19.11.2012 Views

software training courses 2010 corsi di addestramento ... - EnginSoft

software training courses 2010 corsi di addestramento ... - EnginSoft

software training courses 2010 corsi di addestramento ... - EnginSoft

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Healing the swine flu with<br />

modeFRONTIER<br />

One of the hot topics of the winter<br />

2009 that probably will be remembered<br />

is the outbreak of the so-called “swine<br />

flu”. The new virus A-H1N1 captured<br />

the attention of the Italian me<strong>di</strong>a,<br />

which literally bombarded the<br />

population with daily reports on the<br />

number of deaths, the severity of this<br />

virus and other alarms based on the<br />

opinion of some “epidemiology<br />

experts”, sprea<strong>di</strong>ng in this way the<br />

fear within the population.<br />

During the first weeks of autumn some<br />

sentences such as “We will have an<br />

extraor<strong>di</strong>nary peak of flu <strong>di</strong>ffusion between Christmas and<br />

the new year” or “we will be the victim of a new pandemia<br />

with many deaths” were pronounced.<br />

How is it possible to pre<strong>di</strong>ct such an “apocalyptic” scenario<br />

so many weeks in advance? The truth is that it is extremely<br />

<strong>di</strong>fficult, especially when no previous knowledge on the virus<br />

behavior is available. However, in epidemiology some simple<br />

mathematical models have been developed and used for<br />

many years; they are mainly based on or<strong>di</strong>nary <strong>di</strong>fferential<br />

equations (shortly ODEs).<br />

Probably, the most known model is the so-called SIR model,<br />

where the population, which is supposed to be large and<br />

homogeneous enough, is <strong>di</strong>vided into three groups<br />

(Susceptible, Infected and Recovered), accor<strong>di</strong>ng to their<br />

status (see [4]). Strong simplifications are present in this<br />

model which can be applied as scale level; in some cases it<br />

could lead to poor results. For this reason, there is a variety<br />

of SIR based models which remove some of these<br />

Newsletter <strong>EnginSoft</strong> Year 6 n°4 - 35<br />

A photo of the A-H1N1 virus (left) and a swine (right). They do not look so dangerous…<br />

Figure 2: The solution of a classical SIR model: the three categories S, I and R are plotted versus time,<br />

expressed in weeks. It is clear that the <strong>di</strong>sease has a peak between the first and the second week and that the<br />

maximum number of ill people is around 150 over 1000. In this case we adopted the following values for the<br />

parameters (β=10, ν=5 and the number of initial infected is 1.98 for 1000 persons).<br />

simplifications in an attempt to be closer to reality. In this<br />

work, we suggest to add a new category to the standard SIR<br />

model in order to consider the fact that unfortunately, some<br />

infected people may <strong>di</strong>e. The resulting model can be<br />

expressed as:<br />

This is a non-linear system of first<br />

order ODEs; the four categories used to<br />

classify the population are S =<br />

Susceptible, I = Infected, R =<br />

Recovered, D = Dead and they are<br />

expressed in percentage terms. For this<br />

reason the sum of all the categories<br />

has to be always equal to one. The<br />

parameters β, ν and δ are constants<br />

which determine the evolution of the <strong>di</strong>sease. The results<br />

strongly depend on the numerical values of these parameters.<br />

Specifically the peak value of the infected and the week of<br />

the year when it will appear, which<br />

are important information to have<br />

in advance, can be really <strong>di</strong>fficult to<br />

capture if there is not a rigorous<br />

estimation of the above mentioned<br />

parameters.<br />

Obviously, it is mandatory to know<br />

the initial con<strong>di</strong>tions before solving<br />

the system: in other words we have<br />

to know the number of susceptible,<br />

infected, recovered and dead<br />

persons at time zero, when we want<br />

to begin our simulation.<br />

The solution of such equations is<br />

always done, exclu<strong>di</strong>ng trivial cases,<br />

through numerical techniques which<br />

have been expressively defined to<br />

tackle this kind of problem. To

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!