Assignment 2 [10 points]

Consider the SIR model with demography with birth - and death rate 𝜇. The SIR equations,
where all symbols have their standard meaning, are
!"
= 𝜇 − 𝛽𝑆𝐼 − 𝜇𝑆;
!#
!$
= +𝛽𝑆𝐼 − 𝛾𝐼 − 𝜇𝐼;
!#
!%
= +𝛾𝐼 − 𝜇𝑅.
!#

Consider now the case of paediatric vaccination, where a fraction 𝑝 of newborns are
vaccinated and therefore protected from infection.

1 Adopt the standard SIR model from above to include this paediatric vaccination. (1
point)
!"
= 𝜇(1 − 𝑝) − 𝛽𝑆𝐼 − 𝜇𝑆,
!#
!$
!#
= +𝛽𝑆𝐼 − 𝛾𝐼 − 𝜇𝐼,
!%
!#
= +𝛾𝐼 + 𝜇𝑝 − 𝜇𝑅.

By a change of variable, 𝑆 = 𝑆 & (1 − 𝑝), 𝐼 = 𝐼 & (1 − 𝑝), 𝑅 = 𝑅& (1 − 𝑝) − 𝑝, and writing your
model from (2) in terms of 𝑆 & , 𝐼 & , 𝑅& , you find that it has exactly the same form as the
standard SIR equations, but with one important modification.

2 Proof that with this change of variables you can indeed bring the model with
paediatric vaccination in exactly the same form as the standard SIR equations (2
points).
First, substitute into the model:
!" !
(1 − 𝑝) !#
= 𝜇(1 − 𝑝) − 𝛽𝑆 & 𝐼 & (1 − 𝑝)' − 𝜇𝑆 & (1 − 𝑝),
!$ !
(1 − 𝑝) !#
= +𝛽𝑆 & 𝐼 & (1 − 𝑝)' − 𝛾𝐼 & (1 − 𝑝) − 𝜇𝐼 & (1 − 𝑝),
!% !
(1 − 𝑝) = +𝛾𝐼 & (1 − 𝑝) + 𝜇𝑝 − 𝜇(𝑅& (1 − 𝑝) − 𝑝),
!#
!% !
Second, simplify the third equation for !#
:
!" !
(1 − 𝑝) !#
= 𝜇(1 − 𝑝) − 𝛽𝑆 & 𝐼 & (1 − 𝑝)' − 𝜇𝑆 & (1 − 𝑝),
!$ !
(1 − 𝑝) !#
= +𝛽𝑆 & 𝐼 & (1 − 𝑝)' − 𝛾𝐼 & (1 − 𝑝) − 𝜇𝐼 & (1 − 𝑝),
!% !
(1 − 𝑝) !# = +𝛾𝐼 & (1 − 𝑝) − 𝜇𝑅& (1 − 𝑝),
Finally, divide all equations by (1 − 𝑝):
!" !
!#
= 𝜇 − 𝛽𝑆 & 𝐼 & (1 − 𝑝) − 𝜇𝑆 & ,
!$ !
!#
= +𝛽𝑆 & 𝐼 & (1 − 𝑝) − 𝛾𝐼 & − 𝜇𝐼 & ,
!% !
!#
= +𝛾𝐼 & − 𝜇𝑅& ,
QED

3 What is the major difference between the model from (2) and the standard SIR model?
And what does this mean for the dynamics of the model? (2 points)
The most important difference is that 𝛽 if now scale by (1 − 𝑝), so effectively
making it smaller. This means for the dynamics that 𝑅( is reduced by a factor (1 −
 𝑝), so if p is large enough, 𝑅( can be pushed below 1, and making the disease go
extinct.

4 The basic reproductive ratio for measles is 17. What fraction of the infant population
should be vaccinated to prevent measles from spreading in the population. Explain
why? (2 points)
)*
17(1 − 𝑝) < 1 , so, 𝑝 > )+ = 0,94 . This is based on assignment (4) where we
found that 𝑅( is reduced by (1 − 𝑝), so choosing 𝑝 such that 𝑅( is smaller than 1
has the desired effect, because in standard SIR with demography where 𝑅( is
smaller than 1 always results in the disease going extinct, without having some
endemic state.

Childhood diseases, for which paediatric vaccination is applied, are off course
characterised by a strong seasonal forcing effect.

5 Mention at least three possible features of the SIR dynamics in the presence of
seasonal forcing? (1point)
1) recurring yearly epidemics; 2) harmonic resonance; 3) subharmonic resonance,
periodic doublings, transitions to chaos.

Finally, consider dynamic variability in childhood disease incidence in real data. The
graph below is based on Figure 5.16 from Keeling and Rohani, Case reports for measles
in London 1944 to 1988. The black line demonstrates weekly reported cases, with the
gray line depicting the per capita birth rate. The dashed grey line demonstrates effective
birth rate, correcting for vaccination, that started in 1968.
Birth Rate
Disease IncidenceDisease Incidence

Normalized Birth Normalized

Rate

6 Describe the types of dynamics that you observe, and relate this back to what you
know about SIR models with seasonal forcing, and what you discussed w.r.t.
vaccination. (2 points)
Until 1950, we see harmonic resonance, an outbreak every year. From 1950 –
1968 there is subharmonic resonance, with an outbreak every two years. This is
probably due to the lower birth rate that pushed the dynamics through a
bifurcation leading to periodic doubling. From 1968 onwards we find more
periodic doublings and a transition to chaotic dynamics. The vaccination pushes
the effective birthrate way down, bringing the dynamics in this regime.

Assignment 3 [10 points]

1 What is stochastic extinction, and when is this most likely to happen? (2 points)
This is when in a stochatistic discrete event model by chance the amount of
infected individuals drops to zero, making the disease go extinct (even when in a
continuous model it would not do that). This most likely happens in small
 populations, diseases that undergo large amplitude oscillations (due to e.g. strong
seasonal forcing), and diseases with small 𝑅( .

In an invading scenario, where a single infected individual enters a fully susceptible

population, stochastic extinction can also occur.

2 What is meant with stochastic extinction in this scenario, and what is the probability
of this actually happening (no need to derive of proof this)? (1 points)
Chance that an invading infected individual recovers before passing the infection
)
to a secondary case. The chance for this to happen is % .
"

Next assume additional periodic forcing, which will lead to periodic outbreaks in large
enough populations.

3 What happens with the observed dynamics in smaller and smaller populations. You
may assume a scenario including immigrants. Take into account the concept of Critical
Community Size. (2 points)
In large populations we expect to see behaviour as in continuous models,
recurring infections due to the seasonal forcing. As the population size goes down,
the recurring infections are there, but with added noise, and for smaller
populations the noise start to disrupt the nice continuous pattern. If the
population size drops below the CCS, stochastic extinction can happens, so the
infection dies out, and for years nothing may happen. However, due to immigrants,
a new epidemic could be started, leading to isolated -in time- outbreaks (like in
the data for Iceland shown during lectures).

4 What is a metapopulation model in infectious disease modelling? (1 point)

A model where you subdivide the population in distinct subpopulation (cities,
counties, etc), each having independent dynamics, and some limited form of
interaction.

Consider two subpopulations with a one-way coupling, meaning population 1 is coupled

to population, but not vice-versa. Also assume that population 1 is large enough the be
described as fully deterministic.

5 Assuming that population 2 is also deterministic, formulate the SIR equations with
demography for these coupled populations. You may assume equal birth/date rates
and recovery rates in both populations (1 point)
!,# - !,$ - -
= 𝜇𝑁) − 𝛽) 𝑋) # − 𝜇𝑋) , = 𝜇𝑁' − 𝛽' 𝑋' 7 $ + 𝜌') # 9 − 𝜇𝑋' ,
!# .# !# .$ .$
!-# -# !-$ -$ -#
!#
= +𝛽) 𝑋) . − 𝛾𝑌) − 𝜇𝑌) , !#
= +𝛽' 𝑋' 7. + 𝜌') . 9 − 𝛾𝑌' − 𝜇𝑌' ,
# $ $
!/# !/$
!#
= +𝛾𝑌) − 𝜇𝑍) , !#
= +𝛾𝑌' − 𝜇𝑍' .

For these two coupled deterministic models, now assume that an infection is introduced
in population 1, and that population 2 is fully susceptible. Also assume that this infection
has a basic reproductive ratio that allows it to spread and cause an epidemic

6 Explain in words the dynamics in this coupled model. (1 point)

Population 1 will develop a standard epidemic, and then settle into the endemic
state. In population 2 the epidemic will also immediately develop, due to the
deterministic coupling. Moreover, analysis shows, as demonstrated in the lectures,
that the early dynamics in population 2 is exponential growth with a rate 𝛽' − 𝛾' .
 Since the dynamics in 2 in coupled to those of 1, it is hard to say without
simulating how the infection in 2 would develop.

Finally assume that population 2 is too small to be considered fully deterministic. Replace
the model population 2 now with a stochastic SIR model.

7 What do you observe now? Take the strength of coupling between population 1 and
2 into account in your discussion (2 points)
a. If the coupling 𝜌') is small enough, there is a high probability that the disease
will not spread in population 2
b. For larger coupling 𝜌') the pathogen can spread, but their can be a significant
delay of spreading of the disease in population 2 with respect to the epidemic
in population 1