Application of the Modified Adomian Decomposition Method on a Mathematical Model of COVID-19

In this study, we constructed and analysed a mathematical model of COVID-19 in order to comprehend the transmission dynamics of the disease. The re-production number ( C R ) was calculated via the next generation matrix method. We also used the Lyaponuv method to show the global stability of both the disease free and endemic equilibrium points. The results showed that the disease-free equilibrium point is globally asymptotically stable if 1 C R < and the endemic equilibrium point is globally asymptotically stable if 1 C R > . We further used the Adomian decomposition method and the modified Adomian decomposition method to obtain the solutions of the model. Numerical analysis of the model was done using Sagemath 9.0 software.


Introduction
Mathematical disease models are important tools in analyzing the spread and the control of infectious diseases.These models are based on dividing the host population into compartments, each containing individuals that are identical in terms of their status with respect to the disease under consideration.For example, in the SIR model there are three compartments namely Susceptibles (S), Infections (I) and Removed (R).From the compartments, the set of equations will arise.The equations specify how the sizes of the compartment change overtime, and are linear and nonlinear.Solutions of these equations will yield, S(t) the size (R) are used to represent the classes.Individuals are recruited into the susceptible class through the rate Θ.The susceptible population is exposed to the disease through contact with symptomatic and asymptomatic infectious individuals.
The parameters s ρ and a ρ represent the effective contact rates for individu- als in the symptomatic and asymptomatic infectious classes, respectively.There is a fraction of individuals who use face masks in the population and it is given as 0 1 κ < ≤ , whereas 0 1 ω < ≤ represents the expected decrease in the risk of infection as a result of using face masks.The exposed individuals progress to infectious classes at the rate σ .A fraction 0 1 q < ≤ of the exposed shows no symptoms and they proceed to asymptomatic infectious class whereas the remaining fraction ( ) shows symptoms of the disease and hence proceeds to the symptomatic class.The symptomatic infectious are hospitalized at the rate of π and are quarantined at the rate of γ .The asymptomatic infectious recover at the rate τ .The quarantined are hospitalized at the rate ε and they recover at the rate ν whereas the hospitalized recover at the rate β .There is a natural death rate of δ for individuals in all classes.Additionally, individuals in the symptomatic, hospitalized, asymptomatic, and quarantined classes have disease-induced death rates of 1 α , 2 α , 3 α , and 4 α , respectively.The summary of the description of the variables and parameters is given in Table 1.
Using the symbols and variables described in Table 1 we draw the compartmental model that shows the progression of the disease, given in Figure 1.
Using the variables, parameters and the compartmental model we derive the system of ordinary differential model equations as follows: The total population is denoted by N and given as, and d d N t is given by ( ) The natural death rate , , , The disease induced death rates for the symptomatic, hospitalized, Asymptomatic and quarantined individuals respectively

Positivity of Solutions
Since the variables of model equation system (1) represent humans, it is important to show that they are positive.
Theorem 1 If the initial values are given by ( ) Thus S is positive since ( ) is positive and the exponential function is always positive.Using the same method, we can prove the rest of the equations of system Equation (1) and show that ( )

Boundedness of the Solutions
Theorem 2 All positive solutions presented in Theorem (1) are bounded.Proof: In the absence of the disease, i.e.
Hence, the positive solutions of model Equation (1) are bounded.

Existence and Stability of Equilibrium Points
The equilibrium points are calculated by equating the left hand side of equation system (1) to zero.This leads to the following system of equations, )

The Disease Free Equilibrium Point ( fep d )
In the absence of infection 0 Solving the equations of system 5, we obtain ,0,0,0,0,0,0

The Reproduction Number of the Model
We compute the basic reproduction number C R of the model system (1) by following [10].The basic reproduction number measures the average number of new infections generated by a single infected person in a completely susceptible population.Using the next generation matrix method, we consider the classes that are actively transmitting the disease.These are: Next, we find the matrices i F and i V which are the rate of appearance of new infections in compartment i and transfer of individuals into and out of compartment i by all other means, respectively, given ) ( ) ( ) Then taking partial derivatives of both i F and i V on the disease-free equilibrium point we get, ( ) ( ) ( ) Next we calculate the inverse of V which is given by: The next generation matrix is given as the following product: and we calculate the eigenvalues of 1 FV − as follows: )( ) The reproduction number is given by the largest eigenvalue of the determinant of the matrix )( ) Equation ( 7

Global Stability of the Disease-Free Equilibrium Point
The global stability of the disease-free equilibrium R < , the disease-free equilibrium is globally asymptotically stable in Ω and unstable if where, ( )( ) From Equation ( 9), the derivative is given as The expansion of Equation ( 11) yields, ) ( ) ( ) Equation ( 12) can be simplified to s a Principle [11], we conclude that the disease-free equilibrium ( fep D ) of the model of COVID-19 is globally asymptotically stable in Ω whenever 1 C R < .

The Endemic Equilibrium Point
The endemic equilibrium point is denoted by , , , , , ,

Global Stability Analysis of the Endemic Equilibrium Point
The Lyapunov asymptotic stability theorem is used to prove the global asymptotic stability of the endemic equilibrium point.Using the method for constructing the Lyapunov function discussed in [12], we formulate the Lyapunov function for model Equation (1).
) ) , , , , , , ln The derivative of L along the solutions of the model in Equation ( 1) is given by the expression: From Equation ( 4), all solutions of Equation ( 13 L is positive definite and L  is negative definite, hence the function L is the Lyapunov function for model Equation ( 1) and the endemic equilibrium ep E is globally asymptotically stable by the Lyapunov asymptotic stability analysis [13].Hence the proof.

Adomian Decomposition Method
In this section we apply the Adomian decomposition method proposed by George Adomian in the mid 1980's, [14] to solve the system Equation (1).

Basic Concepts of the Adomian Decomposition Method
Let us consider the initial value problem expressed as where L is the linear operator, N is the nonlinear operator and R is the remaining linear part.By defining the inverse operator of L as 1 L − , we introduce it on both sides of Equation ( 17) to get, [ ] Solving for u in Equation ( 18) leads to, where, ( ) The Adomian Decomposition Method assumes that the unknown function u can be expressed by an infinite series of the form, where the component n u will be determined recursively.This method also defines the nonlinear term by the Adomian polynomials.More precisely, the ADM assumes that the nonlinear operator can be decomposed by an infinite series of polynomials given by, ( ) where n A 's are the Adomian's polynomials defined as, ( ) , , , , and are calculated using the formular in [15] 0 0 Substituting Equation (20) and Equation ( 21) into Equation ( 19) and using the fact that R is a linear operator we obtain, ( ) ( ) Therefore the formal recurrence algorithm could be defined by, The n-term approximation of the solution is given by, ( ) ( ) The advantage of this method is that it solves problems in a direct way and in uncomplicated manner without linearization, perturbation or any unpreferable assumptions that may change the physical behaviour of the problem under discussion.

The Modified Adomian Decomposition Method
Here, we present the Modified Adomian Decomposition Method (MADM) which we use to solve the system of equations arising from the mathematical model of COVID-19.In [16], the authors proposed the modification of the ADM by introducing the terms into the calculations of the standard ADM.To understand the procedure of MADM we consider Equation (23) and insert to obtain: ( ) ( ) So the recursive algorithm for MADM is defined as, ( In this method we set 1 0 u = and 1 p = so that we solve for the coefficients n a 's for 0,1, 2, n =  .The approximation of the solution is found by replacing the coefficients in the solution equation given by: The advantage of using MADM is that it reduces the computational size of the problem being solved because it involves the calculation of 0 A only.

Solution of the Mathematical Model of COVID-19 Using the Modified Adomian Decomposition Method
Applying Equation ( 27) on each equation of system (1), we obtain the recursive relationship for each equation as follows: ( ) ( ) ( ) ( ) ( ) ( ) and setting 1 p = we find the n a 's for 1, 2,3, n =  .and replace in Equation (29) to write the solution for equation system (1).

Results and Discussion
The results obtained from solving system Equation (1) using ADM are compared with the results obtained using MADM.We will use COVID-19 data for Zambia [17] as initial values and are given as ( )

Conclusion
The model of COVID-19 has been formulated and its dynamic behaviour inves-tigated.We showed that the population classes are non-negative.Using the next generation matrix, we calculated the basic reproduction number, C R , which is useful in guiding control strategies.By constructing Lyapunov functions, we proved global stability of disease free and endemic equilibrium points.If the basic reproduction number is less than 1, all solutions converge to the disease free equilibrium point and the disease dies out from the population.When the basic reproduction number is greater than 1, the endemic equilibruim point is globally stable, meaning that the disease will persist and the number of infected individuals tends to be a positive constant.Lastly, it is well known that analytical solutions of nonlinear ordinary differential equations are difficult to find.In this paper, we solved nonlinear and linear Ordinary Differential Equations (ODEs) using the ADM and MADM.It can be seen from numerical solutions in Figure 2 that we demonstrated the ability of ADM and MADM in solving ODEs.The series solutions converge very rapidly and from the graphs in Figure 2, we can conclude that the ADM and MADM are very efficient and accurate methods in solving nonlinear mathematical models.

3 )Figure 1 .ρFraction of the exposed who showed symptomsπ
Figure 1.Compartmental model of the transmission of COVID-19.Table 1. Symbols and description of parameters.Parameters/Variables Description S ) consists of two reproduction numbers.The first reproduction number defines the number of new COVID-19 cases generated from the symptomatic infected individuals in class s I .The second reproduction number is number of new COVID-19 cases generated from the asymptomatic infectious individuals in class a I .Hence the reproduction number is written as, .

Theorem 4 If 1 CR
> , then the endemic equilibrium point ep E of model Equation (1) is globally asymptotically stable in the region Ω.
natural death rate, δ is calculated by taking the reciprocal of the average life expectancy (in months).In Zambia the life expectancy is 64.70 years [18]of the parameters are estimated from the literature as given in Table 2. Using the initial values and the parameter values we draw the graphs using SageMaths 9.0 software.The results are shown in Figures 2(a)-(g).

Figure 2 .
Figure 2. Numerical analysis of MADM and ADM solutions of the model of COVID-19 (12)g Equation(12)for calculating the Adomian polynomials for