurable and 0u(t)1, for t[0,t1]}. According to the stochastic control theory in the book[26] of ksendal, we need to find an optimal control variable u(t) that minimizes the objective functional when the initial state is x0. We define the expectation of the initial state x0 as

    E0,x0[t10(AE(t)+BI(t)+Cu(t))dt]. (5.1)

    Let's assume that there is a fixed constant ˉu(t) in the deterministic problem that ˉu(t)1 with u(t)ˉu(t) a.s. The class of admissible control laws is

    Π={u(t):uisadaptedand0u(t)1,a.s.}. (5.2)

    In order to obtain a solution of stochastic control, we define the expectation of the system at time t and a fixed value of x as follows:

    Js(t,x,u)=Et,x[t10(AE(t)+BI(t)+Cu(t))dt]. (5.3)

    Now, let's define the value function to be

    V(t,x)=infu()ΠJs(t,x,u)=Js(t,x,u).

    We now define the control law of minimizing the expected value of Js:ΠR+ given by (5.3). The present solution formulated is the solution of the stochastic analogue we now describe for the optimal control problem.

    Given the system (1.2) and Π as in (5.2) with Js as in (5.3), find the value of the function

    U(t,x)=infu()ΠJs(t,x,u), (5.4)

    and an objective function

    u=arginfu()ΠJs(x,u(t))Π.

    By the following theorem, the optimal control u(t) is obtained.

    Theorem 4. A solution to the optimal control problem presented in problem (5.2) is of the form

    u=min{1,max{1bI((UIUR)aIC1),0}}. (5.5)

    Proof. We calculate LU(t):

    LU(t)=[ΛμSβ1SI(1+α1S)(1+γI)]US(t)+[β1SI(1+α1S)(1+γI)(μ+φ)E]UE(t)+[φEβ2I1+α2IauI1+buI(μ+ν)I]UI(t)+(β2I1+α2I+auI1+buIμR)UR(t)+σ21S22USS(t)+σ22E22UEE(t)+σ23I22UII(t)+σ24R22URR(t)+σ1σ2SE2USE(t)+σ1σ3SI2USI(t)+σ1σ4SR2USR(t)+σ2σ3EI2UEI(t)+σ2σ4ER2UER(t)+σ3σ4IR2UIR(t). (5.6)

    Applying the Hamilton-Jacobi-Bellman theory[26], the minimum of (5.4) can be obtained as

    infu()Π[AE+BI+Cu+LU].

    In order to obtain the optimal control solution, consider the following expression:

    AE(t)+BI(t)+Cu(t)+LU(t). (5.7)

    Take the partial derivative of (5.7) with respect to u and set it equal to 0. Thus, the equation is obtained,

    CUIaI(1+buI)2+URaI(1+buI)2=0. (5.8)

    Considering the bounds of u, we can get an expression for u(t).

    In this section, we illustrate the theoretical results with example. By using Milstein's method[27], the discrete equations of system (1.3) are described by

    {Sk+1=Sk+(ΛμSkβ1SkIk(1+α1Sk)(1+γIk))Δt+σ1Skξ1kΔt+σ21Sk2(ξ21k1)Δt,Ek+1=Ek+(β1SkIk(1+α1Sk)(1+γIk)(μ+φ)Ek)Δt+σ2Ekξ2kΔt+σ22Ek2(ξ22k1)Δt,Ik+1=Ik+(φEkβ2Ik1+α2IkauIk1+buIk(μ+ν)Ik)Δt+σ3Ikξ2kΔt+σ23Ik2(ξ22k1)Δt,Rk+1=Rk+(β2Ik1+α2Ik+auIk1+buIkμR)Δt+σ4Rkξ2kΔt+σ24Rk2(ξ22k1)Δt,

    where ξ1k,ξ2k,ξ3k, and ξ4k(k=1,2,) are independent Gaussian random variables subject to N(0,1), and σi(i=1,2,3,4) is the intensity of white noise.

    We choose Λ=1.2,μ=0.004,β1=0.0134,β2=0.025,α1=0.09,α2=0.02,γ=0.015, φ=0.019,ν=0.02,a=0.052,b=0.01, the initial value S(0)=58,E(0)=15,I(0)=20,R(0)=20, and the step size Δt=0.01.

    In Figure 1, we choose u=0.66,σ1=σ2=0.05,σ3=0.04,σ4=0.1 to get that Rs0=1.0029>1, satisfying the condition of Theorem 2. The result of the graph is consistent with our conclusion in Theorem 2.

    Figure 1.  The solution S(t),E(t),I(t),R(t) of the model and its density function diagram.

    Figure 2 shows the stochastic epidemic system (1.2) with u=0.66,σ1=0.3,σ2=0.49, σ3=0.49, σ4=0.3, and we get that Re0=0.9945<1, which satisfies the conditions of Theorem 3; this is consistent with our conclusion in Theorem 3. When the intensities of white noises are sufficiently large, the disease of the stochastic epidemic system (1.2) is extinct.

    Figure 2.  The extinction of the solution E(t),I(t),R(t) of the model as u=0.66.

    Figure 3(a) shows the extinction image when u takes the variable and the other parameters take the same values as in Figure 2, while Figure 3(b) shows the corresponding trend of u varies with time t.

    Figure 3.  The extinction of the solution E(t),I(t),R(t) for varying the treatment control u and the trend of u changing with time t.

    The dynamic behaviors of a class of SEIR epidemic models are studied. First, we prove the existence and uniqueness of a global positive solution for the stochastic model. Second, we explore that the positive solution of the model has a stationary distribution, and we investigate the sufficient conditions for the extinction of the stochastic SEIR epidemic system. Furthermore, we aimed to minimize the total cost of infection and treatment expenses by studying optimal control strategies. The existence of optimal solutions is proved by using the stochastic maximum principle, and the dynamic behavior of the model affected by u is studied. It can be found through experiments that the disease becomes extinct faster when u takes variable values than when it takes constant values, and the number of infections has significantly decreased. In addition, we present the trend of u over time t when the disease is extinct. Based on the changing trends, the public health system can dynamically adjust treatment strategies.

    It is shown by detailed theoretical analysis that environmental white noise can control the spread of diseases to some extent, and different proportions of control therapies can be used at different times to achieve the purpose of controlling infectious diseases with the least cost. This provides a theoretical basis for the actual control of infectious diseases.

    Although we have studied the optimal treatment control of infectious diseases and can implement different proportions of treatment control according to different time periods. It is very difficult to find the optimal control measures because the dynamics of disease transmission are very complex and influenced by many factors. In addition, in practice, measures to control epidemics cannot be singular, and multiple measures should be considered to jointly control the spread of diseases. Therefore, the control effect of implementing multiple measures together should be studied in combination with reality.

    Jinji Du: conceptualization, investigation, methodology, formal analysis, original draft preparation; Chuangliang Qin: validation, formal analysis, writing-review and editing; Yuanxian Hui: resources, writing-review and editing. All authors have read and approved the final version of the manuscript for publication.

    This work is supported by the National Natural Science Foundation of China (No. 11971127), the Key Research Project of Higher School in Henan Province (Nos. 23B110014 and 25B110019) and the Science and Technology Innovation Youth Project of Zhumadian (No. QNZX202310).

    The authors declare that they have no competing of interests regarding the publication of this paper.



    [1] Z. Hu, W. Ma, S. Ruan, Analysis of SIR epidemic models with nonlinear incidencerate and treatment, Math. Biosci., 238 (2012), 12–20. https://doi.org/10.1016/j.mbs.2012.03.010 doi: 10.1016/j.mbs.2012.03.010
    [2] M. Khan, Q. Badshah, S. Islam, I. Khan, S. Shafie, S. Khan, Global dynamics of SEIRS epidemic model with non-linear generalized incidences and preventive vaccination, Adv. Differ. Equ., 2015 (2015), 88. https://doi.org/10.1186/s13662-015-0429-3 doi: 10.1186/s13662-015-0429-3
    [3] F. Rihan, H. Alsakaji, Dynamics of a stochastic delay differential model for COVID-19 infection with asymptomatic infected and interacting people: case study in the UAE, Results Phys., 28 (2021), 104658. https://doi.org/10.1016/j.rinp.2021.104658 doi: 10.1016/j.rinp.2021.104658
    [4] M. Abdoon, A. Alzahrani, Comparative analysis of influenza modeling using novel fractional operators with real data, Symmetry, 16 (2024), 1126. https://doi.org/10.3390/sym16091126 doi: 10.3390/sym16091126
    [5] C. Qin, J. Du, Y. Hui, Dynamical behavior of a stochastic predator-prey model with Holling-type Ⅲ functional response and infectious predator, AIMS Mathematics, 7 (2022), 7403–7418. https://doi.org/10.3934/math.2022413 doi: 10.3934/math.2022413
    [6] T. Kar, S. Jana, Application of three controls optimally in a vector-borne disease-mathematical study, Commun. Nonlinear Sci., 18 (2013), 2868–2884. https://doi.org/10.1016/j.cnsns.2013.01.022 doi: 10.1016/j.cnsns.2013.01.022
    [7] K. Hattaf, M. Rachik, S. Saadi, Y. Tabit, N. Yousfi, Optimal control of tuberculosis with exogenous reinfection, Applied Mathematical Sciences, 3 (2009), 231–240.
    [8] F. Gumaa, M. Abdoon, A. Qazza, R. Saadeh, M. Arishi, A. Degoot, Analyzing the impact of control strategies on VisceralLeishmaniasis: a mathematical modeling perspective, Eur. J. Pure Appl. Math., 17 (2024), 1213–1227. https://doi.org/10.29020/nybg.ejpam.v17i2.5121 doi: 10.29020/nybg.ejpam.v17i2.5121
    [9] J. Tchuenche, S. Khamis, F. Agusto, S. Mpeshe, Optimal control and sensitivity analysis of an influenza model with treatment and vaccination, Acta Biotheor., 59 (2011), 1–28. https://doi.org/10.1007/s10441-010-9095-8 doi: 10.1007/s10441-010-9095-8
    [10] K. Okosun, R. Ouifki, N. Marcus, Optimal control analysis of a malaria disease transmission model that includes treatment and vaccination with waning immunity, Biosystems, 106 (2011), 136–145. https://doi.org/10.1016/j.biosystems.2011.07.006 doi: 10.1016/j.biosystems.2011.07.006
    [11] H. Alsakaji, Y. El-Khatib, F. Rihan, A. Hashish, A stochastic epidemic model with time delays and unreported cases based on Markovian switching, Journal of Biosafety and Biosecurity, 6 (2024), 234–243. https://doi.org/10.1016/j.jobb.2024.08.002 doi: 10.1016/j.jobb.2024.08.002
    [12] X. Zhang, X. Liu, Backward bifurcation of an epidemic model with saturated treatment, J. Math. Anal. Appl., 348 (2008), 433–443. https://doi.org/10.1016/j.jmaa.2008.07.042 doi: 10.1016/j.jmaa.2008.07.042
    [13] S. Jana, S. Nandi, T. Kar, Complex dynamics of an SIR epidemic model with saturated incidence rate and treatment, Acta Biotheor., 64 (2016), 65–84. https://doi.org/10.1007/s10441-015-9273-9 doi: 10.1007/s10441-015-9273-9
    [14] M. Ali, F. Guma, A. Qazza, R. Saadeh, N. Alsubaie, M. Althubyani, et al., Stochastic modeling of influenza transmission: insights into disease dynamics and epidemic management, Partial Differential Equations in Applied Mathematics, 11 (2024), 100886. https://doi.org/10.1016/j.padiff.2024.100886 doi: 10.1016/j.padiff.2024.100886
    [15] R. Upadhyay, A. Pal, S. Kumari, P. Roy, Dynamics of an SEIR epidemic model with nonlinear incidence and treatment rates, Nonlinear Dyn., 96 (2019), 2351–2368. https://doi.org/10.1007/s11071-019-04926-6 doi: 10.1007/s11071-019-04926-6
    [16] M. Ali, S. Alzahrani, R. Saadeh, M. Abdoon, A. Qazza, N. Al-kuleab, et al., Modeling COVID-19 spread and non-pharmaceutical interventions in South Africa: a stochastic approach, Scientific African, 24 (2024), e02155. https://doi.org/10.1016/j.sciaf.2024.e02155 doi: 10.1016/j.sciaf.2024.e02155
    [17] F. Rihan, H. Alsakaji, C. Rajivganthi, Stochastic SIRC epidemic model with time-delay for COVID-19, Adv. Differ. Equ., 2020 (2020), 502. https://doi.org/10.1186/s13662-020-02964-8 doi: 10.1186/s13662-020-02964-8
    [18] Q. Yang, D. Jiang, N. Shi, C. Ji, The ergodicity and extinction of stochastically perturbed SIR and SEIR epidemic models with saturated incidence, J. Math. Anal. Appl., 388 (2012), 248–271. https://doi.org/10.1016/j.jmaa.2011.11.072 doi: 10.1016/j.jmaa.2011.11.072
    [19] S. Gani, S. Halawar, Optimal control analysis of deterministic and stochastic SIS epidemic model with vaccination, Int. J. Solids Struct., 12 (2017), 251–263.
    [20] X. Mao, G. Marion, E. Renshaw, Environmental Brownian noise suppresses explosions in population dynamics, Stoch. Proc. Appl., 97 (2002), 95–110. https://doi.org/10.1016/S0304-4149(01)00126-0 doi: 10.1016/S0304-4149(01)00126-0
    [21] X. Mao, Stochastic differential equations and applications, Chichester: Horwood Publishing, 2008.
    [22] L. Wang, D. Zhou, Z. Liu, D. Xu, X. Zhang, Media alert in an SIS epidemic model with logistic growth, J. Biol. Dynam., 11 (2017), 120–137. https://doi.org/10.1080/17513758.2016.1181212 doi: 10.1080/17513758.2016.1181212
    [23] R. Khasminskii, Stochastic stability of differential equations, 2 Eds., Heidelberg: Spinger-Verlag, 2011. https://doi.org/10.1007/978-3-642-23280-0
    [24] V. Rico-Ramirez, U. Diwekar, Stochastic maximum principle for optimal control under uncertainty, Comput. Chem. Eng., 28 (2004), 2845–2849. https://doi.org/10.1016/j.compchemeng.2004.08.001 doi: 10.1016/j.compchemeng.2004.08.001
    [25] S. Khare, K. Mathur, K. Das, Optimal control of deterministic and stochastic Eco-epidemic food adulteration model, Results in Control and Optimization, 14 (2024), 100336. https://doi.org/10.1016/j.rico.2023.100336 doi: 10.1016/j.rico.2023.100336
    [26] B ksendal, Stochastic differential equations: an introduction with applications, Berlin: Springer, 2003. https://doi.org/10.1007/978-3-642-14394-6
    [27] D. Higham, An algorithmic introduction to numerical simulation of stochastic differential equations, SIAM Rev., 43 (2001), 525–546. https://doi.org/10.1137/S0036144500378302 doi: 10.1137/S0036144500378302
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(656) PDF downloads(102) Cited by(0)

Figures and Tables

Figures(3)  /  Tables(1)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog