key: cord-0793513-0rovtkcm
authors: Fatima, BiBi; Zaman, Gul; Alqudah, Manar A.; Abdeljawad, Thabet
title: Modeling the pandemic trend of 2019 Coronavirus with optimal control analysis
date: 2020-12-09
journal: Results Phys
DOI: 10.1016/j.rinp.2020.103660
sha: e41ecddd77fc940ca2261fada8941cd315503b5f
doc_id: 793513
cord_uid: 0rovtkcm

In this work, we propose a mathematical model to analyze the outbreak of the Coronavirus disease (COVID-19). The proposed model portrays the multiple transmission pathways in the infection dynamics and stresses the role of the environmental reservoir in the transmission of the disease. The basic reproduction number [Formula: see text] is calculated from the model to assess the transmissibility of the COVID-19. We discuss sensitivity analysis to clarify the importance of epidemic parameters. The stability theory is used to discuss the local as well as the global properties of the proposed model. The problem is formulated as an optimal control one to minimize the number of infected people and keep the intervention cost as low as possible. Medical mask, isolation, treatment, detergent spray will be involved in the model as time dependent control variables. Finally, we present and discuss results by using numerical simulations.

The 2019 novel Coronavirus (COVID-19) is a new virus that causes respiratory illness in people. This virus probably, originally, emerged from an animal source but now seems to be spreading from person-to-person. On 12th December 2019, a pneumonia case of unknown etiology was reported in Wuhan City, Hubei Province, China, and on 31st December 2019, the disease outbreak was reported to World Health Organization (WHO) [1] [2] [3] [4] .

Cases have been detected in most countries worldwide and community spread is being detected in a growing number of countries. On March 11, the COVID-19 outbreak was characterized as a pandemic by the WHO [5, 6] . 

The reservoir for 2019 nCoronavirus (COVID-19) is denoted as W. We divide the population into six compartment: S p (t) is susceptible people; E p (t) is the exposed people; I p (t) is infectious people; infectious but asymptotic class A p (t); hospitalized H p (t); remover or recovery class R p (t). b is the birth rate and µ 0 is the death rate. The susceptible people will be infected through sufficient contact with W and I p and the transmission rate were defined by β p , β w . The transmissibility of A p was α times I p and that of hospitalize was q time I p .

This section, describe the (COVID-19) virus transmission model between reservoir and people and from people to people. This model contains a composition of differential equations. The compartmental deterministic mathematical model can be represented by nonlinear system of ordinary differential equations: dS p (t) dt = b − β p I p S p − β p αA p S p − β p qH p S p − β w WS p − µ 0 S p , dE p (t) dt = β p I p S p + β p αA p S p + β p qH p S p + β w WS p − (κ + µ 0 )E p , S p (0) > 0, E p (0) ≥ 0, I p (0) ≥ 0, A p (0) ≥ 0, H p (0) ≥, R p (0) ≥, W(0) ≥ 0, β p shows transmission per unit time, q shows the approximate transmissibility of hospitalized patient, κ is the progression at which individuals go to infectious class, ρ is the moving rate from exposed class E p to infectious class I p , (1 − ρ) is that of transmission to asymptotic class A p . The rate at which infected individuals are hospitalize is γ a and γ 1 is the recovery rate beyond hospitalization. The recovery rate of hospitalized patient is γ 2 , δ is the life time of virus reservoir.

We discuss qualitative study of the proposed model. For this we find equilibria of the model (1). In order to find the disease free equilibrium of the proposed model (1), we set the right hand side of all equations equal to zero and set E p = I p = A p = H p = R p = W = 0, we get F 0 is given by

, and it occur when the disease present in the population where:

A simple but effective measure of the transmissibility of an infectious disease is given by the basic reproduction number R 0 , defined as the total number of secondary infections produced by introducing a single infective in a completely susceptible population. In general, for simple epidemic models, if R 0 is greater than unity, an epidemic will occur while if R 0 is less than unity, an outbreak will most likely not occur.

J o u r n a l P r e -p r o o f

To find R 0 for our proposed model(1) we use the method of Driessche and Watmough [19] , we have

R 0 is therefore the spectral radius of next generation matrixH = FV −1 . Thus R 0 for our proposed model (1) becomes

where

The basic reproduction number R 0 of the proposed model consists of four parts, which represent the four different rout of transmission one from the exposed population, second is from the infected population, third from hospitalized population and fourth from the environmental reservoir to the susceptible population. The transmission from this four population classes shape the overall disease risk of COVID-19 pandemic.

In this section, we present sensitivity analysis of a few parameters which are used in the proposed model (1). This will makes it easier for us to know that parameters that have highly effect on the reproductive number. For this analysis we apply the technic given in [20] . Sensitivity index of basic reproductive number R 0 , is given by

where h is parameter. For our model (1) sensitivity analysis is given by: We show the local asymptotic stability of disease free equilibrium point and endemic equilibria of the system (1) in the following theorems.

Proof. The Jacobian matrix of the system at DFE point (S 0 , 0, 0, 0, 0, 0), is given by

Hence, the characteristic equation of matrix 3 is given by

According to Routh-Herwitz criteria, all the roots of the characteristic polynomial P(ζ) have negative real parts, which complete the proof [21, 22] . Theorem 2. If R 0 > 1, then the endemic equilibrium point E * is locally asymptotically stable, unstable for R 0 < 1.

Proof. Linearization of the model(1) around endemic equilibrium point E * is given by

where

Using elementary row transformation we get the following matrix :

where

. It is clear that all of the eigen values λ i , for all i = 1, 2, 3, 4, of J * have negative real parts for R 0 > 1 which completes the proof.

Theorem 3. For R 0 < 1 the disease free equilibrium of the system is stable globally asymptotically, unstable for R 0 > 1.

Proof. We define the following Lyapunov function, and show that this function satisfy the condition of lyapunov function that is function is positive definite and its derivative is negative definite,

Here d i where i = 1, 2, 3, 4, 5 are arbitrary constant, which are determined later by differentiating equation (7) , and using the system (1) we have

By choosing the positive parameter d 1 = d 2 = d 3 = QQ 1 , d 4 = 1 Q 2 , d 5 = µ 0 and after interpretation we get, [23, 24] , and E p = I p = A p = H p = W = 0. All the condition of Lyapunov function are satisfied that function is positive definite and its derivative is negative definite. Thus the disease free equilibrium is globally asymptotically stable in F 0 .

Theorem 4. If R 0 > 1, then the endemic equilibrium point E * is globally asymptotically stable and unstable otherwise.

J o u r n a l P r e -p r o o f Proof. To prove the global asymptotic stability of the proposed model (1) at endemic equilibrium E * , we use castilo chevez method [26, 27] let us consider the subsystem of (1):

Taking the Jacobean as well as the additive compound matrix of order 2 for the above system (8), which may take the form is given by:

Consider the function Q(χ) = Q(S p , E p ,

Now

where Let (c 1 , c 2 , c 3 ) be a vector in R 3 and the . of (c 1 , c 2 , c 3 ) given by,

Now we take the Lozinski measure described by [28] ,

Hence model (1) is globally asymptotically stable.

We solved the proposed deterministic model using Runge-e-Kutta method of order four [25] . This verify our analytical results. The variable and parameter value in Table. 1 were used for simulation. For the purpose of illustrations, we assumed some parameters values. The choice of parameters are taken in the way as to be biologically feasible. The time interval is taken 0 − 250 units with initial population for susceptible people S p (t), exposed people E p (t), infected with COVID-19 I p (t), asymptomatic people A p (t), hospitalize people H p (t), and recovered people R p (t), reservoir for COVID-19 W(t).

The application of Runge-e-Kutta method of order 4th on the proposed model leads to the following system:

J o u r n a l P r e -p r o o f

Step1:

Step2: for i = 1, 2...n − 1.

Step 3: for i = 1, 2, 3, ..., n − 1, write S *

When we run the above algorithm by using Matlab software, we get the graphs presented in Fig.3 and Fig.4 , which represent the dynamics of susceptible population (S p (t)); Exposed population (E p (t)); infected with COVID-19 (I p (t)); asymptomatic population (A p (t)); hospitalized population (H p (t)); recovered population (R(t)); and reservoir (W(t)). The biological interpretation of this results show that if R 0 < 1, then the susceptible population decreases, then become stable and shows that there will be always stable susceptible population. The dynamics of exposed, infected, asymptomatic, hospitalize , recover and reservoir for COVID-19 conclude that the number of these populations will decrease and reach to zero, which show the stability of the proposed model. Exposed Population (Ep) 14 J o u r n a l P r e -p r o o f

We formulate control strategies on the basis of sensitivity analysis and dynamic of the proposed model. The maximum sensitivity index parameter is (β p , β w ) whose value is (0.9398437 , 0.601562) increase in this parameter by 10 percent would increase the threshold quantity by( 9.939 and 6. 01562. Therefore to control the spread of the disease we need to minimize this parameters by taking the control variable u 1 (t) and u 2 (t) representing (awareness about medical mask, hand washing and isolation of infected and non infected people). Moreover the parameters , δ, µ 0 decrease the threshold quantity by 10 percent by increasing this parameter, to increase this we use the control variables u 3 (t), u 4 (t) representing oxygen therapy, mechanical ventilation and detergent spray.

Our goal here are to reduce COVID-19 in the population through increasing the number of recovered person R(t) and decreasing the number of infectious I(t), and hospitalized H(t), environmental reservoir W(t) by applying the time dependent control variables u 1 (t), u 2 (t), u 3 (t), u 4 (t).

i. u 1 (t) is the time dependent control variable representing the awareness about medical mask hand washing.

ii. u 2 (t) is the time dependent control variable representing isolation of infected people.

iii. u 3 (t) is the time dependent control variable representing oxygen therapy mechanical ventilation.

iv. u 4 (t) represent the time dependent control variable for environmental reservoir i,e detergent spray.

By using this control variables our optimal control problem which is modified version of (1) become

with initial condition

The goal here is to show that it is possible to implement time dependent control measures while minimizing the cost of implementation of those techniques [29] . We choose the objective (cost) function by 15 J o u r n a l P r e -p r o o f

dt. (16) In equation (16) ν 1 , ν 2 , ν 3 , ν 4 , ν 5 , ν 6 , ν 7 , ν 8 , represent weight constant. The weight constant ν 1 , ν 2 , ν 3 , ν 4 , represent relative cost of infectious person I p , asymptomatic person A p , hospitalized person H p and reservoir W while ν 5 , ν 6 , ν 7 , ν 8 , represents the associated cost of control variables . 1 2 ν 5 u 2 1 , 1 2 ν 6 u 2 2 , 1 2 ν 7 u 2 3 , 1 2 ν 8 u 2 4 describes self care, isolation, medical treatment, and detergent spray.

Our purpose is to find an optimal control pair u *

dependent on system (3), we define the control set,

Let us take the control system (15) along initial condition at time t = 0 and reveal the presence of the control problem. Where as bounded Lebesgue measurable controls, positive initial conditions and positive bounded solutions to the state system occur [30] . To assest the optimal solution, we go back to the optimal control problem (15), (16) . First we use the Lagrangian and Hamiltonian considering the optimal control problem (15) and (16) . We define the Lagrangian in the following equation, L(S p , E p , I p , A p , H p , R p , W p u 1 , u 2 , u 3 , u 4 ) = ν 1 I p + ν 2 A p + ν 3 H p + ν 4 W + 1 2 (ν 5 u 2 1 (t) + ν 6 u 2 2 (t) + ν 7 u 2 3 (t)

For the smallest value of the Lagrangian, we determine Hamiltonian H as,

Hence in this way presence of control problem, we consider the following consequent.

To proceed further, first we show that these control (u 1 (t), u 2 (t), u 3 (t), u 4 (t)) are exists. We follow the result of , which demonstrate Theorem 5. There exist an optimal control u * = (u * 1 , u * 2 , u * 3 , u * 4 ) ∈ U, to the control problem as stated in Eqs (15)- (16) .

Proof. In order to prove the presence of an optimal control, using the result in [31] . Since the control variables and the state variables are positive values. It is also noted that the control variables set U is convex and closed by statement. Moreover the control system is bounded which state the compactness of the problem. The integrand in , ν 1 I p + ν 2 A p + ν 3 H p + ν 4 W + 1 2 (ν 5 u 2 1 (t) + ν 6 u 2 2 (t) + ν 7 u 2 3 (t) + ν 8 u 2 4 (t) is also convex with respect to control set U. Which guarantee about the existence of the optimal control (u * 1 , u * 2 , u * 3 , u * 4 ). 16 J o u r n a l P r e -p r o o f

In order to characterize an optimal solution to (15) and (16) . First we use the Lagrangian and Hamiltonian considering the optimal control problem (15) and (16) . Indeed the Lagrangian illustrate the optimal control problem is presented by the following equation,

We define the associated Hamiltonian (H), therefore using the notion λ = (λ 1 , λ 2 , λ 3 , λ 4 , λ 5 , λ 6 ) and F = (F 1 , F 2 , F 3 , F 4 , F 5 , F 6 , F 7 ) then, For the smallest value of the Lagrangian, we determine Hamiltonian H for the optimal control problem as,

where, x = (S p , E p , I p , A p , H p , R p W), λ = (λ 1 , λ 2 , λ 3 , λ 4 , λ 5 , λ 6 , λ 7 ).

and F(x, u) = F 1 (x, u), F 2 (x, u), F 3 (x, u), F 4 (x, u), F 5 (x, u), F 6 (x, u), F 7 (x, u). Following the Pontryagin'n Maximum Principle [31, 32] for finding the optimal solution to the proposed control problem (15) . Using (x * , u * ) as a notation for the optimal solution then,

The maximality condition H(t, x * (t), u * (t), λ(t))∂x = max u 1 ,u 2 ,u 3 ,u 4 ∈[0,1] H(x * (t), u 1 , u 2 , u 3 , u 4 , λ(t)); (20) define the transversally condition as λ(t f ) = 0 (21)

J o u r n a l P r e -p r o o f = 0.0071; κ = 0.00041; β w = 0.0000123; b = 0.003907997; α = 0.98; γ 1 = 0.0000404720925; γ a = 0.000431; γ 2 = 0.00135; δ = 0.017816; ρ = 0.00007; φ 1 = 0.05; φ 2 = 0.06; q = 0.00997; µ 0 = 0.014567125;. This parameters are chosen in such away that are biologically more feasible. Furthermore, the weight constants are assumed to be ν 1 = 0.6610000; ν 2 = 0.54450; ν 3 = 0.0090030; ν 4 = 0.44440; ν 6 = 0.3550; ν 7 = 0.67676; ν 8 = 0.999. The obtained results are presented from figure 5-6. Figures 5-6 shows the dynamic of susceptible, exposed individuals, infected individuals, asymptomatic individuals, hospitalize individuals , recovered individuals, and reservoir with and with out control. 

In every disease the role of threshold parameter is very important for transmission potential of diseases. We have found threshold quantity R 0 by using the next generation matrix method. We have used Routh Hurwitz criteria for the local stability of the proposed model, while for the global stability we have used the Lyapunov function theory and geometrical approach. We further, have used optimal control strategy to minimize infected people and maximize the number of recovered people in the population. Medical mask, isolation, treatment and detergent spray have been in

Writing original draft, review and editing

World Health Organization. Novel Coronavirus-China, Disease outbreak news : Update

Wuhan coronavirus real-time update

China Confirms Human-to-Human Transmission of 2019-nCoV

Transmission of 2019-nCoV Infection from an asymptomatic contact in Germany

From SARS-CoV to Wuhan 2019-nCoV Outbreak: Similarity of Early Epidemic and Prediction of Future

Middle East respiratory syndrome coronavirus: a casecontrol study of hospitalized patients

Clinical course and outcomes of critically ill patients with Middle East respiratory syndrome coronavirus infection

WHO statement regarding the outbreak of novel coronavirus (2019-nCoV)

Contributions to the mathematical theory of epidemics

A fractional-order epidemic model with time-delay and nonlinear incidence rate

Numerical simulations of a delay model for immune system-tumor interaction

Fractional order mathematical modeling of COVID-19 transmission

A novel algorithm for time-fractional foam drainage equation

Application of a hybrid method for systems of fractional order partial differential equations arising in the model of the one-dimensional Keller-Segel equation

Nowcasting and forecasting the potential domestic and international spread of the 2019-nCoV outbreak originating in Wuhan, China: a modelling study

Novel coronavirus 2019-nCoV: early estimation of epidemiological parameters and epidemic predictions, medRxiv

Estimation of the transmission risk of 2019-nCoV and its implication for public health interventions

Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission

Modeling and control of the hepatitis b virus spreading using an epidemic model

Stability analysis and optimal vaccination of an SIR epidemic model

Global dynamics of vector borne disease with horizontal transmission in host population

The stability of dynamical system

Stability of nonautonomous system

The transmission dynamic and optimal control of acute and chronic hepatitis B

A geometric approach to global-stability problems

Mathematical approaches for emerging and reemerging infectious diseases: an introduction

Logrithmic norms and projections applied to linear differential system

Analyzing the MERS disease control strategy through an optimal control problem

Ordinary Differential Equation

North-Holl, Amsterdam, The Netherlands optimal control problem

The pontryagin maximum principle and optimal economic growth problems, Proceedings of the Steklov institute of mathematics

This research was funded by the Deanship of Scientific Research at Princess Nourah bint Abdulrahman University through the Fast-track Research Funding Program.

Theorem 6. Let the optimal state variables and control variables are denoted by S * p , E * p , I * p , A * p , H * p R * p , W * be optimal state (u * 1 , u * 2 , u * 3 , u * 4 ) for the optimal control problem (15) and (16) . Then the set of adjoint variables λ(t) satisfyingthe transversality conditions (Boundary conditions) is define as,More over, the controls variables u *Proof: The adjoint system (22) comes from the direct application of the Pontryagin Maximum Principle (20) , while the transversal conditions are the direct consequences of λ(T) = 0. For the set of optimal functions u * 1 , u * 2 , u * 3 and u * 4 , we used ∂H ∂u . We solve the optimality system numerically in the subsequent section. Because it would be easy in understanding for the reader rather then analytical results. The optimality system are characterized by the control system (15), the adjoint system (22) , boundary (terminal) conditions, together with the optimal control functions. Clearly, the simulation carried out justified our control strategies to minimize the infected population, asymptomatic population, hospitalize and reservoir, and to maximize the susceptible and recovered population as shown in Fig.5-6. 

Here we solve the optimal control system (15) to see the impact of medical mask, isolation, treatment and detergent spray by using the Runge-Kutta method of order four. We use forward Runge-kutta procedure to solve the state system 16 with initial condition in time [0, 50]. Now to solve the adjoint system 22 we use backward Runge-kutta procedure with transversality condition and the solution of system 16. We use the following parameters for the simulation purposes: 18 J o u r n a l P r e -p r o o f

The authors declare that there are no conflicts of interest regarding the publication of this paper.