23
GIS and its GIS and its applications in Health applications in Health ران ه ت ی ک ش ز پ وم ل ع گاه ش ن دا ت ش هدا ت کده ش ن دا ران ه ت ی ک ش ز پ وم ل ع گاه ش ن دا ت ش هدا ت کده ش ن دا ر ماه ا;pma& د ر ماه ا;pma& د1384 1384 ل ع ده: ن ه ه د0 اراي ل ع ده: ن ه ه د0 اراي0 ی0 ی ت ش دو ق ح ر کب ا ت ش دو ق ح ر کب ا

GIS and its applications in Health

Embed Size (px)

DESCRIPTION

GIS and its applications in Health. دانشکده بهداشت دانشگاه علوم پزشکی تهران آذر ماه 1384 ارايه دهنده: عل ي اکبر حقدوست. Definition. Lilienfeld 1976 in “ Fundamental epidemiology ” - PowerPoint PPT Presentation

Citation preview

Page 1: GIS and its applications in Health

GIS and its applications in GIS and its applications in HealthHealth

دانشکده بهداشت دانشگاه علوم پزشکی تهراندانشکده بهداشت دانشگاه علوم پزشکی تهران13841384آذر ماه آذر ماه

اکبر حقدوست اکبر حقدوستييارايه دهنده: علارايه دهنده: عل

Page 2: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Definition

Lilienfeld 1976 in “Fundamental epidemiology”

EpidemiologyEpidemiology: study of distribution of disease, or pathological condition in human populations and factors that influence this distribution

What does mean “distribution” in above definition?

We should explore the distribution in timetime and spacespace

John Snow project

Page 3: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

�Example

Page 4: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Poisson DistributionPoisson Distribution

k!

eλk)P(x

λk

Assumptions:Assumptions:

1.1.Constant proportion in areaConstant proportion in area

2.2.Independent distributionIndependent distribution

Page 5: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Flying bombs strikes, south London

آيا بمبهاي آلمان در جنگ جهاني دوم آن گونه كه ادعا مي نمود قدرت تفكيك باال در هدف قرار

دادن مواضع را داشت؟

Page 6: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Flying bombs strikes, south London

Number of bombs in cells

# of cells 0 1 2 3 4 5 Total

observed 229 211 93 35 7 1 576

expected 227.5 211.3 98.2 30.4 7.1 1.54 576

Page 7: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Flying bombs strikes, south London

How can we test if the distribution of bombs was random?

x

Sm

x

xxobs

dfmwithondistributifollowsmx

S

x

SModelPoissonIn

xm

i

i

i

ii

x

x

2

1

222

22

2

)1()(

exp

)exp(

)1()1(

1:

Page 8: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

This method is robust even for small sample numbers

You can estimate variance even based on the group data

Limitation

It does not taking into account distances between events

Flying bombs strikes, south London

Page 9: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

A

Nearest neighbor method

Minimum distance between observation I and (n-1)

Checks the distances between events

a b

cd

Page 10: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Exponential (e)

nx

n

n

xe

ne

)1lim(

)1

1lim(

Page 11: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Exponential (e)

nqnnn en

nqqp )1()1(

Page 12: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

R

r

Page 13: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

A

r 2

1

Probability that a random point within the boundaries of the large circle falls within the boundaries of the small circle

Probability that a random point within the boundaries of the large circle does not fall within the boundaries of the small circle

A

r 2

Probability that n random points within the boundaries of the large circle does not fall within the boundaries of the small circle

An

rn

eA

r2

)1(2

Page 14: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Probability that at least one of n random points within the boundaries of the large circle falls

within the boundaries of the small circle

A

rn

e

2

1

Page 15: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Expected median, mean and variance based on these explanations (nearest neighbor analysis)

2068.0)var(5.0)(

)2log(47.047.0)(

n

Arwith

n

Armean

n

Armedian

Page 16: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

example

دارد؟ وجود ارتباطي هيچ شكل اين نقاط بين آيا

0

5

10

15

0 5 10 15

x

y

Page 17: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

نقاط ضعف روش نزديكترين گزينه

Very large and small area (A) may reduce the significance

0

5

10

15

20

25

30

35

40

45

50

0 5 10 15 20 25 30 35 40 45 50

x

y

0

5

10

0 5 10

x

y

Page 18: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

BoundaryBoundary influenceinfluence

Page 19: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Transformed map

Uniform distribution is violated in nearly all situations.

Redraw a map so that the density of the individuals at risk is equalized over the area under study, called a cartogram

Page 20: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Original map

Page 21: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Cartogram

Page 22: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Limitations in using cartogram

1. We need to have an accurate map to illustrate the population density

2. The population at risk is not usually distributed exactly the same as the whole population

3. It needs power full computer to generate a cartogram

Page 23: GIS and its applications in Health

GIS workshop in TUMSGIS workshop in TUMS A.A. HaghdoostA.A. Haghdoost

Alternative method

Case control approach

It means we can compare the spatial distribution of cases versus controls using Poisson models