roles of remote sensing for influenza risk prediction and early

35
Roles of Remote Sensing for Influenza Risk Prediction and Early Warning Roles of Remote Sensing for Influenza Risk Prediction and Early Warning Richard Kiang, Radina Soebiyanto, Farida Adimi NASA Goddard Space Flight Center Richard Kiang, Radina Soebiyanto, Farida Adimi NASA Goddard Space Flight Center GEO Health & Environment Community of Practice Workshop Centre National d’Études Spatiales (CNES) Paris, France, 27-29 July 2010 GEO Health & Environment Community of Practice Workshop Centre National d’Études Spatiales (CNES) Paris, France, 27-29 July 2010

Upload: others

Post on 12-Sep-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Roles of Remote Sensing for Influenza Risk Prediction and Early

Roles of Remote Sensing for Influenza Risk Prediction and Early Warning

Roles of Remote Sensing for Influenza Risk Prediction and Early Warning

Richard Kiang, Radina Soebiyanto, Farida AdimiNASA Goddard Space Flight Center

Richard Kiang, Radina Soebiyanto, Farida AdimiNASA Goddard Space Flight Center

GEO Health & Environment Community of Practice WorkshopCentre National d’Études Spatiales (CNES)

Paris, France, 27-29 July 2010

GEO Health & Environment Community of Practice WorkshopCentre National d’Études Spatiales (CNES)

Paris, France, 27-29 July 2010

Page 2: Roles of Remote Sensing for Influenza Risk Prediction and Early

Epidemic-prone acute respiratory diseases have no borders, and can be spread rapidly around the world. Global, coordinated surveillance & control efforts are essential.

Epidemic-prone acute respiratory diseases have no borders, and can be spread rapidly around the world. Global, coordinated surveillance & control efforts are essential.

2003 SARSSpread to 37 countries in weeks

2004 H5N1 Avian InfluenzaSpread to 62 countries since 2004. There are still frequent outbreaks in Indonesia, Egypt, and some Southeast Asian countries.

2009 H1N1 PandemicSpread to 48 countries in a month despite heightened public awareness and substantial preventive and control efforts

Page 3: Roles of Remote Sensing for Influenza Risk Prediction and Early

Cilia being invaded by flu virusSource: National Geographic Source: CDC

hemaglutinin

neuraminidase

Page 4: Roles of Remote Sensing for Influenza Risk Prediction and Early

Antigenic driftmutations in HA & NA

Antigenic shiftnovel genes through reassortment

Animals to humansjumping across species

Seasonal epidemicnew strains continue to appear

Pandemicse.g., 1918 H1N1 Spanish Flu, 1957 H2N2 Asian Flu, 1968 H3N2 Hong Kong Flu

Pandemic potentialsH5N1 Avian Flu, 2009 H1N1 “Swine Flu” pandemic

Genetic & Antigenic Variation Among Influenza Viruses

Genetic & Antigenic Variation Among Influenza Viruses

Page 5: Roles of Remote Sensing for Influenza Risk Prediction and Early

First appeared in Hong Kong in 1996-1997, HPAI has spread to approximately 60 countries. More than 250 million poultry were lost.

Worldwide the mortality rate is 53%.

Co-infection of human and avian influenza in humans may produce deadly strains of viruses through genetic reassortment.

On average one major pandemic occurred in each century. 90 years have passed since the 1918 pandemic (0.675M deaths in the US, and 21-50M deaths worldwide).

H5N1 AI — THE PROBLEMH5N1 AI — THE PROBLEM

[email protected]

Page 6: Roles of Remote Sensing for Influenza Risk Prediction and Early

DISTRIBUTION OF H5N1 HUMAN CASESDISTRIBUTION OF H5N1 HUMAN CASES

Source: WHO. Cases from 2003 to June 19, 2008.

[email protected]

Page 7: Roles of Remote Sensing for Influenza Risk Prediction and Early

Highly Pathogenic AI Cases Since January 2010Highly Pathogenic AI Cases Since January 2010

FAO EMPRESFAO EMPRES

Page 8: Roles of Remote Sensing for Influenza Risk Prediction and Early

HUMANS

POULTRY TRADE

wild birdsdomestic birds

ducks & geese

poultry, products, feed, waste, personnel,

equipment

BIRD TRADE MIGRATORY BIRDS

POULTRYSectors 1&2 Sectors 3&4

H5N1 TRANSMISSION PATHWAYSH5N1 TRANSMISSION PATHWAYS

LPAI spill over

HPAI spill back

[email protected]

human flu virus

pandemic strain

reassortment

?

Page 9: Roles of Remote Sensing for Influenza Risk Prediction and Early

Asia43% thru poultry14% thru mig. birds

Analysis of Global Spread of H5N1 through Phylogenetic Evidence, Poultry & Bird Trades,

And Bird Migration Data

Analysis of Global Spread of H5N1 through Phylogenetic Evidence, Poultry & Bird Trades,

And Bird Migration Data

Africa 25% thru poultry38% thru mig. birds

Europe87% thru mig. birds

US

Most likely thru poultry to surrounding countries first, then thru migratory birds to US mainland

Source: Kilpatrick et al., PNAS 2006.

[email protected]

Page 10: Roles of Remote Sensing for Influenza Risk Prediction and Early

Objective 4

Objective 3 Objective 2

Objective 1

Page 11: Roles of Remote Sensing for Influenza Risk Prediction and Early

OBJECTIVESOBJECTIVES

Perform empirical AI outbreak risk analyses based on outbreak history, environmental parameters, and socio-economic factors.

Identify spatiotemporal risk for AI outbreaks based on wetlanddistributions, prevalence of bird species, flyways of migratory birds, surface characteristics, and socioeconomic factors.

Model the spread of AI virus from large commercial poultry farms to small and backyard farms under typical environmental and socioeconomic conditions.

Model weekly influenza-like illness cases based on observed and forecast meteorological parameters for regions in the US and other tropical countries.

[email protected]

Page 12: Roles of Remote Sensing for Influenza Risk Prediction and Early

What environmental and socio-economical factors may contribute to highly pathogenic AI outbreaks?

Page 13: Roles of Remote Sensing for Influenza Risk Prediction and Early

Poultry Outbreaks, Human Cases, Wet Markets, And Distribution Centers

Poultry Outbreaks, Human Cases, Wet Markets, And Distribution Centers

January – February 2006Based on Media & Publicly Available Information

January – February 2006Based on Media & Publicly Available Information

[email protected]

Page 14: Roles of Remote Sensing for Influenza Risk Prediction and Early

Histograms of Distance from Neighborhoods With/without Outbreaks to Other Locations

Histograms of Distance from Neighborhoods With/without Outbreaks to Other Locations

Log

(N+1

)Lo

g (N

+1)

meters

Page 15: Roles of Remote Sensing for Influenza Risk Prediction and Early

What areas around wetlands may have higher risks for AI outbreaks?

Page 16: Roles of Remote Sensing for Influenza Risk Prediction and Early

NAMRU-2 Bird Surveillance Sites on Java

Page 17: Roles of Remote Sensing for Influenza Risk Prediction and Early

NAMRU-2 Bird Surveillance StudyNAMRU-2 Bird Surveillance StudyThe role of migratory birds in the spread of H5N1 remains under

considerable debates.

In Indonesia, migratory pathways are only known for shorebirds (East Asian-Australasian flyway) and migratory ducks and geese (East Asian & Central Asian flyways).

4067 birds comprising of 98 species and 23 genera were collected in 2006-2007.

Most common birds: striated heron, common sandpiper, and domestic chicken.

6%3% 14%

Page 18: Roles of Remote Sensing for Influenza Risk Prediction and Early

(continued)(continued)

RNA was extracted from swabs; RT-PCR was conducted for H5N1 genes; antibodies was detected using hemagglutination inhibition and other tests.

Species with the highest seropostive rates in each category are Muschovy duck (captive), striated heron (non-migratory) and Pacific golden plover (migratory).

16% of the captive birds (duck, swan, pigeon, etc.) showed H5N1antibody.

Infected captive birds can be asymptomatic.

In Indonesia, the role of migratory birds in H5N1 transmission is limited.

Page 19: Roles of Remote Sensing for Influenza Risk Prediction and Early

ASTER False-Color, Google Earth And Land Use Maps Around Indramayu

ASTER False-Color, Google Earth And Land Use Maps Around Indramayu

Page 20: Roles of Remote Sensing for Influenza Risk Prediction and Early

Supervised ClassificationSupervised Classification

Page 21: Roles of Remote Sensing for Influenza Risk Prediction and Early

EU’s & UK’s Practice:

3 km protection zone10 km surveillance zonelarger restricted zone

Buffer zones can be established to limit the spread of H5N1 around wetlands and the nearby farmlands Buffer zones can be established to limit the spread of H5N1 around wetlands and the nearby farmlands

Page 22: Roles of Remote Sensing for Influenza Risk Prediction and Early

How do AI viruses spread on and off farms, within and across poultry sectors, and into the environment?

Page 23: Roles of Remote Sensing for Influenza Risk Prediction and Early

Densely Populated Sector I Poultry Production Area

Google Earth image

Page 24: Roles of Remote Sensing for Influenza Risk Prediction and Early

Detection of H5N1 Infection on a Poultry FarmDetection of H5N1 Infection on a Poultry Farm

Highly pathogenic AI infection on a poultry farm cannot be detected immediately.

Some infected poultry may not look very sick.

In a poultry house with 20,000 chickens, an infection of <1% may not be detected in a walkthrough.

Using a SEIR model, it can be shown that it may take 4-5 days to detect an outbreak.

Before an infection is detected, viruses continue to spread onfarm and off farm, through service personnel, equipment, materials, and the poultry that have been shipped out.

[email protected]

Page 25: Roles of Remote Sensing for Influenza Risk Prediction and Early

On-Farm and Off-Farm Spread of H5N1On-Farm and Off-Farm Spread of H5N1

[email protected]

Page 26: Roles of Remote Sensing for Influenza Risk Prediction and Early

Within and Across-Sector Spread of H5N1Within and Across-Sector Spread of H5N1

[email protected]

Page 27: Roles of Remote Sensing for Influenza Risk Prediction and Early

How does seasonality vary geographically? How is influenza transmission influenced by the environment? How can this be used for forecasting and pandemic early warning?

Page 28: Roles of Remote Sensing for Influenza Risk Prediction and Early

Latitudinal variability in influenza transmission pattern

Experimental findings on the effect of meteorological factors in influenza transmission, virus survivorship and host susceptibility

Viboud et al. (2006). PLoS Med 3(4):e89

Empirical Evidences of Environmental Influences On Influenza Transmissions

Empirical Evidences of Environmental Influences On Influenza Transmissions

Page 29: Roles of Remote Sensing for Influenza Risk Prediction and Early

Hong KongHong Kong

Land surface temperature

Air temperature

Rainfall

Relative humidity

Dew point

Evaporation

Pressure

Solar irradiance

Sunshine hours

Windspeed

Page 30: Roles of Remote Sensing for Influenza Risk Prediction and Early

Time series for environmental parameters and weekly seasonal influenza cases

Hong KongHong Kong

Training Prediction

Page 31: Roles of Remote Sensing for Influenza Risk Prediction and Early

Maricopa County, ArizonaMaricopa County, Arizona

Land surface temperature

Air temperature

Rainfall

Relative humidity

Dew point

Pressure

Solar irradiance

Windspeed

Page 32: Roles of Remote Sensing for Influenza Risk Prediction and Early

Time series for environmental parameters and weekly seasonal influenza cases

Maricopa County, Arizona

Maricopa County, Arizona

Training Prediction

Page 33: Roles of Remote Sensing for Influenza Risk Prediction and Early

Time series for environmental parameters and weekly seasonal influenza cases

New York CityNew York City

Training Prediction

Page 34: Roles of Remote Sensing for Influenza Risk Prediction and Early

Epidemic-prone acute respiratory diseases have no borders, and can be spread rapidly around the world. Internationally coordinated surveillance & control efforts are essential. Better understanding of the influenza seasonality and the environmental effects on transmission will help the global surveillance and control efforts.

Epidemic-prone acute respiratory diseases have no borders, and can be spread rapidly around the world. Internationally coordinated surveillance & control efforts are essential. Better understanding of the influenza seasonality and the environmental effects on transmission will help the global surveillance and control efforts.

Page 35: Roles of Remote Sensing for Influenza Risk Prediction and Early

Thank You!

[email protected]