Remote Sensing and GIS: Tools for the Prediction of Epidemic for the

Intervention Measure

Gouri Sankar BhuniaDepartment Of Vector Biology & Control, Rajendra Memorial Research Institute Of Medical Sciences (ICMR), Agamkuan,

Patna – 800 007, Bihar, India

India Geospatial Forum, 8th February, 2011

What is ‘Kala-azar’? Kala-azar/Visceral leishmaniasis - vector-borne diseases

- protozoan parasite (Leishmania donovani)- Ph. argentipes

VL exists in 88 countries on five continents

Estimated 300,000 cases occurred annually and more than 60% of the global burden account for India, Nepal, and Bangladesh

Bihar contributes to over 80-85% of National Kala-Azar Burden.

32 Dist. Out of 38 dist. have endemic foci for Kala-Azar cases.

Disease affects mostly poorest of the poor specially Mushar Community.

Role of RS in Kala-azar Control

• RS technology - tool for the surveillance of habitats, densities of vector species and even prediction of the incidence of diseases.

• To recover continuous fields of air temperature, humidity, and vapor pressure deficit from remotely sensed observations have significant potential for disease vector monitoring and related epidemiological applications.

• Role of RS by its synoptic coverage, high repetivity, bird eye view, inaccessibility, cost effectiveness

Role of GIS in Kala-azar Control

Seamlessly integrates disparate types of information sources Environmental Conditions Substance Characteristics Fate/Transport Exposure/Latency

Supports multidisciplinary analysis using a systems approach Environmental Economic Demographic

GIS architecture is ideal for handling the complexities of a large number of spatially distributed variables

Provides a vehicle for improving our understanding of the contextual relationships between factors

Objectives

Examining disease distribution and its relation with the environmental factors and vector distribution in a Kala-azar endemic region in Bihar, India

Study Area

BIHARBIHAR

Data Used Ground Survey Data –

i. Disease Incidence Report 2005-2010ii. Adult sandflies (Ph. argentipes) were collected

from indoors (Human dwelling & cattle shed) using CDC light trap

iii. Indoor climate (e.g., Temperature and Relative humidity)

Satellite Data –

Landsat-5 TM (Path/Row-141/42; DOP-22/10/09)

Methodology

Field SurveyField

SurveySatellite

dataSatellite

dataTopographical

SheetTopographical

Sheet

Pre-processing

Pre-processing

Spectral AnalysisSpectral Analysis

SAVISAVI

LSTLST

Entomological Data

Entomological Data

Climatic Data

Climatic Data

WIWI

Statistical

Analysis

Epidemiological Data

Epidemiological Data

Integration in GIS PlatformIntegration in GIS Platform

Geo-statistical Analysis

Geo-statistical Analysis

Model Development for sandfly Prediction

Model Development for sandfly Prediction

LULCLULC

Esti

mati

on

of

Associa

tion

Esti

mati

on

of

Associa

tion

RESULTS

Temporal distribution of cases and deaths of the study area

Calculation of mean centre and directional distribution of disease

Monthly distribution of sandfly density in the study site

P. argentipes

Percent Sergentomiya Percent P. papatasi Percent

Male (M) 126 36.95 80 54.05 9 60.00

Female (F) 215 63.05 68 45.95 6 40.00

Total 341 100% 148 100% 15 100%

Relative abundance (%)

67.66 29.37 2.98

M:F Ratio 1:1.59 1:1.18 1:1.50

Sandfly characteristics of the study area

Sandfly prediction based on inside room climate data

Predictor

Variables

Coefficients

(95% CI)

SE(βs) T-statistic

p-value

Intercept -87.26 (-108.88 - -66.33) 10.09 -8.65 <0.0000

Room temperature 1.42 (2.25 – 0.62) 0.39 3.69 0.0012

Room relative humidity (RH)

0.84 (1.13 – 0.56) 0.13 6.19 <0.0000

R2 = 0.80; p-value <0.0000

Relationship between maximum, minimum and mean SAVI value with sandfly density

Wetness Index (WI) map of the Muzaffarpur district

Land Surface Temperature (LST) of the study area, derived from Landsat- 5 TM

Prediction of sandfly based environmental variables derived from remote sensing

technology

Predictor

Variables

Standard Error

(βs) T-statistic p-value

Intercept 6.283 -1.746 0.006

Minimum SAVI 1.986 0.175 2.02 0.050Mean SAVI 3.284 0.228 2.636 0.015

Mean LST 0.198 0.251 1.965 0.013

Minimum WI 0.052 0.423 4.282 0.000

Maximum WI 0.041 0.301 2.825 0.010

R2 = 0.85; p-value<0.001

Land use/land cover (LULC) map of the study area

Association between land use/land cover classes with the presence/absence of

sandfly

VariablesPresence of LULC

classesChi-square test

(X2)P-value

Settlement 10 9.44 0.002Marshy land 16 3.92 0.048Moist fallow 23 0.75 0.384River 2 18.47 0.000Sand 3 6.63 0.010Surface waterbody 16 9.93 0.002Vegetation 9 4.33 0.040Agricultural/crop land

24 0.00 1.00

Discussion and Conclusion

Maximum number of sandfly species are recorded in the month of September-October, whereas, minimum number recorded from the month of January-February

Standard deviation of ellipse shows that disease are distributed from eastern to western direction

Inside room temperature and humidity play an important role for the breeding and propagation of sandfly distribution

The predictive value of this remote sensing map based on LST, SAVI and WI indices data appears to be better for the forecast of the disease risk areas.

Multivariate regression analysis showed that minimum SAVI, Mean SAVI, mean LST, minimum WI, maximum WI highly significant to predict the sandfly density.

Analysis of land use/land cover features revealed that adult sand fly density was significantly associated with land cover variables (e.g., settlement, surface water body, moist fallow, vegetation, sand and river).