In this research endeavor, we undertake a comprehensive analysis of a compartmental model for the monkeypox disease, leveraging the Atangana–Baleanu fractional derivative framework. Our primary objective is to investigate the effectiveness of a range of control strategies in containing the transmission of this infectious ailment. The parameterization of the model is executed meticulously via the application of the maximum likelihood estimation technique. Our study involves a rigorous mathematical analysis of the considered model, which encompasses an exploration of the existence and uniqueness of solutions, as well as the establishment of conditions ensuring the compactness and continuity of these solutions. Subsequently, we embark on an extensive stability analysis of the model, complemented by the computation of both the effective and basic reproduction numbers. These calculations are instrumental in illuminating the long-term behavior of the epidemic. Additionally, we perform a sensitivity analysis of the basic reproduction number to discern the influence of various factors on disease transmission dynamics. To derive our numerical results, we implement the Adams–Bashforth predictor–corrector algorithm tailored for the Atangana–Baleanu fractional derivatives. We employ this numerical technique to facilitate the simulation of the model under a spectrum of fractional-order values, offering a visual representation of our findings. Our study underscores the pivotal roles of infection awareness, vaccination campaigns, and effective treatment in significantly curtailing disease transmission, thus contributing valuable insight to the field of epidemiology.

## I. INTRODUCTION

The monkeypox disease caused by the virus MPVX (the monkeypox virus) was detected for the very first time in the year 1958, but it was confused with smallpox until the beginning of the 1970s when it was clearly identified as monkeypox. Since the year 2002, specifically after May, at least 15 000 reports of monkeypox-infected cases have been recorded in every continent except for Antarctica. The infection is transmitted through direct contact. It can spread from infected rodents to humans and from infected humans to symptomatic ones. Since the outbreak of monkeypox, in several parts of Africa, researchers have started working on its transmission dynamics (TeWinkel, 2019; Bankuru , 2020; Leandry and Mureithi, 2023). According to the observation, the population of synanthropic rodents has rapidly grown in Africa, which has thereby led to the spread of the infection in various regions. Monkeypox, according to the WHO, is a variant of Arthur pox virus belonging to the same family as that of smallpox; however, the noted symptoms of monkeypox in comparison with smallpox are not that contagious and extreme. The most evidently observed symptoms of monkeypox include headaches, muscular pain, lymph nodes becoming swollen, fever, tiredness, rashes, and lesions (Usman and Adamu, 2017). Normally, the rate of mortality due to the infection is 10% of the infected population. A major number of deaths are recorded among children who are young primarily within the age group of 10 years. Five to twenty days is generally considered as the incubation time period. A total number of 86 516 confirmed cases of monkeypox have been recorded all over 113 countries as of March 18, 2023. In a span of 1 year, from May 2022 to 2023, the infection has widely spread across various regions of Asia, America, Africa, Europe, and Oceania. This was the first instance of the infection out-breaking rapidly in countries outside Central and West Africa. As of now, no absolutely clear and effective treatment methods have been found to control the transmission of the infection; however, vaccination does play a significant role in disease prevention as observed in a considerable number of cases. Antiviruses like vaccinia immune globulin, Tecovirimat, and Brincidofovir are proven effective in controlling the disease spread to a significant extent. Earlier, the vaccination given for smallpox had served 85% efficiently in containing the monkeypox infection; however, due to the successful eradication of smallpox worldwide, this control strategy can no longer be implemented. Due to lack of attention, not much study has been done on the transmission dynamics of this disease in the past.

Fractional calculus has garnered notable attention in recent years, emerging as a pivotal domain of mathematics distinguished by its rigorous applications across diverse facets of mathematical epidemiology (Atangana and Owolabi, 2018; Atangana and Qureshi, 2019; Biswas , 2022; Li , 2021; Baleanu , 2018). This prominence is attributed to the prevalence of fractional-order differential equations, which leverage non-local operators (Toufik and Atangana, 2017; Solís-Pérez , 2018). Such operators are quintessential and widely adopted tools in the realm of mathematical modeling. The salient distinction of fractional-order systems is their capacity to encompass a broader spectrum of problems, offering enhanced clarity and precision compared to traditional integer-order systems when investigating an array of real-world challenges within the purview of mathematical epidemiology (Liu , 2020; Majee , 2023). Furthermore, an inherent advantage of fractional operators, in contrast to their ordinary counterparts (Pandey 2013; Prasad and Bali, 2023; Saifuddin , 2016, 2017; Samanta, 2017; Samanta 2013; Stauch 2011), is their intrinsic hereditary properties, rendering them particularly appealing in scientific applications (Abro, 2020; Vijayalakshmi and Roselyn Besi, 2022).

Generally, when we consider fractional calculus, it is just an extended version of ordinary calculus (Agaba 2017; Ahmed 2021; Agusto, 2017; Ayinla 2021; Biswas 2017a; Cai 2017) where the order of the derivatives or integrals can be taken arbitrary as real or complex values (Yu , 2023; Huang and Yu, 2023; Parmar , 2023; Tasman, 2015; Zuo 2022). The motivation behind replacing ordinary calculus with fractional operators is that the latter can prove to be much more efficient in modeling various real-world problems specifically when the dynamics of the system keep changing with respect to the inherent constraints defined for the given system (Jan , 2019; Chinnathambi , 2021; Kumar , 2019) Also, fractional derivatives can be local and non-local in nature. Local fractional derivatives have recently found great relevance in mathematical modeling primarily due to their ability to retain certain significant characteristics of ordinary derivatives; however, these local operators do tend to lose the memory property, which is a usual hereditary characteristic of non-local fractional-order derivatives making them more significant (Panhwer , 2022; Abro , 2023). That is why, we have used the Atangana–Baleanu fractional derivative for our analysis as it is non-local in nature and has Mittag–Leffler function as its kernel; it is nonsingular as well (Aslam , 2021; Atangana and Koca, 2016; Qureshi and Jan, 2021; Srivastava , 2021; Roy and Pop, 2021).

Mathematical modeling involving fractional calculus has been a very significant area of study for scientists and epidemiologists working in this field (Din 2022; Du 2013; Lia 2017; Moore 2019; Muhammad Altaf and Atangana, 2019). Some of the prominent research in this domain can be found in the works of (Alzubaidi , 2023; Somma , 2019; Adel , 2023; Adom-Konadu, 2023). Mathematical models described by a system of fractional differential equations tend to possess a greater degree of freedom than ordinary differential equations (Dye and Wolpert, 1988; Geweke, 1991; Haario 2006; Haario 2001; Khan and Atangana, 2022; Laine, 2008; Lenhart and Workman, 2007; Okosun and Makinde, 2014); that is why they are considered more efficient in studying the dynamics of any particular disease for a given set of data as well as for capturing memory effects (Baleanu , 2020; Ali , 2021). In the present times, a wide range of fractional derivatives have been defined and employed for the mathematical modeling of diseases primarily due to their much more advanced characteristics, which helps in a better analysis of various practical real-world scenarios (Podlubny, 1999; Samko , 1993). Previously, various mathematical and statistical techniques had been used to discuss the impact of non-local operators and the fading memory effect by employing fractional differential equations. Various operators with fractional order have been implemented using local and non-local kernels (Jajarmi and Baleanu, 2018; Odibat, 2006, 2010). The Caputo and Riemann–Liouville fractional operators have one drawback; the kernel function used a singular despite being non-local. This singularity can be a hindrance when solving real-world problems. To overcome this drawback, Atangana and Baleanu defined a fractional order with a new operator, which involves the Mittag–Leffler function as its kernel, which is nonsingular in nature. So, in addition to all the properties of Caputo and Riemann–Liouville fractional derivatives, the kernel being nonsingular helps in a better understanding of the system's dynamics while dealing with practical mathematical models (Rahman , 2021; Rezapour , 2020).

A lot of work has been previously done by researchers to study the dynamics of monkeypox infection using both deterministic and fractional systems of differential equations. Okyere and Ackora-Prah (2023) in their work showed the impacts of using fractional-order derivatives on various compartments of their model under study. The influence of fractional derivatives in the dynamics of the monkeypox disease was highlighted, and the numerical simulation showed the conditions under which the infection could be eradicated in a span of five days. Adel (2023) in their work had investigated the dynamics of a novel fractional-order monkeypox model with optimal control. Their results provided a better insight into various previously used prevention and control strategies and suggested some new strategies for controlling disease transmission. Peter (2022, 2023) emphasized the mathematical modeling of the monkeypox disease using fractional derivatives; they determined the equilibrium points, conducted the stability analysis, and formulated various optimal control strategies for the prevention and control of the infection. They offered numerous parameters for controlling the infection, which will be of tremendous help to the community's policymakers. In their research, Alharbi (2022) described a novel fractional model of monkeypox and did the stability analysis to get a better understanding of the crucial features of the monkeypox virus, thereby helping the decision-makers in containing the disease transmission.

The distinctive contribution of our research is grounded in the novel approach we have employed to formulate a compartmental model for the monkeypox disease. In this innovative model, we subdivide the human population into distinct classes, predicated on their awareness or lack thereof regarding the infection. This stratification based on the awareness factor represents a noteworthy departure from conventional modeling practices, where the human population has traditionally been treated as a uniform entity in prior studies. Consequently, our research underscores the paramount significance of incorporating awareness dynamics into the modeling of the monkeypox disease. Furthermore, we have introduced vaccination as a pivotal control parameter within our model to assess its efficacy in curbing the transmission of the monkeypox disease. Within this novel framework, we have meticulously crafted a compartmental model for monkeypox, employing fractional derivatives as a mathematical foundation. Our research extends beyond model formulation, delving deeply into the intricate dynamics of this infectious disease. Our exploration advances the understanding of monkeypox transmission dynamics by introducing these distinctive elements, which have hitherto been underrepresented in the literature.

The structural organization of the remaining part of this paper is as follows. In Sec. II, we expound upon the foundational tenets and essential concepts of fractional calculus, providing the necessary theoretical underpinning for our subsequent modeling endeavors. Section III encompasses the pivotal formulation of our model. Initially, we construct a deterministic monkeypox model utilizing a system of ordinary differential equations. Subsequently, we refine this model by introducing fractional-order differential equations, employing the Atangana–Baleanu derivative framework [as outlined in the works of Baleanu (2018), Atangana and Owolabi (2018), and Danane (2020)]. Section IV delves into a comprehensive exploration of the fundamental properties intrinsic to our model. These properties encompass the determination of the invariant region, the establishment of conditions guaranteeing the existence and uniqueness of solutions, and the assurance of compactness and continuity of the model. Section V is dedicated to an exhaustive analysis of the model's equilibrium points, computation of both basic and effective reproduction numbers, and a detailed stability analysis. In Sec. VI, we elucidate the intricacies of model calibration, followed by an in-depth sensitivity analysis presented in Sec. VII. Section VIII offers a comprehensive presentation of our numerical simulations, accompanied by graphical representations of our results, thus providing a visual dimension to our findings. Finally, in Sec. IX, we draw a definitive conclusion, summarizing the key outcomes and insights derived from our research endeavors.

## II. FUNDAMENTAL DEFINITIONS OF FRACTIONAL CALCULUS

Here, we shall discuss some fundamental definitions and concepts of fractional calculus (Abdeljawad , 2020; Kumar , 2022).

*Definition 1*. Let us consider a function h(t) and a scalar q such that $ 0 < q < 1$; then, we can define the Atangana–Baleanu derivative in the Caputo sense (ABC derivative) in the following manner:

The normalization function is denoted by $ G ( q )$, and it satisfies $ G ( 0 ) = G ( 1 ) = 1$. Here, $ E q$ denotes the Mittag–Leffler function of a single parameter defined in the following manner: $ E q ( u ) = \u2211 k = 0 \u221e u k \Gamma ( q k + 1 )$, where $ 0 < q < 1$.

*Definition 2*. Let us consider a function h(t) and a scalar q such that $ 0 < q < 1$, we can define the Atangana–Baleanu integral that corresponds to the ABC derivative in the following manner:

*Definition 3*. Here, we shall give the definition of the Laplace transform of the Atangana–Baleanu fractional derivative taken in the Caputo sense.

## III. MODEL FORMULATION

Here, we shall develop a compartmental model to describe the dynamics of monkeypox disease. For this, we have divided the human population into nine compartments, namely, susceptible unaware humans *S _{U}*, susceptible aware humans

*S*, exposed humans

_{A}*E*, asymptomatically infected humans

_{H}*I*, symptomatically infected humans

_{A}*I*, hospitalized humans

_{H}*H*, recovered humans

_{H}*R*, unaware humans that took the vaccination

_{H}*V*, and vaccinated aware humans

_{U}*V*. Now coming to the rodent population, we have subcategorized it into two compartments, which are susceptible vector (rodents)

_{H}*S*and infected vector (rodents)

_{V}*I*. So, we have constructed our model with a total of 11 state variables.

_{V}Humans and rodents are recruited at a constant rate of *π _{H}* and

*π*, respectively, into the population.

_{V}*λ*denotes the fraction of susceptible humans that are aware. $ \beta , \beta 1 , \beta 2$, and

*β*

_{3}denote the effective contact rate between rodents and unaware humans, unaware humans and other humans, aware humans and other humans, and infected rodents and other symptomatic rodents, respectively. Here,

*ρ*

_{1}denotes the movement rate of asymptomatically infected humans to either the hospitalized class or recovered class.

*θ*

_{2}denotes fraction of asymptomatically infected humans that are hospitalized and the remaining gets recovered.

*η*

_{1}denotes the rate of infection of exposed humans. The fraction of these infected humans that become symptomatic is given by

*θ*

_{1}, and the remaining ones move into the asymptomatically infected class. Next,

*α*

_{1}gives the rate of treatment given to symptomatic infected human population of which a fraction

*η*

_{2}gets recovered and the remaining are further hospitalized. The rate of vaccination given to the human population is denoted by

*u*

_{1}and the efficiency rate of the vaccine by

*σ*

_{1}.

*θ*

_{3}is the fraction of aware humans that are vaccinated. r denotes the rate of recovery of hospitalized humans.

*μ*and

_{h}*μ*are the natural mortality rates of the human and vector population, respectively, whereas

_{v}*δ*and

*delta*

_{1}denote the disease induced death rate of the human and the rodent population, respectively.

*N*is the total human population and

_{H}*N*is the total vector (rodent) population. They satisfy the following equations:

_{V}## IV. BASIC PROPERTIES

The parameters of the above model (2) are assumed to be strictly non-negative. Also, since this model takes into consideration the living population, the state variables are also assumed to be non-negative at the initial time t = 0.

### A. Invariant region

**Lemma 1**. The initial data are assumed as follows: $ P H ( 0 ) \u2265 0 , \u2009 P V ( 0 ) \u2265 0$, where $ P H ( t )$ and $ P V ( t )$ are given by

So, under the ABC derivative, the two closed sets $ P H ( 0 ) \u2265 0 , P V ( 0 ) \u2265 0$ are positively invariant for the model (2).

**Proof**. To obtain the dynamics of the model (2) for the human and vector population, we shall sum up the compartmental equations for these two populations separately in the following manner (Atangana and Owolabi, 2018):

Now, for the given initial conditions and t > 0, the solutions of the model (3) lying in *P _{H}* and

*P*continue to stay in

_{V}*P*and

_{H}*P*, respectively. This shows that the regions

_{V}*P*and

_{H}*P*will serve as an attractor for all the solutions belonging to $ R 9 +$ and $ R 2 +$ and hence is positively invariant.

_{V}### B. Existence and uniqueness of the solution

*S*, we will get the following results:

_{U}Thus, observing the system of equations we got above, we can conclude that $ Q 1 , Q 2 , Q 3 , Q 4 , Q 5 , Q 6 , Q 7 , Q 8 , Q 9$, *Q*_{10}, and *Q*_{11} do satisfy the Lipschitz condition. Here, $ c 1 , c 2 , c 3 , c 4 , c 5 , c 6 , c 7 , c 8 , c 9 , c 10 , c 11$ represent the corresponding Lipschitz constants for the equations given above.

**Theorem.**

*The solution of the monkeypox fractional compartmental model (2) will be unique in nature if for*$ 0 \u2264 t \u2264 T$

*, it will satisfy the condition as follows:*

**Proof**. Previously, we had established the boundedness of $ S U ( t ) , \u2009 S A ( t ) , \u2009 E H ( t ) , \u2009 I A ( t ) , \u2009 I H ( t ) , \u2009 H H ( t ) , \u2009 R H ( t ) , \u2009 V U ( t ) , \u2009 V A ( t ) , \u2009 S V ( t )$, and $ I V ( t )$. We have also observed from Eqs. (4) and (6) that

*Q*

_{1},

*Q*

_{2},

*Q*

_{3},

*Q*

_{4},

*Q*

_{5},

*Q*

_{6},

*Q*

_{7},

*Q*

_{8},

*Q*

_{9},

*Q*

_{10}, and

*Q*

_{11}evidently satisfy the Lipschitz criterion. By employing the set of equations given in Eq. (6) along with a recursive formula, we get

*j*, in view of triangle inequality and using the set of Eqs. (6), we deduce that

Here, the symbol $ U i$ is given by the expression $ U i = 1 \u2212 q G ( q ) r i + x q G ( q ) \Gamma ( q ) r i < 1$ according to the hypothesis.

So, from the above results, we can observe that the sequences $ S U k , S A k , E H k , I A k , I H k , H H k , R H k , V U k , V A k , S V k , I V k$ are Cauchy sequences in the Banach space K(Z) that we had defined earlier. So, by nature, they are uniformly convergent, and hence, as we apply on these sequences the limit as $ n \u2192 \u221e$ we will obtain exactly one solution, which establishes the uniqueness criterion.

Therefore, through this we can finally conclude that the solution for the given system of fractional differential equations for the monkeypox model (3) exists and is unique.

### C. Compactness and continuity

Let us first define an operator F as follows: $ F = ( F 1 , F 2 , F 3 , F 4 , F 5 , F 6 , F 7 , F 8 , F 9 , F 10 , F 11 ) : \mathbb{Z} \u2192 \mathbb{Z}$

We will now go on to prove the compactness and continuity of the set $ B = ( B 1 , B 2 , B 3 , B 4 , B 5 , B 6 , B 7 , B 8 , B 9 , B 10 , B 11 )$ defined above.

$C$ being a positive constant. Therefore, $ | | B ( S U , S A , E H , I A , I H , H H , R H , V U , V A , S V , I V ) | | \u2264 C$.

This establishes the uniform bounded property of the operator $B$.

As $ t 1 \u2192 t 2$, we get $ | | B ( B 1 , B 2 , B 3 , B 4 , B 5 , B 6 , B 7 , B 8 , B 9 , B 10 , B 11 ) ( t 2 ) \u2212 B ( B 1 , B 2 , B 3 , B 4 , B 5 , B 6 , B 7 , B 8 , B 9 , B 10 , B 11 ) | | ( t 1 ) | | \u2192 0$. Therefore, $B$ does not depend on $ ( S H , E H , I A , I H , T H , P H , R H , S R , I R , S V , I V ) .$ This establishes the equicontinuity of the operator $B$.

In order to establish the compactness and continuity parts, we employ the following theorem:

$ Arzela \u2212 Ascoli \u2009 Theorem :$ Let us consider a region Ω be in $\u2102$, and let $F$ be a family of complex-valued functions on Ω that is pointwise bounded. Then every sequence $ { f n}$ in $F$ has a subsequence that converges to a continuous function on Ω, the convergence being uniform on compact subsets. Since $B$ is completely continuous, according to the Arzela–Ascoli theorem, it is relatively compact. This completes the result.

## V. EQUILIBRIUM POINTS AND THEIR STABILITY ANALYSIS

Here, we shall perform a rigorous analysis of the model (2) to find the disease-free and endemic equilibrium points and further perform their stability analysis.

### A. Disease-free equilibrium point

To find the disease-free equilibrium (DFE) point, we shall equate the right-hand sides of the equations of the model (3) to 0. The infected variables are also equated to zero to compute this equilibrium point. On doing this, the DFE point that we get for the system of fraction differential equations (3) is given by the following manner.

### B. Basic reproduction number and effective reproduction number

#### 1. Basic reproduction number

*I*. We shall denote the right-hand side of the infected classes of the model (3) using two matrices $F$ and $V$ as defined below such that

_{V}*F*and

*V*:

#### 2. Effective reproduction number

To assess the transmission dynamics of the disease under investigation in a timely and scientifically rigorous manner, we employ the concept of the effective reproduction number, denoted as *R _{t}*. Similar to the basic reproduction number, $ R 0$, we estimate

*R*by multiplying the fraction of the susceptible host population by $ R 0$. Estimating

_{t}*R*effectively necessitates consideration of several key parameters, including the number of infected cases, the chosen serial time interval, and the time of symptom onset in infected individuals.

_{t}*R*is a crucial metric for understanding the current state of the epidemic. When

_{t}*R*equals 1, it signifies that the disease is in an endemic phase, where the number of infected cases remains stable over time. Conversely, when

_{t}*R*is less than or equal to 1, it indicates that the number of infected cases is declining, eventually leading to the eradication of the disease from the population under specific conditions.

_{t}*R*essentially represents the average number of secondary cases generated by a single primary case within a population at a given point in time, which allows us to assess the potential for disease transmission. It is important to note that

_{t}*R*is dynamic and may change as the immunity of the population changes over time due to various factors, including vaccination and natural infection. In our analysis, we make the assumption that the distribution of the serial interval follows a gamma distribution with a mean of 8.5, which is a common approach in epidemiological modeling. In Fig. 1, we have calculated the daily values of the effective reproduction number,

_{t}*R*, using the moving average method. This plot illustrates the changing dynamics of disease transmission over time. Our model, which incorporates specific control measures, suggests that the infection rate declines to below 1 within a few months. This decline in

_{t}*R*indicates a reduction in the spread of the disease, indicating that, under the applied measures, it is possible to control and manage the epidemic over a defined time period.

_{t}### C. Local stability of the disease-free equilibrium point

**Theorem.** *The DFE point E _{0} is locally asymptotically stable if the reproduction number* $ R 0 < 1$

*and is unstable if*$ R 0 > 1$ (Ghosh 2016; Pal 2015).

**Proof**. For the system (2), the Jacobian matrix J computed at the DFE point

*E*

_{0}is given as follows:

We can observe that the eigenvalues of the Jacobian matrix at the disease-free equilibrium point are all negative. The eigenvalues being all negative indicate that the disease-free equilibrium point $ E 0$ is locally asymptotically stable whenever $ R 0 < 1$. From this, we can conclude that the transmission of the monkeypox disease can be contained within the given population under the condition *R*_{0} being less than unity and the initial population size taken for the fractional-order system lies in the basin of attraction of the equilibrium point $ E 0$ The proof is thus established.

### D. Attractive solution

**Theorem.** *An attractive solution will exist for the fractional-order model (2) provided that the zero solution* $ \Phi ( t ) = 0$ *will satisfy the equation* $ lim t \u2192 \u221e y 0 = 0$ *for* $ | | y 0 | | < \u03f5$.

**Proof**. We have previously shown that pure model (2) possesses a solution, which is unique. We can now define the zero solution as follows:

Thus, $ | | ( S U , S A , E H , I A , I H , H H , R H , V U , V A , S V , I V ) | | \u2264 \u03f5 \u21d2 lim t \u2192 \u221e ( S U , S A , E H , I A , I H , H H , R H , V U , V A , S V , I V ) ( t ) = 0$.

This shows that the solution will be attractive in nature and therefore asymptotically stable.

### E. Hyer–Ulam's stability

The investigation into the stability of functional equations can be traced back to a seminal inquiry posed by Stanislaw Ulam ca. 1940, with the primary focus on the stability of homomorphisms within groups. Approximately 1 year later, D. H. Hyer embarked on the exploration of this inquiry by leveraging the concept of additive mappings between Banach spaces. He harnessed the “contraction mapping theorem” to establish foundational results, marking the inception of significant advancements in the examination of the problem initially proposed by Ulam. Subsequently, in 1978, another distinguished researcher named Rassias undertook this question and extended and generalized the proofs and outcomes originally put forth by Hyer. The concept of the Hyer–Ulam stability has seen extensive research application in the realms of both ordinary and fractional differential equations. In the context of mathematical modeling for diseases, the analysis of stability assumes paramount importance. Moreover, the Hyer–Ulam stability framework is highly regarded as an indispensable tool for scrutinizing the dynamics of nonlinear fractional models. This form of stability analysis proves exceptionally valuable when addressing a wide spectrum of practical problems for which obtaining an exact solution is a formidable challenge. In such scenarios, the Hyer–Ulam stability framework facilitates the approximation of solutions, enabling the dynamic analysis of fractional-order systems.

In order to determine the stability of the solution of the model (2), we consider a particular kind of differential inequality described below.

**Theorem.**

*Let us consider the matrix*$H$

*as follows:*

**Proof**. We have already shown that the fractional-order system (3) has a unique solution. Now, let us consider two different solutions of the fractional-order model (2) of the form:

*C*denotes the maximum value of the Lipschitz constants, i.e., $ C = max { c 1 , c 2 , c 3 , c 4 , c 5 , c 6 , c 7 , c 8 , c 9 , c 10 , c 11}$. So,

Keeping in mind, the definition of *h*_{1}, *h*_{2}, *h*_{3}, *h*_{4}, *h*_{5}, *h*_{6}, *h*_{7}, *h*_{8}, *h*_{9}, *h*_{10}, and *h*_{11}, we can observe that the above expression will converge to zero. For the above matrix, the eigenvalues are as follows: $ e 1 = 0 , \u2009 e 2 = 0 , \u2009 e 3 = 0 , \u2009 e 4 = 0 , \u2009 e 5 = 0 , \u2009 e 6 = 0 , \u2009 e 7 = 0 , \u2009 e 8 = 0 , \u2009 e 9 = 0 , \u2009 e 10 = 0$ and $ e 11 = h 1 + h 2 + h 3 + h 4 + h 5 + h 6 + h 7 + h 8 + h 9 + h 10 + h 11$. We get the spectral radius $ S = max { | e s | , s = 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11} = h 1 + h 2 + h 3 + h 4 + h 5 + h 6 + h 7 + h 8 + h 9 + h 10 + h 11 < 1$. Therefore, we can say that the fractional-order monkeypox model is Hyer–Ulam's stable as a result of the discrete dichotomy shown above.

## VI. MODEL CALIBRATION USING MAXIMUM LIKELIHOOD ESTIMATOR

Maximum likelihood estimation (MLE) stands as a pivotal and widely endorsed statistical tool employed for parameter estimation, as documented by Myung (2003), Portet (2020), and Zeng and Lin (2007). Its robustness is particularly evident when dealing with nonlinear models and data sets that do not conform to the normal distribution. To employ MLE, one must have a likelihood function, which serves as the foundation for making statistical inferences. In practice, when armed with a set of observed data and a mathematical model of interest, the challenge lies in deriving a suitable probability density function that accurately represents the data generation process. This endeavor can be viewed as an inverse problem and can be effectively addressed through the formulation of a likelihood function. Detailed information on the calibration of the model using the maximum likelihood estimation can be found in the Appendix, accompanied by a visual representation of the data set. In our specific application, we employ MLE to calibrate model (1) to the annual incidence data of monkeypox cases in the United States. The parameters under estimation include *β*, *β*_{1}, *β*_{2}, *β*_{3}, *η*_{1}, and *η*_{2}. Figure 2 illustrates the curve representing the model fitting achieved through the maximum likelihood estimation technique. This fitting process is employed to estimate the aforementioned parameters based on the recently recorded cases of monkeypox infection in the United States. Furthermore, Table I provides a comprehensive listing of the parameter values employed within our monkeypox model. This meticulous calibration using MLE enables us to refine the model's parameters and align it more closely with the empirical data, thus enhancing our understanding of the dynamics of monkeypox transmission in the United States (Table II).

Parameters . | Descriptions . | Values . | Sources . |
---|---|---|---|

λ | Fraction of susceptible humans that are aware | 0.25 | Assumed |

π _{H} | Constant rate of recruitment for humans | π* $ N H ( 0 ) day \u2212 1$ _{H} | |

μ _{h} | Human mortality rate | $ 0.0105 \u2009 day \u2212 1$ | The World Bank (2012) |

π _{V} | Constant rate of recruitment for vector | $ 0.2 \u2009 day \u2212 1$ | Peter (2022) |

μ _{v} | Vector mortality rate | $ 0.002 \u2009 day \u2212 1$ | Peter (2022) |

δ | Death rate of humans due to monkeypox | 0.003 25 day^{−1} | Assumed |

δ_{1} | Death rate of rodents due to monkeypox | 0.0064 day^{−1} | Assumed |

α_{1} | Movement rate from symptomatic infected human population | 0.08 day^{−1} | Assumed |

β | Effective contact rate of rodents to unaware humans | 0.000 24 day^{−1} | Estimated |

β_{1} | Effective contact rate of unaware humans to other humans | 0.3380 day^{−1} | Estimated |

β_{2} | Effective contact rate of aware humans to other humans | 0.000 002 3 day^{−1} | Estimated |

β_{3} | Effective contact rate of rodents to rodents | 0.1073 day^{−1} | Estimated |

ρ_{1} | Movement rate from asymptomatically infected humans | 0.001 | Assumed |

η_{1} | Movement rate from exposed humans | 0.92 day^{−1} | Estimated |

η_{2} | Fraction of symptomatic humans that are recovered | 0.43 day^{−1} | Estimated |

θ_{1} | Fraction of exposed humans that become symptomatic | 0.99 | Assumed |

θ_{2} | Fraction of asymptomatically infected humans that are hospitalized | 0.5 | Assumed |

θ_{3} | Fraction of aware humans that are vaccinated | 0.05 | Assumed |

u_{1} | Vaccination rate of humans | 0–1 | Assumed |

σ_{1} | Efficiency rate of vaccine | 0–1 | Assumed |

r | Recovery rate of hospitalized individuals | 0.036 | Assumed |

Parameters . | Descriptions . | Values . | Sources . |
---|---|---|---|

λ | Fraction of susceptible humans that are aware | 0.25 | Assumed |

π _{H} | Constant rate of recruitment for humans | π* $ N H ( 0 ) day \u2212 1$ _{H} | |

μ _{h} | Human mortality rate | $ 0.0105 \u2009 day \u2212 1$ | The World Bank (2012) |

π _{V} | Constant rate of recruitment for vector | $ 0.2 \u2009 day \u2212 1$ | Peter (2022) |

μ _{v} | Vector mortality rate | $ 0.002 \u2009 day \u2212 1$ | Peter (2022) |

δ | Death rate of humans due to monkeypox | 0.003 25 day^{−1} | Assumed |

δ_{1} | Death rate of rodents due to monkeypox | 0.0064 day^{−1} | Assumed |

α_{1} | Movement rate from symptomatic infected human population | 0.08 day^{−1} | Assumed |

β | Effective contact rate of rodents to unaware humans | 0.000 24 day^{−1} | Estimated |

β_{1} | Effective contact rate of unaware humans to other humans | 0.3380 day^{−1} | Estimated |

β_{2} | Effective contact rate of aware humans to other humans | 0.000 002 3 day^{−1} | Estimated |

β_{3} | Effective contact rate of rodents to rodents | 0.1073 day^{−1} | Estimated |

ρ_{1} | Movement rate from asymptomatically infected humans | 0.001 | Assumed |

η_{1} | Movement rate from exposed humans | 0.92 day^{−1} | Estimated |

η_{2} | Fraction of symptomatic humans that are recovered | 0.43 day^{−1} | Estimated |

θ_{1} | Fraction of exposed humans that become symptomatic | 0.99 | Assumed |

θ_{2} | Fraction of asymptomatically infected humans that are hospitalized | 0.5 | Assumed |

θ_{3} | Fraction of aware humans that are vaccinated | 0.05 | Assumed |

u_{1} | Vaccination rate of humans | 0–1 | Assumed |

σ_{1} | Efficiency rate of vaccine | 0–1 | Assumed |

r | Recovery rate of hospitalized individuals | 0.036 | Assumed |

Variables . | Descriptions . | Values . | Sources . |
---|---|---|---|

$ N H ( 0 )$ | Initial total human population | 332 403 650 | U.S. Department of Commerce (2023) |

$ S U ( 0 )$ | Initial susceptible unaware human population | $ 2.29 \xd7 10 7$ | Estimated |

$ S A ( 0 )$ | Initial susceptible aware human population | $ N H ( 0 ) \u2212 S U ( 0 ) \u2212 E H ( 0 ) \u2212 I A ( 0 ) \u2212 I H ( 0 ) \u2212 V U ( 0 ) \u2212 V A ( 0 )$ | |

$ E H ( 0 )$ | Initial population of exposed humans | 206 | Estimated |

$ I A ( 0 )$ | Initial population of humans infected with asymptomatic monkeypox | 0 | Assumed |

$ I H ( 0 )$ | Initial population of humans infected with symptomatic monkeypox | 0 | Assumed |

$ H H ( 0 )$ | Initial population of hospitalized humans | 0 | Assumed |

$ R H ( 0 )$ | Initial population of recovered humans | 0 | Assumed |

$ N V ( 0 )$ | Initial total rodent population | $ S V ( 0 ) + I V ( 0 )$ | |

$ S V ( 0 )$ | Initial susceptible rodent population | $ 1.82 \xd7 10 7$ | Estimated |

$ I V ( 0 )$ | Initial population of rodent that are infected with monkeypox | $ 2.3041 \xd7 10 4$ | Estimated |

Variables . | Descriptions . | Values . | Sources . |
---|---|---|---|

$ N H ( 0 )$ | Initial total human population | 332 403 650 | U.S. Department of Commerce (2023) |

$ S U ( 0 )$ | Initial susceptible unaware human population | $ 2.29 \xd7 10 7$ | Estimated |

$ S A ( 0 )$ | Initial susceptible aware human population | $ N H ( 0 ) \u2212 S U ( 0 ) \u2212 E H ( 0 ) \u2212 I A ( 0 ) \u2212 I H ( 0 ) \u2212 V U ( 0 ) \u2212 V A ( 0 )$ | |

$ E H ( 0 )$ | Initial population of exposed humans | 206 | Estimated |

$ I A ( 0 )$ | Initial population of humans infected with asymptomatic monkeypox | 0 | Assumed |

$ I H ( 0 )$ | Initial population of humans infected with symptomatic monkeypox | 0 | Assumed |

$ H H ( 0 )$ | Initial population of hospitalized humans | 0 | Assumed |

$ R H ( 0 )$ | Initial population of recovered humans | 0 | Assumed |

$ N V ( 0 )$ | Initial total rodent population | $ S V ( 0 ) + I V ( 0 )$ | |

$ S V ( 0 )$ | Initial susceptible rodent population | $ 1.82 \xd7 10 7$ | Estimated |

$ I V ( 0 )$ | Initial population of rodent that are infected with monkeypox | $ 2.3041 \xd7 10 4$ | Estimated |

## VII. SENSITIVITY ANALYSIS

In this section, we conduct a sensitivity analysis for the given fractional-order model (2). The primary objective of this analysis is to ascertain the key parameters that exert the most significant influence on the control of disease transmission (Marino 2008). This approach is based on the works of Peter (2023), Baithalu (2023), and Zarin (2022) and combines two well-established techniques: Latin hypercube sampling (LHS) and partial rank correlation coefficient (PRCC) multivariate analysis. To perform this sensitivity analysis, we randomly sample 1000 parameter sets. Our focus is on identifying the parameters that have a substantial impact on the basic reproduction number, $ R 0$. From the analysis of Fig. 3, it becomes evident that the parameters *β*, *β*_{1}, *α*_{1}, and *η*_{1} exhibit sensitivity concerning the basic reproduction number, $ R 0$. Notably, parameters *β*, *β*_{1}, and *η*_{1} demonstrate positive sensitivity indices, implying that increases in these parameters result in higher values of $ R 0$. This outcome is counterproductive to our goal of disease transmission control, as it exacerbates the spread of the infection. Thus, elevating these parameters would not be an effective strategy for containment. Conversely, a negative sensitivity index is observed for parameter *α*_{1}. This implies that as *α*_{1} increases, signifying improved treatment for symptomatic infected individuals, the value of $ R 0$ decreases. This inverse correlation with *α*_{1} is a promising finding. In practical terms, enhancing *α*_{1} by providing more effective treatment to symptomatic individuals, leading to their transition into the recovered or hospitalized classes, results in a decline in the value of $ R 0$. This reduction in the basic reproduction number is indicative of the potential for disease eradication within the environment. In summary, increasing *α*_{1} emerges as a scientifically sound and effective control strategy for curbing the spread of the monkeypox infection. Identifying these significant parameters provides valuable insights into the factors that have the most substantial influence on the dynamics of disease transmission within the population. This information is instrumental in designing effective control strategies and interventions to mitigate the spread of the disease.

## VIII. NUMERICAL SIMULATION

In this section, we delve into the analysis of various control strategies designed to mitigate the propagation of the monkeypox infection. Drawing upon the findings from our numerical simulations, we aim to identify the most efficacious single or combined control strategy that could contribute to the eradication of the infection while maintaining environmental safety. For our numerical experiments, we employed the Adams–Bashforth predictor–corrector method. The computational framework was adapted from the approach outlined in (Biswas , 2022; Kumar, 2022; Khan , 2019). This method allowed us to approximate solutions for the fractional-order system, specifically within the context of the Atangana–Baleanu derivative, providing us with a precise and detailed analysis of the dynamics of the monkeypox infection.

In Fig. 4, we investigate the ramifications of varying the contact rate parameter, *β*, between the rodent population and unaware humans, on the dynamics of three distinct human populations in the context of the monkeypox infection: asymptomatic infected individuals (*I _{A}*), symptomatic infected individuals (

*I*), and hospitalized individuals (

_{H}*H*). At $ \beta = 0.024$, we observe a substantial escalation in the number of infected and hospitalized humans approximately 30 days into the simulation. The maximum values for

_{H}*I*,

_{A}*I*, and

_{H}*H*reach 3500, 125 000, and 75 000, respectively. Upon decreasing the value of

_{H}*β*, significant changes become evident in the number of cases across all three compartments. When $ \beta = 0.000 \u2009 24$, the number of infected and hospitalized cases approaches zero, indicating a substantial improvement compared to the previous scenario. Therefore, diminishing the contact rate between the rodent population and the unaware human population emerges as an efficacious strategy for disease transmission control.

In Fig. 5, we analyze the impact of varying the contact rate parameter *β*_{1} on the dynamics of three distinct human populations concerning the monkeypox infection: asymptomatic infected individuals (*I _{A}*), symptomatic infected individuals (

*I*), and hospitalized individuals (

_{H}*H*). When $ \beta 1 = 0.338$, there is a noticeable upsurge in the number of infected and hospitalized individuals, with the peaks occurring roughly 20 days into the simulation. The maximum values for

_{H}*I*,

_{A}*I*, and

_{H}*H*reach approximately 35, 1270, and 760, respectively, after around 100 days. As

_{H}*β*

_{1}is progressively reduced, a marginal decline in the number of infected and hospitalized individuals is observed. For instance, at $ \beta 1 = 0.000 \u2009 338$, the reduction is modest. After 100 days, the maximum number of asymptomatic infected (

*I*), symptomatic infected (

_{A}*I*), and hospitalized (

_{H}*H*) individuals becomes approximately 31, 1100, and 680. Thus, reducing the contact rate between unaware individuals and other humans does result in a decrease in the number of infected cases, but the reduction is not substantial. Consequently, this strategy may not be deemed one of the most effective approaches for controlling the spread of monkeypox infection.

_{H}Figure 6 illustrates the influence of vaccination parameter *u*_{1} on the dynamics of three distinct human populations within the context of monkeypox infection: asymptomatic infected individuals (*I _{A}*), symptomatic infected individuals (

*I*), and hospitalized individuals (

_{H}*H*). When $ u 1 = 0$, signifying the absence of vaccination, the model reveals a substantial upsurge in the number of infected and hospitalized humans approximately 30 days into the simulation. After 100 days, these populations reach 184, 7500, and 4000 for

_{H}*I*,

_{A}*I*, and

_{H}*H*, respectively. The introduction of vaccination into the population results in a notable alteration in disease transmission dynamics. As the vaccination rate increases, a discernible decline in the prevalence of cases within each of the three compartments is observed. At $ u 1 = 0.75$, there is a subtle but significant reduction in the number of infected and hospitalized cases when compared to the scenario with no vaccination. After 100 days, the maximum number of cases in the

_{H}*I*,

_{A}*I*, and

_{H}*H*compartments reaches 2, 200, and 100, respectively. Hence, providing vaccination to both infected individuals and those requiring hospitalization can be considered an efficacious strategy for mitigating the spread of the monkeypox infection.

_{H}In Fig. 7, we have conducted an investigation into the impact of varying the fractional-order parameter, *q* on the dynamics of monkeypox infection. Specifically, we have analyzed the changes in the numbers of asymptomatically infected individuals (*I _{A}*), symptomatically infected individuals (

*I*), and hospitalized cases (

_{H}*H*). Our findings suggest that the choice of the fractional-order parameter

_{H}*q*plays a crucial role in influencing the progression of the monkeypox infection. When

*q*is set to 0.99, we observe a noteworthy upsurge in the number of infected and hospitalized cases approximately 30 days into the simulation. After 100 days, the recorded values show a substantial increase, reaching around 35 for

*I*, 1270 for

_{A}*I*, and 760 for

_{H}*H*. This rapid escalation in cases indicates the highly contagious nature of monkeypox in this fractional order. Conversely, as we gradually decrease the fractional order

_{H}*q*, we notice a concomitant decrease in the number of infected and hospitalized cases. Notably, at

*q*= 0.8, a slight surge in cases is observed after approximately 50 days, though this surge is relatively modest. Under this scenario, the peak numbers in the compartments for asymptomatically infected, symptomatically infected, and hospitalized cases are limited to 3, 180, and 50, respectively. These results signify a notable improvement in controlling the transmission of the infection when compared to the higher

*q*value. This underscores the significance of reducing the fractional-order parameter (

*q*) as a highly recommended strategy for mitigating the spread of the monkeypox disease. This strategy appears effective in reducing the overall number of infected and hospitalized cases, indicating its potential as a valuable intervention in the management and containment of this infectious disease.

## IX. CONCLUSION

In the current investigation, we have employed an innovative compartmental model to examine the dynamics of monkeypox disease. Within this model, we have subdivided the human population into two distinct compartments, distinguishing between those who are informed about the infection (the “aware” population) and those who remain uninformed (the “unaware” population). This approach has allowed us to conduct a comprehensive analysis of the transmission dynamics. The primary focus of our study encompasses the human and rodent populations, as they are the key categories under scrutiny. Our model was rigorously parameterized using empirical data on monkeypox-infected cases in the United States. The graphical representations of our model's outputs reveal a notable alignment with the collected empirical data, affirming the reliability of our model for the estimation of various parameters. These parameters, in turn, facilitate enhanced predictions concerning the dynamics of the monkeypox infection. In the course of our research, we have computed both the basic and effective reproduction numbers and established conditions for the stability of the disease-free equilibrium point. Furthermore, our sensitivity analysis of the basic reproduction number has yielded insightful conclusions regarding the impact of various factors on the transmission of the infection within the population. Additionally, we have demonstrated the Hyer–Ulam stability for our model. After a thorough examination of the asymptotic behavior of the disease-free equilibrium point, we have reached the critical conclusion that the complete eradication of monkeypox from the system is only achievable when the basic reproduction number, $ R 0$, remains strictly below one. This condition holds, assuming that the initial population size of infected humans and rodents falls within the basin of attraction of the disease-free equilibrium point.

Our research outcomes have elucidated that the implementation of diverse control strategies, such as the dissemination of infection awareness, widespread vaccination campaigns, and the provision of efficacious treatments for infected individuals, holds paramount significance in mitigating the monkeypox epidemic and curtailing its propagation within the population. Employing numerical simulations, we have discerned the critical role played by minimizing the effective contact rate between individuals who are informed about the infection and those who remain unaware, as well as between the human and rodent populations, in effectively modulating disease transmission dynamics. These findings have the potential to offer valuable insights to governmental and community policymakers. They may use this knowledge to refine existing policies and develop novel, evidence-based approaches for enhancing the control of monkeypox spread, ultimately safeguarding public health and well-being.

## ACKNOWLEDGMENTS

The authors express their gratitude to the associate editor and the learned reviewers whose comments and suggestions have helped to improve this paper. Research of Santanu Biswas is supported by Dr. D. S. Kothari Postdoctoral Fellowship under the University Grants Commission scheme [Ref. No. F.4-2/2006 (BSR)/MA/19-20/0057].

## AUTHOR DECLARATIONS

### Conflict of Interest

The authors have no conflicts to disclose.

### Author Contributions

A. Santanu Biswas and B. Humaira Aslam have contributed equally to this work.

**Santanu Biswas:** Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (equal); Resources (lead); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). **Humaira Aslam:** Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (lead); Resources (equal); Writing – original draft (lead); Writing – review & editing (equal). **Pankaj Kumar Tiwari:** Conceptualization (supporting); Supervision (supporting); Writing – review & editing (supporting).

## DATA AVAILABILITY

The data are taken from https://www.commerce.gov and https://data.worldbank.org. Some parameter values have been taken from the work of Peter (2022), and the remaining ones are either assumed or estimated.

The data sets generated during and/or analyzed during the current research work are available from the corresponding author upon reasonable request.

### APPENDIX: MAXIMUM LIKELIHOOD ESTIMATATION

Suppose for any data vector $ u = ( u 1 , u 2 , u 3 , \u2026 . , u m )$ and a parameter vector $ v = ( v 1 , v 2 , v 3 \u2026 . , v m )$, let f(u|v) be a probability density function such that with respect to this the likelihood function can be defined as $ L ( v | u ) = f ( u | v )$, which represents the likelihood function of the parameter variable v, providing that the recorded data u are already given. After collecting the necessary data and determining the likelihood function, the next step would be to reach certain statistical conclusions about the population and the study. Since different values of the parameter variable correspond to different probability distributions, we are just interested in finding or estimating those parameter values that are indexed by the probability distribution underlying the given data set. The home likelihood estimation technique is based on the principle of estimating the parameter vector value that will maximize the obtain likelihood function. On exploring the multidimensional parameter space, the parameter vectors obtained as a result are known as the maximum likelihood estimate and are expressed as $ v MLE = ( v 1 , MLE , v 2 , MLE , \u2026 , v k , MLE )$. So, basically, MLE is just a method of finding a suitable probability distribution that will make the observed set of data most probable See Fig. 8.

## REFERENCES

*Applications of Centre Manifold Theory*

_{0}and its role on global stability

*Mathematical Approaches for Emerging and Re Emerging Infectious Diseases: An Introduction*

_{0}in models for infectious diseases in heterogeneous populations

^{+}T-cells

**41**, 101325 (

*Optimal Control Applied to Biological Models*

*Fractional Integrals and Derivatives: Theory and Applications*