You are on page 1of 15

Seasonal Influenza

Matlab Simulation Experiment


Asia Renee Suarez

Table of Contents
Introduction...................................................................................................... 3
SEIR Model....................................................................................................... 4
Formulas and Initial Code.................................................................................6
SEIR MODEL AND INFLUENZA...........................................................................8
Further Experiments....................................................................................... 10
Conclusion...................................................................................................... 13

Introduction
On average, anywhere between 5% and 20% of the United
States population are affected by the seasonal influenza every year.
There are several different types of Influenza: Influenza A, Influenza B,
and Influenza C. Each strain of influenza has its own severity, but
typically they are similarly transmitted through bodily fluids. What is
incredibly interesting about influenza is that it is completely possible
for a person to actually be contagious but not yet know as the
symptoms have yet to appear. In this way, influenza can easily be
accidently spread by someone. This is one reason why the disease is
continuously an issue in populations like that of the United States.
Typically most individuals recover fully from influenza, but there are
limited cases where it has caused death; however, this is a small
portion of the population as deaths, if they occur, usually impact either
the elderly or infants. Currently the only way to prevent the disease is
a yearly vaccine. Typically the vaccine is based on the most popular
strain from the previous year. It is important to note that influenza is
not completely 100% preventable with the vaccine because there are
always new and changing strains of the disease, which makes it
extremely difficult to prevent.
Through the use of MATLAB, I aim to simulate seasonal influenza
with realistic data provided by the CDC. I plan to simulate this with

special real values initially, but to truly understand how certain factors
effect the disease, I will be testing select variables individually and
comparing them to a control group.

SEIR Model
In order to simulate Influenza, it is necessary to use the SIER
Model below:

Essentially the model is a way to track individuals as they move


through the different stages of a disease. S represents the number of
individuals that are susceptible to becoming infected. E represents the
number of individuals that have been exposed to someone who is
infected with the disease. I represents the number of individuals that
are currently infected with the disease. For the sake of this simulation,
R represents the number of individuals that have recovered from the
disease, and are no longer capable of being either susceptible or

infected by the disease. Also, N represents the total population at a


given time. There are specific variables that represent rates that
describe how quickly an individual can move from group to group. First,
it is important to note mu,

, which represents the mortality rate of

the population. In this particular version of the SEIR Model, only


susceptible females are capable of giving birth and every group is
capable of dying from a non-disease related cause. Next, there is beta,
, which represents how often a susceptible individual will come into
contact with an infected person. Additionally, sigma,

, represents

the rate at which an exposed individual becomes infected. This model


assumes that every individual that becomes exposed to an infected
individual will also become infected. Lastly, gamma,

, represents

the rate at which an individual recovers and becomes completely


resistant of the disease.

Formulas and Initial Code


Numerical Equations
This SEIR Model shows the movement of people over time, which
means that the equations that represent this change are very much
time dependent. The variables, S,E,I,R, and N are continuously
changing. These are the following related equations.

1.

S (t + t)
I (t)
= ( NS ( t ) )
S(t )
t
N

2.

E ( t + t )
(t )I ( t )
=S
( + )E ( t )
t
N

3.

I (t + t)
=E ( t ) + ( + )I (t)
t

4.

R (t+ t)
=I ( t )R(t )
t

5.

N=S+ E+ I + R

The following assumptions are made:


The birth and death rate are the same.
The death rate does not change due to the disease.
An individual that becomes exposed will definitely become
infected.
Once a person is recovered, they can no longer get the disease
again.

Initial MATLAB Code Shell Without Variable Values


%SEIR MODEL
%Set time limits:
T= ; %maximum time
dt= ; %change in time step
clockmax=ceil(T/dt);
%rates
b= ;
sig= ;
g= ;
mu= ;

%beta
%sigma
%gamma
%mu

%Initial Values
S=;
E=;
I=;
R=0; %R=0 Initially because there are no recovered individuals at the start of the disease
N=S+I+E+R;
%Place to store values
sSave=zeros(1,clockmax);
eSave=sSave;
iSave=sSave;
rSave=sSave;
nSave=sSave;
tSave=sSave;
%Find values over time

for clock=1:clockmax
t=clock*dt;
%Changes
S2E=dt*b*(I/N)*S;
E2I=dt*sig*E;
I2R=dt*g*I;
%New Values
S=S+mu*(N-S)-S2E;
E=E+S2E-E2I-(mu+sig)*E;
I=I+E2I-I2R+(mu+g)*I;
R=R+I2R-mu*R;
N=S+E+I+R;
%Store new values
sSave(clock)=S;
eSave(clock)=E;
iSave(clock)=I;
rSave(clock)=R;
nSave(clock)=N;
tSave(clock)=t;
end
figure
plot(tSave,sSave,'y',tSave,eSave,'b',tSave,iSave,'r',tSave,rSave,'g')

SEIR MODEL AND INFLUENZA


Through research, I attempted to find the values of variables that
could most accurately replicate a logical representation of an Influenza
simulation within the population of the United States. I decided to
simulate this based on a maximum time of 730 days. From the CDC, I
found that a person can be contagious for about 8 days, so I chose

8
=0.21918 . It takes approximately 17 days for all symptoms to
365

subside so I chose

17
=.046575 .
365

From the World Bank, I found that

the mortality rate in the United States is .008 so I chose

=.008 . As

for the value of

, which represents how often a susceptible

individual will come into contact with an infected person I chose an


arbitrary value so that

=.410959 . Initially, the number of susceptible

individuals is set to the current population of the United States so


S=318900000 . The number of exposed and recovered individuals is
set to 0. The number of infected individuals is set to 25 so

Results
The following graph was produced:

I =25 .

It is clear that this graph holds logical results. As the number of


susceptible individuals is decreasing, the number of exposed, infected,
and recovered individuals is increasing in turn. Notice that the number
of susceptible individuals does not flat line after it plunges between
days 200 and 300. This is because the number of susceptible
individuals is always increasing per day due to births. I believe this is
also why the number of recovered individuals also appears to increase.
After the program is run, the new total population is

N=6.3006108

which is almost double the size of the original population. We do see


the expected spike in infected and exposed.

Further Experiments
Now that we know what the expected graph should be given the
researched data, we can now attempt to see how changes in select
variables can affect the movements from group to group in this SEIR
Model.
Experiment 1: Change in

The goal of this particular experiment is to see if how often a


susceptible individual will come into contact with an infected person
actually matters. Does this affect the graph? I believe that a higher
will cause more individuals to become infected and a lower
will cause striking fewer individuals to become infected.
Using the parameters:
=0.16 5
=0.41095 9
=0. 9
The following graphs were produced:
When

=0.16 5 :

When

=0.41095 9 :

When

=0. 9 :

Results of Experiment 1
values
0.16 5
0.41095 9
0.9

Maximum Number
of Infected
7
4.328210
9.6656107
1.328910 8

It is very clear that the values of


model. We see that when

Time Frame at
Occurrence
500-600 days
250-350 days
150-250

do in fact impact the

is large not only does it affect more

people, but it also effects the time at which the maximum number of
individuals infected has reached its upmost limit. When

is at its

lowest value, the event where the number of infected individuals


reaches it max occurs much later than the other. This makes sense
logically. If an individual is more likely to come into contact with an
infected individual, he must be more likely to become infected.
Similarly, if an individual comes into contact with less infected
individuals, he will be less likely to become infected.

Conclusion
In conclusion, I believe I was able to simulate the Seasonal
Influenza with data from both the World Bank and the Center for
Disease and Control, as the graph seemed to display accurate data.
However, I wish that I could have found a way to make the disease
more cyclic as in real life an individual can contract the disease more
than once. Through my experiment, I found that how often a
susceptible individual will come into contact with an infected person
does play a roll in how quickly the disease spreads and also how many
people are infected. This leads me to believe that a possible
quarantine, which would significantly reduce the number of exposed,
would surely help control the disease, as it would result in less
exposure and in turn less transmission.
In the future, I would like to test a few more variables depending
on conditional statements. For example, I would like to place a
conditional statement on

where a randomly generated number, 0

or 1, can determine if an individual can become infected or not by


being exposed. Additionally, I would like to try to add the vaccine for
the disease to the simulation.

You might also like