MATH0011 Lecture 8

Contents

MATH0011
Numbers and Patterns in
Nature and Life

Lecture 8

I f ti
Infectious
disease
di
modeling
d li I

Examples
E
l off IInfectious
f ti
Diseases
Di
Basic Epidemic Modeling
SIR Model
Modeling on the SARS outbreak in Amoy
Gardens

http://147.8.101.93/MATH0011/
2

Infectious disease modeling

Infectious disease modeling

Epidemics:

Black Death

Europe lost 1/3 of population in

1347 - 1350.
Great Plague of London, 166466.
Killed more than 75,000 of total
population of 460,000.

Epidemics:

rat flea

AIDS
AIDS virus

Bovine
spongiform
encephalopathy
(mad cow
disease)

Influenza

Killed 25 million in 1918-19 in

Europe.

SARS

Chicken flu

Coronavirus

influenza virus
3

MATH0011 Lecture 8

Infection
occurs

Infectious disease modeling

Individuals
disease
state : Susceptible

Mathematical models can help us to predict

the rate of spread of the disease
whether we shall have an outbreak
the effects of different disease control
strategies (e.g., should everyone be
required to be immunized?)

Epoch :

Latent

tA

First
transmission
to another
susceptible

Latency to
infectious
transition

Immune /
Removed

Infectious

tB

tC

Incubation period
Serial interval
Symptoms
appear

tD

tE

time
Individual no
longer
infectious to
susceptibles
(recovery or
removal)

Note : tD is constrained to lie in the interval (tB , tE), so

tD > tC (as shown) and tD < tC are both possible.
5

Basic Assumptions of the simplest

epidemic model, the SIR model

The total population is divided into three classes:

Susceptible class

Infective class

Removed class

Population size is large and constant

No birth, death, immigration or emigration.

No latency

Only 3 stages: susceptible, infective, removed.

Homogeneous
g
mixing
g

Each pair of individuals has equal probability of

coming into contact with one another (this is
reasonable for a school or households in a building).
7

MATH0011 Lecture 8

Basic Assumptions of the SIR model

Basic Assumptions of the SIR model

Divide the total population N into three groups:
St + It + Rt = N
I) Susceptible
S
tibl class,
l
St = S(t) = no. off susceptibles
tibl

The members of the population progress

through the three classes in the following
order
order.

those who may catch the disease but currently are not
infected.

II) Infective class, It = I(t) = no. of infectives

those who are infected with the disease and currently
contagious.

III) Removed class, Rt = R(t) = no. of removals

Susceptible

those who cannot get the disease or pass the disease to

others, because they either have recovered permanently,
are naturally immune, are isolated, or have died.

Immune/
Removed
Rt

Infectious
It

St

SIR model
10

Basic Assumptions of the SIR model

Basic Assumptions of the SIR model

Disease spreads when a susceptible

individual comes in contact with an
infected individual and subsequently
becomes infected.

Susceptible
St

Assuming homogeneous mixing, the mass

action principle says that the number of
encounters between susceptibles and
infectives is given by the product St It.

St It

Infectious
It

However, only a proportion of the contacts

between susceptibles and infectives result in
infection. Hence, in the next time interval,

St +1 = St St It .
11

12

MATH0011 Lecture 8

Basic Assumptions of the SIR model

Basic Assumptions of the SIR model

During one time step, the infective class grows

by the addition of the newly infected. At the
same time, some infectives recover or die, and
so progress to the removed stage of the
disease.
The removal rate measures the proportion of
the infective class that ceases to be infective,
and thus moves into the removed class,, in one
step.
It
Recovered / immune

Susceptible

St It

St

Infectious

It

Recovered /
immune

It

Rt

Therefore, we have

I t +1 = I t + St I t It ,
Rt +1 = Rt + I t

Infectious
It

Clearly the removed class increases in size by

exactly the same amount that infected class
d
decreases.

Rt
13

14

Exercise 1 Find St, It and Rt when t =2.

If we let S = St+1 St and similarly for I and

R, then the dynamics of the functions S, I
governed by
y the following
g equations.
q
and R are g

Exercise 2 Use Excel to compute St, It and Rt for t

from 0 to 80 and plot the graph of St, It and Rt .

S = St +1 St = St It ,
I = It +1 It = St It It ,
R = Rt +1 Rt = It .
For example, if I0 = 1, S0=1000, R0=0, =0.001
and = 0.1, then S1=1000 0.001(1000)(1)=999
and I1= 1+ 0.001(1000)(1) 0.1(1) = 1.9 ~ 2.
15

16

MATH0011 Lecture 8

Threadshold Values and Critical Parameters

Lets find out when I is zero.

I = S t I t I t = ( S

We will say an epidemic occurs if I > 0 for some time t

(i.e., if at some time t, the number of infectives grows).
If I < 0 for all times, then the size of the infective class
does not increase and no wider outbreak of the
disease takes place.
Therefore, it is important to know when

I = S t It It
= ( S t ) I t
is positive, zero, or negative.

) It

I = 0 once It = 0 (as the population is disease

free).
Now assume It > 0, then we have

If S t >
, then I > 0 .

If S t =
, then I = 0 .

If S t <
, then I < 0 .

17

18

P0 =

Note that St is a non-increasing function in t.

Therefore, if S 0 < , then for all t, St < .

Thus, if S0 iis b
Th
below
l
the
h value
l
, then
h II < 0 ffor

all times, and the disease decreases in the

population.

However, when S0 > , the number of infectives

will grow and an epidemic occurs

occurs. In other

words, we have an outbreak if and only if

P0 =

S0 > 1

19

S0

The above expression is called the basic

reproduction number of the infection.
infection

P
=
S
=
(
S ) ( ) from a more
Lets consider 0 0
0
biological viewpoint.
The term S0 I0 measures the number of
individuals that become infected at the outset of an
p
epidemic.
If we divide by I0 , we obtain a per-infective
measurement: S0 is the number of individuals
who become infected by contact with a single ill
individual during the initial time step.
20

MATH0011 Lecture 8

Actually, if we introduce one infective into a wholly

susceptible population S0 , this ill individual may
eventually infect many more than S0 others, because an
infective may remain contagious for many time steps
steps.
For example, if an infective remains contagious with
certain disease for about 7 days. Then, using a time step
of 1 day, this infective would infect about
(S0 ) (7) susceptibles over the course of a week.
Moreover, if the period of contagion lasts 7 days, then
each day
y we expect
p
roughly
g y or approximately
pp
y 1 7 of the
total number of infectives to move from the infective class
It into the removed class Rt .
Just enter

1 day

2 days

3 days

4 days

Because the removal rate measures the

fraction of the infective class cured during a
single time step, we have found a good
estimate for ; we take
= 1 0 .1429
7

5 days 6 days

At the same time, we have found a good

interpretation for 1 : it is the average
duration of infectious period, i.e. the duration
of staying in infective class I before moving
into the removed class R .
In fact, we can estimate for real diseases
by observing infected individuals and
determining the mean infectious period 1
first.

21

In summary, we have

1
P = ( S )

no. of new cases arising from one average duration
=

.
infective per unit time
of infection

22

Thus, P0 is interpreted as the

average number of secondary
i f ti
infections
that
th t would
ld b
be produced
d
d
by an infective in a wholly
susceptible population of size S0 .

23

Note that, from this point of view, the critical

value of P0 = 1 makes good biological sense.
If P0 > 1, then a primary case of disease
induces more than one secondary case off the
illness, the size of the infective class increases,
and an epidemic results.
If P0 = 1, then a diseased individual produces
only one new case of the disease, and no
epidemic can occ
occur;
r there can be no gro
growth
th in
the number of infectives.
When P0 < 1, the disease dies out.
24

MATH0011 Lecture 8

Continuous model

Basic Reproductive Number (R0)

In a population
if P0 > 1 : epidemic
id i
if P0 = 1 : endemic stage

if P0 < 1 : sucessful control of infection

For completely susceptible population,

measles
l : P0 = 15-20
15 20
smallpox : P0 = 3-5

SARS : P0 = ???

Note that so far we are using discrete time

intervals (e.g. one day). If we let the time interval
t be
to
b very small,
ll say, one second.
d Th
Then II is
i
almost equal to the instantaneous change of I.
Therefore, one may replace I by dI/dt, which is
the rate of change of I (also called the derivative
of I). Similarly, we may replace S and R by
dS/dt and dR/dt respectively.
respectively
With these notations, our system of equations
can be replaced by the following system of
ordinary differential equations.

25

A system of three ordinary differential

equations describes the SIR model:

26

dS = I(t )S(t )
dt

dI = I(t )S(t ) I(t )

dt
dR = I (t )
dt
where is the infection rate and is the
removal rate of infectives.

27

Given such a system of differential

equations, one would like to solve it (i.e.
find functions S, I and R which satisfy the
equations).
equations)
In general, this system of differential
equations does not have any closed form
solution. However, for given and , we
can solve the systems of differential
equations by using some mathematical
software like Matlab.
In general, the graphs of the S, I, R have
the following shapes.
28

MATH0011 Lecture 8

Some Mathematics Courses

Graphs of S, I, R functions

To learn more about differential calculus, you

may take the following courses.
courses
MATH0201-Basic Calculus
MATH1804-University Mathematics A
MATH1805-University Mathematics B
To learn more about differential equations, you
may take MATH2405-Differential Equations or
MATH3406-Introduction to Partial Differential
Equations

Typical dynamics for the SIR model.

29

Case study: The Hong Kongs

SARS Outbreak in 2003

30

Since November 2002 (and perhaps earlier) an

outbreak of a very contagious atypical pneumonia
(now named Severe Acute Respiratory Syndrome,
SARS) initiated in the Guangdong Province of
China.
This outbreak started a world-wide epidemic after a
medical doctor from Guangzhou infected several
persons at a hotel in Hong
p
g Kong
g around February
y
21st, 2003.
At the very beginning of SARS outbreak, SARS was
believed to be a disease transmitted by respiratory
droplets through close person-to-person contact.
31

Respiratory droplets are relatively large-sized

particles and thus cannot travel long distances
through air and therefore this mode of
transmission cannot account for the rapid and
wide
id spread
d off disease
di
att Amoy
A
Gardens,
G d
a
housing estate in the Kowloon district of Hong
Kong.

At that time, one may therefore ask if the mode

of transmission is airborne.

32

MATH0011 Lecture 8

It is well-known that influenza is an airborne disease.

In the 4th March 1978 issue of the British Medical
Journal, there was a report with detailed statistics of
a flu epidemic in a boys boarding school with a total
of 763 boys.
boys
Of these 512 were confined to bed during the
epidemic, which lasted from 22nd January to 4th
February 1978. It seems that one infected boy
initiated the epidemic. The SIR model was applied
byy J.D. Murray,
y to studyy the spread
p
the flu epidemic
p
in this school and he found that
= 0.00218, = 0.440 and hence
P0 = xS0 = 0.00218 x 763 0.44 = 3.78.

Application of SIR model to the

Amoy Garden outbreak

At that time, in order to see whether SARS is likely

to be an airborne disease,, Dr. W.K. Ching,
g, Dr.
S.K.Chung, and Dr. Patrick Ng of the HKU
Mathematics Department studied the spread of
SARS epidemic at a high-rising residential building
(Block E of the Amoy Gardens) by applying the
SIR model.

33

Since from 26th March 2003 to 30th March 2003,

most of these confirmed cases are from Block E of
the Amoy Gardens, we shall assume all of them
are actually from Block E.

34

There are 33 floors, with 8 flats on

each floor in Block E. Assuming that 3
people are living in each flat, we
estimated that there were
33 x 8 x 3 = 792 residents
living in Block E.
Therefore, we set S(0)= 792 I(0) ,
where I(0) is an unknown at that time.

35

It is believed that one infected resident (a super

spreader) initiated the epidemic when he visited
and stayed with his brother's family, so we
assume that 0 < I(0) 4.
Note that the infected number of residents I(t) at
time t is unobservable (why?)
We assume R(t) = the cumulative number of
residents in Block E confirmed with SARS
symptoms at time t. It is because once a resident
was confirmed with SARS symptoms, he would
be removed and could not infect other residents.
36

MATH0011 Lecture 8

Below are the cumulative numbers of confirmed

cases from 26th March 2003 to 30th March 2003.
Day

1
(26/3)

2
(27/3)

3
(28/3)

4
(29/3)

5
(30/3)

Confirmed
cases

R(1)
7

R(2)
22

R(3)
56

R(4)
78

R(5)
112

The system of differential equations can be solved

by standard mathematical software once we have
fixed , , S(0) and I(0). In particular, we can find
the predicted Re(t) curve and compare it with the
observed R values.
values
The key idea of estimating the model parameters is
to choose the parameters , and I(0) such that
they minimize the square of errors between the
observed data and the model predictions, i.e.
5

min (R(i ) Re(i ))2 0 < , 1, and 1 I(0) 4.

Remark: On 31 March, Block E of Amoy Gardens

was put in isolation and on 1 April, the government
relocated all the residents to two holiday camps
in Sai Kung and Chai Wan for isolation.

, ,I (0) i =1

Here Re(i) is obtained by solving the differential

equations numerically using a mathematical software.
37

In this way, we find that when = 0.001875 and

= 0.975 and I(0) = 4 (and hence P0 = 1.52), the
error between the predicted and the observed
values
l
off R is
i minimum.
i i
1
(26/3)

2
(27/3)

3
(28/3)

4
(29/3)

5
(30/3)

Confirmed
cases

R(1)
7

R(2)
22

R(3)
56

R(4)
78

R(5)
112

Estimated
cases

Re(1)
13

Re(2)
27

Re(3)
47

Re(4)
77

Re(5)
117

Day

38

39

For SARS, in the Block E of Amoy Gardens

scenario, it was found that = 0.001875, =
0.975 and P0 = 1.52.
For influenza,
influenza in the broadening school
scenario, Murray found = 0.00218, = 0.44
and P0 = 3.78.
Note that the value of the infection parameter
in these two cases are very close to each other.
This suggests that both epidemics may have
similar ways of transmission.
At that time, a report of this finding was sent to
the WHO.
40

10

MATH0011 Lecture 8

The mystery of Amoy Gardens remains unsolved

even though there are several competing theories.
Therefore, the mode of transmission is still not very
clear. SARS appears
pp
to be transmitted mainly
y by
y
person-to-person contact. However, it could also be
transmitted by contaminated objects, air, or by other
unknown ways

Reference
n

Mathematical Models in Biology, An

Introduction E.S.
Introduction,
E S Allman and J.A.
J A Rhodes,
Rhodes
Cambridge University Press, 2004.

For reference, see the recent

book At the Epicentre,
published by the Hong Kong
University Press.
41

42

11