Main
Main
Main
a r t i c l e i n f o a b s t r a c t
Article history: We propose a compartmental mathematical model for the spread of the COVID-19 disease with special
Received 18 April 2020 focus on the transmissibility of super-spreaders individuals. We compute the basic reproduction number
Accepted 22 April 2020
threshold, we study the local stability of the disease free equilibrium in terms of the basic reproduction
Available online 27 April 2020
number, and we investigate the sensitivity of the model with respect to the variation of each one of its
2010 MSC: parameters. Numerical simulations show the suitability of the proposed COVID-19 model for the outbreak
34D05 that occurred in Wuhan, China.
92D30
© 2020 Elsevier Ltd. All rights reserved.
Keywords:
Mathematical modeling of COVID-19
pandemic
Wuhan case study
Basic reproduction number
Stability
Sensitivity analysis
Numerical simulations
https://doi.org/10.1016/j.chaos.2020.109846
0960-0779/© 2020 Elsevier Ltd. All rights reserved.
2 F. Ndaïrou, I. Area and J.J. Nieto et al. / Chaos, Solitons and Fractals 135 (2020) 109846
2. The proposed COVID-19 compartment model the disease induced death rates due to infected, super-spreaders,
and hospitalized individuals, respectively. At each instant of time,
Based on a 2016 model [10], and taking into account the ex-
dF (t )
istence of super-spreaders in the family of corona virus [11], we D(t ) := δi I (t ) + δ p P (t ) + δh H (t ) = (2)
propose a new epidemiological compartment model that takes dt
into account the super-spreading phenomenon of some individu- gives the number of death due to the disease. The transmissibility
als. Moreover, we consider a fatality compartment, related to death from asymptomatic individuals has been modeled in this way since
due to the virus infection. In doing so, the constant total popula- it was not apparent their behavior. Indeed, at present, this question
tion size N is subdivided into eight epidemiological classes: sus- is a controversial issue for epidemiologists. A flowchart of model
ceptible class (S), exposed class (E), symptomatic and infectious (1) is presented in Fig. 1.
class (I), super-spreaders class (P), infectious but asymptomatic
class (A), hospitalized (H), recovery class (R), and fatality class (F). 3. Qualitative analysis of the model
The model takes the following form:
One of the most significant thresholds when studying infectious
⎧ dS
⎪
⎪ = −β NI S − l β HN S − β
P
S, disease models, which quantifies disease invasion or extinction in
⎪
⎪ dt N
a population, is the basic reproduction number [12]. In this section
⎪
⎪
⎪
⎪ dE
β NI S + l β HN S + β NP S − κ E,
we obtain the basic reproduction number for our model (1) and
⎪
⎪ =
⎪ dt
⎪
study the locally asymptotically stability of its disease free equilib-
⎪
⎪ rium (see Theorem 1).
⎪ dI
⎪ = κρ1 E − (γa + γi )I − δi I,
⎪
⎪ dt
⎪
⎪ 3.1. The basic reproduction number
⎪
⎪ dP
⎨ = κρ2 E − (γa + γi )P − δ p P,
dt (1) The basic reproduction number, as a measure for disease spread
⎪
⎪ dA
⎪
⎪ = κ (1 − ρ1 − ρ2 )E, in a population, plays an important role in the course and control
⎪
⎪ dt of an ongoing outbreak. It can be understood as the average num-
⎪
⎪
⎪
⎪
dH
= γa (I + P ) − γr H − δh H, ber of cases one infected individual generates, over the course of
⎪
⎪ dt
⎪
⎪ its infectious period, in an otherwise uninfected population. Using
⎪
⎪ dR the next generation matrix approach outlined in van den Driess-
⎪
⎪ = γi (I + P ) + γr H,
⎪
⎪ dt che and Watmough [13] to our model (1), the basic reproduction
⎪
⎪
⎩ dF number can be computed by considering the below generation ma-
= δi I + δ p P + δh H, trices F and V, that is, the Jacobian matrices associated to the rate
dt
of appearance of new infections and the net rate out of the corre-
with β quantifying the human-to-human transmission coefficient
sponding compartments, respectively,
per unit time (days) per person, β quantifies a high transmis- ⎡ ⎤
sion coefficient due to super-spreaders, and l quantifies the rela- 0 β βl β
tive transmissibility of hospitalized patients. Here κ is the rate at ⎢0 0 0 0⎥
JF = ⎣
0⎦
and
which an individual leaves the exposed class by becoming infec- 0 0 0
tious (symptomatic, super-spreaders or asymptomatic); ρ 1 is the 0 0 0 0
⎡ ⎤
proportion of progression from exposed class E to symptomatic in- κ 0 0 0
fectious class I; ρ 2 is a relative very low rate at which exposed in- ⎢−κρ1 i 0 0 ⎥
dividuals become super-spreaders while 1 − ρ1 − ρ2 is the progres- JV = ⎣ ⎦,
−κρ2 0 p 0
sion from exposed to asymptomatic class; γ a is the average rate at 0 −γa −γa h
which symptomatic and super-spreaders individuals become hos-
pitalized; γ i is the recovery rate without being hospitalized; γ r is where
the recovery rate of hospitalized patients; and δ i , δ p , and δ h are i = γa + γi + δi , p = γa + γi + δ p and h = γr + δh . (3)
F. Ndaïrou, I. Area and J.J. Nieto et al. / Chaos, Solitons and Fractals 135 (2020) 109846 3
Table 1
Values of the model parameters corresponding to the situation of Wuhan, as discussed in Section 5, for which
R0 = 0.945.
R(t ) = γi 0 I (s ) + P (s ) ds + γr 0 H (s )ds
t t
(5) + ( κ + i )h + ( h + i ) p ,
⎪
⎩ βρ1 h
F (t ) = δi 0 I (s )ds + δ p 0 P (s )ds + δh 0 H (s )ds.
t t t
a3 = κ (1 − R0 )(h p + h i + i p ) + κ p
i
Furthermore, since the total population size N is constant, one
β ρ2 h 1 1
has +κi + κ p βγa l ρ1 +
p h i
S(t ) = N − [E (t ) + I (t ) + P (t ) + A(t ) + H (t ) + R(t ) + F (t )]. (6)
1 1
+κi βγa l ρ2 + + i h p ,
Therefore, the local stability of model (1) can be studied h p
through the remaining coupled system of state variables, namely,
a4 = κi h p (1 − R0 ).
the variables E, I, P, and H in (1). The Jacobian matrix associated to
these variables of (1) is the following one:
⎡ ⎤ Moreover, we also compute, in terms of R0 , the following ex-
−κ β lβ β pression:
⎢κρ1 −i 0 0 ⎥
JM = ⎣ ⎦, (7) a1 a2 − a3 = (1 − R0 )(κ + i )κi + (1 − R0 )(κ + h + p )κ p
κρ2 0 − p 0
0 γa γa −h βρ1 βγa l ρ1
+ ( κ + p + i ) + κ p
where ϖi , ϖp , and ϖh are defined in (3). The eigenvalues of the ma- p i
β ρ2 βγa l ρ2
trix JM are the roots of the following characteristic polynomial:
+ ( κ + p + i ) + κi
Z (λ ) = λ4 + a1 λ3 + a2 λ2 + a1 λ + a4 , p p
βγa l ρ1 κ βγa l ρ2 κ
where + ( κ + h + i ) + ( κ + h + p )
h h
a1 = κ + h + i + p , + ( κ + i )h + ( h + i ) p .
a2 = −βκρ1 − β κρ2 + κh + κi + h i + κ p
+ h p + i p , From these previous expressions, it is clear that if R0 < 1, then
the conditions of the Liénard–Chipard test are satisfied and, as a
a3 = −βγa κ l ρ1 − βγa κ l ρ2 − βκρ1 h − β κρ2 h − βκρ1 p
consequence, the disease free equilibrium is stable. In the case
−β κρ2 i + κh i + κh p + κi p + h i p , when R0 > 1, we have that a4 < 0 and, by using Descartes’ rule
of signs, we conclude that at least one of the eigenvalues is posi-
a4 = −βγa κ l ρ2 i − βγa κ l ρ1 p − β κρ2 i h
tive. Therefore, the system is unstable. In conclusion, we have just
− βκρ1 h p + κh i p . proved the following result:
4 F. Ndaïrou, I. Area and J.J. Nieto et al. / Chaos, Solitons and Fractals 135 (2020) 109846
Fig. 2. Number of confirmed cases per day. The green line corresponds to the real data obtained from reports [5,20,21] while the black line (I + P + H) has been obtained
by solving numerically the system of ordinary differential Eq. (1), by using the Matlab code ode45. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
Fig. 3. Number of confirmed deaths per day. The red line corresponds to the real data obtained from reports [5,20,21] while the black line has been obtained by solving
numerically, using the Matlab code ode45, our system of ordinary differential Eq. (1) to derive D(t) given in (2). (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.)
6. Conclusions and discussion reality of the Wuhan outbreak (see Fig. 2) and predicting a di-
minishing on the daily number of confirmed cases of the disease.
Classical models consider SIR populations. Here we have taken This is in agreement with our computations of the basic reproduc-
into consideration the super-spreaders (P), hospitalized (H), and fa- tion number in Section 4 that, surprisingly, is obtained less than 1.
tality class (F), so that its derivative (see formula (2)) gives the Moreover, it is worth to mention that our model fits also enough
number of deaths (D). Our model is an ad hoc compartmental well the real data of daily confirmed deaths, as shown in Fig. 3.
model of the COVID-19, taking into account its particularities, some Our theoretical findings and numerical results adapt well to the
of them still not well-known, giving a good approximation of the real data and it reflects or reflected the reality in Wuhan, China.
6 F. Ndaïrou, I. Area and J.J. Nieto et al. / Chaos, Solitons and Fractals 135 (2020) 109846
The number of hospitalized persons is relevant to give an estimate [2] Ndaïrou F, Area I, Nieto JJ, Silva CJ, Torres DFM. Mathematical modeling of Zika
of the Intensive Care Units (ICU) needed. Some preliminary simula- disease in pregnant women and newborns with microcephaly in Brazil. Math
Methods Appl Sci 2018;41:8929–41. doi:10.1002/mma.4702.
tions indicate that this would be useful for the health authorities. [3] Rachah A, Torres DFM. Dynamics and optimal control of Ebola transmission.
Our model can also be used to study the reality of other coun- Math Comput Sci 2016;10:331–42. doi:10.1007/s11786- 016- 0268- y.
tries, whose outbreaks are currently on the rise. We claim that [4] Brauer F, Castillo-Chavez C, Feng Z. Mathematical models in epidemiology.
New York: Springer-Verlag; 2019.
some mathematical models like the one we have proposed here [5] COVID-19 Coronavirus Pandemic. 2020. https://www.worldometers.info/
will contribute to reveal some important aspects of this pandemia. coronavirus/repro, Accessed March 26.
Of course, this investigation has some limitations, being the [6] Chen T-M, Rui J, Wang Q-P, Cui J-A, Yin L. A mathematical model for simulat-
ing the phase-based transmissibility of a novel coronavirus. Infect Dis Poverty
first on the relative recent spread of the new coronavirus and
2020;9(1):24. doi:10.1186/s40249- 020- 00640- 3.
therefore the limited data accessible at the beginning of this study. [7] Maier BF, Brockmann D. Effective containment explains subexponential growth
In the future, we can develop further this prototype. Even with in recent confirmed COVID-19 cases in China. Science 2020;8. doi:10.1126/
science.abb4557.
these shortcomings, the model can be useful due to the high rel-
[8] Trilla A. One world, one health: the novel coronavirus COVID-19 epidemic. Med
evance of the topic. Finally, we suggest new directions for further Clin (Barc) 2020;154(5):175–7. doi:10.1016/j.medcle.2020.02.001.
research: [9] Wong G, Liu W, Liu Y, Zhou B, Bi Y, Gao GF. MERS, SARS, and Ebola: the role of
super-spreaders in infectious disease. Cell Host Microbe 2015;18(4):398–401.
1. the transmissibility from asymptomatic individuals; doi:10.1016/j.chom.2015.09.013.
[10] Kim Y, Lee S, Chu C, Choe S, Hong S, Shin Y. The characteristics of middle
2. compare, in the near future, our results with other models;
eastern respiratory syndrome coronavirus transmission dynamics in South Ko-
3. consider sub-populations related to age, gender, etc.; rea. Osong Public Health Res Perspect 2016;7:49–55. doi:10.1016/j.phrp.2016.
4. introduce preventive measures in this COVID-19 epidemic and 01.001.
for future viruses; [11] Alasmawi H, Aldarmaki N, Tridane A. Modeling of a super-spreading event of
the MERS-corona virus during the Hajj season using simulation of the existing
5. integrate into the model some imprecise data by using fuzzy data. Int J Stat Med Biol Res 2017;1:24–30.
differential equations; [12] van den Driessche P. Reproduction numbers of infectious disease models. In-
6. include the viral load of the infectious into the model. fect Dis Model 2017;2:288–303. doi:10.1016/j.idm.2017.06.002.
[13] van den Driessche P, Watmough J. Reproduction numbers and sub-threshold
These and other questions are under current investigation and endemic equilibria for compartmental models of disease transmission. Math
Biosci 2002;180:29–48. doi:10.1016/S0025-5564(02)00108-6.
will be addressed elsewhere. [14] Gantmacher FR. The theory of matrices, 1. Providence, RI: AMS Chelsea Pub-
lishing; 1998.
Funding [15] Liénart A, Chipart H. Sur le signe de la partie réelle des racines d’une équation
algébrique. J Math Pures Appl (6 ème série) 1914;10:291–346.
[16] Powell DR, Fair J, LeClaire RJ, Moore LM, Thompson D. Sensitivity analysis of an
This research was funded by the Portuguese Foundation for infectious disease model. In: Boston M, editor. Proceedings of the international
Science and Technology (FCT) within project UIDB/04106/2020 system dynamics conference; 2005.
[17] Chitnis N, Hyman JM, Cushing JM. Determining important parameters in the
(CIDMA). Ndaïrou is also grateful to the support of FCT through spread of malaria through the sensitivity analysis of a mathematical model.
the Ph.D. fellowship PD/BD/150273/2019. The work of Area and Ni- Bull Math Biol 2008;70:1272–96. doi:10.1007/s11538- 008- 9299- 0.
eto has been partially supported by the Agencia Estatal de Inves- [18] Rodrigues HS, Monteiro MTT, Torres DFM. Sensitivity analysis in a dengue epi-
demiological model. Conference papers in mathematics. Hindawi, editor; 2013.
tigación (AEI) of Spain, cofinanced by the European Fund for Re-
doi:10.1155/2013/721406. Vol. 2013, Art. ID 721406
gional Development (FEDER) corresponding to the 2014–2020 mul- [19] Aguilar J.B., Faust G.S.M., Westafer L.M., Gutierrez J.B.. Investigating the
tiyear financial framework, project MTM2016-75140-P. Moreover, impact of asymptomatic carriers on COVID-19 transmission. Preprint.
10.1101/2020.03.18.20037994
Nieto also thanks partial financial support by Xunta de Galicia un-
[20] de la Salud O.P.. Alerta epidemiológica nuevo coronavirus (ncov).
der grant ED431C 2019/02. 2020a. https://www.paho.org/hq/index.php?option=com_docman&view=
download&category_slug=coronavirus- alertas- epidemiologicas&alias=
Declaration of Competing Interest 51351- 16- de- enero- de- 2020- nuevo- coronavirus- ncov- alerta- epidemiologica- 1
&Itemid=270&lang=es, accessed on January 16.
[21] de la Salud O.P.. Actualización epidemiológica nuevo coronavirus (2019-ncov).
The authors declare that they have no known competing finan- 2020b. https://www.paho.org/hq/index.php?option=com_docman&view=
cial interests or personal relationships that could have appeared to download&category_slug=coronavirus- alertas- epidemiologicas&alias=
51355- 20- de- enero- de- 2020- nuevo- coronavirus- ncov-actualizacion-
influence the work reported in this paper. epidemiologica-1&Itemid=270&lang=es, accessed on January 20.
References
[1] Djordjevic J, Silva CJ, Torres DFM. A stochastic SICA epidemic model for HIV
transmission. Appl Math Lett 2018;84:168–75. doi:10.1016/j.aml.2018.05.005.