Modelling Infectious Diseases 1

Modelling Infectious Diseases

Mathematical Models of Infectious Diseases

Prof. Dr. Niel Hens

Modelling Infectious Diseases 2
Prologue
Introduction

Introduction

There are countless of infectious agents that can infect human,

animal and plant hosts.

They can be transmitted directly between hosts via respiratory air

droplets or bodily fluids.

They can also be transmitted indirectly through an intermediary

source, for instance via mosquitoes, ticks, rodents, environmental
particles or contaminated blood products.

Infectious agents evolve and transform while new agents emerge

regularly, implying their supply can be considered infinite.
Modelling Infectious Diseases 3
Prologue
Introduction

Introduction

Microscopically small infectious agents with relatively short life

spans, which replicate within their hosts are called microparasites
such as viruses, bacteria and fungi.

Larger infectious agents with relatively longer life spans called

macroparasites such as parasitic worms.

Many infectious agents live inside or on the surface of their hosts’

bodies without causing illness or even discomfort. In fact, hosts even
depend for their survival on infectious agents (e.g., bacteria in the
human gut).

Infectious agents causing disease in their hosts are often referred to

as pathogens.

Infectious diseases are caused by pathogens, which are transmissible

between hosts, either directly or indirectly.
Modelling Infectious Diseases 4
Prologue
Introduction

Introduction

An infectious disease is often characterized by its

transmission route

transmission potential

natural history
incubation time

latent or pre-infectious period

infectious period

acquired immunity

symptomatic/asymptomatic
Modelling Infectious Diseases 5
Prologue
Natural History

Natural history

The Infectious Disease Process:

Keeling and Rohani (2008)

Modelling Infectious Diseases 6
Prologue
Transmission Routes

Transmission routes

Close contacts transmission (Varicella, Influenza, Mumps,

Tuberculosis, Measles, ...)

Sexual transmission (Hepatitis B, HIV, HPV, ...)

Parenteral transmission (Hepatitis B and C, HIV, ...)

Orofecal transmission (Hepatitis A, Cholera, Rotavirus, ...)

Vertical transmission (Hepatitis B and C, HIV, Syphilis, ...)

Vectorborne transmission (Malaria, Dengue, Leishmaniasis, ...)

...
Modelling Infectious Diseases 7
Prologue
Historic Pandemics

Historic pandemics

Plague of Justinian, from 541 to 750, killed between 50% and 60%
of Europe’s population
The Black Death of 1347 to 1352 killed 25 million in Europe over 5
years (estimated to be between 25 and 50% of the populations of
Europe, Asia, and Africa - the world population at the time was 500
million)
The introduction of smallpox, measles, and typhus to the areas of
Central and South America by European explorers during the 15th
and 16th centuries caused pandemics among the native inhabitants
Between 1518 and 1568 disease pandemics are said to have caused
the population of Mexico to fall from 20 million to 3 million
Modelling Infectious Diseases 8
Prologue
Historic Pandemics

Historic pandemics

The first European influenza epidemic occurred between 1556 and

1560, with an estimated mortality rate of 20%

Smallpox killed an estimated 60 million Europeans during the 18th

century

In the 19th century, tuberculosis killed an estimated one-quarter of

the adult population of Europe; by 1918 one in six deaths in France
were still caused by TB.

The Influenza Pandemic of 1918 (or the Spanish Flu) killed 25-50
million people

UNAIDS estimates that in 2013, 35 million people live with HIV and
1.5 million people died due to AIDS worldwide
Mathematical Models of Infectious Diseases 9

Part I

Mathematical Models of Infectious

Diseases
Mathematical Models of Infectious Diseases 10

Outline
2 Introduction

3 Compartmental Models
Discrete time models
A mathematical intermezzo
Kermack and McKendrick’s Model
The SIR model with vital dynamics
The SIR model with vital dynamics and vaccination

Building your own model

Transmission within multiple subpopulations
Who Acquires Infection From Whom?

4 Epilogue
Mathematical Models of Infectious Diseases 11
Introduction

Mathematical modelling of infectious diseases

Purposes:
prediction: requires the inclusion of known complexities and
population-level heterogeneity

understanding: investigating the factors that drive dynamics

Building a model presents a trade-off:

accuracy: reproduce what is observed and predict future dynamics

transparency: ability to understand how model components influence

the dynamics and interact

flexibility: ease of adapting the model to new situations

Mathematical Models of Infectious Diseases 12
Introduction

Mathematical modelling of infectious diseases

Limitations:
models present a simplification of reality

chance events of infectious disease transmission hinder perfect

prediction

A good model:
suited to its purpose: simple as possible, but no simpler

balance accuracy, transparency, flexibility

parametrisable from available data

Mathematical Models of Infectious Diseases 13
Introduction

Mathematical modelling of infectious diseases

Daniel Bernoulli was the first to present a mathematical model for

smallpox in 1760

Since then many people have developed models to describe

infectious disease dynamics, see e.g. Bailey (1975); Anderson and
May (1991); Grenfell and Dobson (1995); Daley and Gani (1999);
Hethcote (2000)

see e.g. historical perspectives by Klaus Dietz.

Several textbooks are available; amongst which Vynnycky and White

(2010) provides an excellent introduction to mathematical modelling
Mathematical Models of Infectious Diseases 14
Introduction

Contact between individuals

Predicting the number of infections at time t + 1 based on the

circumstance at time t
The force of infection λ
the per capita rate at which a susceptible individual contracts
infection

it is assumed proportional to the number of infectious persons at

time t and depending on how the contact structure is assumed to
change with population size N it is given by:

λt = βIt
λt = βIt /Nt
Mathematical Models of Infectious Diseases 15
Introduction

Contact between individuals

The number of new infections at time t + 1 is given by λt St and

thus:
It+1 = βSt It
It+1 = βSt It /Nt
This is referred to as the mass action principle
density-dependent transmission: It+1 = βSt It :
- as the population size increases, so does the contact rate

- mostly applicable to plant and animal diseases (homogeneity)

frequency-dependent transmission: It+1 = βSt It /Nt :

- the contact rate is assumed constant regardless of a change in
population size

- mostly applicable to human and vectorborne diseases (heterogeneity)

Mathematical Models of Infectious Diseases 16
Introduction

Contact between individuals

Note that when in a constant population, both density- and

frequency-dependent are equivalent

It+1 = βSt It /Nt = βSt It /N = β̃St It

In what follows I will use It+1 = βSt It without loss of generality

unless specified differently.
Mathematical Models of Infectious Diseases 17
Introduction

Contact between individuals

Heterogeneity

airborne infections: age - example: children at school have more

contacts with children of the same age

sexually transmitted infections: age and sexual behavior

temporal heterogeneity
seasonality,

week vs weekend,

holiday vs non-holiday, . . .

...
Mathematical Models of Infectious Diseases 18
Introduction

Contact between individuals, plants, herds, . . .

Interdisciplinary Perspectives on Infectious Diseases 3

(a) (b) (c)

(d) (e) (f)

Figure 1: Examples of networks used in epidemiology. (a) Contacts between 22 intravenous drug users, as recorded in [3]; squares refer to
primary contacts. Given that the identity of contacts is known, they can be interlinked. (b) Caricature of a snowball sampling algorithm,
squares are primary contacts, diamonds are secondary, and circles are tertiary contacts. Given that the identity of contacts is known they can
be linked. (c) Example of a configuration model network. Each individual has a prescribed degree distribution, which gives rise to “half-
Mathematical Models of Infectious Diseases 19
Introduction

Modeling frameworks
compartmental models
the population is subdivided into broad subgroups (compartments)

individuals are tracked collectively

roughly either deterministic or stochastic (probabilistic)

deterministic models describe what happens ‘on average’ in a
population

stochastic models allow the number of individuals who move between

compartments to vary through chance

transmission dynamic or static models

metapopulation models

network models

microsimulation or agent-based models

Mathematical Models of Infectious Diseases 20
Compartmental Models

Rules of engagement

Entity Symbol 1 Symbol 2

maternal immune M (number) mpi (proportion)
susceptibles S (number) s (proportion)
exposed/latent* E (number) e (proportion)
infected I (number) i (proportion)
recovered R (number) r (proportion)
total population size N (number) -
age-stratified pop. size N(a) (number) -
life expectancy L (dec. number) -
survivor function - m (proportion)
recovery rate ν (rate) -
force of infection λ (rate) -
transmission parameter β (rate) -
basic reproduction number R0 (dec. number) -
*depending on the context e refers to exp(1) = 2.71828 . . .

Table: Glossary of the most important symbols in this section.

Mathematical Models of Infectious Diseases 21
Compartmental Models

The basic SIR compartmental model

State variables
S: susceptible individuals
I : infectious individuals (able to transmit the infectious agent)
R: removed/recovered ind. (not transmitting the infectious agent)
N = S + I + R: total and constant population size

Parameters
β: transmission rate
ν: recovery rate

Force of infection
follows the law of mass action: λt = βIt

→ lifelong immunity after infection

Mathematical Models of Infectious Diseases 22
Compartmental Models

Model formulation
SIR model:

'$ '$ '$

λ - ν -
S I R
&% &% &%

Extensions
maternal immunity/passive immunity: MSIR

exposed or latent infection (pre-infectious): SEIR

waning of immunity: SIRS

susceptible upon recovery: SIS

infected for life: SI

Mathematical Models of Infectious Diseases 23
Compartmental Models
Discrete time models

The discrete time deterministic SIR model

SIR compartmental model

Equations:

St+1 = St − λt St
It+1 = It + λt St − νIt
Rt+1 = Rt + νIt

with Nt+1 = St+1 + It+1 + Rt+1 = St + It + Rt = Nt .

λt = βIt and ν are risks

risks are related to rates as follows:

risk = 1 − e −rate

if the ‘rate’ is small:

risk ≈ rate
Mathematical Models of Infectious Diseases 24
Compartmental Models
Discrete time models

The discrete time stochastic SIR model

The number of newly infected cases arises from a stochastic process

with mean βIt St

The number of newly recovered individuals arises from a stochastic

process with mean νt It
What distribution would you assume for both these quantities?
Mathematical Models of Infectious Diseases 24
Compartmental Models
Discrete time models

The discrete time stochastic SIR model

The number of newly infected cases arises from a stochastic process

with mean βIt St

The number of newly recovered individuals arises from a stochastic

process with mean νt It
What distribution would you assume for both these quantities?
a Poisson distribution

a binomial distribution

What are the (dis)advantages for both distributions?

Mathematical Models of Infectious Diseases 25
Compartmental Models
Discrete time models

The discrete time deterministic SIR model

When do you expect the number of new infections to decrease?

Focus on the second equation:

It+1 = It + βIt St − νIt
clearly It+1 = It if βSt = ν
The epidemic will
decrease if St < ν/β

increase if St > ν/β

ν/(βN) is called the relative removal rate

Nβ/ν = NβD is called the basic reproduction number R0

Here D denotes the average infectious period

Consequently β = R0 /(ND)
Mathematical Models of Infectious Diseases 26
Compartmental Models
Discrete time models

R-code for the discrete stochastic SIR model

# R website : http://cran.r-project.org
# Initial parameters
N = 1E6 # Total population
I = 1 # Number of Infectious at time 0
S = N-1 # Number of Susceptibles at time 0
R = 0 # Number of Recovered at time 0

# Transmission parameters
R0 = 2.48 # Basic Reproduction number
nu = 1/6 # Recovery rate (in days)
b = nu*R0/N # Infection rate (in days)

# Initial states
Sold = S # Number of Susceptibles at time t=0
Iold = I # Number of Infectious at time t=0
Rold = R # Number of Recovered at time t=0
Mathematical Models of Infectious Diseases 27
Compartmental Models
Discrete time models

R-code discrete stochastic SIR model: option 1: Poisson

# Placeholders
Svec =Sold; Ivec = Iold; Rvec = Rold
stop = FALSE
# Loop - continue until stop = TRUE
while (!stop){
Ih = rpois(1,b*Iold*Sold); print(Ih)
Rh = rpois(1,nu*Iold)
if ((Iold+Ih-Rh)<=0) break
Sold = Sold-Ih
if (Sold<=0){Rold=N; break}
Iold = Iold+Ih-Rh
Rold = Rold+Rh
Svec = c(Svec,Sold)
Ivec = c(Ivec,Iold)
Rvec = c(Rvec,Rold)
if (Iold==0){stop=T}
}
Mathematical Models of Infectious Diseases 28
Compartmental Models
Discrete time models

R-code discrete stochastic SIR model: option 2: Binomial

# Placeholders
Svec =Sold; Ivec = Iold; Rvec = Rold
stop = FALSE
# Loop - continue until stop=TRUE
while (!stop){
Ih = rbinom(1,Sold,(1-exp(-b*Iold)))
print(Ih)
Rh = rbinom(1,Iold,(1-exp(-nu)))
Sold = Sold-Ih
Iold = Iold+Ih-Rh
Rold = Rold+Rh
Svec = c(Svec,Sold)
Ivec = c(Ivec,Iold)
Rvec = c(Rvec,Rold)
if (Iold==0){stop=T}
}
Mathematical Models of Infectious Diseases 29
Compartmental Models
Discrete time models

Class exercise

Run the discrete SIR model 100,000 times

Keep track of the final size

Study the distribution of final sizes. What do you observe?

How does this distribution change with increasing R0 ?

What happens if R0 decreases below 1?

Mathematical Models of Infectious Diseases 30
Compartmental Models
Discrete time models

The basic reproduction number

Recall R0 = NβD, the basic reproduction number

What does R0 represent?

Consider the total number of new infections in the population

between time t and t + 1:
βIt St

At the start of an epidemic, say t = 0: I0 = 1 and S0 = N and thus

the total number of new infections between t = 0 and t = 1 equals

βN

By the end of the infectious period of duration D time units, the

infectious person would have infected βND individuals

Therefore R0 is the number of secondary cases caused by a single

infective introduced into a wholly susceptible population of size N
during the infective’s infectious period.
Mathematical Models of Infectious Diseases 31
Compartmental Models
Discrete time models

The basic reproduction number

R0 constitutes a threshold:
if R0 > 1 then the epidemic can grow

if R0 ≤ 1 then the epidemic will die out

Using R0 , one defines the number of effective contacts by each

person per unit time by
ce = R0 /D.

Therefore β = ce /N is the “per capita number of effective contacts

made by a given individual per unit time”, or equivalently “the per
capita rate at which two specific individuals come into effective
contact per unit time”

For a given pathogen, it is difficult to define an effective contact

More on this later on

Mathematical Models of Infectious Diseases 32
Compartmental Models
Discrete time models

Discrete time models: an alternative notation

Other notation:

St+δt = St − λt St δt
It+δt = It + λt St δt − νIt δt
Rt+δt = Rt + νIt δt

with δt symbolizing the size of a small time step.

The time step in a discrete time model could be very important.

Why?

Next:

→ consider continuous time models

→ focus on deterministic continuous time models

Mathematical Models of Infectious Diseases 33
Compartmental Models
Discrete time models

From discrete to continuous time models

Reordering the equation:

St+δt − St
= −λt St
δt
It+δt − It
= λt St − νIt
δt
Rt+δt − Rt
= νIt
δt
Mathematical Models of Infectious Diseases 33
Compartmental Models
Discrete time models

From discrete to continuous time models

Limit for δt → 0:
St+δt − St
lim = −λt St
δt→0 δt
It+δt − It
lim = λt St − νIt
δt→0 δt
Rt+δt − Rt
lim = νIt
δt→0 δt
Mathematical Models of Infectious Diseases 34
Compartmental Models
Discrete time models

From discrete to continuous time models

Using derivatives:
dSt
= −λt St
dt
dIt
= λt St − νIt
dt
dRt
= νIt
dt
Mathematical Models of Infectious Diseases 35
Compartmental Models
A mathematical intermezzo

Population dynamics: Malthus (1798)

Evolution is determined by the balance between births and deaths:

dx
= bx − mx = (b − m)x ≡ rx.
dt

For r = 0, steady population:

x(t) = x(0).

For r > 0, exponential growth:

x(t) = x(0) e |r |t .

For r < 0, exponential decay:

x(t) = x(0) e −|r |t .

Mathematical Models of Infectious Diseases 36
Compartmental Models
A mathematical intermezzo

Population dynamics: Verhulst (1836)

Some limiting factor must prevent the population to grow indefinitely:
dx  x
= rx 1 − .
dt K

All solutions scale to one:

t → r −1 t, x → Kx.

Analytical solution:

x(0) e t
x(t) = .
1 + x(0)(e t − 1)

Horizontal asymptote:

lim x(t) = 1.
t→+∞
Mathematical Models of Infectious Diseases 37
Compartmental Models
A mathematical intermezzo

Linear stability analysis

Steady states are points where evolution stops:

dx
= F (x) = x (1 − x) = 0 −→ x = 0, 1.
dt

Small deviation from 0 increases:

x(t) = δx(t) → δx(t) ∝ e t ,

the steady state is unstable.

Small deviation from 1 decreases:

x(t) = 1 − δx(t) → δx(t) ∝ e −t ,

the steady state is stable.

Mathematical Models of Infectious Diseases 38
Compartmental Models
Kermack and McKendrick’s Model

A Contribution to the Mathematical Theory of Epidemics

more than a contribution, a basis...

”Reference may here be made to the work of Ross and Hudson (1915-17)
in which the same problem is attacked. The problem is here carried to a
further stage, and it is considered from a point of view which is in one
sense more general.[...]”

Kermack and McKendrick (1927)

Mathematical Models of Infectious Diseases 39
Compartmental Models
Kermack and McKendrick’s Model

Kermack-McKendrick’s model

The SIR model without vital dynamics

(and without a general infectious period distribution)

State variables
S: Susceptible individuals
I : Infectious individuals (able to transmit the infectious agent)
R: Removed/Recovered ind. (not transmitting the infectious agent)
N = S + I + R: Total (and constant) population

Parameters
β: transmission rate
ν: recovery rate

Force of infection (frequency dependent mass action)

follows the law of mass action: λ(t) = β I (t)
N
Mathematical Models of Infectious Diseases 40
Compartmental Models
Kermack and McKendrick’s Model

The SIR model

'$ '$ '$
λ - ν -
S I R
&% &% &%

dS(t) I
= −βS ,
dt N
dI (t) I
= βS − νI ,
dt N
dR(t)
= νI ,
dt
with N = S + I + R.
Note that the time dependence is not always made explicit in the
notation; here in the right-hand side (RHS) of the equation.
Mathematical Models of Infectious Diseases 41
Compartmental Models
Kermack and McKendrick’s Model

Analytic approximation of the incidence

Let s, i, r denote the proportion of population in the susceptible,

infectious and removed classes, respectively:
S I R
s= , i= , r= ,
N N N
β
τ = νt, σ = ν.

The equations for the SIR model can be written as

ds(τ )
= −σsi,
dτ
di(τ )
= σsi − i,
dτ
dr (τ )
= i,
dτ
with s + i + r = 1.
Mathematical Models of Infectious Diseases 42
Compartmental Models
Kermack and McKendrick’s Model

Analogy with ecology

For very slow recovery, σ >> 1, the system reduces to

ds
= −σsi,
dτ
di
= σsi,
dτ
with s + i = 1.

It is equivalent to two logistic equations

ds
= −σs(1 − s),
dτ
di
= σ(1 − i)i.
dτ
Mathematical Models of Infectious Diseases 43
Compartmental Models
Kermack and McKendrick’s Model

Analytic approximation of the incidence

Substituting i = 1 − s − r in 1st and 3rd equation of system our SIR

model:
dr
= 1−s −r
dτ
ds
= −σs
dr
with initial values s(0) = s0 and r (0) = 0, one obtains

dr
= 1 − s0 e −σr − r .
dτ
Assuming σr  1 and developing the exponential term
dr 1
≈ 1 − s0 + (s0 σ − 1)(r ) + s0 σ 2 r 2 := F (r ).
dτ 2
Mathematical Models of Infectious Diseases 44
Compartmental Models
Kermack and McKendrick’s Model

Analytic approximation of the incidence

Analytic solution
√
1  √ −q 
r (t) = σs 0 − 1 + −q tanh( t − ϕ)
σ 2 s0 2
where √ p
−q = (σs0 − 1)2 + 2s0 i0 σ 2
and  σs − 1 
0
ϕ = tanh−1 √
−q
Mathematical Models of Infectious Diseases 45
Compartmental Models
Kermack and McKendrick’s Model

Analytic approximation of the incidence

Finally, since
dr
i(t) =
dt
1 √  √−q 
−2
i(t) = −q cosh t − ϕ
2s0 σ 2 2
Application: Plague-related deaths in Bombay (1905-1906)
Mathematical Models of Infectious Diseases 46
Compartmental Models
Kermack and McKendrick’s Model

Kermack and McKendrick: theory of epidemics

i(t) = 890 cosh−2 (0.2t − 3.4)

Mathematical Models of Infectious Diseases 47
Compartmental Models
Kermack and McKendrick’s Model

Kermack and McKendrick: the threshold phenomenon

Let’s focus now on the second equation:

di(t)
= βsi − νi,
dt
= (βs − ν)i.

If the initial fraction of the susceptible population is less than ν/β

then we have di(t)dt < 0 and the epidemic dies out:

ν/β is called the relative removal rate

The inverse of the relative removal rate is called the basic

reproduction number
A pathogen will invade if and only if R0 > 1.
Mathematical Models of Infectious Diseases 48
Compartmental Models
Kermack and McKendrick’s Model

Kermack and McKendrick: phase-plot

β = 0.2856, ν = 1/7, R0 = 2

0.5
s0 = 0.99
s0 = 0.9
s0 = 0.8
s0 = 0.7

0.4
s0 = 0.6
s0 = 0.5
Fraction of infectious

0.3
0.2
0.1
0.0

0.2 0.4 0.6 0.8 1.0

Fraction of susceptible

Application: Vaccination can be used to reduce the proportion of

susceptible population below 1/R0 for disease eradication.
This matter will be discussed later.
Mathematical Models of Infectious Diseases 49
Compartmental Models
Kermack and McKendrick’s Model

Kermack and McKendrick: the threshold phenomenon

Having defined R0 ,

assuming an entirely susceptible population s0 ≈ 1, the threshold

phenomenon can be re-expressed as:

A pathogen will invade if and only if R0 > 1.

Note that in a stochastic model this is not true: ‘can invade’

Mathematical Models of Infectious Diseases 50
Compartmental Models
Kermack and McKendrick’s Model

Kermack and McKendrick: final size and R0

Total fraction of infected population according to R0

1

0.9

0.8

0.7
Fraction of infected

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
R0
Mathematical Models of Infectious Diseases 51
Compartmental Models
Kermack and McKendrick’s Model

Some first conclusions

Assumption:

the time course of the epidemic is sufficiently fast not to be

influenced by population demography.

Conclusion:

extinction of the pathogen

Question:
How to explore the long-term persistence
and endemic dynamics of an infectious disease?

Inclusion of demographic processes

Mathematical Models of Infectious Diseases 52
Compartmental Models
Kermack and McKendrick’s Model

SIR model with demography: model formulation

'$ '$ '$
λ - ν -
- S I R
&% &% &%
µ µ µ
? ? ?

dS(t) I
= B − βS − µS,
dt N
dI (t) I
= βS − νI − µI ,
dt N
dR(t)
= νI − µR.
dt
Mathematical Models of Infectious Diseases 53
Compartmental Models
Kermack and McKendrick’s Model

SIR model with demography: model formulation

Working with constant population:

S + I + R = N ⇒ B = µN

Let s, i, r denote the proportion of population in the susceptible,

infectious and removed classes, respectively.

ds(t)
= µ − βsi − µs,
dt
di(t)
= βsi − νi − µi,
dt
dr (t)
= νi − µr .
dt
Mathematical Models of Infectious Diseases 54
Compartmental Models
Kermack and McKendrick’s Model

Equilibrium values
Setting the right-hand side of the system to 0:
µ − βsi − µs = 0,
βsi − νi − µi = 0,
νi − µr = 0.
Existence of two equilibria:
The trivial equilibrium (disease free equilibrium):
E0 = (1, 0, 0)

The endemic equilibrium:

E1 = (s1 , i1 , r1 )
µ + ν µ  β  
= , − 1 , 1 − s1 − i1 .
β β µ+ν
β
E1 is positive ⇔ > 1.
µ+ν
Mathematical Models of Infectious Diseases 55
Compartmental Models
Kermack and McKendrick’s Model

Threshold value

Focus on the second equation of the system:

di(t)
= βsi − νi − µi
dt
β = 5, µ = 1/75, ν = 2 ⇒ R0 = 2.49

We have

0.20
di(t)
0.15
> 0 ⇔ βs −(ν+µ) > 0
dt I

0.10

di(t) ν+µ
0.05

>0⇔s>
dt β
0.00

0.2 0.4 0.6 0.8 1.0

S
Mathematical Models of Infectious Diseases 56
Compartmental Models
Kermack and McKendrick’s Model

Threshold value: the basic reproduction number

Intuitively:
β: the transmission rate
the mean number of new cases produced by an infectious individual
per time unit.

An infected individual remains infectious for 1/(µ + ν) time unit

β
During that time period, a typical infectious individual produces µ+ν
infections in a totally susceptible population. Thus,
β
R0 = .
µ+ν
Mathematical Models of Infectious Diseases 57
Compartmental Models
Kermack and McKendrick’s Model

Threshold value: the basic reproduction number

If R0 > 1, an epidemic occurs and the solution of the system

approaches the endemic equilibrium:
 1  
µ 1 
E1 = , 1− , 1 − s1 − i1 .
R0 µ + ν R0
Mathematical Models of Infectious Diseases 58
Compartmental Models
Kermack and McKendrick’s Model

R code for solving dynamical systems

R package desolve (Soetaert et al, 2010)

Solving Initial value differential equations in R

Calling the library

library(deSolve)

Function ode in R, general call:

ode(y=vector containing state variables,
times=vector containing the time units for integration,
func=function containing the model equations,
parms=vector containing the model parameters)
Mathematical Models of Infectious Diseases 59
Compartmental Models
Kermack and McKendrick’s Model

The SIR model in R

Definition of the ODE system:

Derivatives of state variables are expressed as a list

SIR<-function(t,state,parameters)
{
with(as.list(c(state, parameters)),
{
N=S+I+R
dS = mu*N-beta*I*S/N - mu*S
dI = beta*I*S/N - nu*I - mu*I
dR = nu*I -mu*R
list(c(dS, dI, dR))
})
}
Mathematical Models of Infectious Diseases 60
Compartmental Models
Kermack and McKendrick’s Model

The SIR model in R

Parameter definition/initialisation
N=50000
R0=2; mu0=1/75; nu0=1/(1/24)
beta0=R0*(mu0+nu0)
parameters = c(mu=mu0,beta=beta0,nu=nu0)

Initial values
state = c(S=N-1,I=1,R=0)

Time range:
from T0 = 0 to Tmax = 1000 by step of 0.01
times<-seq(0,1000,by=0.01)

Solution of Initial value differential equations

out<-as.data.frame(
ode(y=state,times=times,func=SIR,parms=parameters)
)
Mathematical Models of Infectious Diseases 61
Compartmental Models
Kermack and McKendrick’s Model

Summary
library(deSolve)
N=50000;R0=2; mu0=1/75; nu0=1/(1/24)
beta0=R0*(mu0+nu0)
parameters = c(mu=mu0,beta=beta0,nu=nu0)
state = c(S=N-1,I=1,R=0)
times = seq(0,1000,by=0.01)
SIR<-function(t,state,parameters)
{
with(as.list(c(state, parameters)),
{
N=S+I+R
dS <- mu*N-beta*I*S/N - mu*S
dI <- beta*I*S/N - nu*I - mu*I
dR <- nu*I -mu*R
list(c(dS, dI, dR))
})
}
require(deSolve)
out<-as.data.frame(ode(y=state,times=times,func=SIR,parms=parameters))
Mathematical Models of Infectious Diseases 62
Compartmental Models
Kermack and McKendrick’s Model

Number of susceptible individuals

tail(out)
time S I R
99996 999.95 25166.63 164.4616 24668.91
99997 999.96 25166.63 164.4615 24668.91
99998 999.97 25166.63 164.4615 24668.91
99999 999.98 25166.63 164.4615 24668.91
100000 999.99 25166.63 164.4615 24668.91
100001 1000.00 25166.63 164.4615 24668.91

Thus, for R0 = 2, the proportion of susceptible individuals at

equilibrium is around 50%.
Mathematical Models of Infectious Diseases 63
Compartmental Models
Kermack and McKendrick’s Model

Number of infected individuals

50000
v=1/(1/24) (2 weeks)
40000
30000
susceptible

20000
10000
0

0 200 400 600 800 1000

time
Mathematical Models of Infectious Diseases 64
Compartmental Models
Kermack and McKendrick’s Model

The basic reproduction number and vaccination programs

Assumptions
SIR model with demography

a proportion p of each cohort of new-born is vaccinated immediately

at birth

Model representation
Mathematical Models of Infectious Diseases 65
Compartmental Models
Kermack and McKendrick’s Model

Model formulation
'$ '$ '$
µ(1 − p)N λ - ν - µpN
- S I R
&% &% &%
µ µ µ
? ? ?

dS(t) I
= µ(1 − p)N − βS − µS,
dt N
dI (t) I
= βS − νI − µI ,
dt N
dR(t)
= µpN + νI − µR.
dt
Mathematical Models of Infectious Diseases 66
Compartmental Models
Kermack and McKendrick’s Model

Model formulation

Working with constant population:

S + I + R = N ⇒ B = µN.

Let s, i, r denote the proportion of population in the susceptible,

infectious and removed classes, respectively.

ds(t)
= µ(1 − p) − βsi − µs,
dt
di(t)
= βsi − νi − µi,
dt
dr (t)
= µp + νi − µr .
dt
Mathematical Models of Infectious Diseases 67
Compartmental Models
Kermack and McKendrick’s Model

Equilibrium Values
Setting the right-hand side of the system to 0:

µ(1 − p) − βsi − µs = 0,
βsi − νi − µi = 0,
µp + νi − µr = 0.
Existence of two equilibria
The trivial equilibrium (disease free equilibrium):

E0 = (1 − p, 0, p)

The endemic equilibrium:

E1 = (s1 , i1 , r1 )
 
µ+ν µ
= , ((1 − p)R0 − 1) , 1 − s1 − i1 .
β β

E1 is positive ⇔ (1 − p)R0 > 1

Mathematical Models of Infectious Diseases 68
Compartmental Models
Kermack and McKendrick’s Model

Threshold value
Vaccination efficacy:
The efficacy of vaccination programs is governed by the relationship
between vaccination coverage and the basic reproduction number.

The proportion of vaccinated new-born population should attain the

critical value
1
pc = 1 −
R0
for disease eradication. This quantity is also know as the herd
immunity threshold

Herd immunity refers to the indirect protection experienced by

unvaccinated individuals resulting from immune individuals in the
population

This leads us to the definition of the effective reproduction number

Re = p s R0 ,
where ps is the proportion susceptible to infection
Mathematical Models of Infectious Diseases 69
Compartmental Models
Kermack and McKendrick’s Model

Threshold value

−3
x 10

7
proportion of infectious population

0
20
15 1
0.8
10 0.6
5 0.4
Basic reproduction ratio (R0) 0.2
0 Proportion of vaccinated new−born (p)
0

Proportion of infected population regarding different values of R0 and the

proportion of vaccinated population.
Mathematical Models of Infectious Diseases 70
Compartmental Models
Kermack and McKendrick’s Model

SIR model with vaccination in R

N=50000
R0=2; mu0=1/75; nu0=1/(1/24); p0=0.4
beta0=R0*(mu0+nu0)
parameters = c(mu=mu0,beta=beta0,nu=nu0,p=p0)
state = c(S=N-1,I=1,R=0)
times = seq(0,1000,by=0.01)
SIRv=function(t,state,parameters)
{
with(as.list(c(state, parameters)),
{
N=S+I+R
dS = mu*(1-p)*N-beta*I*S/N - mu*S
dI = beta*I*S/N - nu*I - mu*I
dR = nu*I -mu*R+mu*p*N
list(c(dS, dI, dR))
})
}
out = as.data.frame(ode(y=state,times=times,func=SIRv,parms=parameters))
Mathematical Models of Infectious Diseases 71
Compartmental Models
Kermack and McKendrick’s Model

Results

12000
1e+05

10000
8e+04

8000
Number susceptible

Infected individuals

6000
6e+04

4000
4e+04

2000
2e+04

0 200 400 600 800 2e+04 6e+04 1e+05

Time Susceptible ind

Mathematical Models of Infectious Diseases 72
Compartmental Models
Building your own model

Building your own model

Consider a constant population and an infectious agent with the

following characteristics:

New-born individuals are passively immune (maternal antibodies) for

time period 1/γ after which they are susceptible to infection

Force of infection: mass action principle

Newly infected individuals enter into a latent phase (duration 1/κ):

these individuals are infected but not infectious; they were Exposed
to the infectious agent.

Life-long immunity

What does the flow diagram look like?

Mathematical Models of Infectious Diseases 73
Compartmental Models
Building your own model

Building your own model: an MSEIR example

State: Maternal Susceptible Latent Infected Immune
Time Scale: (Months) (Years) (Days) (Days) (Life long)

µ  γ  λ  κ  ν 
Birth - M

- S - E - I
  
- R

µ µ µ µ µ
? ? ? ? ?
Figure: Illustration of the MSEIR model. The individuals enter to the
population protected by maternal antibodies after loosing those they move into
the susceptible class, after infection to the exposed class (infected but not
infectious), after a latent period to the infectious class and after recovering to
the immune class.

ODE system?
Mathematical Models of Infectious Diseases 74
Compartmental Models
Building your own model

MSEIR example: ODE System

dM(t)
= µN − (γ + µ)M
dt
dS(t)
= γM − (λ + µ)S
dt
dE (t)
= λS − (µ + κ)E
dt
dI (t)
= κE − (µ + ν)I
dt
dR(t)
= νI − µR.
dt
where λ = β NI .
In this case the basic reproductive number is given by:
κβ
R0 = .
(κ + µ)(ν + µ)
Mathematical Models of Infectious Diseases 75
Compartmental Models
Building your own model

MSEIR example

Basic reproduction number: R0 =2

Duration of maternal immunity: 3 months

Duration of the latent period: 2 days

Duration of the infectious period: 1 week

Mathematical Models of Infectious Diseases 75
Compartmental Models
Building your own model

MSEIR example

Basic reproduction number: R0 =2

Duration of maternal immunity: 3 months

Duration of the latent period: 2 days

Duration of the infectious period: 1 week

Parameters:
R0 (κ+µ)(ν+µ)
β= κ

γ = 1/(3/12); µ = 1/75; ν = 52; κ = 1/(2/365)

Work with numbers:

Implementation in R?
Mathematical Models of Infectious Diseases 76
Compartmental Models
Building your own model

Summary
N=1E5
mu0=1/75; sigma0=1/(3/12);kappa0=1/(2/365);nu0=1/(7/365);
beta0=R0*(kappa0+mu0)*(nu0+mu0)/kappa0
parameters = c(mu=mu0, sigma=sigma0,kappa=kappa0,nu=nu0,beta=beta0)
state = c(M=1, S=N-2,E=0, I=1,R=0)
times=seq(0,1000,by=0.01)

MSEIR=function(t,state,parameters)
{
with(as.list(c(state, parameters)),
{
N=M+S+E+I+R
dM= mu*N-(mu+sigma)*M
dS = sigma*M-beta*I*S/N - mu*S
dE = beta*I*S/N-(kappa+mu)*E
dI = kappa*E - nu*I - mu*I
dR = nu*I -mu*R
list(c(dM, dS, dE, dI, dR))
})
}
Mathematical Models of Infectious Diseases 77
Compartmental Models
Building your own model

Summary

out=as.data.frame(ode(y=state,times=times,func=MSEIR,parms=parameters))

par(mfrow=c(1,2))
#### Time series plot
plot(range(times),c(0,N),xlab="time (years)",
ylab="Fractions",type="n",ylim=c(0,N))
lines(out$time,out$M,lwd=2,col=4)
lines(out$time,out$S,lwd=2,col=1)
lines(out$time,out$E,lwd=2,col=5)
lines(out$time,out$I,lwd=2,col=2)
lines(out$time,out$R,lwd=2,col=3)
legend(mean(times),N,c("Maternal","Susceptible","Exposed",
"Infected","Recovered"),text.col=c(4,1,5,2:3))
##### Phase plan
plot(out$S,out$I ,type="l",main=" ", xlab="Susceptible ", ylab="Infectio
Mathematical Models of Infectious Diseases 78
Compartmental Models
Building your own model

The MSEIR model results

12000
1e+05 Maternal
Susceptible

10000
Exposed
8e+04

Infected
Recovered

8000
6e+04

Infectious
Fractions

6000
4e+04

4000
2e+04

2000
0e+00

0 200 400 600 800 2e+04 6e+04 1e+05

time (years) Susceptible

Mathematical Models of Infectious Diseases 79
Compartmental Models
Building your own model

Class exercise

Implement vaccinating children in the MSEIR model

Code the model using proportions rather than numbers

Code an extension of the model in which recovered individuals

become susceptible again: MSEIRS

What would happen if one switches to a resolution of days for the

time scale?
Mathematical Models of Infectious Diseases 80
Compartmental Models
Transmission within multiple subpopulations

Transmission within multiple subpopulations

The law of mass action as stated before relies on the hypothesis that
infected and susceptible individuals mix homogeneously.
β is thus considered as age and time independent.

Objective Taking into account the social structures in the

population.
Assumptions
The population is constructed from multiple subpopulations.

the transmission process is governed by the mixing patterns in the

population
Mathematical Models of Infectious Diseases 81
Compartmental Models
Who Acquires Infection From Whom?

Who Acquires Infection From Whom?

The transmission process is governed by the mixing pattern commonly
represented by the WAIFW ’Who Acquires Infection From Whom’ matrix.

A first example

'$ '$
βab -
βaa A B βbb
within group &% βba &%

mixing rate
between groups
mixing rate

Figure: Mixing patterns between two subpopulations: βaa is the mixing rate
within group A, βbb is the mixing rate within group B. βab and βba are the
mixing rates between the two groups.
Mathematical Models of Infectious Diseases 82
Compartmental Models
Who Acquires Infection From Whom?

The WAIFW matrix:

 
βaa βab
C= .
βba βbb
Mathematical Models of Infectious Diseases 83
Compartmental Models
Who Acquires Infection From Whom?

Application 1: The SIS transmission model for Gonorrhea

Gonorrhea is a disease transmitted through the population by

heterosexual contacts.

Immunity to reinfection does not exist.

The two subpopulations, males and females, interact according to

the following mixing matrix
 
0 βfm
C= .
βmf 0
Mathematical Models of Infectious Diseases 84
Compartmental Models
Who Acquires Infection From Whom?

Application 1: The SIS transmission model for Gonorrhea

The SIS transmission model for Gonorrhea (Capasso, 2008):

dSf
= −βfm Sf (t)Im (t) + νf If (t)
dt
dIf
= βfm Sf (t)Im (t) − νf If (t)
dt
dSm
= −βmf Sm (t)If (t) + νm Im (t)
dt
dIm
= βmf Sm (t)If (t) − νm Im (t).
dt
Assuming Nf = Sf + If and Nm = Sm + Im
dIf
= βfm (Nf − If (t))Im (t) − νf If (t),
dt
dIm
= βmf (Nm − Im (t))If (t) − νm Im (t).
dt
Mathematical Models of Infectious Diseases 85
Compartmental Models
Who Acquires Infection From Whom?

SIS model for Gonorrhea: endemic equilibrium

System:

βfm (Nf − If (t))Im (t) − vf If (t) = 0,

βmf (Nm − Im (t))If (t) − vm Im (t) = 0.

After some calculations, and noting the relative removal rates as

ρf = νf /βf , ρm = νm /βm

Nf Nm − ρf ρm N f N m − ρf ρm
If (∞) = and Im (∞) = .
ρf + N + N m ρm + N + N f
This equilibrium is positive ⇔ Nf Nm − ρf ρm > 0 or
 
N f βf N m βm
× = R0,f × R0,m > 1.
vf vm
Mathematical Models of Infectious Diseases 86
Compartmental Models
Who Acquires Infection From Whom?

SIS Model for Gonorrhea in R

parameters <- c(beta1=0.000003,beta2=0.000006,v1=0.007,v2=0.05,N1=10000,N2=15000)

state <- c(IF=1,IM=0)
Gonorrhea<-function(t,state,parameters){
with(as.list(c(state, parameters)),{
dIF <- beta1*(N1-IF)*IM-v1*IF
dIM <- beta2*(N2-IM)*IF-v2*IM
list(c(dIF,dIM))})
}
times<-seq(0,3000,by=0.1)

beta2.i<-0.000006*c(0.25,0.5,1)
beta2.i
y1mat<-y2mat<-matrix(0,length(times),length(beta2.i))
for(i in 1:length(beta2.i)){
parameters <- c(beta1=0.000003,beta2=beta2.i[i],v1=0.007,v2=0.05,N1=10000,N2=15000)
require(deSolve)
out <- as.data.frame(ode(y=state,times=times,func=Gonorrhea,parms=parameters))
y1mat[,i]<-out$IF
y2mat[,i]<-out$IM}
Mathematical Models of Infectious Diseases 87
Compartmental Models
Who Acquires Infection From Whom?

SIS Model for Gonorrhea: results

female male

8000

8000
6000 R0_m=1.80

6000
I_female

R0_m=0.90

I_male
4000

4000
2000

2000
R0_m=0.45
0

0 500 1500 2500 0 500 1500 2500

times times
Mathematical Models of Infectious Diseases 88
Compartmental Models
Who Acquires Infection From Whom?

Application 2: SIR - 2 interacting subpopulations

Model structure: Conversely to gonorrhea model, the out-diagonal
elements of the WAIFW matrix are positive
 
β11 β12
C= .
β21 β22
representing contacts within and across subpopulations.

µ1  λ1  ν1 
Birth - S

- I

- R

µ µ µ
@1 @1 @1
R
@ R
@ R
@

µ2  λ2  ν2 
Birth - S
µ
- I

- R

2 µ2 µ2
@ @ @
R
@ R
@ R
@
Figure: Illustration of a SIR model with two subpopulations.
Mathematical Models of Infectious Diseases 89
Compartmental Models
Who Acquires Infection From Whom?

Application 2: SIR - 2 interacting subpopulations

Force of infection
Σ2j=1 βij Ij Si = λi Si .

ODE system

dSi (t)
= − Σ2j=1 βij Ij Si + Ni µi − µi Si


dt
dIi (t)
Σ2j=1 βij Ij Si − (µi + νi )Ii


=
dt
dRi (t)
= νi Ii − µi Ri .
dt
We will now consider the symmetric case setting β12 = β21 = α.
Mathematical Models of Infectious Diseases 90
Compartmental Models
Who Acquires Infection From Whom?

Application 2: implementation in R
alpha<-0
parameters <- c(beta11=0.05,beta12=alpha,beta21=alpha,beta22=0.05,
nu1=1/30,nu2=1/30,mu=0.001)
state <- c(S1=0.8,I1=0.2,R1=0,S2=0.8,I2=0.2,R2=0)
times<-seq(0,10000,by=0.01)

SIRtwo<-function(t,state,parameters)
{
with(as.list(c(state, parameters)),
{
dS1 <- -(beta11*I1+beta12*I2)*S1+mu-mu*S1
dI1 <- (beta11*I1+beta12*I2)*S1-nu1*I1-mu*I1
dR1 <- nu1*I1 - mu*R1
dS2 <- -(beta21*I1+beta22*I2)*S2+mu-mu*S2
dI2 <- (beta21*I1+beta22*I2)*S2-nu2*I2-mu*I2
dR2 <- nu2*I2 - mu*R2
list(c(dS1,dI1,dR1,dS2,dI2,dR2))
})
}
Mathematical Models of Infectious Diseases 91
Compartmental Models
Who Acquires Infection From Whom?

Application 2: implementation in R

times<-seq(0,10000,by=0.01)
require(deSolve)
out <- as.data.frame(ode(y=state,times=times,func=SIRtwo,
parms=parameters))

par(mfrow=c(1,1))
plot(out$S1,log(out$I1),type="l",main=" ",
xlab="S",ylab="log(I)",ylim=c(-10,0),xlim=c(0,1))
legend(0,-7.5,c("alpha=0"),lty=c(1:4))
Mathematical Models of Infectious Diseases 92
Compartmental Models
Who Acquires Infection From Whom?

Application 2: results

0
−2
−4
log(I)

−6

alpha=0
−8

alpha=0.025
alpha=0.05
alpha=0.075
−10

0.0 0.2 0.4 0.6 0.8 1.0

S
Mathematical Models of Infectious Diseases 93
Compartmental Models
Who Acquires Infection From Whom?

Transmission over age and time in the SIR model

The model with interacting subpopulations can be seen as an

age-time dependent SIR model or age-structured SIR model

The population is divided into a finite number of age groups

interacting with each other
i Ni )   
P
i (µ- λ1 ν1
Birth S1 - I1 - R1
  
µ1 µ1 µ1
η1 @
@
R η1 R
@
@ η1 @@
R

 λ2  ν2 
? ? ?
S2 - I2 - R2
 µ2 µ µ
2 2
@ @ @
R
@ R
@ R
@
Figure: Illustration of age structured SIR model with two age groups.
Mathematical Models of Infectious Diseases 94
Compartmental Models
Who Acquires Infection From Whom?

Transmission over age and time in the SIR model

A general approach: partial differential equations

Each differential equation represent the change (over time and age)
in the compartment.

∂S(a,t) ∂S(a,t)
 ∂a + ∂t = −(λ(a, t) + µ(a))S(a, t),





∂I (a,t)
∂a + ∂I ∂t
(a,t)
= λ(a, t)S(a, t) − (ν + α + µ(a))I (a, t),




 ∂R(a,t) + ∂R(a,t) = νI (a, t) − µ(a)R(a, t),

∂a ∂t

where N(a, t) = S(a, t) + I (a, t) + R(a, t) and S(0, T ) = B(t), the

number of births all susceptible to infection.

Quite complicated to solve

Mathematical Models of Infectious Diseases 95
Compartmental Models
Who Acquires Infection From Whom?

The Cohort Age Structured model

For a population with K age groups, the system of ordinary

differential equations for the first age group (i = 1) in an
age-structured SIR model is given by

dS1 (t) X
− ΣK


= j=1 β1j Ij S1 + (µi Ni ) − µ1 S1 − η1 S1 ,
dt
i
dI1 (t)
ΣK


= j=1 β1j Ij S1 − ν1 I1 − η1 I1 ,
dt
dR1 (t)
= ν1 I1 − µ1 R1 − η1 R1 .
dt
Mathematical Models of Infectious Diseases 96
Compartmental Models
Who Acquires Infection From Whom?

The Cohort Age Structured model

Denoting ηi is the rate at which individuals of age-class i pass from

Si , Ii , Ri to Si+1 , Ii+1 , Ri+1 , we have for i = 2 to K

dSi (t)
= − ΣK


j=1 βij Ij Si + ηi−1 Si−1 − µi Si − ηi Si ,
dt
dIi (t)
ΣK


= j=1 βij Ij Si + ηi−1 Ii−1 − (νi + µi )Ii − ηi Ii ,
dt
dRi (t)
= νi Ii − µi Ri + ηi−1 Ri−1 − ηi Ri .
dt
This model is a cohort age-structured model (CAS-model) with the
disadvantage that people can instantaneously grow older.
Mathematical Models of Infectious Diseases 97
Compartmental Models
Who Acquires Infection From Whom?

The Realistic Age Structured model

A better alternative to the CAS-model is the RAS-model: Realistic

Age-structured Model

In the RAS-model the SIR model is applied to each age-class of one

year and after one year people instantaneously move to the next
age-class

The model resembles the educational system were children move to

another class after one year

The RAS-model consists of the following two-step iteration:

Assuming one-year age-groups, let {Si (t), Ii (t), Ri (t)} denote the
number of susceptible, infected and recovered individuals of age
i = 0, . . . K − 1 at time t (in years).
Mathematical Models of Infectious Diseases 98
Compartmental Models
Who Acquires Infection From Whom?

The Realistic Age Structured model

Step 1: Given initial values

{Si (t), Ii (t), Ri (t)} = {Si (t0 ), Ii (t0 ), Ri (t0 )}, i = 0, . . . K − 1 we solve the
following set of ODEs:

dSi (t)

 dt = −(λi (t) + µi )Si (t),
dIi (t)
dt = λi (t)Si (t) − (ν + µi )Ii (t),
 dRi (t) = νI (t) − µ R (t),

dt i i i

to obtain {Si (t + 1), Ii (t + 1), Ri (t + 1)}, i = 0, . . . K − 1 after one year.

Mathematical Models of Infectious Diseases 99
Compartmental Models
Who Acquires Infection From Whom?

The Realistic Age Structured model

Step 2: Individuals are then shifted by one year:

{Si (t + 1), Ii (t + 1), Ri (t + 1)} →

{Si+1 (t + 1), Ii+1 (t + 1), Ri+1 (t + 1)}, i = 0, . . . , K − 2

all newborns B are assumed susceptible to infection:

{S0 (t + 1), I0 (t + 1), R0 (t + 1)} = {B, 0, 0}.

This process is iterated throughout the time period of interest.

Mathematical Models of Infectious Diseases 100
Compartmental Models
Who Acquires Infection From Whom?

Two examples (PDE and RAS)

Goeyvaerts et al. (2015); Kovac et al. (2018)

Mathematical Models of Infectious Diseases 101
Epilogue

Summary

We have introduced the basic building blocks of mathematical

models of infectious diseases

Models are a simplification of reality and model assumptions are

crucial in the development of mathematical models
However, several questions still remain:
How do we inform models using empirical data?

How can models be compared?

...
Mathematical Models of Infectious Diseases 102
Epilogue

Contact details

Niel Hens: [email protected]

Acknowledgements:

Mathieu Andraud

Olivier Lejeune

More in:

Hens, N., Shkedy, Z., Aerts, M., Faes, C., Van Damme, P. and Beutels, P.
“Modeling Infectious Disease Parameters Based on Serological and Social
Contact Data: A Modern Statistical Perspective”
Springer-Verlag New York. 2012
Mathematical Models of Infectious Diseases 103
Epilogue

Selected Book references

Becker (1989)

Anderson and May (1991)

Capasso (2008)

Keeling and Rohani (2008)

Vynnycky and White (2010)

Halloran et al. (2010)

Diekmann et al. (2012)

...
Mathematical Models of Infectious Diseases 104
Epilogue

Anderson, R. and May, R. (1991). Infectious Diseases of Humans: Dynamics and

Control. Oxford University Press, Oxford.
Bailey, N. (1975). The Mathematical Theory of Infectious Diseases and its
Applications. Charles Griffin and Company, London.
Becker, N. G. (1989). Analysis of infectious disease data. London, Chapman and Hall.
Capasso, V. (2008). Mathematical Structures of Epidemic Systems. Springer.
Daley, D. and Gani, J. (1999). Epidemic Modelling: An Introduction. Cambridge
University Press.
Diekmann, O., Heesterbeek, H., and Britton, T. (2012). Mathematical Tools for
Understanding Infectious Disease Dynamics. Princeton University Press.
Goeyvaerts, N., Leuridan, E., Faes, C., Van Damme, P., and Hens, N. (2015).
Multi-disease analysis of maternal antibody decay using non-linear mixed models
accounting for censoring. Statistics in Medicine, 34(20):2858–2871.
Grenfell, B. and Dobson, A. (1995). Ecology of Infectious Disease in Natural
Populations. Cambridge, UK: Cambridge University Press.
Halloran, E., Longini, I., and Struchiner, C. (2010). Design and Analysis of Vaccine
Studies. Springer-verlag.
Hethcote, H. (2000). The mathematics of infectious diseases. SIAM REview,
42(4):599–653.
Keeling, M. and Rohani, P. (2008). Modeling Infectious Diseases in Humans and
Animals. Princeton University Press.
Mathematical Models of Infectious Diseases 105
Epilogue

Kermack, W. O. and McKendrick, A. G. (1927). A contribution to the mathematical

theory of epidemics. Proceedings of the Royal Society of London, Series A,
115:700–721.
Kovac, T., Haber, T., Reeth, F. V., and Hens, N. (2018). Heterogeneous computing
for epidemiological model fitting and simulation. BMC Bioinformatics, 19(1):101.
Malthus, T. (1798). An Essay on the Principle of Population. Johnson, J.
Verhulst, P. F. (1836). Notice sur la loi que la population suit dans son accroissement.
Corr. Mat. et Phys., 10:113–121.
Vynnycky, E. and White, R. (2010). An Introduction to Infectious Disease Modelling.
Oxford University Press.