Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Innovation in observation: early outbreak detection
using remote sensing
Elena N. Naumova
Holistics for Health 2013
Tufts University Schools of Engineering and Medicine
August 26-28, 2013
InForMID -Tufts Initiative for the Forecasting and Modeling of Infectious Diseases
Why we need outbreak predictions?
Focus on prevention and community development strategies:
² to minimize damages ² to optimize resource allocation ² to set priorities ² to better tailor environmental and health policies ² to build resilient communities
2
RS for disease early warning
Vector-borne infections: Rift Valley Fever Malaria Dengue
Enteric infections: Cholera Rotavirus Cryptosporidiosis
Jagai JS, Sarkar R, Castronovo D, Kattula D, Ward H, Kang G, Naumova EN. Seasonality of Rotavirus in South Asia: A Meta-Analysis Approach. PLoS ONE. 2012;7(5):e38168. Jagai JS, Castronovo DA, Monchak J, Naumova EN. Seasonality of cryptosporidiosis: a meta-analysis approach. Environmental Research. 2009 May; 109(4):465-78.
Health Alerts
Already available: ² dew point
temperature ² wind chill index ² air quality score ² pollen counts Extreme weather events warnings:
² ice storms ² hurricanes ² tornados
4
Innovations in Outbreak Detection
New concepts: ² Outbreak
signatures ² Combinatorial
decomposition ² Real-time
forecasting ² Data systems
and structures
5 Naumova EN, O'Neil E, MacNeill I. INFERNO: a system for early outbreak detection and signature forecasting. MMWR Morb Mortal Wkly Rep. 2005 Aug 26;54 Suppl:77-83.
Anticipate Detect Alert Alarm
What is needed to detect an outbreak?
Data, Models, Research Integration: more and better high quality = less noise (outcome-specific)
on time = less delay (allow for time to react)
6
RS: higher resolution, higher frequency, and targeted products to enable time-series dynamic disease studies
Anticipate Detect Alert Alarm
Properties: Coverage Specificity Sensitivity Compatibility Comparability Accessibility Availability Completeness Timeliness Spatially and
temporally aligned
Structural missingness
Data Structures Data sources:
Medical care facilities (hospitals, rehabilitation centers, ambulatory clinics, drug dispensaries)
Diagnostic testing facilities (laboratories, mobile diagnostic units)
Social-care facilities (day care, assisted living and nursing homes, hospice services)
Correctional facilities Schools/Work places Locations of intentional
screening
Data streams: Death records Forensic records Laboratory tests Chief complains for
emergency room visits Medical service and
equipment use records Insurance claims Pharmaceutical sales
records Absenteeism reports Hotline calls Website queries Media news clips
Data Structures (cont.) Supportive data:
Demographic profiles Vaccination coverage
records Calendars of social activities Environmental samples Population migration and
displacement patterns Land use, climate, and
meteorological information
Domestic animals and wildlife surveillance
Data curators: Public health institutions Local and regional
departments of vital statistics
National health statistics institutions
Insurance industry Pharmaceutical industry Governmental and
nongovernmental organizations Academic institutions
Fefferman NH, Naumova EN. Innovation in Observation: A Vision for Early Outbreak Detection. Emerging Health Threats. 2010.
How to decide what you need?
time in days
GI r
ate
per 1
00,0
00
0 10 20 30 40 50 60
0
2
4
6
8
turb
idity
(NTU
)
0.5
1
1.5
turbidity
Outbreak period
The time series of daily rate of non-specific gastroenteritis and effluent water turbidity around the outbreak period 03/28/93 - 04/ 24/93.
Naumova EN, Egorov AI, Morris RD, Griffiths JK. The elderly are a sensitive subpopulation for waterborne Cryptosporidium infection: hospitalizations for gastroenteritis in the elderly before and during the 1993 Milwaukee cryptosporidiosis outbreak. EID. 2003. 9(4):418-25.
Data streams: Hospital records Data source: City Hospital Data curator: Billing office Supporting Data: Census, MWRA
Outbreak of cryptosporidiosis in Milwaukee, WI
Outbreak detection: a time series approach
…these models have an underlying assumption of stationarity... They are not well-suited for data exhibiting strong asymmetry or sudden bursts of very large amplitude at irregular time points.
Success = f(A3, C3, E3) A3: Aligned, Available, Accessible C3: Clean, Complete, Comparable E3: Evidence-based, Efficient,
Effective
Research Integration: more and better
² Different implications for
different regions or subpopulations
² When this distinction has been ignored, the quality of prediction is in jeopardy for over or under estimation.
13
Disease Rate vs Disease Risk
Risk = hazard × vulnerability = P(negative consequences) × V[f(exposure, susceptibility, coping capacity)]
low high lo
w
hig
h
RATE
RIS
K
What should be done?
² Translational Research: move from research lab to operational settings; assess the utility, effectiveness, and cost-efficiency; have a sense of required investment
² Community Participation: evaluate the potential for behavioral change and develop strategies for building resilient communities
14
low high
low
high
RATE
RIS
K
Where to invest?
Focus on prevention and community development strategies and select health outcomes:
² Burden is high ² Preventable ² Seasonal
Data-to-Knowledge Discovery Disruptive technologies that share four important
characteristics: ² high rate of technology change, ² broad potential scope of impact, ² large economic value that could be affected, ² substantial potential for disruptive economic impact.
15
Methodology: more and better
² Outbreak definitions and signatures by: ² pathogen ² source/mode/dose of exposure ² affected population
² Amplification effects and environmental triggers ² Optimized targeted monitoring ² Predictive mathematical, statistical, computational
modeling ² Simulation suites ² Games for Health ² Interdisciplinary vision
16
Relevant Publications 1. Fefferman NH, Naumova EN. Innovation in Observation: A Vision for Early Outbreak
Detection. Emerging Health Threats. 2010. 2. O’Neill EA, Naumova EN. Defining Outbreak: Breaking out of the confusion. JPHP.
2007, Dec; 28(4):442-55. 3. Naumova EN, O'Neil E, MacNeill I. INFERNO: a system for early outbreak detection
and signature forecasting. MMWR Morb Mortal Wkly Rep. 2005 Aug 26;54 Suppl:77-83. 4. Fefferman NH, Naumova EN. Combinatorial decomposition of an outbreak signature.
Math Biosciences. 2006. 202(2):269-87. 5. Jagai JS, Sarkar R, Castronovo D, Kattula D, Ward H, Kang G, Naumova EN. Seasonality
of Rotavirus in South Asia: A Meta-Analysis Approach. PLoS ONE. 2012;7(5):e38168. doi: 10.1371/journal.pone.0038168.
6. Wenger JB, Naumova EN. Seasonal synchronization of influenza in the United States older adult population. PLoS ONE. 2010 (on-line)
7. Naumova EN, MacNeill IB. Seasonality assessment for biosurveillance systems. In: Advances in Statistical Methods for the Health Sciences: Applications to Cancer and AIDS Studies, Genome Sequence Analysis, and Survival Analysis. Edited by N. Balakrishnan, Jean-Louis Auget, M. Mesbah, Geert Molenberg. Birkhauser, Boston. 2006; (pp. 437-450)
8. Lofgren E, Fefferman NH, Naumov YN, Gorski J, Naumova EN. Influenza seasonality: underlying causes and modeling theories review. Journal of Virology. 2007; 81(11): 5429-36.
9. Naumova EN, Jagai J, Matyas B, DeMaria A, MacNeill IB, Griffiths JK. Seasonality in six enterically transmitted diseases and ambient temperature. Epidemiology & Infections. 2007. 135(2):281-92.
10. Lofgren E, Fefferman NH, Doshi M, Naumova EN. Assessing seasonal variations in multisource surveillance data: annual harmonic regression. D. Zeng et al. (eds.): BioSurveillance 2007. Lectures Notes in Computer Science (LNCS) 4506, Springer. 2007; pp. 114-123.
11. Lofgren ET, Wenger JB, Fefferman NH, Bina D, Gradis S, Bhattacharyya S, Naumov YN, Gorski J, Naumova EN. Disproportional effects in populations of concern for pandemic influenza: insights from seasonal epidemics in Wisconsin, 1967-2004. Influenza and Other Respiratory Viruses. 2010; 4(4): 205-12 [Epub ahead of print]
12. Castronovo DA, Chui K, Naumova EN. Visualization of spatio-temporal disease
patterns with dynamic maps. Environmental Health. 2009. (on-line)
Acknowledgements Collaborators & Former students: Magaly Koch (BU), Irene Bosch (MIT), Alexander Liss Nina Fefferman (Rutgers) Denise Castronovo (Mapping Sustainability, Inc USA) Kenneth Chui, Steve Cohen, Julia Wenger, Jesse McEntee, Jeneius Aminawung (Tufts University) Funding Source NIH-NIAID, Grant # NO1AI50032 NIH-NIAID, Grant # U19AI062627 Data Sources Centers for Medicare and Medicaid Services (CMS) MA PHD