Department of Epidemiology and Public Health

Health Systems Research and

Dynamical Modelling Unit

Introduction to SEIR Models

Nakul Chitnis

Workshop on Mathematical Models of Climate Variability,

Environmental Change and Infectious Diseases
Trieste, Italy
8 May 2017
Outline

SI Model

SIS Model

The Basic Reproductive Number (R0 )

SIR Model

SEIR Model

2017-05-08 2
Mathematical Models of Infectious Diseases

Population-based models
I Can be deterministic or stochastic
I Continuous time
• Ordinary differential equations
• Partial differential equations
• Delay differential equations
• Integro-differential equations
I Discrete time
• Difference equations
Agent-based/individual-based models
I Usually stochastic
I Usually discrete time

2017-05-08 3
Mathematical Models of Infectious Diseases

Population-based models
I Can be deterministic or stochastic
I Continuous time
• Ordinary differential equations
• Partial differential equations
• Delay differential equations
• Integro-differential equations
I Discrete time
• Difference equations
Agent-based/individual-based models
I Usually stochastic
I Usually discrete time

2017-05-08 3
Outline

SI Model

SIS Model

The Basic Reproductive Number (R0 )

SIR Model

SEIR Model

2017-05-08 4
SI Model

Susceptible-Infectious Model: applicable to HIV.

rβI/N
S I

dS I
= −rβS
dt N
dI I
= rβS
dt N
S: Susceptible humans
I: Infectious humans
r: Number of contacts per unit time
β: Probability of disease transmission per contact
N: Total population size: N = S + I.

2017-05-08 5
Analyzing the SI Model

The system can be reduced to one dimension,

dI I
= rβ(N − I) ,
dt N
with solution,
I0 N
I(t) = ,
(N − I0 )e−rβt + I0
for I(0) = I0 .
Equilibrium Points:

Idfe = 0
Iee = N

2017-05-08 6
Analyzing the SI Model

The system can be reduced to one dimension,

dI I
= rβ(N − I) ,
dt N
with solution,
I0 N
I(t) = ,
(N − I0 )e−rβt + I0
for I(0) = I0 .
Equilibrium Points:

Idfe = 0
Iee = N

2017-05-08 6
Numerical Solution of SI Model

1000

800
Infectious Humans

600

400

200

0
0 5 10 15 20
Time (Years)
With r = 365/3 years−1 , β = 0.005, N = 1000, and I(0) = 1.

2017-05-08 7
Definition of Transmission Parameters

Note that in some models, usually of diseases where contacts

are not well defined, rβ (the number of contacts per unit time
multiplied by the probability of disease transmission per
contact) are combined into one parameter (often also called β
— the number of adequate contacts per unit time).
For diseases where a contact is well defined (such as sexually
transmitted diseases like HIV or vector-borne diseases like
malaria), it is usually more appropriate to separate the contact
rate, r, and the probability of transmission per contact, β.
For diseases where contacts are not well defined (such as
air-borne diseases like influenza) it is usually more appropriate
to combine the two into one parameter.

2017-05-08 8
Outline

SI Model

SIS Model

The Basic Reproductive Number (R0 )

SIR Model

SEIR Model

2017-05-08 9
SIS Model

Susceptible-Infectious-Susceptible Model: applicable to the

common cold.
γ

rβI/N
S I

dS I
= −rβS + γI
dt N
dI I
= rβS − γI
dt N
γ: Per-capita recovery rate

2017-05-08 10
Analyzing the SIS Model

The system can be reduced to one dimension,

dI I
= rβ(N − I) − γI,
dt N
with solution,
N
rβ · (rβ − γ)
I(t) =   ,
N (rβ−γ) −(rβ−γ)t
1+ rβ I0 − 1 e

for I(0) = I0 .
Equilibrium Points:

Idfe = 0
(rβ − γ)N
Iee =
rβ

2017-05-08 11
Analyzing the SIS Model

The system can be reduced to one dimension,

dI I
= rβ(N − I) − γI,
dt N
with solution,
N
rβ · (rβ − γ)
I(t) =   ,
N (rβ−γ) −(rβ−γ)t
1+ rβ I0 − 1 e

for I(0) = I0 .
Equilibrium Points:

Idfe = 0
(rβ − γ)N
Iee =
rβ

2017-05-08 11
Numerical Solution of SIS Model

1000

800
Infectious Humans

600

400

200

0
0 10 20 30 40 50
Time (Days)
With rβ = 0.5 days−1 , γ = 0.1 days−1 , N = 1000, and I(0) = 1.

2017-05-08 12
Outline

SI Model

SIS Model

The Basic Reproductive Number (R0 )

SIR Model

SEIR Model

2017-05-08 13
 of uncertainty that may be prioritized for urgent res
The Basic Reproductive Number (R0 )
Generation

0 1 2

Initial phase of epidemic (R0 = 3) Disease is ende

Pan-InfORM (2009)
2017-05-08 14
Definition of R0

The basic reproductive number, R0 , is the number of

secondary infections that one infected person would produce
in a fully susceptible population through the entire duration of
the infectious period.
R0 provides a threshold condition for the stability of the
disease-free equilibrium point (for most models):
I The disease-free equilibrium point is locally asymptotically
stable when R0 < 1: the disease dies out.
I The disease-free equilibrium point is unstable when R0 > 1:
the disease establishes itself in the population or an epidemic
occurs.
I For a given model, R0 is fixed over all time.
This definition is only valid for simple homogeneous
autonomous models.
Can define similar threshold conditions for more complicated
models that include heterogeneity and/or seasonality but the
basic definition no longer holds.
2017-05-08 15
Definition of R0

The basic reproductive number, R0 , is the number of

secondary infections that one infected person would produce
in a fully susceptible population through the entire duration of
the infectious period.
R0 provides a threshold condition for the stability of the
disease-free equilibrium point (for most models):
I The disease-free equilibrium point is locally asymptotically
stable when R0 < 1: the disease dies out.
I The disease-free equilibrium point is unstable when R0 > 1:
the disease establishes itself in the population or an epidemic
occurs.
I For a given model, R0 is fixed over all time.
This definition is only valid for simple homogeneous
autonomous models.
Can define similar threshold conditions for more complicated
models that include heterogeneity and/or seasonality but the
basic definition no longer holds.
2017-05-08 15
Definition of R0

The basic reproductive number, R0 , is the number of

secondary infections that one infected person would produce
in a fully susceptible population through the entire duration of
the infectious period.
R0 provides a threshold condition for the stability of the
disease-free equilibrium point (for most models):
I The disease-free equilibrium point is locally asymptotically
stable when R0 < 1: the disease dies out.
I The disease-free equilibrium point is unstable when R0 > 1:
the disease establishes itself in the population or an epidemic
occurs.
I For a given model, R0 is fixed over all time.
This definition is only valid for simple homogeneous
autonomous models.
Can define similar threshold conditions for more complicated
models that include heterogeneity and/or seasonality but the
basic definition no longer holds.
2017-05-08 15
Definition of R0

The basic reproductive number, R0 , is the number of

secondary infections that one infected person would produce
in a fully susceptible population through the entire duration of
the infectious period.
R0 provides a threshold condition for the stability of the
disease-free equilibrium point (for most models):
I The disease-free equilibrium point is locally asymptotically
stable when R0 < 1: the disease dies out.
I The disease-free equilibrium point is unstable when R0 > 1:
the disease establishes itself in the population or an epidemic
occurs.
I For a given model, R0 is fixed over all time.
This definition is only valid for simple homogeneous
autonomous models.
Can define similar threshold conditions for more complicated
models that include heterogeneity and/or seasonality but the
basic definition no longer holds.
2017-05-08 15
Evaluating R0

R0 can be expressed as a product of three quantities:

  
Number of Probability of  
Duration of
R0 =  contacts   transmission 
infection
per unit time per contact

For SIS model:

1
R0 = r × β ×
γ

2017-05-08 16
Reproductive Numbers

The (effective) reproductive number, Re , is the number of

secondary infections that one infected person would produce
through the entire duration of the infectious period.
Typically, but not always, Re is the product of R0 and the
proportion of the population that is susceptible.
Re describes whether the infectious population increases or
not. It increases when Re > 1; decreases when Re < 1 and is
constant when Re = 1. When Re = 1, the disease is at
equilibrium.
Re can change over time.
The control reproductive number, Rc , is the number of
secondary infections that one infected person would produce
through the entire duration of the infectious period, in the
presence of control interventions.

2017-05-08 17
Reproductive Numbers

The (effective) reproductive number, Re , is the number of

secondary infections that one infected person would produce
through the entire duration of the infectious period.
Typically, but not always, Re is the product of R0 and the
proportion of the population that is susceptible.
Re describes whether the infectious population increases or
not. It increases when Re > 1; decreases when Re < 1 and is
constant when Re = 1. When Re = 1, the disease is at
equilibrium.
Re can change over time.
The control reproductive number, Rc , is the number of
secondary infections that one infected person would produce
through the entire duration of the infectious period, in the
presence of control interventions.

2017-05-08 17
Reproductive Numbers

The (effective) reproductive number, Re , is the number of

secondary infections that one infected person would produce
through the entire duration of the infectious period.
Typically, but not always, Re is the product of R0 and the
proportion of the population that is susceptible.
Re describes whether the infectious population increases or
not. It increases when Re > 1; decreases when Re < 1 and is
constant when Re = 1. When Re = 1, the disease is at
equilibrium.
Re can change over time.
The control reproductive number, Rc , is the number of
secondary infections that one infected person would produce
through the entire duration of the infectious period, in the
presence of control interventions.

2017-05-08 17
Reproductive Numbers

The (effective) reproductive number, Re , is the number of

secondary infections that one infected person would produce
through the entire duration of the infectious period.
Typically, but not always, Re is the product of R0 and the
proportion of the population that is susceptible.
Re describes whether the infectious population increases or
not. It increases when Re > 1; decreases when Re < 1 and is
constant when Re = 1. When Re = 1, the disease is at
equilibrium.
Re can change over time.
The control reproductive number, Rc , is the number of
secondary infections that one infected person would produce
through the entire duration of the infectious period, in the
presence of control interventions.

2017-05-08 17
Reproductive Numbers

The (effective) reproductive number, Re , is the number of

secondary infections that one infected person would produce
through the entire duration of the infectious period.
Typically, but not always, Re is the product of R0 and the
proportion of the population that is susceptible.
Re describes whether the infectious population increases or
not. It increases when Re > 1; decreases when Re < 1 and is
constant when Re = 1. When Re = 1, the disease is at
equilibrium.
Re can change over time.
The control reproductive number, Rc , is the number of
secondary infections that one infected person would produce
through the entire duration of the infectious period, in the
presence of control interventions.

2017-05-08 17
Evaluating Re

  
Number of Probability of  
Duration of
Re (t) =  contacts   transmission 
infection
per unit time per contact
 
Proportion of
×  susceptible 
population

For SIS model:

S(t)
Re (t) = R0 ×
N (t)
rβS(t)
= .
γN (t)

2017-05-08 18
The Basic Reproductive Number (R0 )

[Link]

2017-05-08 19
Outline

SI Model

SIS Model

The Basic Reproductive Number (R0 )

SIR Model

SEIR Model

2017-05-08 20
SIR Model

Susceptible-Infectious-Recovered Model: applicable to measles,

mumps, rubella.

rβI/N γ
S I R

dS I
= −rβS
dt N
dI I
= rβS − γI
dt N
dR
= γI
dt
R: Recovered humans
with N = S + I + R.

2017-05-08 21
Analyzing the SIR Model

Can reduce to two dimensions by ignoring the equation for R

and using R = N − S − I.
Can no longer analytically solve these equations.
Infinite number of equilibrium points with I ∗ = 0.
Perform phase portrait analysis.
Estimate final epidemic size.

2017-05-08 22
R0 for the SIR Model

  
Number of Probability of  
Duration of
R0 =  contacts   transmission 
infection
per unit time per contact

1
R0 = r × β ×
γ
rβ
=
γ
If R0 < 1, introduced cases do not lead to an epidemic (the
number of infectious individuals decreases towards 0).
If R0 > 1, introduced cases can lead to an epidemic
(temporary increase in the number of infectious individuals).

rβ S(t)
Re (t) =
γ N
2017-05-08 23
Phase Portrait of SIR Model
THE MATHEMATICS OF INFECTIOUS DISEASES 605

0.8 σ=3
infective fraction, i

0.6

0.4

0.2

0 ↑
0 0.2 1 0.4
smax = σ 0.6 0.8 1
susceptible fraction, s
Hethcote (2000)

2017-05-08 Fig. 2 Phase plane portrait for the classic SIR epidemic model with contact number σ = 3. 24
Numerical Solution of SIR Model

1000

800
Susceptible
Infectious
Humans

600 Recovered

400

200

0
0 20 40 60 80 100
Time (Days)
With rβ = 0.3 days−1 , γ = 0.1 days−1 , N = 1000, and
S(0) = 999, I(0) = 1 and R(0) = 0.

2017-05-08 25
Numerical Solution of SIR Model

1000

800
Susceptible
Infectious
Humans

600 Recovered

400

200

0
0 20 40 60 80 100
Time (Days)
With rβ = 0.3 days−1 , γ = 0.1 days−1 , N = 1000, and
S(0) = 580, I(0) = 20 and R(0) = 400.

2017-05-08 26
Human Demography

Need to include human demographics for diseases where the

time frame of the disease dynamics is comparable to that of
human demographics.
There are many different ways of modeling human
demographics
I Constant immigration rate
I Constant per-capita birth and death rates
I Density-dependent death rate
I Disease-induced death rate.

2017-05-08 27
Endemic SIR Model

Λ rβI/N γ
Birth S I R

µ µ µ

Death Death Death

dS I
= Λ − rβS − µS
dt N
dI I
= rβS − γI − µI
dt N
dR
= γI − µR
dt
N =S+I +R

Λ: Constant recruitment rate

µ: Per-capita removal rate
2017-05-08 28
Analyzing the Endemic SIR Model

Can no longer reduce the dimension or solve analytically.

There are two equilibrium points: disease-free and endemic

Λ Λ(γ + µ)
Sdfe = See =
µ rβµ
Λ(rβ − (γ + µ))
Idfe = 0 Iee =
rβ(γ + µ)
γΛ(rβ − (γ + µ))
Rdfe = 0 Ree =
rβµ(γ + µ)

Can perform stability analysis of these equilibrium points and

draw phase portraits.

2017-05-08 29
R0 for the Endemic SIR Model

  
Number of Probability of  
Duration of
R0 =  contacts   transmission 
infection
per unit time per contact

1
R0 = r × β ×
γ+µ
rβ
=
γ+µ

If R0 < 1, the disease-free equilibrium point is globally

asymptotically stable and there is no endemic equilibrium
point (the disease dies out).
If R0 > 1, the disease-free equilibrium point is unstable and a
globally asymptotically stable endemic equilibrium point exists.
2017-05-08 30
Numerical Solution of Endemic SIR Model

1000

800
Susceptible
Infectious
Humans

600 Recovered

400

200

0
0 20 40 60 80 100
Time (Days)
With rβ = 0.3 days−1 , γ = 0.1 days−1 , µ = 1/60 years−1 ,
Λ = 1000/60 years−1 , and S(0) = 999, I(0) = 1 and R(0) = 0.

2017-05-08 31
Numerical Solution of Endemic SIR Model

1000
Susceptible
Infectious
Recovered
800
Humans

600

400

200

0
0 20 40 60 80 100 120
Time (Years)
With rβ = 0.3 days−1 , γ = 0.1 days−1 , µ = 1/60 years−1 ,
Λ = 1000/60 years−1 , and S(0) = 999, I(0) = 1 and R(0) = 0.

2017-05-08 32
Outline

SI Model

SIS Model

The Basic Reproductive Number (R0 )

SIR Model

SEIR Model

2017-05-08 33
SEIR Model

Susceptible-Exposed-Infectious-Recovered Model: applicable to

measles, mumps, rubella.

Λ rβI/N ε γ
Birth S E I R

µ µ µ µ

Death Death Death Death

E: Exposed (latent) humans

ε: Per-capita rate of progression to infectious state

2017-05-08 34
SEIR Model

dS I
= Λ − rβS − µS
dt N
dE I
= rβS − εE
dt N
dI
= εE − γI − µI
dt
dR
= γI − µR
dt

with
N = S + E + I + R.

2017-05-08 35
R0 for the Endemic SEIR Model
  
Number of Probability of  
Duration of
R0 =  contacts   transmission 
infection
per unit time per contact
 
Probabililty of
× surviving 
exposed stage

1 ε
R0 = r × β × ×
γ+µ ε+µ
rβε
=
(γ + µ)(ε + µ)
If R0 < 1, the disease-free equilibrium point is globally
asymptotically stable and there is no endemic equilibrium
point (the disease dies out).
If R0 > 1, the disease-free equilibrium point is unstable and a
globally asymptotically stable endemic equilibrium point exists.
2017-05-08 36
Extensions to Compartmental Models

Basic compartmental models assume a homogeneous

population.
Divide the population into different groups based on infection
status:
M : Humans with maternal immunity
S: Susceptible humans
E: Exposed (infected but not yet infectious) humans
I: Infectious humans
R: Recovered humans.
Can include time-dependent parameters to include the effects
of seasonality.
Can include additional compartments to model vaccinated and
asymptomatic individuals, and different stages of disease
progression.
Can include multiple groups to model heterogeneity, age,
spatial structure or host species.
2017-05-08 37
References

O. Diekmann, H. Heesterbeek, and T. Britton, Mathematical Tools for

Understanding Infectious Disease Dynamics.
Princeton Series in Theoretical and Computational Biology. Princeton University
Press, Princeton, (2013).
H. W. Hethcote, “The mathematics of infectious diseases”, SIAM Review 42,
599–653 (2000).
M. J. Keeling and P. Rohani, Modeling Infectious Diseases in Humans and
Animals.
Princeton University Press, Princeton, (2007).

2017-05-08 38