Title: Comment modeliser la propagation d
1Comment modeliser la propagation depidemies a
lechelle mondiale ?
- Marc Barthelemy
- CEA, France
Forum Theorie 02/2008
2Outline
- Motivations
- Global scale metapopulation model
- Theoretical results
- Predictability
- Epidemic threshold, arrival times
- Applications
- SARS
- Pandemic flu Control strategy and Antivirals
3Theoretical physics and interdisciplinarity
- Theoretical physics in other fields
- Biology, ecology, epidemiology, economy, social
sciences,
- Tools, concepts and methods
- Data analysis scaling laws, data collapse
- Modeling minimal model, relevant parameters,
4Epidemiology an interdisciplinary field
Transportation systems
Social networks
Urbanism
Biology, virology
5Epidemiology vs. Etiology Two levels
- Microscopic level (bacteria, viruses)
compartments - Understanding and killing off new viruses
- Quest for new vaccines and medicines
- Macroscopic level (communities, species)
- Integrating biology, movements and interactions
- Vaccination campaigns and immunization strategies
6Epidemiology past and current
- Human movements and disease spread
- Complex movement patterns different means,
different scales (SARS) Importance of networks
Black death Spatial diffusion Model
721st century Networks and epidemics
Epidemics spread on a contact network
Different scales, different networks
- Individual Social networks (STDs on sexual
contact network) NB Sometimes no network !
- Intra-urban Location network (office, home,
shops, )
- Inter-urban Railways, highways
- Global Airlines (SARS, Flu, )
8Airline network and epidemic spread
Complete IATA database - 3100 airports
worldwide - 220 countries - 20,000
connections - wij passengers on connection
i-j - gt99 total traffic
9Modeling in Epidemiology
10Metapopulation model
j
l
Baroyan et al, 1969 40 russian cities Rvachev
Longini, 1985 50 airports worldwide Grais
et al, 1988 150 airports in the US Hufnagel
et al, 2004 500 top airports worldwide
Colizza, Barrat, Barthelemy Vespignani, PNAS
(2006)
11Metapopulation model
Reaction-diffusion models FKPP equation
Inner city term
Travel term
Flahault Valleron (1985) Hufnagel et al, PNAS
2004, Colizza, Barrat, Barthelemy, Vespignani
PNAS 2006, BMB, 2006. Theory Colizza
Vespignani, Gautreau al,
12Metapopulation model
- Theoretical studies
- Predictability ?
- Epidemic threshold ?
- Arrival times distribution ?
13Predictability
One outbreak realization
Another outbreak realization ? Effect of noise ?
?
?
?
?
?
?
14Overlap measure
Similarity between 2 outbreak realizations
Overlap function
time t
time t
time t
time t
15Predictability
no degree fluctuations no weight fluctuations
degree heterogeneity
weight heterogeneity
Colizza, Barrat, Barthelemy Vespignani, PNAS
(2006)
16Predictability
- degree heterogeneity
- decreases predictability
l
wjl
j
- Weight heterogeneity
- increases predictability !
- Good news Existence of preferred channels !
- Epidemic forecast, risk analysis of containment
strategies
17Theoretical result Threshold
- Reproductive number for a population
- moments of the degree distribution
- travel probability
Travel restrictions not efficient !!!
- Effective for reproductive number for a network
of populations
Colizza Vespignani, PRL (2007)
18Theoretical result average arrival time
- Ansatz for the arrival time at site t (starting
from s)
- Population of city k
- Traffic between k and l
- Transmissibility
- Set of paths between s and t
Gautreau, Barrat, Barthelemy, JSTAT (2007)
19Applications
- 1. SARS test of the model
- 2. Control strategy testing antivirals
20Application SARS
pop i
pop j
- refined compartmentalization
- parameter estimation clinical data local
fit - geotemporal initial conditions available
empirical data - modeling intervention measures standard
effective modeling
21SARS predictions
22SARS predictions (2)
Colizza, Barrat, Barthelemy Vespignani, bmc
med (2007)
23More from SARS - Epidemic pathways
- For every infected country
-
- where is the epidemic coming from ?
- - Redo the simulation for many disorder
realizations - (same initial conditions)
- - Monitor the occurrence of the paths
- (source-infected country)
24(No Transcript)
25SARS- what did we learn ?
- Metapopulation model, no tunable parameter
- good agreement with WHO data !
- Existence of pathways
- confirms the possibility of epidemic forecasting
! - Useful information for control strategies
26Application of the metapopulation model effect
of antivirals
- Threat Avian Flu
- Question use of antivirals
- Best strategy for the countries ?
- Model
- Etiology of the disease (compartments)
- MetapopulationTransportation mode (air travel)
27Predictions pandemic flu
28Effect of antivirals
- Comparison of strategies (travel restrictions not
efficient)
- Baseline reference point (no antivirals)
- Uncooperative each country stockpiles AV
- Cooperative each country gives 10 (20) of
its own stock
29Effect of antivirals Strategy comparison
Best strategy Cooperative !
Colizza, Barrat, Barthelemy, Valleron,
Vespignani, PLoS Med (2007)
30Conclusions and perspectives
- Global scale (metapopulation model)
- Pandemic forecasting
- Theoretical problems (reaction-diffusion on
networks) - Smaller scales-country, city (?)
- Global level simplicity due to the dominance
of air travel - Urban area ? Model ? What can we say about the
spread of a disease ? Always more data available
31Collaborators and links
- Collaborators
- - A. Barrat (LPT, Orsay)
- - V. Colizza (ISI, Turin)
- - A.-J. Valleron (Inserm, Paris)
- - A. Vespignani (IU, Bloomington)
-
- Collaboratory
- PhD students
- - P. Crepey (Inserm, Paris)
- - A. Gautreau (LPT, Orsay)
http//cxnets.googlepages.com
marc.barthelemy_at_cea.fr
32The End
33Geographic spread
Epidemics starting in Hong Kong
34Geographic spread
Epidemics starting in Hong Kong
35Geographic spread
Epidemics starting in Hong Kong
36Collaborators and links
- Collaborators
- - A. Barrat (LPT)
- - V. Colizza (IU ISI)
- - A.-J. Valleron (Inserm)
- - A. Vespignani (IU)
-
- Collaboratory
- PhD students
- - P. Crepey (Inserm)
- - A. Gautreau (LPT)
http//cxnets.googlepages.com
marc.barthelemy_at_gmail.com
37Theoretical result average arrival time
- Test on the world-wide air travel network
Gautreau, Barrat, Barthelemy, JSTAT (2007)
38SARS predictions (3) - False positives
Country Median 90 CI
Japan 30 9-114
Netherlands 2 1-10
United Arab Emir. 2 1-11
Bangladesh 2 1-14
Bahrain 1 1-20
Cambodia 1 1-8
Nepal 1 1-20
Brunei 1 1-8
Israel 1 1-13
Mauritius 1 1-32
Saudi Arabia 1 1-11
Colizza, Barrat, Barthelemy Vespignani, bmc
med (2007)
39Effect of antivirals
- Flu type disease Compartments
40Discussion scale vs. complexity
World
Modeling ?
City
41Predictability
- Predictability Reliability of epidemic forecast
- Robustness of epidemic scenario versus noise
- Measure of statistical similarity of the
history of epidemic outbreak with -
- - same initial conditions
- - different noise realization
42Global spread Summary
- Metapopulation model
- Importance of heterogeneity-high predictability
- Possibility of forecasting existence of epidemic
pathways
- Why does it work ?
- Existence of a dominant network air travel !!!
- Other scales ? (country, urban area, )
43Epidemic spread at the country level Detecting
robust patterns
- Detecting epidemic pathways from epidemic data
- Case-studies flu in the USA (1972-2002) and
France (1984-2004)
44Epidemic spread Detecting robust patterns
?
- Detecting epidemic pathways from epidemic data
45Epidemic spread Detecting robust patterns
New cases per 100,000 inhabitants and per Unit
time
week
- Correlation between A and B ?
- Effect of transportations modes ? Existence of a
dominant mode ?
46Epidemic spread Detecting robust patterns
New cases per 100,000 inhabitants and per Unit
time
week
- Correlation between A and B Pearson coeff
47Detecting robust patterns
- Value of the pearson-comparison value ?
Correlations due to constraints
Null model Reshuffling
Maximal correlation Synchronized peaks
Crepey and Barthelemy, Amer. J. Epidemiology
(2007)
48Detecting robust patterns
- Average X for the correlation between CA and the
other states
- Existence of a robust channel large ltXgt and
small dispersion
49Detecting robust patterns
- Existence pairs with large P (USA)
- Correlations with states
- California
- Illinois
- New York
50Detecting robust patterns
- Existence of pairs with large P (France)
Bretagne IdF Rhone-Alpes
51Detecting robust patterns - relation with
transport modes
- Relation between P and transportation modes
(Multivariate analysis air travel, distance,
temperature)
- Existence of a dominant mode for inter state
spread - in the US !
52Detecting robust patterns (France)
- Relation between P and transportation modes
(Multivariate analysis air travel, car, train)
Parameter Coefficient
Distance -9.220e-02
Car traffic 2.645e-01
Train traffic 2.043e-01
No dominant transportation mode !
53Summary
- Two factors
- Heterogeneity of flows (epidemic pathways) at
all scales - Existence of a dominant transportation mode
depends on the scale - Global scale air travel
- USA air travel (interstate case)
- France car, train, equivalent