==== Front Sci Rep Sci Rep Scientific Reports 2045-2322 Nature Publishing Group UK London 37385997 37192 10.1038/s41598-023-37192-z Article Role of immigration and emigration on the spread of COVID-19 in a multipatch environment: a case study of India Das Tanuja 1 Bandekar Shraddha Ramdas 2 Srivastav Akhil Kumar 3 Srivastava Prashant K 4 Ghosh Mini minighosh@vit.ac.in 5 1 grid.266820.8 0000 0004 0402 6152 Department of Mathematics and Statistics, University of New Brunswick, Fredericton, Canada 2 grid.89336.37 0000 0004 1936 9924 Department of Integrative Biology, University of Texas at Austin, Austin, TX USA 3 grid.462072.5 0000 0004 0467 2410 Mathematical and Theoretical Biology, BCAM - Basque Center for Applied Mathematics, Bilbao, Spain 4 grid.459592.6 0000 0004 1769 7502 Department of Mathematics, Indian Institute of Technology Patna, Patna, 801103 India 5 grid.412813.d 0000 0001 0687 4946 Division of Mathematics, School of Advanced Sciences, Vellore Institute of Technology, Chennai, India 29 6 2023 29 6 2023 2023 13 1054624 12 2022 17 6 2023 © The Author(s) 2023 https://creativecommons.org/licenses/by/4.0/ Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Human mobility has played a critical role in the spread of COVID-19. The understanding of mobility helps in getting information on the acceleration or control of the spread of disease. The COVID-19 virus has been spreading among several locations despite all the best efforts related to its isolation. To comprehend this, a multi-patch mathematical model of COVID-19 is proposed and analysed in this work, where-in limited medical resources, quarantining, and inhibitory behaviour of healthy individuals are incorporated into the model. Furthermore, as an example, the impact of mobility in a three-patch model is studied considering the three worst-hit states of India, i.e. Kerala, Maharashtra and Tamil Nadu, as three patches. Key parameters and the basic reproduction number are estimated from the available data. Through results and analyses, it is seen that Kerala has a higher effective contact rate and has the highest prevalence. Moreover, if Kerala is isolated from Maharashtra or Tamil Nadu, the number of active cases will increase in Kerala but reduce in the other two states. Our findings indicate that the number of active cases will decrease in the high prevalence state and increase in the lower prevalence states if the emigration rate is higher than the immigration rate in the high prevalence state. Overall, proper travel restrictions are to be implemented to reduce or control the spread of disease from the high-prevalence state to other states with lower prevalence rates. Subject terms Mathematics and computing Infectious diseases http://dx.doi.org/10.13039/100019635 Indian Institute of Technology, Patna http://dx.doi.org/10.13039/100009368 University of New Brunswick http://dx.doi.org/10.13039/100019904 Vellore Institute of Technology, Chennai http://dx.doi.org/10.13039/100008562 University of Texas at Austin http://dx.doi.org/10.13039/501100014371 Basque Center for Applied Mathematics issue-copyright-statement© Springer Nature Limited 2023 ==== Body pmcIntroduction The pattern and spread of diseases such as COVID-19 may vary from place to place in a vast country like India, which is divided geographically into many states. Many factors may contribute to such irregular patterns among these different states. These factors include travel, population density, literacy, community behaviour, healthcare infrastructure, etc. Hence, the spread in different states may follow a different pattern. Therefore, in this study, we propose a multi-patch COVID-19 model to address these concerns and study the impacts of individuals’ mobility along with other important factors such as limited medical resources and inhibitory behaviour of individuals due to the disease prevalence. The novel Coronavirus SARS-CoV-2 first emerged in Wuhan, China1 and has spread worldwide, and become a fatal pandemic. As per WHO2, this disease spreads via nasal discharge and saliva droplets when one sneezes or coughs. It is spreading in an unforeseen way in various parts of the world, and different variants of the same have been reported since then. Whether a country is developed, developing, or poor, the cases have been rising and declining in an unprecedented way. This could be due to several reasons like the mobility of individuals, low vaccination, improper testing drives, no compulsion to practice of social distancing or to wear face masks, no governmental interventions, lack of medical resources and political disputes, etc. Travel plays a key role in helping the spread of infectious diseases by the introduction of a pathogen in a new area. In the study by3, travel or migration is emphasized to be the potential force in disseminating infectious diseases. With an ongoing pandemic that spreads through direct contact and where hygiene plays a significant role, frequent movement of individuals from a higher prevalence area to a remote novel village or town with minimal prevalence can increase the risk of viral spillover. Considering the worst hit states of India, which are Kerala, Maharashtra and Tamil Nadu summing up to a population of more than 217 million, human mobility would certainly contribute to accelerating the spread, as individuals carrying the virus have different genetic makeup and varied immunological experiences. Looking into the aspect of the rate at which the virus traveled the world and impacted the entire human livelihood, the study in4 covered the impact of mobility restrictions on disease spread, in the most populous nation, China. Their findings suggested that there could be a possible delay in epidemic peak provided two regions with higher risks were locked down. Various studies on the mathematical modeling of COVID-19 have been conducted with various aspects since the beginning of the pandemic. Most of these works are based on the classic compartmental model5. A brief literature review on different SEIR models and other extended compartmental models, as well as the COVID-19 disease origin, COVID-19 transmission with and without interventions, has been discussed by Anirudh6 and Meehan et al.7. Working on a network model, it is shown that the disease transmission rate of COVID-19 is higher than SARS8. To estimate the unreported infective in China, an SEIR model with unreported infective has been proposed9. Bandekar and Ghosh developed a compartmental model with reported and unreported cases along with quarantine and hospitalization to estimate infective detection rates for India and three of its states10. Some interventions are found to be of prime importance in studying these compartmental models. In11, the authors worked on SEIRD model and developed the system by means of partial differential equations coupled with a heterogeneous diffusion model. A very interesting study in12 presents the co-infection dynamics of Zika and SARS-CoV-2 capturing sexual transmission of Zika in the model. It further shows that the inclusion of time-dependent controls pertaining to Zika and COVID-19 mitigation strategies could significantly reduce the co-infection burden. In a similar context, the study by13 deals with HBV and COVID-19 co-infection dynamics with optimal control. Forecasting disease spread is emphasized to identify various determinants and effectively implement intervention strategies14. Authors emphasized the importance of face masks, sanitation, social distancing, and other non-pharmaceutical interventions to combat the spread of the disease15–18. Similarly, Srivastav et al. provided predictions on the disease dynamics considering face-mask efficacy, quarantine, and hospitalization facilities19. Along with these interventions, the availability of adequate medical resources is a significant concern in pandemic-affected countries. The number of cases is increasing, and there are insufficient beds, ventilators, and oxygen cylinders to accommodate the infected individuals. In addition, the front-line workers are struggling due to the lack of an adequate number of PPE kits20. As per report21, India had one bed per 2000 individuals before the epidemic hit the country. A detailed study based on the limitation of medical resources is presented for India, Brazil, and the USA, with an analysis on backward bifurcation when the number of infective exceeds a threshold value22. Therefore, quarantine, precautionary measures, and a limited treatment facility are three crucial factors affecting the prevalence of COVID-19 disease. In the study23 also impacts of several non-pharmaceutical measures on the population dynamics of COVID-19 in Nigeria is analysed. Their findings suggested that a detection rate of 0.8 per day with approximately 55% population abiding by social distancing norms could lead to a significant decrease in new cases and hence the prevalence of the disease. In this study, a multi-patch model is developed for COVID-19. Several studies for various other infectious diseases have been conducted using the multi-patch models8,24–27. In24, the authors have worked on a multi-patch dengue model depicting spatial and temporal transmission and effectiveness of mosquito control strategies in Kolkata, India. The study concluded that controls exhibiting higher environment persistence performed better and connectedness between different regions makes control strategies effective. In25, Hsieh et al. proposed a multi-patch model to study the impact of people’s mobility during the epidemic of influenza spreading while different patches had different infection levels. In a similar context, the authors in the study26 worked on a multi-patch dengue model focusing on the influence of the Wolbachia bacterium in the optimal control study. The impact of mobility on the spatial spread of dog rabies between two patches involving different disease prevalence levels was studied by Liu et al. in27, by developing a two-patch SEIRS compartmental model. The significance of short-term human mobility is studied by framing a theoretical two-patch model in28. Additionally, a cost-effective strategy and optimal control analysis was applied to provide better insights into the control of disease spread considering economic aspects as well. The impact of migration on infectious diseases involving two patch systems29 and n-patch system30 is performed to show the mobility impact on spatial distribution. In a multi-patch, flux-based model, the geometry of migrations between patches is determined along with the influx of the infective into a patch31. The study shows the effect of time delay between the influx of the infective into a patch and the basic reproduction number. The global stability of a multi-group SIR and a delayed multi-group SIS model with patch structure is investigated32,33. These studies conclude that migration and incidence delays do not affect disease dynamics. The study by Zhang et al. focused on a multi-patch epidemic model with distributed delays considering both deterministic and stochastic scenarios34. They used nonlinear incidence rates and considered the waning immunity of the recovered population. The dynamics of a susceptible-infected-susceptible model along with a detailed analysis of the asymmetric connectivity matrix associated with the model has been studied35. According to the study, the basic reproduction number decreases as the infection spreads. The sufficient conditions for disease extinction and persistence have been established by working on a two-patch SIR model with varying population sizes and regime switching36. The investigation of COVID-19 transmission in a multi-patch environment is interesting and crucial to draw significant conclusions about the impact of mobility on disease spread. However, there has not been much progress in this direction. Meng et al. investigated a multi-patch SEIR compartmental model of COVID-19 disease spread37. They assumed a three-patch model to numerically explain the effect of migration and quarantine rate on the basic reproduction number. In this study, we consider a multi-patch model which accounts for the quarantine, saturated treatment, and nonlinear incidence rate. The nonlinear incidence rate accounts for the inhibitory behaviour of healthy individuals to avoid contracting the disease. The inter-patch movement is allowed in susceptible, exposed, asymptomatic, and recovered compartments only. Individuals who are symptomatic, under medical supervision, or have been quarantined are presumed unable to move. The main purpose of this study is to understand the effect of the movement of populations and its impact on the spread of COVID-19. As the movement is also impacted by various other factors during COVID-19, we shall consider specific cases of movement of populations in this work. We have also accounted for the saturated treatment in the model reflecting the limitation of the medical resources during the disease outbreak. A practical example is exhibited numerically by a particular 3-patch model and performing data fitting and predictive analysis for that. The paper is organized as follows. In the following section, a multi-patch model is proposed and analysed. The existence and stability of disease-free equilibrium and endemic equilibrium are examined. In sections “results”, “parameter estimation and short time prediction”, “discussion”, we explore three different cases: when both emigration and immigration are allowed when only emigration is allowed, and when neither emigration nor immigration is allowed. Section “conclusion” is devoted to estimating parameters using data from three states of India, Kerala, Tamil Nadu, and Maharashtra, as an example. The short-term and long-term predictions are provided in the following section. In the end, we summarize the results and conclude our work. The n-patch SEIQHR model for COVID-19 As discussed in the introduction, we consider an n-patch model for COVID-19. Assume that the total population is divided into n- patches. Further, in each patch, the population is subdivided into seven compartments: susceptible, exposed, asymptomatic, quarantined, symptomatic, hospitalized or home isolated, and recovered. Therefore, we choose a metapopulation system consisting of arbitrary n- patches where Si, Ei, Iai, Qi, Isi, Hi, and Ri are corresponding compartments in the i-th patch (see Table 1). Hence the total population N is the sum of individuals in all n-patches, i.e., N=∑i=1n(Si+Ei+Iai+Qi+Isi+Hi+Ri)=∑i=1nNi.Table 1 Description of variables considered in the model system (1). Variable Description Si Compartment of susceptible individuals in the i-th patch Ei Compartment of exposed individuals in the i-th patch Iai Compartment of asymptomatic individuals in the i-th patch Qi Compartment of quarantined individuals in the i-th patch Isi Compartment of symptomatic individuals in the i-th patch Hi Compartment of hospitalized or home isolated individuals under medical supervision in the i-th patch Ri Compartment of recovered individuals in the i-th patch Ni Total population in the i-th patch In the following, we shall describe each compartment in the i-th patch along with the relevant assumptions for developing the model (1). All newborns are assumed to be susceptible and enter the susceptible compartment Si with a constant birth rate Λi. Susceptible individuals interact with COVID-19 infected individuals, thereby becoming exposed and leaving the susceptible compartment and joining the exposed compartment. As is the case for COVID-19, an infected individual may or may not exhibit symptoms. So, both symptomatic and asymptomatic can transmit the infection. Information about the spread and prevalence of the disease has an inhibitory effect on the healthy population. Hence, the interaction term is modeled as a nonlinear saturated type function, similar to the one studied by Capasso and Serio38 accounting for this inhibitory behaviour. The interaction between susceptible and symptomatic individuals occurs in each patch in the form, βsi1+msiIsiIsi which is saturated by βsimsi due to the effect of inhibition. Similarly, the interaction between susceptible and asymptomatic individuals also occurs in each patch in the form, βai1+maiIaiIai which is saturated by βaimai due to the effect of inhibition. Here βsi and βai are infection rates corresponding to symptomatic and asymptomatic populations. Here msi and mai are constants representing inhibitory effects in symptomatic and asymptomatic populations, respectively, due to information in the population. Individuals in exposed compartment Ei are exposed to infection but are yet to be infectious. Within two weeks (on average ten days for COVID-1939) after exposure, the exposed one becomes infectious and leaves the exposed compartment. Individuals in the asymptomatic compartment Iai are infected but do not exhibit symptoms. They are identified through contact tracing40. It is assumed that a fraction of these will be quarantined based on information if they come in contact with an infected individual, and those who exhibit symptoms in this compartment will move to the symptomatic compartment. Individuals in quarantined compartment Qi are assumed to remain in quarantine and thus do not spread disease. If they develop symptoms during quarantine, they can either join the symptomatic compartment or recover and move to the recovered compartment. Individuals who start showing symptoms move to the symptomatic compartment Isi from the exposed compartment. According to the government advisory, symptomatic infected individuals must be either self-isolated at home or hospitalized. So, individuals leave this compartment either via hospitalization and receiving medical attention, which is considered saturated to emphasize that all individuals seeking medical attention may not receive it41 or by going into self-isolation at a rate of αsiIsi. As mentioned above, this compartment receives inflow from Ei, Iai, and Qi. Individuals in compartment Hi are either hospitalized and receiving medical attention or in home isolation and may or may not be receiving medical attention. As these individuals are under observation, they are isolated from the rest of the population and cannot spread infection. They either recover and join the recovered compartment or die due to disease infection. The last compartment, Ri contains recovered individuals. These individuals are assumed to have short-term immunity and thus cannot be infected again during the modeling period. It is assumed that the movement from one patch to another is allowed in compartments Si, Ei, Iai and Ri only. We denote alk, blk, clk and dlk as the traveling rates from patch k to patch l of susceptible, exposed, asymptomatic, and recovered individuals, respectively. Individuals in other compartments i.e., Isi and Hi, cannot move as they are severely infected. The following table summarizes all the parameters considered in patch i, described above for each population compartment (Table 2).Table 2 Explanation of parameters included in ith patch of system (1). Parameters Description Values Λi Constant recruitment rate of individuals in susceptible compartment Vary βsi Effective contact rate between susceptible and symptomatic individuals Estimated βai Effective contact rate between susceptible and asymptomatic individuals 0.00005–0.0002 msi Rate constant representing inhibitory constant for symptomatic individuals 0.1 mai Rate constant representing inhibitory constant for asymptomatic individuals 0.1 μi Natural mortality rate of individuals 0.003 σi Disease infection rate 0.1619 γi Quarantining rate of asymptomatic individuals Estimated θi Rate of showing symptom in asymptomatic individuals 0.01–0.0819,42 ηi Rate of showing symptom in quarantined individuals 0.001 αsi Isolation rate of symptomatic individuals Estimated αi Hospitalization rate of symptomatic individuals 0.02–0.119 νi Saturation constant representing limitation of medical resources 0.01 μsi Disease mortality rate of symptomatic individuals 0.005219 μhi Disease mortality rate of hospitalised individuals 0.004219 ρi Recovery rate of self-quarantined individuals 0.0002–0.00219,43 ξi Recovery rate of hospitalised/self-isolated individuals 1141,19 εi Proportion of exposed individuals not showing clinical symptoms 0.07–0.3319,39 The inter and intra patch interactions and mobility of populations in patch i and patch j are represented by Fig. 1.Figure 1 Flow diagram of the multi-patch model (1) showing interactions between patch i and patch j. Mathematically, we express the interactions of individuals in patch i with other patches, j=1,2,…,n such that j≠i with the help of the following system of differential equations. Note that this representation is true for all n- patches (i.e. for each ith patch with i=1,2,…,n).1 dSidt=Λi-μiSi-βsi1+msiIsiSiIsi-βai1+maiIaiSiIai+∑j=1,j≠in(aijSj-ajiSi),dEidt=βsi1+msiIsiSiIsi+βai1+maiIaiSiIai-(σi+μi)Ei+∑j=1,j≠in(bijEj-bjiEi),dIaidt=εiσiEi-μiIai-γiIai-θiIai+∑j=1,j≠in(cijIaj-cjiIai),dQidt=γiIai-(ηi+ρi+μi)Qi,dIsidt=(1-εi)σiEi-(μsi+μi)Isi+ηiQi+θiIai-αsiIsi-αiIsi1+νiIsi,dHidt=αsiIsi+αiIsi1+νiIsi-(μhi+μi)Hi-ξiHi,dRidt=ρiQi+ξiHi-μiRi+∑j=1,j≠in(dijRj-djiRi). We denote the mobility matrices as A=[aij]n×n, B=[bij]n×n, C=[cij]n×n, D=[dij]n×n with aii=bii=cii=dii=0 ∀i=1,2,…,n. Theorem 2.1 Denote, Λ¯=∑i=1nΛi and μ¯=min{μi}, i=1,2,…,n. The basic feasible region of system (1) is given by,Γ:={(S1,E1,Ia1,Q1,Is1,H1,R1,…,Sn,En,Ian,Qn,Isn,Hn,Rn)∈R+7n|0≤N(t)≤Λ¯μ¯,0≤Si≤S0i,i=1,2,…,n}, where S0i is defined in the proof. Proof Consider the system (1) with all non negative initial values holding Ni(0)≥0,i=1,2,…,n. It can be easily verified using similar arguments as described by Srivastav et al.19 that all the compartments in each patch remain non negative for all positive time if the initial values lie in R+7n . Further, from model (1) the total population N(t) satisfies,dNdt=∑i=1ndNidt≤∑i=1nΛi-∑i=1nμiNi<Λ¯-μ¯N, where Λ¯=∑i=1nΛi and μ¯=min{μi}. Therefore, N(t)≤(N(0)-Λ¯μ¯)e-μ¯t+Λ¯μ¯. This implies N(t)≤Λ¯μ¯ when N(0)≤Λ¯μ¯. Similarly, as in37, the first equation of system (1) is written as,dSidt≤Λi-μiSi+∑j=1,j≠in(aijSj-ajiSi)=(b-G0S)i, where S=[Si]n×1, b=[Λi]n×1 and2 G0=[δij(μi+∑j=1,j≠inaji)-(1-δij)aij]n×n. and δij denotes the Kronecker delta, defined as δij=1 for i=j and δij=0 for i≠j. Hence, dSidt≤0 for Si=(G0-1b)i=Si0 (say) for each i=1,2,…,n. Since G0 is a nonsingular M- matrix, G0-1 is non negative and (G0-1b)i≥0 (see44 for details) for all i=1,2,…,n. This completes the proof. □ Here, this result emphasizes that all compartments representing the different stages of the disease in the population must not be negative and they should also remain bounded. A negative and unbounded population makes no sense in populations. Results As mentioned earlier, the primary purpose of this study is to find out the impact of the movement of individuals in different patches. For the control of COVID-19 spread, various non-pharmaceutical interventions related to mobility, such as local confinement zones, no entry without negative tests, or even a complete travel ban, have been applied by healthcare personnel. Because of this, we shall consider our proposed model under three different cases for the ith patch- (i) when both immigration and emigration are allowed, (ii) when only emigration is allowed, and (iii) when there is complete isolation of the local patch from rest of the population. Here it is assumed that, in all three cases, the other patches are connected. In the following, we will analytically study these cases. Case 1: both immigration and emigration are allowed In this section, we consider the case where both immigration and emigration are allowed in each patch i (i=1,2,…,n), which means the mobility parameters aij,aji,bij,bji,cij,cji,dij,dji≠0 for j≠i, j=1,2,…,n. This means that individuals can both move from patch i to any other patch and also other patch individuals can join the individuals in patch i. Under this assumption, we obtain the following results that hold for ith patch. The unique disease-free equilibrium (DFE) for each patch i (i=1,2,…,n) of the system (1) is E0(i)=((G0-1b)i,0,0,0,0,0,0) and the unique disease-free equilibrium of the multi-patch system is given as, E0(multi)=((G0-1b)1,0,0,0,0,0,0,(G0-1b)2, 0,0,0,0,0,0,…,(G0-1b)n,0,0, 0, 0, 0, 0). Here, G0 is as defined in (2), b=[Λi]n×1, and (G0-1b)i is i th element of [G0-1b]n×1. The basic reproduction number is given by 3 R0(multi)=ρ(V1V11-1V22-1+V2V11-1), where V11=[δij(σi+μi+∑j=1,j≠inbji)-(1-δij)bij]n×n, V22=[δij(μi+γi+θi+∑j=1,j≠incji)-(1-δij)cij]n×n, V1=δij(σiSi0εiβai+σiSi0βsiμsi+μi+αi+αsi(εiγiηiηi+ρi+μi+εiθi))n×n=[v1ij]n×n, and V2=δij(σiSi0βsi(1-εi)μsi+μi+αi+αsi)n×n=[v2ij]n×n, i=1,2,…,n, j=1,2,…,n. The DFE E0(multi)=(S10,0,0,0,0,0,0,S20,0,0,0,0,0,0,…,Sn0,0,0,0,0,0,0) is locally asymptotically stable for R0(multi)<1 and unstable for R0(multi)>1. The DFE, E0(multi) is globally asymptotically stable provided R0(multi)<1 and νi=0 for each i=1,2,…,n. The endemic equilibrium E∗(multi) always exists for R0(multi)>1. The endemic equilibrium E∗(multi)=(S1∗,E1∗,Ia1∗,Q1∗, Is1∗,H1∗,R1∗,…,Sn∗,En∗,Ian∗,Qn∗,Isn∗,Hn∗,Rn∗) may exist for R0(multi)<1. The condition of existence is that if 4 (A0+A1D[Isi∗]+A2D[Isi∗2]+A3D[Isi∗3]+A4D[Isi∗4])Is∗=0 has a positive real root, where the coefficients are given in section “effect of mobility during epidemic” of the supplementary document. Case 2: only emigration is allowed In this section, we explore the case when immigration of individuals is not allowed but emigration is allowed in patch i (for any fixed i,  i=1,2,…,n), i.e., aij=bij=cij=dij=0 and aji,bji,cji,dji≠0 for j≠i, j=1,2,…,n. Thus, in this case, only individuals from patch i can move out and go to the any other patch j but the individuals from other patches cannot enter the patch i. Under this assumption, we obtain the following results that hold for ith patch. The DFE E~0(i)=(S~i0,0,0,0,0,0,0)=(Λiμi+∑j=1,j≠inaji,0,0,0,0,0,0) exists always in Γ for each i (i=1,2,…,n). The basic reproduction number in patch i is given by 5 R~0(i)=v1ii(σi+μi+∑j=1,j≠inbji)(μi+γi+θi+∑j=1,j≠incji)+v2ii(σi+μi+∑j=1,j≠inbji),i=1,2,…,n. Again, when immigration of individuals is prohibited into patch i (i=1,2,…,n), the infection caused by an individual during its symptomatic infectious period is given by R~0,s(i)=βsiS~i0μsi+μi+αsi+αi. The infection caused by an individual initially asymptomatic is given by R~0,a(i)=βaiS~i0μi+γi+θi+∑j=1,j≠incji+θiμi+γi+θi+∑j=1,j≠incji+γiμi+γi+θi+∑j=1,j≠incjiηiμi+ρi+ηiR~0,s(i), where the last term reflects the route by which an individual initially asymptomatic can become symptomatic. Hence, R¯0(i) in (5) can be expressed as R~0(i)=σiσi+μi+∑j=1,j≠inbji(εR~0,a(i)+(1-ε)R~0,s(i)), where the factors σiσi+μi+∑j=1,j≠in,bji, for i=1,2,…,n, indicate the probability of not dying during the exposed period and so becoming infectious. The DFE, E~0(i) is locally asymptotically stable if R~0(i)<1 and unstable if R~0(i)>1. E~0(i) is globally asymptotically stable if R~0(i)<1 and νi=0. Considering the Lyapunov function, L2=k1iEi+k2iIai+k3iQi+k4iIsi, we obtain the required conditions. The endemic equilibrium is given by, E~∗(i)=(S~i∗,E~i∗,I~ai∗,Q~i∗,I~si∗,H~i∗,R~i∗) for each i, where the components are E~i∗=μi+γi+θi+∑j=1,j≠incjiεiσiI~ai∗, Q~i∗=γiI~ai∗ηi+ρi+μi, I~ai∗=(li1+li2I~si∗)εiI~si∗(1+νiI~si∗)(εili3+(1-εi)∑j=1,j≠incji), S~i∗=(σi+μi+∑j=1,j≠inbji)E~i∗βsi1+msiI~si∗I~si∗+βai1+maiI~ai∗I~ai∗,H~i∗=αsiI~si∗+αiI~si∗1+νiI~si∗μhi+μi+ξi, R~i∗=ρiQ~i∗+ξiH~i∗μi+∑j=1,j≠indji, and I~si∗ satisfies 6 ai4I~si∗4+ai3I~si∗3+ai2I~si∗2+ai1I~si∗+ai0=0. The coefficients in the above expression are, ai4=li22li4li3(βsimsi+βaimai+(μi+∑j=1,j≠inaji)msimai), ai3=li2li3[(μi+∑j=1,j≠inaji)li4{msi(νili3+maili1)+mai(li2+msili1)}+li4{li1(βsimai+ βaimsi)+li5}-Λiνili3(βsimai+βaimsi)], ai2=1li3[(μi+∑j=1,j≠inaji)li4{msili2li3+(νili3+maili1)(li2+msili1)+maili1li2}+ li4{li2(βsili3+βaili1)+li1li5}-Λili3(νili5+li2(βsimai+βaimsi)}], ai1=li4[(μi+∑j=1,j≠inaji)(li2+msili1+νili1+maili12li3)+li1li3(βsili3+βaili1)]-Λi(li5+νi(βsili3+βaili1))], ai0=(μi+∑j=1,j≠inaji)li1li4-Λi(βsili3+βaili1)=(μi+∑j=1,j≠inaji)li1li4(1-R~0(i)), and li1=μsi+μi+αsi+αi, li2=νi(μsi+μi+αsi), li3=(μi+γi+θi)1-εiεi+ηiγiηi+ρi+μi+θi, li4=(σi+μi+∑j=1,j≠inbji)μi+γi+θi+∑j=1,j≠incjiεiσi and li5=βsi(νiLi3+maili1)+βai(msili1+li2). The number of possible positive real roots of Eq. (6) depends on coefficients ai4-ai0 as follows Table 3. Endemic equilibrium, E~∗(i) exists for R~0(i)>1 and may or may not exist for R~0(i)<1 for each i. As the endemic equilibrium E~∗(i) exists if Eq. (6) has a real positive root, and the number of endemic equilibrium points will be equal to the number of real positive roots of (6). In Table 3 it is shown that a positive real I~si∗ exists for R~0>1. So E~∗(i) exists for R~0(i)>1. Again since real positive I~si∗ may or may not exist for R~0<1 (by Table 3), so E~∗(i) may or may not exist for R~0(i)<1. Endemic equilibrium E~∗(i) becomes locally asymptotically stable if 7 b~i1b~i2b~i3b~i4+b~i0b~i2b~i3+2b~i0b~i1b~i4>b~i0b~i32b~i4+b~i12b~i42+b~i1b~i22+b~i02,b~i2b~i3b~i4+b~i0b~i4>b~i22+b~i1b~i42,b~i3b~i4>b~i2,b~i4>0,b~i0>0. holds, where b~i4=-(Ji11+Ji22+Ji33+Ji44+Ji55),b~i3=-Ji23Ji32+Ji22Ji33+Ji22Ji44+Ji33Ji44-Ji25Ji52+Ji22Ji55+Ji33Ji55+Ji44Ji55+Ji11(Ji22+Ji33+Ji44+Ji55),b~i2=-Ji13Ji21Ji32+Ji23Ji32Ji44-Ji22Ji33Ji44-Ji15Ji21Ji52+Ji25Ji33Ji52+Ji25Ji44Ji52-Ji25Ji32Ji53+Ji23Ji32Ji55-Ji22Ji33Ji55-Ji22Ji44Ji55-Ji33Ji44Ji55+Ji11(Ji23Ji32-Ji33Ji44+Ji25Ji52-Ji33Ji55-Ji44Ji55-Ji22(Ji33+Ji44+Ji55)),b~i1=Ji15Ji21Ji33Ji52+Ji15Ji21Ji44Ji52-Ji25Ji33Ji44Ji52-Ji15Ji21Ji32Ji53+Ji25Ji32Ji44Ji53-Ji25Ji32Ji43Ji54-Ji23Ji32Ji44Ji55+Ji22Ji33Ji44Ji55+Ji13Ji21Ji32(Ji44+Ji55)+Ji11(-Ji25Ji33Ji52-Ji25Ji44Ji52+Ji25Ji32Ji53+Ji33Ji44Ji55-Ji23Ji32(Ji44+Ji55)+Ji22(Ji44Ji55+Ji33(Ji44+Ji55))),b~i0=-Ji15Ji21(Ji33Ji44Ji52-Ji32Ji44Ji53+Ji32Ji43Ji54)-Ji13Ji21Ji32Ji44Ji55+Ji11(Ji25Ji33Ji44Ji52-Ji25Ji32Ji44Ji53+Ji25Ji32Ji43Ji54+Ji23Ji32Ji44Ji55-Ji22Ji33Ji44Ji55), and Ji11=-(μi+βsiI~si∗1+msiI~si∗+βaiI~ai∗1+maiI~ai∗+∑j=1,j≠inaji)=-ΛiS~i∗, Ji13=-βaiS~i∗(1+maiI~ai∗)2, Ji15=-βsiS~i∗(1+msiI~si∗)2, Ji21=βsiI~si∗1+msiI~si∗+βaiI~ai∗1+maiI~ai∗, Ji22=-(σi+μi+∑j=1,j≠inbji), Ji23=βaiS~i∗(1+maiI~si∗, Ji25=βsiS~i∗(1+msiI~si∗)2, Ji32=εiσi, Ji33=-(μi+γi+θi+∑j=1,j≠incji), Ji43=γi, Ji44=-(ηi+ρi+μi), Ji52=(1-εi)σi, Ji53=θi, Ji54=ηi, Ji55=-(μsi+μi+αsi+αi(1+νiI~si∗)2). Table 3 Number of possible positive real roots of Eq. (6). Cases Coefficients Number of possible ai4 ai3 ai2 ai1 ai0 Positive roots R~0(i)<1 + +/− +/− +/− + 4/2/0 R~0(i)>1 + +/− +/− +/− – 3/1 Theorem 3.1 Suppose, for a fixed i, aij=bij=cij=dij=0 for j≠i, j=1,2,…,n and R~0(i)=1βa=βai∗=(σi+μi+∑j=1,j≠inbji)(μi+γi+θi+∑j=1,j≠incji)-(σiSi0βsiμsi+μi+αi+αsi(εiγiηiηi+ρi+μi+εiθi))+v2ii(μi+γi+θi+∑j=1,j≠incji)σiSi0εi i.e., Then system (1) undergoes forward (transcritical) bifurcation if ν<νi∗ and a backward bifurcation if ν>νi∗ where, νi∗= li1εi2σi2(maiβsi+msiβai∗)βai∗Si02(μi+γi+θi+∑j=1,j≠incji)2-li12εiσimsiβai∗Si0μi+γi+θi+∑j=1,j≠incji+βsi(σi+μi+∑j=1,j≠inbji)+(σi+μi+∑j=1,j≠inbji)2msili1αi(σi+μi+∑j=1,j≠inbji-βai∗Si0εiσiμi+γi+θi+∑j=1,j≠incji)2. Remark 3.1 One of the necessary conditions for non existence of E~∗(i) for R~0(i)<1 is νi=0. Because the DFE E~0(i) is global asymptotically stable here and so disease will extinct for R~0(i)<1 under the condition νi=0. Besides since the endemic equilibrium E~∗(i) always exists for R~0(i)>1, so disease will persist for R~0(i)>1. This behaviour of the disease system is found due to taking saturated hospital facilities in the system (for details, see41). Case 3: neither immigration nor emigration is allowed In this section, we consider the case where both immigration and emigration are not allowed in patch i, for any fixed i,  i=1,2,…,n. Hence, aij=aji=bij=bji=cij=cji=dij=dji=0 for j≠i, and j=1,2,…,n. Then the fixed ith patch is separated from the rest of the multipatch system and acts as a single patch. Thus, in this case, no inter-patch movement is allowed and hence all the patches are isolated. In this case, we obtain the following results for the fixed ith patch. The DFE E¯0(i)=(S¯i0,0,0,0,0,0,0)=(Λiμi,0,0,0,0,0,0) exists always in Γ. The basic reproduction number in patch i is given by 8 R¯0(i)=v1ii(σi+μi)(μi+γi+θi)+v2ii(σi+μi),i=1,2,…,n. Similarly as the above case, R¯0(i) in (8) can be expressed as R¯0(i)=σσ+μ(εR¯0,a(i)+(1-ε)R¯0,s(i)). The DFE, E¯0(i) is locally asymptotically stable if R¯0(i)<1, globally asymptotically stable if R¯0(i)<1 and νi=0 and unstable if R¯0(i)>1. The endemic equilibrium is given by E¯∗(i)=(S¯i∗,E¯i∗,I¯ai∗,Q¯i∗,I¯si∗,H¯i∗,R¯i∗), where (S¯i∗,E¯i∗,I¯ai∗, Q¯i∗,I¯si∗,H¯i∗ ,R¯i∗)=(S~i∗,E~i∗,I~ai∗,Q~i∗,I~si∗,H~i∗,R~i∗)|aji=bji=cji=dji=0. Endemic equilibrium, E¯∗(i) exists for R¯0(i)>1 and may or may not exist for R¯0(i)<1. Endemic equilibrium E¯∗(i) becomes locally asymptotically stable if 9 b¯i1b¯i2b¯i3b¯i4+b¯i0b¯i2b¯i3+2b¯i0b¯i1b¯i4>b¯i0b¯i32b¯i4+b¯i12b¯i42+b¯i1b¯i22+b¯i02,b¯i2b¯i3b¯i4+b¯i0b¯i4>b¯i22+b¯i1b¯i42,b¯i3b¯i4>b¯i2,b¯i4>0,b¯i0>0. holds, where (b¯i4,b¯i3,b¯i2,b¯i1,b¯i0)=(b~i4,b~i3,b~i2,b~i1,b~i0)|aji=bji=cji=dji=0. Remark 3.2 R~0(i)in(5)1, so the disease will survive in the system in all the following cases.If mobility rates are increased between patches i.e., p12=p21=p13=p31=p23=p32=0.05 (gives R0(multi)=1.96>1), then the number of active cases will decrease in patch 1 and will increase in patch 2 and patch 3. The blue curve represents this outcome in Fig. 3 for all three patches. If immigration rate in patch 1 is higher than emigration rate i.e., p12=p13=0.01, p21=p31=p23=p32=0.001 (gives R0(multi)=1.55>1), the number of active cases will increase in patch 1 and will decrease in patch 2 and patch 3. The green curve represents this outcome in Fig. 3. If emigration rate from patch 1 is higher than immigration rate i.e., p21=p31=0.01, p12=p13=p23=p32=0.001 (gives R0(multi)=3.6>1), then the number of active cases will decrease in patch 1 and will increase in patch 2 and patch 3. The magenta curve represents this outcome in Fig. 3. Figure 3 Plot of predicted active cases in all three patches for different mobility rates. Black curve: estimation curve, Red curve: prediction curve when all patches are connected with the same mobility rate i.e., p12=p21=p23=p32=p13=p31= 0.001; Blue curve: prediction curve when mobility rates within all patches are increased i.e., p21=p31=p12=p13=p23 =p32=0.05; Green curve: prediction curve when immigration rate in patch 1 is higher than emigration rate i.e., p12=p13 =0.01, p21=p31=p23=p32=0.001; Magenta curve: prediction curve when emigration rate from patch 1 is higher than immigration rate i.e., p21=p31=0.01, p12=p13=p23=p32=0.001. In the above, we have explained the scenario where all patches are connected directly via mobility, and the rates of mobility are varied. We note from the outcome that increased mobility in all patches reduces active cases in patch 1 but increases cases in the other two states. Further, increased emigration in Kerala minimizes the burden on Kerala but causes a wave in Tamil Nadu. Similarly, if we reduce immigration from Kerala, it significantly reduces the load on both Maharashtra and Tamil Nadu. These observations are interesting as, by restricting the mobility of the high load state, we can significantly control the spread of the disease. Now we study the system behaviour when mobility is restricted between any two patches or among all the patches. For this, we assume βs2=0.00006 and βs3=0.00008 and remaining all parameters are fixed as mentioned in Tables 2 and 4 and mobility rates are same p12=p21=p23=p32=p13=p31=0.001 in all the patches. Then we obtain R0(multi)=2.539>1. Therefore, the disease will persist in all three patches. The red curve represents the active case number in each patch in Fig. 4. Now, we consider the following cases to establish the mobility impact on disease persistence and extinction.If mobility between patch 1 and patch 2 is only restricted i.e., p12=p21=0, p23=p32=p13=p31=0.001 (implies R0(multi)=2.558), then the number of active cases will increase in patch 1 and patch 3 and will decrease in patch 2 only. The blue curve represents this result in Fig. 4. If mobility between patch 1 and patch 3 is only restricted i.e., p12=p21=p23=p32=0.001, p13=p31=0 (implies R0(multi)=2.558), then the number of active cases will increase in patch 1 and patch 2 and will decrease in patch 3. The green curve exhibits this outcome in Fig. 4. If mobility is restricted between all the patches i.e., p12=p21=p23=p32=p13=p31=0 (implies R¯0(1)=2.57, R¯0(2)=0.89, R¯0(3)=0.86 evaluated using (8)), then the number of active cases will increase in patch 1 and will decrease in patch 2 and patch 3. The magenta curve represents this outcome in Fig. 4. Figure 4 Plot of predicted active cases in all three patches for different mobility rates. Black curve: estimation curve, Red curve: prediction curve when all patches are connected with same mobility rate i.e., p12=p21=p23=p32=p13=p31= 0.001; Blue curve: prediction curve when no mobility between patch 1 and patch 2 i.e., p12=p21=0, p23=p32=p13= p31=0.001; Green curve: prediction curve when no mobility between patch 1 and patch 3 i.e., p12=p21=p23=p32=0.001, p13=p31=0; Magenta curve: prediction curve when all patches are isolated i.e., p12=p21=p13=p31=p23=p32=0. Hence, we conclude if mobility is restricted among all patches, the disease load will be reduced in all the patches. Further, if mobility is restricted from a higher prevalence state (Kerala), then the disease load in that state will be low. Hence, we conclude that travel restrictions from high-prevalence states may reduce or control disease spread in other states. Discussion In this paper, we propose and analyse a multi-patch model of COVID-19 disease transmission where each patch consists of seven compartments. Patches are connected through the inter-patch movement of individuals in susceptible, exposed, asymptomatic, and recovered compartments. We also accounted for the inhibitory effects of healthy individuals on incidence rate and the limitation of medical resources in the model system. We studied the multi-patch system having two or more patches. One of the important factors considered in our work is saturated incidence rates which are saturated by βsimsi and βaimai where βsi, βai represent the effective contact rates between susceptible and infected and msi, mai represent the early precaution measures. To the best of our knowledge, no multi-patch model of COVID-19 disease is analysed by taking a saturated incidence rate. The basic reproduction number of the multi-patch system is calculated in Eq. (3). We establish the existence of disease-free equilibrium and endemic equilibrium. One of the main goals of this work is to capture the mobility effect in disease spreading during the epidemic period. Therefore we study three cases-(i) when both immigration and emigration are allowed, (section “case 1: both immigration and emigration are allowed”), (ii) when only emigration is allowed (section “case 2: only emigration is allowed”) and (iii) when there is complete isolation of the local patch from rest of the population (section “case 3: neither immigration nor emigration is allowed”). It is observed that basic reproduction number of case 2 is less than case 1 (Theorem 3.2) and case 3 (Remark 3.2), which signifies that the basic reproduction number decreases as mobility rates increase. The multi-patch model is designed numerically into a three-patch model where the data fitting is provided with the help of an example of three states in India, taken as the three patches. Our objective is to study the COVID-19 disease outbreak in these three states. As per reported cases in MoHFW, Kerala is the highest prevalence state in India of COVID-19 disease at the end of July 2021. So we have selected Kerala as patch 1, two neighbouring states of Kerala- Maharashtra as patch 2, and Tamil Nadu as patch 3. We have used the data sets of active COVID-19 cases47 from July 14th to August 20th, 2021. Since the active individuals are those who are clinically reported as COVID-19 patients, the H compartment of our model is fitted with the data (Fig. 2a,c,e). We estimate the parameters βsi, the effective contact rate between susceptible and symptomatic, γi, quarantined rate of asymptomatic individuals, and αsi, quarantined rate of symptomatic individuals, which are given in Table 4 for all the three patches. By looking at the estimated values, it is observed that the effective contact rate between susceptible and symptomatic is quite high, and the quarantined rate of symptomatic is quite low in patch 1. Whereas in patch 3 contact rate between susceptible and symptomatic is very low, and the quarantined rate of symptomatic is high. The quarantined rate of asymptomatic is very low in patch 2. Therefore, patch 1 is considered the high prevalence state, and patch 3 is considered the lowest prevalence state among these three patches. With the help of the estimated parametric values, we have predicted the number of active cases in all three patches for the next 22 days, i.e., from August 21st to September 11th, 2021. It is observed that the active cases graph is decaying in all patches in Fig. 2b,d,f. We further explore the impact of mobility of COVID-19 spread, choosing different mobility rates between patches. It is observed that when there is mobility restriction between Kerala and either Maharashtra or Tamil Nadu, the number of active cases increases in Kerala (Fig. 4) and decreases in other states. Therefore, it is noted that when mobility is prohibited from the high prevalence state to lower prevalence states, the number of active cases number will increase in the high prevalence state. A similar result is seen if the immigration rate is higher than the emigration rate in the high-prevalence state. However, an increase in mobility rates reduce the disease prevalence in high prevalence state (Kerala) and increase disease outbreak in lower prevalence states (Maharashtra, Tamil Nadu) (Fig. 3). Also, the number of active cases will decrease in the high prevalence state and increase in the lower prevalence states if the emigration rate is higher than the immigration rate in the high prevalence state. We also exhibit the impact of mobility on COVID-19 disease extinction (Fig. 4). For this, we have reduced the effective contact rate between susceptible and symptomatic in the lower prevalence states, Maharashtra and Tamil Nadu. It is later shown that the disease becomes extinct in the lower prevalence states (Maharashtra and Tamil Nadu) if all patches are isolated from each other, that is, no immigration or emigration will happen in any patch. In this case, the disease will persist in only the high prevalence state Kerala (Fig. 4). Conclusion When COVID-19 first emerged in 2020, we observed that every nation had its own set of travel restrictions in place due to the human-to-human transmission of this disease. The movement of individuals from one location to another played a huge role in the spread of disease worldwide. Similar to other countries, the Indian government had imposed several travel restrictions at all levels including both domestic and international travel. India is a huge country with a large population in each of its states. Every day, a significantly large number of people move from one state to another, which is why this movement of people was crucial in the spread of the COVID-19 disease in India. As mentioned earlier, we included the mobility factor and our research demonstrated that if mobility begins in every patch (state in India), the three-patch system transforms into an epidemic system, and disease will persist in every patch. Therefore, the COVID-19 disease outbreak is influenced by the movement of the population in a big way. Herein, it is noted that based on the mobility restrictions the basic reproduction number can increase or decrease, although all rate parameters are fixed. This research offers significant results from numerical experiments that are employed as a specific control. Although the presented results have been confined to only three Indian states, the outcomes are scalable to an even larger number of states. Additionally, similar findings can be obtained for multiple countries using this methodology. E.g. European Union has a comparable demographic zone, making it possible to utilize the present model as a model for other European nations. We have considered the three Indian states as an example and shown the impact of various parameters related to mobility on the spread and control of the disease. Nonetheless, our main results are generic in nature and are independent of the location. The proper setting of parameters helps in predicting the course of disease in any country or state/s. Importance of mobility restrictions Various variants of COVID-19 are detected, thus by imposing mobility restrictions, we can mitigate the introduction and spread of new variants. Mobility restrictions can reduce the transmission of infection by limiting the mobility of infective people across borders. This offers the public health administration a crucial time window to adopt and improve upon domestic control measures like testing, contact tracing, quarantine regulations, and vaccination campaigns, etc. Each country/patch has a limited capacity for treatment. So, mobility restrictions can assist in reducing the additional burden on their healthcare infrastructure. Moreover, it is easy to contain and control disease in a smaller patch rather than in the whole population. Hence, if there is a surge in a particular location (patch), it is advisable to isolate that patch. Future direction Presented study needs to be scaled up to a larger population where one can investigate COVID-19 distribution in different countries rather than states and analyse the impact of international travel restrictions on disease control. We plan to incorporate the impact of vaccination too in our future work. Supplementary Information Supplementary Information. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-023-37192-z. Acknowledgements The authors are grateful to the anonymous referees for their valuable suggestions which improved the manuscript and presentation. Author contributions T.D.: Conceptualization, model design, empirical analysis, evaluation. S.R.B.: Conceptualization, drafted the introduction, interpretation. A.K.S.: Conceptualization, data analysis and interpretation. P.K.S.: Conceptualization, model design, interpretation, reviewed final draft. MG.: Conceptualization, interpretation, reviewed final draft. All authors reviewed and edited the manuscript and supplementary information. Data availability The datasets analysed during the current study are available in the COVID-19 India repository “https://www.covid19india.org/”. Conflict of interest The authors declare no competing interests. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Zhou F Yu T Du R Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: A retrospective cohort study The Lancet 2020 395 1054 1062 10.1016/S0140-6736(20)30566-3 2. URL: https://www.who.int/health-topics/coronavirus. 3. Wilson M Travel and the emergence of infectious diseases Emerg. Infect. Dis. 1995 1 39 46 10.3201/eid0102.950201 8903157 4. Zhou Y Effects of human mobility restrictions on the spread of COVID-19 in Shenzhen, China: A modelling study using mobile phone data Lancet Dig. Health 2020 2 e417 e424 10.1016/s2589-7500(20)30165-5 5. A contribution to the mathematical theory of epidemics. In: Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 115 700–721. 10.1098/rspa.1927.0118 (1927). 6. Anirudh A Mathematical modeling and the transmission dynamics in predicting the COVID-19-what next in combating the pandemic Infect. Disease Modell. 2020 5 366 374 10.1016/j.idm.2020.06.002 7. Meehan MT Rojas DP Adekunle AI Modelling insights into the COVID-19 pandemic Paediatr. Respir. Rev. 2020 35 64 69 32680824 8. Chen T-M Rui J Wang Q-P A mathematical model for simulating the phase-based transmissibility of a novel coronavirus Infect. Dis. Poverty 2020 9 1 8 10.1186/s40249-020-00640-3 31996251 9. Lee C Li Y Kim J The susceptible-unidentified infected-confirmed (SUC) epidemic model for estimating unidentified infected population for COVID-19 Chaos Solitons Fractals 2020 139 110090 10.1016/j.chaos.2020.110090 32834625 10. Bandekar SR Ghosh M Mathematical modeling of COVID-19 in India and its states with optimal control Model. Earth Syst. Environ. 2021 5 1 16 11. Viguerie A Lorenzo G Auricchio F Simulating the spread of COVID-19 via a spatially-resolved susceptible-exposed-infected-recovered-deceased (SEIRD) model with heterogeneous diffusion Appl. Math. Lett. 2021 111 106617 10.1016/j.aml.2020.106617 32834475 12. Omame A Abbas M Onyenegecha CP Backward bifurcation and optimal control in a co-infection model for SARS-CoV-2 and ZIKV Res. Phys. 2022 37 105481 10.1016/j.rinp.2022.105481 13. Omame A Abbas M Modeling SARS-CoV-2 and HBV co-dynamics with optimal control Physica A 2023 615 128607 10.1016/j.physa.2023.128607 36908694 14. Nicola M Alsafi Z Sohrabi C The socio-economic implications of the coronavirus pandemic (COVID-19): A review Int. J. Surg. 2020 78 185 193 10.1016/j.ijsu.2020.04.018 32305533 15. Chu DK Akl EA Duda S Physical distancing, face masks, and eye protection to prevent person-to-person transmission of sars-cov-2 and COVID-19: A systematic review and meta-analysis The Lancet 2020 395 1973 1987 10.1016/S0140-6736(20)31142-9 16. Eikenberry SE Mancuso M Iboi E To mask or not to mask: Modeling the potential for face mask use by the general public to curtail the COVID-19 pandemic Infect. Dis. Model. 2020 5 293 308 32355904 17. Liu X Zhang S COVID-19: Face masks and human-to-human transmission Influenza Other Respir. Viruses 2020 14 472 10.1111/irv.12740 32223072 18. Srivasrav AK Modeling the initial phase of COVID-19 epidemic: The role of age and disease severity in the Basque country, Spain PLoS ONE 2022 17 e0267772 10.1371/journal.pone.0267772 35830439 19. Srivastav AK Tiwari PK Srivastava PK Ghosh M Kang Y A mathematical model for the impacts of face mask, hospitalization and quarantine on the dynamics of COVID-19 in India: Deterministic vs. stochastic Math. Biosci. Eng. 2021 18 182 213 10.3934/mbe.2021010 20. Huang C Wang Y Li X Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China The Lancet 2020 395 497 506 10.1016/S0140-6736(20)30183-5 21. URL: https://data.oecd.org/healtheqt/hospital-beds.htm. 22. Srivastav AK Ghosh M Bandekar SR Modeling of COVID-19 with limited public health resources: A comparative study of three most affected countries Eur. Phys. J. Plus 2021 136 1 26 10.1140/epjp/s13360-021-01333-y 23. Okuonghae D Omame A Analysis of a mathematical model for COVID-19 population dynamics in Lagos, Nigeria Chaos Solitons Fractals 2020 139 110032 10.1016/j.chaos.2020.110032 32834593 24. Senapati A Sardar T Ganguly KS Impact of adult mosquito control on dengue prevalence in a multi-patch setting: A case study in Kolkata (2014–2015) J. Theor. Biol. 2019 478 139 152 10.1016/j.jtbi.2019.06.021 31229456 25. Hsieh Y-H Van den Driessche P Wang L Impact of travel between patches for spatial spread of disease Bull. Math. Biol. 2007 69 1355 1375 10.1007/s11538-006-9169-6 17318677 26. Bock W Jayathunga Y Optimal control of a multi-patch dengue model under the influence of wolbachia bacterium Math. Biosci. 2019 315 108219 10.1016/j.mbs.2019.108219 31229468 27. Liu J Jia Y Zhang T Analysis of a rabies transmission model with population dispersal Nonlinear Anal. Real World Appl. 2017 35 229 249 10.1016/j.nonrwa.2016.10.011 28. Mhlanga A Mupedza TV A patchy theoretical model for the transmission dynamics of SARS-cov-2 with optimal control Sci. Rep. 2022 12 215 10.1038/s41598-022-21553-1 34997115 29. Hethcote HW Qualitative analyses of communicable disease models Math. Biosci. 1976 28 335 356 10.1016/0025-5564(76)90132-2 30. Arino J Van den Driessche P A multi-city epidemic model Math. Popul. Stud. 2003 10 175 193 10.1080/08898480306720 31. Rho Y-A Liebovitch LS Schwartz IB Dynamical response of multi-patch, flux-based models to the input of infected people: Epidemic response to initiated events Phys. Lett. A 2008 372 5017 5025 10.1016/j.physleta.2008.05.065 32288057 32. Muroya Y Enatsu Y Kuniya T Global stability of extended multi-group SIR epidemic models with patches through migration and cross patch infection Acta Mathematica Scientia 2013 33 341 361 10.1016/S0252-9602(13)60003-X 33. Muroya Y Kuniya T Wang J Stability analysis of a delayed multi-group SIS epidemic model with nonlinear incidence rates and patch structure J. Math. Anal. Appl. 2015 425 415 439 10.1016/j.jmaa.2014.12.019 34. Zhang H Xia J Georgescu P Multigroup deterministic and stochastic SEIRI epidemic models with nonlinear incidence rates and distributed delays: A stability analysis Math. Methods Appl. Sci. 2017 40 6254 6275 10.1002/mma.4453 35. Chen S Shi J Shuai Z Wu Y Asymptotic profiles of the steady states for an SIS epidemic patch model with asymmetric connectivity matrix J. Math. Biol. 2020 80 2327 2361 10.1007/s00285-020-01497-8 32377791 36. Liu L Jiang D Hayat T Dynamics of an SIR epidemic model with varying population sizes and regime switching in a two patch setting Physica A 2021 574 125992 10.1016/j.physa.2021.125992 37. Meng, L. & Zhu, W. Analysis of COVID-19 based on SEIR epidemic models in a multi-patch environment. Res. Square (2021) 38. Capasso V Serio G A generalization of the kermack-mckendrick deterministic epidemic model Math. Biosci. 1978 42 43 61 10.1016/0025-5564(78)90006-8 39. Lauer SA Grantz KH Bi Q The incubation period of coronavirus disease 2019 (COVID-19) from publicly reported confirmed cases: Estimation and application Ann. Intern. Med. 2020 172 577 582 10.7326/M20-0504 32150748 40. Aleta, A., Martin-Corral, D., Y Piontti, A. P. et al. Modelling the impact of testing, contact tracing and household quarantine on second waves of COVID-19. Nat. Hum. Behav. 4, 964–971 (2020). 41. Das, T., Srivastava, P. K. & Kumar, A. Nonlinear dynamical behavior of an SEIR mathematical model: Effect of information and saturated treatment. Chaos Interdiscip. J. Nonlinear Sci. 31, 043–104 (2021). 42. Aldila D Ndii MZ Samiadji BM Optimal control on COVID-19 eradication program in Indonesia under the effect of community awareness Math. Biosci. Eng. 2020 17 6355 6389 10.3934/mbe.2020335 33378859 43. Li R Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (sars-cov-2) Science 2020 368 489 493 10.1126/science.abb3221 32179701 44. Berman, A. & Plemmons, R. J. Nonnegative Matrices in the Mathematical Sciences (SIAM, 1994). 45. Horn, R. A. & Johnson, C. R. Topics in Matrix Analysis. (Cambridge University Press, 1991). 46. Horn & RA, J. C. R. Matrix Analysis (Cambridge University Press, 1985). 47. COVID-19, India, data accessed from: https://www.covid19india.org/. on 20th August, 2021 48. MATLAB. 9.7.0.1190202 (R2019b) (The MathWorks Inc., Natick, Massachusetts, 2018).