Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

A data-driven Markov process for infectious disease transmission

Abstract

The 2019 coronavirus pandemic exudes public health and socio-economic burden globally, raising an unprecedented concern for infectious diseases. Thus, describing the infectious disease transmission process to design effective intervention measures and restrict its spread is a critical scientific issue. We propose a level-dependent Markov model with infinite state space to characterize viral disorders like COVID-19. The levels and states in this model represent the stages of outbreak development and the possible number of infectious disease patients. The transfer of states between levels reflects the explosive transmission process of infectious disease. A simulation method with heterogeneous infection is proposed to solve the model rapidly. After that, simulation experiments were conducted using MATLAB according to the reported data on COVID-19 published by Johns Hopkins. Comparing the simulation results with the actual situation shows that our proposed model can well capture the transmission dynamics of infectious diseases with and without imposed interventions and evaluate the effectiveness of intervention strategies. Further, the influence of model parameters on transmission dynamics is analyzed, which helps to develop reasonable intervention strategies. The proposed approach extends the theoretical study of mathematical modeling of infectious diseases and contributes to developing models that can describe an infinite number of infected persons.

Citation: Wang C, Mustafa S (2023) A data-driven Markov process for infectious disease transmission. PLoS ONE 18(8): e0289897. https://doi.org/10.1371/journal.pone.0289897

Editor: Viswanathan Arunachalam, Universidad Nacional de Colombia, COLOMBIA

Received: February 13, 2023; Accepted: July 27, 2023; Published: August 10, 2023

Copyright: © 2023 Wang, Mustafa. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: https://github.com/CSSEGISandData/COVID-19

Funding: This research was funded by the National Natural Science Foundation of China [71671158, 71932002]. The funders had no role in study design data collection and analysis decision to publish or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

1. Introduction

The Coronavirus Disease 2019 (COVID-19) has rapidly stretched across the globe since its discovery. The outbreak of COVID-19 caused serious damage to the lives and properties of the people, interrupted normal productive life, and caused great losses [1–3]. Globally, as of January 9 2023, more than 660 million cases of COVID-19 have been confirmed, which includes more than 6.6 million cases of death [4]. Until now, humans still do not have a good means to effectively fight the virus. The study [5] pointed out that COVID-19 will still be around us for a long time. Therefore, there is an urgent need to find effective methods to capture and predict the transmission process of COVID-19. In addition, the method can be applied to analyze the role and effectiveness of non-drug intervention strategies. Further, more economical and effective strategies for non-drug interventions may be developed based on the analyses.

The COVID-19 transmission process can be considered as containing multiple stages [6, 7]. The stages have different transmission rates but are highly infectious [8, 9]. The initial stages of the transmission process show an exponential increase [10–12]. A large percentage of COVID-19 patients are found to have more severe symptoms [2]. The process of transmission of COVID-19 also seems to be seasonally related [13, 14]. When winter temperatures are extremely low, the transmission of COVID-19 becomes very rapid [15]. However, the simple characteristics of the COVID-19 transmission process are already known. However, it is still important to continue studying the COVID-19 transmission process to closely forecast its development over time. At the same time, there are no drugs or ways to treat COVID-19 rapidly and effectively. Therefore, in the fight against COVID-19, non-drug intervention strategies continue to be used mainly to limit the process of COVID-19 transmission. Hence, after further understanding the process of COVID-19, on the basis of the characteristics of the transmission process, there is a need to develop effective strategies for control and prevention to limit economic losses.

So far, control measures against COVID-19 have attracted much attention, and studies have shown their effectiveness [16–19], which can be divided into two main ways: (1) early identification of cases [20–27]. (2) increase the distance of transmission of the virus, including travel restrictions [28–30], individual and family isolation [19, 20, 31, 32], increased social distance [33–36], and exposure restrictions [37–41]. Moreover, some scholars have studied the effectiveness of other nondrug intervention strategies. Sarkar et al. [42] and Khajanchi et al. [43] demonstrated that quarantining susceptible individuals and contact tracing can control the COVID-19 outbreaks. Kumar Rai et al. [44] and Sarkar et al. [45] studied the effect of environmental contamination on COVID-19 pandemic dynamics considering and not considering vaccination coverage, respectively. Also, Kumar Rai et al. [46] and Khajanchi et al. [47] found that encouraging hospitalization and quarantine through the media helped control the disease’s prevalence.

Since the onset of the COVID-19 epidemic, several mathematical modeling approaches have been used to assess and predict the transmission process of COVID-19. Some scholars study the early diffusion of COVID-19 based on the Poisson processes [48] and the Markov Chain Monte Carlo method [49]. The statistical method based on time series data seems effective [50–53], but it can only follow previous patterns with no ability to predict the changing trend. Modeling based on agent-based simulation techniques, i.e., computing by copying the pattern of individuals (agents) in transmitting the infectious disease, is also a reasonable approach [54–58]. However, this model relies on population-level parameters that are difficult to obtain, such as movement rate, distance, and virus infectivity parameters. Models based on AI techniques are also interesting and effective approaches [59–64], but their effectiveness may be questioned in the absence of sufficient training data set as there are many learning steps they rely on.

Scholars mostly evaluate and predict COVID-19 based on the susceptibility-infection-recovery (SIR) model originally proposed by Kermack and McKendrick in 1927 [65]. This model and its evolved variants have been used by many scholars to simulate or predict the short-term dynamics of infectious diseases [66–70], to assess the impact of intervention strategies [71–74], to evaluate the impact of vaccines [75, 76], and to assess the stability of infectious disease transmission [77, 78]. Similarly, some studies use differential equation methods [79, 80]. Further, to study the random factors in the process, some scholars studied the model combining the Markov process and SIR [81–83] and the model combining the Hidden Markov process with SIR [84–86]. A common feature of these studies is to consider the same state (Susceptible, Infective, or Removal) as a whole during the state transition process. It is easy to see that studying all individuals as a whole will yield different results than studying each individual individually. On the other hand, it is necessary to set a maximum value for the number of susceptible patients in the SIR model. However, the number of susceptible in a region is not constant if inter-regional travel is considered. This setting is not realistic.

Our work is motivated by the fact that existing models used to describe infectious diseases or predict the process of their transmission generally assume that regions are closed and ignore population movements. However, cross-regional movement of populations is very frequent in today’s world, and the impact of population movement can increasingly not be ignored in transmitting and controlling infectious diseases. When population movements are considered, the number of people who can be infected by an infectious disease (or the total population) in an area can be considered variable or infinite. In this paper, we aim to develop a method that allows for the close description and analysis of the complex transmission process of infectious diseases without setting the maximum number of susceptible individuals. The method develops a new infectious diseases Markov model based on the level-dependent Markov process with infinite state space. The model can accurately describe the main characteristics of the infectious disease transmission process by four basic parameters that can be estimated from public data [4]. Based on the proposed infectious diseases Markov model, we propose a simulation method with a heterogeneous structure to obtain the model’s results. Experiments on six COVID-19 cases have shown that the method can well capture the real increasing process of infectious disease-infected patients.

The main contributions of this study are:

  1. In the context of globalization, it is very common for people to travel between different countries and regions. We consider this reality and propose a level-dependent Markov process model with an infinite state space to capture the transmission dynamics of infectious diseases. Unlike the SIR model, the model does not require a predetermined maximum number of susceptible individuals or populations. The proposed approach provides new ideas for research on modeling infectious diseases in the new globalized context.
  2. The mathematical formula for solving the model is given, but the infinite matrix needs to be truncated during the solution. It is difficult to truncate the infinite matrix within the allowed error range. In order to perform the solution quickly, we developed a heterogeneous simulation technique to reflect the situations in which different batch transmissions co-exist.
  3. Several numerical experiments with COVID-19 are given to verify the validity and suitability of the model proposed in this paper. The model proposed in this paper can effectively describe the process of infectious disease transmission that is stable or with turning points. And the error in applying the model to predict the infectious disease transmission process will also be smaller than that of existing models.
  4. We also evaluated the effects of the key parameters on the transmission process over time. The obtained results allow us to evaluate the implemented non-drug intervention strategies and provide the theoretical basis for the design of appropriate non-drug intervention strategies.

2. Materials and method

This section describes the methods adopted in this study. Section 2.1 describes the setting of the research problem and data. Section 2.2 gives the key parameters for constructing the model and how to estimate the parameters. Section 2.3 gives a Markov model with an infinite state space and a mathematical solution to this model, after which a heterogeneous simulation technique is proposed.

2.1. Research setting and data

We developed a new infectious disease Markov model to describe and characterize the diffusion of large-scale epidemics of infectious diseases, using COVID-19 as an example. The model is characterized by level dependence and the inclusion of infinite state space. This model divides the infectious diseases transmission process into two parts: Ⅰ. the generation process of new cases in phase form; Ⅱ. The disappearing process of cases is to cure, self-heal, or die one by one. Assuming that new cases can be detected in a timely manner, the emergence of new cases to their cure, self-healing, or death can be regarded as a service process. Under this assumption, we can view the transmission process of infectious diseases as a level-dependent quasi-birth and death (QBD) process with infinite service desks.

In the following part of simulation verification, we take COVID-19 as a typical example of an infectious disease and simulate different regions of the world that may be of concern: the state of New York, India, Egypt, South Korea, Italy, and Mexico. In selecting the data period, we chose 20 days of data within November 2020 for these countries because we hope our results can reflect the performance of COVID-19 transmission under more than one condition. The 20-day data results from the combined effects of climate, population density, seasonality, and different mitigation measures in different countries. At the end of our study, when we focus on the impact of different containment and mitigation strategies, we assume that some mitigation strategies change the value of transmission parameters to some extent and study the contribution of different parameters to the transmission through this method.

2.2. Parameters and measures

This section simplifies the random factors in the infectious disease transmission process into four key parameters. Based on Markov process theory, the parameters are assumed to obey exponential distribution [13]. The four key parameters can be described specifically:

  1. λ: The rate of infection. The rate of transmission of one batch per infectious disease patient. The two infection batches are assumed to follow an exponential distribution in their interval.
  2. μ: The rate of disappearing. The rate of disappearance of infectious diseases patients who recover themselves, die or are cured. Note that the time spent by patients disappearing also follows an exponential distribution.
  3. d: The batch size of infection. The number of patients included in a batch infected by infectious diseases patients.
  4. k: The starting number of actively infected cases. At the start of the observational period, the number of patients who are suffering from this infectious disease in the observation area.

By using some actual data, we can use the following method to define the following basic parameters: k, λ, d, and μ. Assume that there is a total of m observed epidemic data. And Ni is the total number of active cases reported on day i, ni is the number of newly infected patients increased on day i with comparison to day i-1, and ci is the number of newly disappeared infected patients on day i with comparison to day i-1.

The starting number of the actively infected cases (k) can be obtained directly from the actual statistics without calculation. Then, the parameters λ and d are given in Eq 1. (1) where depending on the actual data, d maybe 0, 1, 2, 3,… Similarly, Eq 2 gives the calculation method for parameter μ.

(2)

The values of these four parameters (λ, μ, d, and k) can be determined by the approach described above. The number of active cases who are suffering from this infectious disease in each period is our most concerned and critical indicator during an epidemic. Therefore, we need to develop a model that uses these four parameters to forecast the increasing process of the active case number.

2.3. Model and computer experiment procedure

An infectious disease Markov model is developed to forecast the increasing process of the active case number. In the model, let the state at the time t be E[N(t)], where the state represents the number of actively infected cases. And according to the infectious disease transmission process, the relationship of the transition between the states of the model ({N(t):t≥0}) is given in Fig 1.

thumbnail
Download:
Fig 1. A graph of the state transfer relationship of the model.

https://doi.org/10.1371/journal.pone.0289897.g001

where τ represents the rate of emergence of the first infected person in an area. It is usually small and does not affect the transmission of infection when it is created.

From Fig 1, it is possible to divide all states into different level sets, defined as follows: (3) (4) (5) (6)

By using these level sets (level n, n = 0,1,2,…), the infinitesimal generator of this infectious disease Markov model ({N(t):t≥0}) is given by (7) where, Ai,j is the rate matrix of the transfer from level i to level j. Ai,i is the square matrix with the same dimension as level i. The dimensions of the other Ai,j can be determined based on Ai,i. This details of Ai,j is presented in the Supporting information (S1 File). Let the initial probability vector ω of this Markov model be (8) where the (k+1)th cell of ω is 1, indicating that the probability of the initial state being k is 1. Suppose pn(t) is the probability that the Markov model is in state n at time t, i.e.

(9)

Then, the transient probability vector (P(t)) of the Markov model at time t can be written as: (10)

According to the Chapman-Kolmogorow equation, P(t) can be obtained from Eq (11).

(11)

Based on the transient probability vector at time t, the average number of active cases at time t can be easily obtained by Eq 12.

(12)

It is easy to see that Q is an infinite matrix. During the solution process, it is necessary to truncate this infinite matrix and keep the error within the allowed range. It is difficult to truncate the matrix Q. For this reason, we developed a heterogeneous computer simulation technique based on this infectious disease transmission model. In the simulation process, the coexistence of different batch sizes of infection was considered, that is, the overall transmission process could be regarded as the union of the infection transmission processes of different batch sizes. The weights of the transmission processes with different infection batches in the combination are different.

The infectious disease transmission process in the coexistence of two infection batches is given in Fig 2. In Fig 2, k is the total number of actively infected cases for all groups at the start of the observation, ki is the starting number of the actively infected cases in the group i, ri is the weight value of the group i, λi is the rate of infection in group i, μi is the rate of disappearing in group i, and di is the batch size of infected patients in group i. The weight value for a given group i is the ratio between the number of actively infected cases in the group i and the total number of actively infected cases at the beginning of the observation period, i.e. . The total infected batch size is obtained by weighting the average infected batch sizes of the two groups, i.e. .

thumbnail
Download:
Fig 2. The Markov model with two groups of infectious batches.

https://doi.org/10.1371/journal.pone.0289897.g002

Note that (13)

We adjusted the value of weight ri for the groups of different batches in the simulation experiment to make them nearly to the real reported values.

3. Results

In this section, using COVID-19 as an example, we describe the transmission of COVID-19 in some regions by applying the methods proposed in this paper and show how to evaluate the effects of interventions. Section 3.1 describes the process of simulating increasing activity cases of COVID-19 using actual data and compares the simulation results with the actual situation. In section 3.2, we experimentally illustrate the effect of each parameter on the COVID-19 transmission process over time.

3.1. Simulation results

In this subsection, COVID-19 was used as a typical example of infectious disease, and the increasing process of COVID-19 activity cases in 6 regions during the 20-day observation period was simulated to reflect the transmission of COVID-19. The key parameters in these examples were obtained by the approach in section 2.2 and based on actual data [4]. Then, the increasing process of COVID-19 active cases in these examples was simulated based on the infectious disease Markov model and heterogeneous simulation techniques proposed in this paper and compared with the real situation.

New York, India, and Egypt were simulated first. The simulation results for New York are presented in the main text, and the simulation results for India and Egypt are presented in the Supporting information (S1 and S2 Tables, S1 and S2 Figs). They represent the transmission process without turning points. Their transmission parameters did not change significantly during the observation period. This is because no intervention strategy was adopted, or the intervention strategy did not significantly change the transmission process of the pandemic during this period.

The epidemic situation in the United States has aroused a lot of attention, and New York is a state of concern in the United States, so first, the epidemic in New York State was simulated. Based on the COVID-19 data reported for New York State, as given in Table 1, the following parameters for the New York State epidemic could be determined. k = 101592, and according to Eqs (1) and (2), the mean infection rate (λ) and disappearing rate (μ) were given as 0.035073852 and 0.006084398, respectively. Then, the increasing number of COVID-19 activity cases in New York State was simulated. The batches of infection were determined to be d1 = 1 and d2 = 2, and the two groups’ infection processes with infectious batches of 1 and 2 had weights (r) of 0.971 and 0.029, respectively. A comparison of the results of the simulation with those of the reported real data is shown in Fig 3.

thumbnail
Download:
Fig 3. The comparison of simulation results with reported data in New York State.

https://doi.org/10.1371/journal.pone.0289897.g003

thumbnail
Download:
Table 1. Reported COVID-19 data for New York State for November 6–25.

https://doi.org/10.1371/journal.pone.0289897.t001

Fig 3 depicts the comparison of the simulation results obtained using our method with the actual COVID-19 transmission process in New York State. The blue solid line represents the simulation results, and the red dashed line represents the actual number of active cases in the Fig 3. The comparison shows that the simulation results of the COVID-19 transmission process obtained according to our proposed method are roughly the same as the actual COVID-19 transmission process using the parameters derived from the actual data.

To further validate the proposed model, we also simulated the epidemic in India and Egypt from November 1 to 20, 2020. The data used for the simulations and the results obtained are shown in the Supporting information (S1 and S2 Tables, S1 and S2 Figs). This shows that the proposed model enables to catch the COVID-19 transmission dynamics well when there is little fluctuation in the transmission parameters.

Unlike the three examples above, the transmission parameters may be turning due to the interventions or the disasters encountered. Therefore, the epidemic in Korea, Italy, and Mexico was further simulated, and there is a turning point (tc) in the growth curve of the activity cases for these three examples. Similar examples with turning points can be found in the literature [16]. The existence of these turning points reflects that the transmission process of COVID-19 in these examples was affected by the intervention strategy or other reasons. Therefore, we also included a turning point (tc) in the simulation of the processes in these three regions.

First, based on the reported data related to COVID-19 in Korea from November 2 to 21, 2020, given in Table 2, a turning point (tc) was found in the COVID-19 transmission process. After calculation, it can be obtained that k = 1825, the average infection rate (λ) and disappearing rate (μ) before the turning point, tc, are 0.062135947 and 0.053096363, respectively. Their values equal 0.09774732 and 0.038994525, respectively, after the turning point, tc. In the simulation, the infection batches were determined to be d1 = 1 and d2 = 2. Before the turning point tc, the weights (r) of the two groups equal 0.940 and 0.060, respectively. And after the turning point tc, they become 0.434 and 0.566, respectively. The simulation results are compared with the reported data in Fig 4.

thumbnail
Download:
Fig 4. The comparison of simulation results with reported data in South Korea.

https://doi.org/10.1371/journal.pone.0289897.g004

thumbnail
Download:
Table 2. Data on COVID-19 for South Korea was reported from the 2nd to the 21st of November.

https://doi.org/10.1371/journal.pone.0289897.t002

Fig 4 shows the comparison of the simulation results obtained using our method with the actual transmission process of COVID-19 in South Korea. The blue solid line in Fig 4 indicates the simulation results, and the red dashed line indicates the number of active cases and tc indicates the turning points in the transmission process. The comparison results show that for the COVID-19 transmission process with a turning point, the simulation results obtained according to our proposed method are approximately the same as the actual data. In this case, the simulation results were obtained using the parameters calculated based on the actual data.

Similarly, to validate that the proposed model can track the dynamics of transmission processes with turning points, the COVID-19 transmission processes in Italy and Mexico were also simulated. The data used for the simulations and the results obtained are shown in the Supporting information (S3 and S4 Tables, S3 and S4 Figs). This indicates that the proposed model can also track the transmission dynamics of the COVID-19 transmission process with a turning point.

According to all the previous simulation results, the following Table 3 is obtained ().

thumbnail
Download:
Table 3. Key parameters in the dissemination of COVID-19 in each country.

https://doi.org/10.1371/journal.pone.0289897.t003

3.2. Influence of parameters

In the following, we assess how three parameters that non-drug intervention strategies may influence impact the infectious disease transmission process over time. The three parameters include k, λ, and d. Note that non-drug interventions can only limit the transmission of the virus and have little impact on patients who are already ill. Therefore, it is assumed that the non-drug intervention strategies that can be taken at present cannot affect μ.

To assess the effect of the parameters, we introduced a new indicator and defined (14) where, θ and denote the value of each parameter before and after changing the nondrug intervention strategy, respectively; and E[Ni(t)] denotes the average number of infected patients with parameter value i.

After that, we first use Egypt as an example and analyze the role of each parameter using Egyptian data. The COVID-19 transmission parameters for Egypt were as follows: k = 1903; λ = 0.085434063; μ = 0.045978143; and two sets of weights (r) of 0.946 and 0.054 for d1 = 1 and d2 = 2, respectively. After that, we assumed that each parameter was changed by the same proportion to half of its original value due to non-drug interventions: = ; = . In order to make = , let d1 = 1 is changed to = 0, d2 = 2 is changed to = 1; = 0.518, = 0.482 are changed from r1 = 0.946, r2 = 0.054, respectively. The trend of ρ with time after changing the parameters is given for Egypt in Fig 5.

thumbnail
Download:
Fig 5. The trend of ρ in Egypt with time when changing k, λ, and d.

https://doi.org/10.1371/journal.pone.0289897.g005

Fig 5 shows the effects of changing k, λ, and d on the COVID-19 transmission process in Egypt. The green, blue, and red lines in Fig 5 represent the trend of the effects of changing k, λ, and d over time, respectively. The results show that change k reduces the number of active COVID-19 cases and the effect does not change over time. Changing λ and d have a small effect on the number of active COVID-19 cases in the early days, and this effect increases significantly over time.

Further, taking South Korea as an example, we analyze the role of each parameter again when there is a turning point in the transmission process. The COVID-19 transmission parameters for South Korea were as follows: tc = 11; k = 1825; λ before and after tc are 0.062135947 and 0.09774732, respectively; μ before and after tc are 0.053096363 and 0.038994525, respectively; the weight of group = 1 is 0.940 before tc; that of group = 2 is 0.060 before tc; the weight of group = 1 is 0.434 after tc; and the weight of group = 2 is 0.566 after tc. Similar to the previous example, the parameter becomes half of the original one. = ; = ; = 1 is changed to = 0, = 2 is changed to = 1 for i = 1, 2; = 0.470, = 0.530, = 0.217, = 0.783 are changed from = 0.940, = 0.060, = 0.434, = 0.566, respectively. The trend of ρ with time after changing the parameters is given for Egypt in Fig 6. The effects of the three parameters in Korea were found to be similar to those in Egypt.

thumbnail
Download:
Fig 6. The trend of ρ in South Korea with time when changing k, λ, and d.

https://doi.org/10.1371/journal.pone.0289897.g006

Fig 6 depicts the effects of changing k, λ, and d on the transmission process of COVID-19 in South Korea. The green, blue, and red lines in Fig 6 represent the trends of the effects of changing k, λ, and d over time, respectively. The results obtained are the same as those in Fig 5. Changing k decreases the number of active COVID-19 cases and the effect does not change over time. Changing λ and d have a small effect on the number of active COVID-19 cases in the early days, and this effect increases significantly over time.

4. Discussion

In Figs 3 and 4 in the main text and Figures in the Supporting information (S1–S4 Figs), it can be seen that there is a good match between the simulation results output based on the proposed method in this paper and the actual data, regardless of whether there are turning points in the transmission process of COVID-19. Figs 5 and 6 shows the role of the model’s key parameters (λ, d, and k) for the model. How a non-drug intervention strategy affects parameter values can be derived by calculating and analyzing parameter values before and after the application of the intervention strategy. The influence of non-drug intervention strategies on the growth process of patients with COVID-19 infection and the transmission process of COVID-19 over time was evaluated using parameters as a medium. In addition, the future increase process of patients infected with COVID-19 can be predicted by extending the length of time of the simulation. Further, the results can be applied to purposefully adjust or develop measures that can effectively control COVID-19 to reduce economic losses due to measure implementation.

From the simulation of stable regions (Fig 3, S1 and S2 Figs) in the observation period: The four key parameters of the proposed Markov model for COVID-19 transmission can be calculated from the actual data [4] (Table 1, S1 and S2 Tables). These parameters comprehensively reflect the complex transmission situation in the observation area, including population density, climate, seasonality, etc. The actual spread process of COVID-19 is a stochastic and complex process, and the patients who are already infected with COVID-19 may infect new patients with different infectious batches. Because the infectious batches in our proposed model can only be integers, it is difficult to characterize the problem of the COVID-19 transmission process by just one infectious disease Markov process. To effectively investigate this problem, it is possible to consider the overall COVID-19 transmission process as the combination of multiple groups that are all independent COVID-19 transmission models. Combining multiple groups of COVID-19 transmission models can capture the dynamics of the increasing number of patients with COVID-19 infection within a region. According to our experimental findings, the increasing dynamics of the number of patients with COVID-19 infection in a region can be effectively captured by considering at least two different groups of COVID-19 transmission models. The heterogeneity of patient transmission does not affect the application of the model.

According to the simulation of regions with turning points (Fig 4, S3 and S4 Figs), we link the turning points with human intervention measures and natural environment changes. This study identifies the turning points in the observation period for several examples through actual data (Table 2, S3 and S4 Tables). By comparing the values of the parameters before and after the turning point, the control effect of the intervention strategy can be assessed. Note that there is a delay period between the time of implementation of the non-drug intervention strategy and the time of the turning point. A particular intervention strategy usually adopted will impact multiple parameters in the COVID-19 transmission process.

We identify the possible causes of turning points by finding out what measures the country took before the turning point. We suppose this change may be due to the relaxation of the social distance restriction in Korea on November 1, 2020. After a few days of transmission, this led to the uncontrollable transmission of the epidemic in the winter. The curfew policy and emergency imposed in some Italian regions after November 5, 2020, could be the reason for the turning point tc = 12 in Italy. On November 11, 2020, Mexico was hit by a hurricane. The reduction of social distance between people caused by centralized placement may be the cause of tc = 7 in Mexico. In this study, the focus is not to prove which measure causes the turning points but to better describe the COVID-19 transmission process through turning points. So, it is acceptable to find the cause of turning points this way. If the reader is interested, the location of the turning point can be found through simple outlier detection and other methods.

At the same time, the parameter changes around the turning point reveal that: (1) The implementation of curfews and emergency state strategies in parts of Italy was effective in controlling the increase in the number of infected patients and the transmission of COVID-19. This indicates that the transmission of COVID-19 can be controlled by adopting the right control approaches. (2) The new social evacuation policy in Korea aimed to control the epidemic further, but it did not work as expected. This may be because it reduces the public’s sense of crisis. This indicates that adopting the wrong control approach may accelerate the spread of COVID-19. (3) Finally, Mexico is an example of how the natural environment can undermine our efforts to control the epidemic. These parameters are used as mediators to establish links between the strategies implemented to control COVID-19 and the increasing process of COVID-19 patient numbers. These links can be used to guide decisions on non-drug intervention strategies.

The parameters for the six examples are given in Table 3. By looking at these parameters, we can clarify the function of the parameters in characterizing the spread of COVID-19. λ, k, and d jointly determine the trend of the COVID-19 transmission process (we believe that non-drug intervention strategies cannot affect μ, so it was not studied). When dλ-μ>0, the COVID-19 activity case number increases, it indicates that the COVID-19 transmission remains severe; conversely, when dλ-μ<0, the COVID-19 activity case number decreases, the COVID-19 transmission is under control. It can also be found that the larger the |dλ-μ| faster the increase or decrease of COVID-19 active cases.

Then, we do two simulations, as shown in Figs 5 and 6. The purpose of these two simulations is to clarify the influence of single parameter change on transmissions. Measures to reduce k (e.g., the earlier discovery of infectious diseases) can immediately reduce the COVID-19 activity case number during the observation period. Moreover, this approach’s influence on the COVID-19 spread is consistent throughout the observation period. This shows the importance of early detection and early implementation of intervention strategies. Comparatively, implementing measures to reduce λ and d is relatively ineffective in limiting the spread of the epidemic at the beginning of the observation period, but this effect becomes greater as the observation period proceeds. This suggests that the long-term impact of reducing the rate and batch of infection on transmission is more significant and increases with time after a widespread and rapid transmission of an epidemic. Therefore, it is more reasonable and economical to adopt different plans to reduce k, λ, and d in the face of different infectious disease transmission situations.

The simulation results of the proposed model were compared with the actual reported data to show that this model can well capture the dynamics of the increasing process of the infectious diseases patient number. In addition, the experimental results of this paper are compared with those of some effective studies based on the SIR model [65–67]. It can be intuitively found that the model proposed in this paper can better capture the transmission process of COVID-19. This is due to the fact that the study was improved in two ways. On the one hand, the models based on SIR assume that the observation area is closed and the total population in the area is constant. This assumption does not align with reality in the current context of frequent international and interregional communication. Moein, Nickaeen [66] also noted this issue. On the other hand, patients in the same state of the SIR model were considered as a whole for the calculation. This assumption is also not fully in line with reality. These two issues are fully considered in the construction of the model in this paper so that better experimental results are achieved.

5. Conclusions & future research directions

5.1. Conclusions

In this study, we propose a new Markov model for infectious diseases that requires the measurement of only four parameters, which has an infinite state space because of the unnecessary setting of maxima. To solve this model, we develop a simulation method for heterogeneous infections. Numerical experiments were conducted to analyze six regions with high transmission of infectious diseases using COVID-19 as an example. These analyses reveal that the model proposed in this paper can effectively capture the spread process of infectious diseases which are similar to COVID-19 and assess the effectiveness of the strategies being implemented to limit infectious diseases. The approach proposed in this paper extends research in infectious disease modeling and provides new ideas for predicting infectious disease transmission and designing effective intervention strategies.

Our study can provide the following recommendations for controlling and limiting COVID-19 transmission:

  1. Outbreaks should be detected early, and widespread spread should be prevented. If COVID-19 does not spread widely in the area, the number of patients is not uniform in different areas. In this case, subregional control (intercity travel restrictions) is an effective approach. This is an effective way to prevent the wider spread of the virus; When the virus has been widely distributed in the observation area, subregional control is not an effective control method, so other more effective control measures should be selected.
  2. According to the two simulations in Figs 5 and 6, we believe that control λ and d can continuously play a role in virus transmission, but the role of λ and d may be different in different regions. How it works is influenced by the region’s climate, population, the epidemic’s spread, and so on. In the actual work of controlling the spread of COVID-19, the role of λ and d in the region can be judged according to the actual data of different regions, and appropriate control strategies can be adopted based on this.
  3. Existing studies have found several ways to reduce the spread of COVID-19, but now countries generally implement multiple control measures simultaneously. We cannot judge from the actual situation how a single control strategy affects k, λ, and d. Considering that COVID-19 may exist in human life for a long time, relevant institutions can clarify the role of a single control strategy through small-scale practical experiments and select appropriate methods for large-scale implementation based on the results of experimental studies.

5.2. Limitations and future research

This study has a few limitations, which can be addressed by future research.

First, the proposed approach has high demands on the devices for computing and storage. This is due to the fact that the model describes the infectious disease transmission process in more detail. If necessary, the calculation process can be optimized to increase the speed of the calculation.

Second, the model was constructed without considering incubation patients or asymptomatic patients. The presence of latent or asymptomatic patients can cause additional infections. Considering the incubation of patients for constructing new models is the focus of our future research.

Finally, some countries have adopted mass vaccination against COVID-19. Some vaccinated people are not infected with infectious diseases9, even if they are exposed to the virus. So, this approach will reduce the transmission of infectious diseases. It is also a good study to construct a new model of infectious disease transmission considering vaccine immunization.

Supporting information

S1 File. The derivation of the model proposed in this paper.

https://doi.org/10.1371/journal.pone.0289897.s001

(DOC)

S1 Table. Data on COVID-19 cases of India from Nov. 1 to 20, 2020.

https://doi.org/10.1371/journal.pone.0289897.s002

(DOC)

S2 Table. Data on COVID-19 cases of Egypt from Nov. 1 to 20, 2020.

https://doi.org/10.1371/journal.pone.0289897.s003

(DOC)

S3 Table. Data on COVID-19 cases of Italy from Nov. 11 to 30, 2020.

https://doi.org/10.1371/journal.pone.0289897.s004

(DOC)

S4 Table. Data on COVID-19 cases of Mexico from Nov. 13 to Dec. 2, 2020.

https://doi.org/10.1371/journal.pone.0289897.s005

(DOC)

S1 Fig. The growth process of confirmed cases of India.

https://doi.org/10.1371/journal.pone.0289897.s006

(DOC)

S2 Fig. The growth process of confirmed cases of Egypt.

https://doi.org/10.1371/journal.pone.0289897.s007

(DOC)

S3 Fig. The growth process of confirmed cases of Italy.

https://doi.org/10.1371/journal.pone.0289897.s008

(DOC)

S4 Fig. The growth process of confirmed cases of Mexico.

https://doi.org/10.1371/journal.pone.0289897.s009

(DOC)

References

  1. 1. Duan H, Wang S, Yang C. Coronavirus: limit short-term economic damage. Nature. 2020;578(7796):515–6. pmid:32099120
  2. 2. Grasselli G, Zangrillo A, Zanella A, Antonelli M, Cabrini L, Castelli A, et al. Baseline characteristics and outcomes of 1591 patients infected with SARS-CoV-2 admitted to ICUs of the Lombardy Region, Italy. Jama. 2020;323(16):1574–81. https://doi.org/10.1001/jama.2020.5394.
  3. 3. Coccia M. The relation between length of lockdown, numbers of infected people and deaths of Covid-19, and economic growth of countries: Lessons learned to cope with future pandemics similar to Covid-19 and to constrain the deterioration of economic system. Science of The Total Environment. 2021;775:145801. https://doi.org/10.1016/j.scitotenv.2021.145801.
  4. 4. CSSE J. Novel Coronavirus (COVID-19) Cases 2022. https://github.com/CSSEGISandData/COVID-19 (accessed December 29th 2020)
    • 5. Cobey S. Modeling infectious disease dynamics. Science. 2020;368(6492):713–4. pmid:32332062
    • 6. Chen T-M, Rui J, Wang Q-P, Zhao Z-Y, Cui J-A, Yin L. A mathematical model for simulating the phase-based transmissibility of a novel coronavirus. Infectious diseases of poverty. 2020;9(1):1–8. https://doi.org/10.1186/s40249-020-00640-3.
    • 7. Wang H, Wang Z, Dong Y, Chang R, Xu C, Yu X, et al. Phase-adjusted estimation of the number of coronavirus disease 2019 cases in Wuhan, China. Cell discovery. 2020;6(1):1–8. pmid:32133152
    • 8. Böhmer MM, Buchholz U, Corman VM, Hoch M, Katz K, Marosevic DV, et al. Investigation of a COVID-19 outbreak in Germany resulting from a single travel-associated primary case: a case series. The Lancet Infectious Diseases. 2020;20(8):920–8. pmid:32422201
    • 9. Wu JT, Leung K, Bushman M, Kishore N, Niehus R, de Salazar PM, et al. Estimating clinical severity of COVID-19 from the transmission dynamics in Wuhan, China. Nature medicine. 2020;26(4):506–10. pmid:32284616
    • 10. Li Q, Guan X, Wu P, Wang X, Zhou L, Tong Y, et al. Early transmission dynamics in Wuhan, China, of novel coronavirus–infected pneumonia. New England journal of medicine. 2020;382:1199–207. pmid:31995857
    • 11. Zhao S, Lin Q, Ran J, Musa SS, Yang G, Wang W, et al. Preliminary estimation of the basic reproduction number of novel coronavirus (2019-nCoV) in China, from 2019 to 2020: A data-driven analysis in the early phase of the outbreak. International journal of infectious diseases. 2020;92:214–7. pmid:32007643
    • 12. Zhao S, Musa SS, Lin Q, Ran J, Yang G, Wang W, et al. Estimating the unreported number of novel coronavirus (2019-nCoV) cases in China in the first half of January 2020: a data-driven modelling analysis of the early outbreak. Journal of clinical medicine. 2020;9(2):388. pmid:32024089
    • 13. Baker RE, Yang W, Vecchi GA, Metcalf CJE, Grenfell BT. Susceptible supply limits the role of climate in the early SARS-CoV-2 pandemic. Science. 2020;369(6501):315–9. pmid:32423996
    • 14. Kissler SM, Tedijanto C, Goldstein E, Grad YH, Lipsitch M. Projecting the transmission dynamics of SARS-CoV-2 through the postpandemic period. Science. 2020;368(6493):860–8. pmid:32291278
    • 15. Xie J, Zhu Y. Association between ambient temperature and COVID-19 infection in 122 cities from China. Science of the Total Environment. 2020;724:138201. pmid:32408450
    • 16. Dehning J, Zierenberg J, Spitzner FP, Wibral M, Neto JP, Wilczek M, et al. Inferring change points in the spread of COVID-19 reveals the effectiveness of interventions. Science. 2020;369(6500):eabb9789. pmid:32414780
    • 17. Hsiang S, Allen D, Annan-Phan S, Bell K, Bolliger I, Chong T, et al. The effect of large-scale anti-contagion policies on the COVID-19 pandemic. Nature. 2020;584(7820):262–7. pmid:32512578
    • 18. Lai S, Ruktanonchai NW, Zhou L, Prosper O, Luo W, Floyd JR, et al. Effect of non-pharmaceutical interventions to contain COVID-19 in China. nature. 2020;585(7825):410–3. pmid:32365354
    • 19. Pei S, Kandula S, Shaman J. Differential effects of intervention timing on COVID-19 spread in the United States. Science advances. 2020;6(49):eabd6370. pmid:33158911
    • 20. Hellewell J, Abbott S, Gimma A, Bosse NI, Jarvis CI, Russell TW, et al. Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts. The Lancet Global Health. 2020;8(4):e488–e96. pmid:32119825
    • 21. Buckee CO, Balsari S, Chan J, Crosas M, Dominici F, Gasser U, et al. Aggregated mobility data could help fight COVID-19. Science. 2020;368(6487):145–6. pmid:32205458
    • 22. He X, Lau EH, Wu P, Deng X, Wang J, Hao X, et al. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nature medicine. 2020;26(5):672–5. pmid:32296168
    • 23. Peak CM, Kahn R, Grad YH, Childs LM, Li R, Lipsitch M, et al. Individual quarantine versus active monitoring of contacts for the mitigation of COVID-19: a modelling study. The Lancet Infectious Diseases. 2020;20(9):1025–33. pmid:32445710
    • 24. Pullano G, Di Domenico L, Sabbatini CE, Valdano E, Turbelin C, Debin M, et al. Underdetection of cases of COVID-19 in France threatens epidemic control. Nature. 2021;590(7844):134–9. pmid:33348340
    • 25. Li R, Pei S, Chen B, Song Y, Zhang T, Yang W, et al. Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (SARS-CoV-2). Science. 2020;368(6490):489–93. pmid:32179701
    • 26. Wang L, Didelot X, Yang J, Wong G, Shi Y, Liu W, et al. Inference of person-to-person transmission of COVID-19 reveals hidden super-spreading events during the early outbreak phase. Nature communications. 2020;11(1):1–6. https://doi.org/10.1038/s41467-020-18836-4.
    • 27. Ferretti L, Wymant C, Kendall M, Zhao L, Nurtay A, Abeler-Dörner L, et al. Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing. Science. 2020;368(6491):eabb6936. pmid:32234805
    • 28. Chinazzi M, Davis JT, Ajelli M, Gioannini C, Litvinova M, Merler S, et al. The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak. Science. 2020;368(6489):395–400. pmid:32144116
    • 29. Della Rossa F, Salzano D, Di Meglio A, De Lellis F, Coraggio M, Calabrese C, et al. A network model of Italy shows that intermittent regional strategies can alleviate the COVID-19 epidemic. Nature communications. 2020;11(1):1–9. https://doi.org/10.1038/s41467-020-18827-5.
    • 30. Kraemer MU, Yang C-H, Gutierrez B, Wu C-H, Klein B, Pigott DM, et al. The effect of human mobility and control measures on the COVID-19 epidemic in China. Science. 2020;368(6490):493–7. pmid:32213647
    • 31. Chang SL, Harding N, Zachreson C, Cliff OM, Prokopenko M. Modelling transmission and control of the COVID-19 pandemic in Australia. Nature communications. 2020;11(1):1–13. https://doi.org/10.1038/s41467-020-19393-6.
    • 32. Salje H, Tran Kiem C, Lefrancq N, Courtejoie N, Bosetti P, Paireau J, et al. Estimating the burden of SARS-CoV-2 in France. Science. 2020;369(6500):208–11. pmid:32404476
    • 33. Giordano G, Blanchini F, Bruno R, Colaneri P, Di Filippo A, Di Matteo A, et al. Modelling the COVID-19 epidemic and implementation of population-wide interventions in Italy. Nature medicine. 2020;26(6):855–60. pmid:32322102
    • 34. Courtemanche C, Garuccio J, Le A, Pinkston J, Yelowitz AJHa. Strong Social Distancing Measures In The United States Reduced The COVID-19 Growth Rate: Study evaluates the impact of social distancing measures on the growth rate of confirmed COVID-19 cases across the United States. Health affairs. 2020;39(7):1237–46. https://doi.org/10.1377/hlthaff.2020.00608.
    • 35. Zhang J, Litvinova M, Liang Y, Wang Y, Wang W, Zhao S, et al. Changes in contact patterns shape the dynamics of the COVID-19 outbreak in China. Science. 2020;368(6498):1481–6. pmid:32350060
    • 36. Nishiura H, Oshitani H, Kobayashi T, Saito T, Sunagawa T, Matsui T, et al. Closed environments facilitate secondary transmission of coronavirus disease 2019 (COVID-19). MedRxiv. 2020. https://doi.org/10.1101/2020.02.28.20029272.
    • 37. Leung NH, Chu DK, Shiu EY, Chan K-H, McDevitt JJ, Hau BJ, et al. Respiratory virus shedding in exhaled breath and efficacy of face masks. Nature medicine. 2020;26(5):676–80. pmid:32371934
    • 38. Chu DK, Akl EA, Duda S, Solo K, Yaacoub S, Schünemann HJ, et al. 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(10242):1973–87. pmid:32497510
    • 39. Liu Y, Ning Z, Chen Y, Guo M, Liu Y, Gali NK, et al. Aerodynamic analysis of SARS-CoV-2 in two Wuhan hospitals. Nature. 2020;582(7813):557–60. pmid:32340022
    • 40. Lunn P, Belton C, Lavin C, McGowan F, Timmons S, Robertson D. Using behavioural science to help fight the coronavirus. ESRI working paper, 2020.
    • 41. Team IC-F. Modeling COVID-19 scenarios for the United States. Nature medicine. 2021;27(1):94–105. pmid:33097835
    • 42. Sarkar K, Khajanchi S, Nieto JJ. Modeling and forecasting the COVID-19 pandemic in India. Chaos Solitons & Fractals. 2020;139:110049. pmid:32834603
    • 43. Khajanchi S, Sarkar K, Mondal J, Nisar KS, Abdelwahab SF. Mathematical modeling of the COVID-19 pandemic with intervention strategies. Results Phys. 2021;25:104285. pmid:33977079
    • 44. Kumar Rai R, Kumar Tiwari P, Khajanchi S. Modeling the influence of vaccination coverage on the dynamics of COVID‐19 pandemic with the effect of environmental contamination. Mathematical Methods in the Applied Sciences. 2023;46(12): 12425–12453. https://doi.org/10.1002/mma.9185.
    • 45. Sarkar K, Mondal J, Khajanchi S. How do the contaminated environment influence the transmission dynamics of COVID-19 pandemic? The European Physical Journal Special Topics. 2022;231(18–20):3697–716. pmid:36033354
    • 46. Rai RK, Khajanchi S, Tiwari PK, Venturino E, Misra AK. Impact of social media advertisements on the transmission dynamics of COVID-19 pandemic in India. Journal of Applied Mathematics and Computing. 2022;68(1):19–44. pmid:33679275
    • 47. Khajanchi S, Sarkar K, Mondal J. Dynamics of the COVID-19 pandemic in India. arxiv. 2020:2005.06286. https://doi.org/10.48550/arXiv.2005.06286.
    • 48. Zhu Y, Chen YQ. On a statistical transmission model in analysis of the early phase of COVID-19 outbreak. Statistics in Biosciences. 2021;13(1):1–17. pmid:32292527
    • 49. Wu JT, Leung K, Leung GM. Nowcasting and forecasting the potential domestic and international spread of the 2019-nCoV outbreak originating in Wuhan, China: a modelling study. The Lancet. 2020;395(10225):689–97. https://doi.org/10.1016/S0140-6736(20)30260-9.
    • 50. Elmousalami HH, Hassanien AE. Day level forecasting for Coronavirus Disease (COVID-19) spread: analysis, modeling and recommendations. arXiv. 2020:2003.07778. https://doi.org/10.48550/arXiv.2003.07778.
    • 51. Atangana A, İğret Araz S. Modeling and forecasting the spread of COVID-19 with stochastic and deterministic approaches: Africa and Europe. Advances in Difference Equations. 2021;2021(1):1–107. pmid:33495699
    • 52. Reddy T, Shkedy Z, Janse van Rensburg C, Mwambi H, Debba P, Zuma K, et al. Short-term real-time prediction of total number of reported COVID-19 cases and deaths in South Africa: a data driven approach. BMC medical research methodology. 2021;21(1):1–11. https://doi.org/10.1186/s12874-020-01165-x.
    • 53. Coccia M. An index to quantify environmental risk of exposure to future epidemics of the COVID-19 and similar viral agents: Theory and practice. Environmental Research. 2020;191:110155. pmid:32871151
    • 54. Shamil M, Farheen F, Ibtehaz N, Khan IM, Rahman MS. An agent-based modeling of COVID-19: validation, analysis, and recommendations. Cognitive Computation. 2021:1–12. pmid:33643473
    • 55. Kerr CC, Stuart RM, Mistry D, Abeysuriya RG, Rosenfeld K, Hart GR, et al. Covasim: an agent-based model of COVID-19 dynamics and interventions. PLOS Computational Biology. 2021;17(7):e1009149. pmid:34310589
    • 56. Sharma A, Bahl S, Bagha AK, Javaid M, Shukla DK, Haleem A. Multi-agent system applications to fight COVID-19 pandemic. Apollo Medicine. 2020;17(5):41. https://doi.org/10.4103/am.am_54_20.
    • 57. Kano T, Yasui K, Mikami T, Asally M, Ishiguro A. An agent-based model of the interrelation between the COVID-19 outbreak and economic activities. Proceedings of the Royal Society A. 2021;477(2245):20200604. pmid:33633491
    • 58. Leung K, Wu JT, Leung GMJNc. Real-time tracking and prediction of COVID-19 infection using digital proxies of population mobility and mixing. Nature communications. 2021;12(1):1–8. https://doi.org/10.1038/s41467-021-21776-2.
    • 59. Mei X, Lee H-C, Diao K-y, Huang M, Lin B, Liu C, et al. Artificial intelligence–enabled rapid diagnosis of patients with COVID-19. Nature medicine. 2020;26(8):1224–8. pmid:32427924
    • 60. Meirom EA, Maron H, Mannor S, Chechik G. How to stop epidemics: controlling graph dynamics with reinforcement learning and graph neural networks. arXiv. 2020:2010.05313. https://doi.org/10.48550/arXiv.2010.05313.
    • 61. Arik S, Li C-L, Yoon J, Sinha R, Epshteyn A, Le L, et al. Interpretable sequence learning for COVID-19 forecasting. Advances in Neural Information Processing Systems. 2020;33:18807–18. https://doi.org/10.48550/arXiv.2008.00646.
    • 62. Sun C, Hong S, Song M, Li H, Wang Z. Predicting COVID-19 disease progression and patient outcomes based on temporal deep learning. BMC Medical Informatics Decision Making. 2021;21(1):1–16. https://doi.org/10.1186/s12911-020-01359-9.
    • 63. Kim M, Kang J, Kim D, Song H, Min H, Nam Y, et al. Hi-COVIDNet: Deep learning approach to predict inbound COVID-19 patients and case study in South Korea. Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining; 2020: 3466–3473. https://doi.org/10.1145/3394486.3412864.
    • 64. Dandekar R, Barbastathis G. Quantifying the effect of quarantine control in Covid-19 infectious spread using machine learning. MedRxiv. 2020. https://doi.org/10.1101/2020.04.03.20052084.
    • 65. Kermack WO, McKendrick AG. A contribution to the mathematical theory of epidemics. Proceedings of the royal society of london Series A, Containing papers of a mathematical physical character. 1927;115(772):700–21. https://doi.org/10.1098/rspa.1927.0118.
    • 66. Moein S, Nickaeen N, Roointan A, Borhani N, Heidary Z, Javanmard SH, et al. Inefficiency of SIR models in forecasting COVID-19 epidemic: a case study of Isfahan. Scientific reports. 2021;11(1):1–9. https://doi.org/10.1038/s41598-021-84055-6.
    • 67. Law KB, Peariasamy KM, Gill BS, Singh S, Sundram BM, Rajendran K, et al. Tracking the early depleting transmission dynamics of COVID-19 with a time-varying SIR model. Scientific reports. 2020;10(1):1–11. https://doi.org/10.1038/s41598-020-78739-8.
    • 68. Cooper I, Mondal A, Antonopoulos CGJC, Solitons , Fractals . A SIR model assumption for the spread of COVID-19 in different communities. Chaos, Solitons & Fractals. 2020;139:110057. pmid:32834610
    • 69. Sun J, Chen X, Zhang Z, Lai S, Zhao B, Liu H, et al. Forecasting the long-term trend of COVID-19 epidemic using a dynamic model. Scientific Reports. 2020;10(1):1–10. https://doi.org/10.1038/s41598-020-78084-w.
    • 70. He S, Peng Y, Sun K. SEIR modeling of the COVID-19 and its dynamics. Nonlinear dynamics. 2020;101(3):1667–80. pmid:32836803
    • 71. Ellison G. Implications of heterogeneous SIR models for analyses of COVID-19. National Bureau of Economic Research Working Paper, 2020. https://doi.org/10.3386/w27373.
      • 72. Toda AA. Susceptible-infected-recovered (sir) dynamics of covid-19 and economic impact. arXiv. 2020:11221. https://doi.org/10.48550/arXiv.2003.11221.
      • 73. López L, Rodo X. A modified SEIR model to predict the COVID-19 outbreak in Spain and Italy: simulating control scenarios and multi-scale epidemics. Results in Physics. 2021;21:103746. pmid:33391984
      • 74. Mwalili S, Kimathi M, Ojiambo V, Gathungu D, Mbogo R. SEIR model for COVID-19 dynamics incorporating the environment and social distancing. BMC Research Notes. 2020;13(1):1–5. https://doi.org/10.1186/s13104-020-05192-1.
      • 75. Bera S, Khajanchi S, Roy TK. Dynamics of an HTLV-I infection model with delayed CTLs immune response. Applied Mathematics and Computation. 2022;430: 127206. https://doi.org/10.1016/j.amc.2022.127206.
      • 76. Dwivedi A, Keval R, Khajanchi S. Modeling optimal vaccination strategy for dengue epidemic model: a case study of India. Physica Scripta. 2022;97(8). https://doi.org/10.1088/1402-4896/ac807b.
      • 77. Khajanchi S, Sarkar K. Forecasting the daily and cumulative number of cases for the COVID-19 pandemic in India. Chaos. 2020;30(7):071101. pmid:32752627
      • 78. Samui P, Mondal J, Khajanchi S. A mathematical model for COVID-19 transmission dynamics with a case study of India. Chaos Solitons Fractals. 2020;140:110173. pmid:32834653
      • 79. Chowell G, Luo R. Ensemble bootstrap methodology for forecasting dynamic growth processes using differential equations: application to epidemic outbreaks. BMC medical research methodology. 2021;21(1):1–18. https://doi.org/10.1186/s12874-021-01226-9.
      • 80. Silva CJ, Cruz C, Torres DF, Munuzuri AP, Carballosa A, Area I, et al. Optimal control of the COVID-19 pandemic: controlled sanitary deconfinement in Portugal. Scientific reports. 2021;11(1):1–15. https://doi.org/10.1038/s41598-021-83075-6.
      • 81. Gourieroux C, Jasiak J. Time varying Markov process with partially observed aggregate data: An application to coronavirus. Journal of econometrics. 2020;232:35–51. pmid:33281272
      • 82. Vivanco-Lira A. Predicting COVID-19 distribution in Mexico through a discrete and time-dependent Markov chain and an SIR-like model. arXiv. 2020:06758. https://doi.org/10.48550/arXiv.2003.06758.
      • 83. Zhang Y, You C, Cai Z, Sun J, Hu W, Zhou X-H. Prediction of the COVID-19 outbreak based on a realistic stochastic model. MedRxiv. 2020. https://doi.org/10.1101/2020.03.10.20033803.
      • 84. Marfak A, Achak D, Azizi A, Nejjari C, Aboudi K, Saad E, et al. The hidden Markov chain modelling of the COVID-19 spreading using Moroccan dataset. Data in brief. 2020;32:106067. pmid:32789156
      • 85. Prasse B, Van Mieghem P. Mobile smartphone tracing can detect almost all SARS-CoV-2 infections. arXiv. 2020:14285. https://doi.org/10.48550/arXiv.2006.14285.
      • 86. Prabhu SM, Subramaniam N. Surveillance of COVID-19 Pandemic using Hidden Markov Model. arXiv. 2020:07609. https://doi.org/10.48550/arXiv.2008.07609.