Global Risk Informatics Microsoft / Gates Foundation Debra
Goldfarb Sr. Director, Technical Computing Industry Strategy
Slide 2
The crisis information gap When the global economic crisis hit
in 2008, world leaders knew they needed to act quickly. They knew
that they needed to take immediate policy actions to protect
communities from downstream impacts on health, nutrition,
education, jobs, and the environment. Agile, targeted responses
required up-to-date evidence of how families were coping with
shocks. Sounds pretty straightforward, no?
Slide 3
Household-level stats take months to collect, and years to
validate!
Slide 4
The information gap is real ? First data becomes available
Slide 5
as are its consequences.
Slide 6
Decision makers had access to real-time data and the tools to
detect the early signals ? Policy-makers and field workers had
models to help uncover the complexities of disease, economic
crises, poverty, civil unrest? We could tailor interventions based
on real data and analysis? We could broadly apply simulation and
modelling to global risk to dramatically change outcomes? But what
if?
Slide 7
Microsoft Gates Foundation Collaboration What are we doing? Why
we care? What will we learn? What are the impacts? How does it
fit?
Slide 8
Guided by the belief that every life has equal value, the Bill
& Melinda Gates Foundation works to help all people lead
healthy, productive lives. In developing countries, it focuses on
improving peoples health and giving them the chance to lift
themselves out of hunger and extreme poverty. In the United States,
it seeks to ensure that all peopleespecially those with the fewest
resourceshave access to the opportunities they need to succeed in
school and life. The Foundation focuses primarily on the bottom 20
The Bill and Melinda Gates Foundation
Slide 9
Slide 10
Malaria today Malaria Burden -2008 863 000 deaths 243 million
cases Half of the world's population is at risk of malaria
Slide 11
Current solution Tools Current: LLINs, IRS, ACTs, accurate
diagnostics Future: vaccine, vector compromise, surveillance tools
Strategies for human behavior change Improve the health systems
infrastructure Economic development Understand climate change
impacts
Slide 12
What motivates the GF? The Goal: Eradication Removal/depletion
of the last malaria parasite on the earth Its been done before:
Smallpox, Rinderpest Guinea Worm, Polio, Measles
Ambiguities/challenges Syndrome vs single disease Animal
reservoirs? Latent infections
Slide 13
Malaria modeling: why technical and high performance computing?
To predict the impact of a particular intervention To explore the
modes of action of specific tools To evaluate transmission patterns
and efforts to reduce them To explore economic and public health
arguments for particular eradication strategies To simulate
approaches to eradication and explore options for achieving it
Slide 14
Malaria Models Transmission models Ross McDonald (transmission)
R 0: The number of new infections that arise from a single one
Within-host models Immunity: partial protection in adult humans who
survive infancy Population models Parasite drug resistance or
insecticide resistance in mosquitoes and then you add in all the
parameters and sub models: biology, climate, human population
models, environmental, technology, complex relationships, food,
etc.
Slide 15
Modern Malaria Models Modern range Simple ODE models
Multiparametric MCMC Simulations Novel modeling approaches Nested
hierarchical models Computational/statistical innovations Network
models of human movement Different assumptions about underlying
biology
Slide 16
Proposed analytical framework incorporates multiple information
sets, enables assessment of vector control interventions
Integration of community inputs into unified framework
Identification of gaps in current intervention set as informant of
TPPs Analytical tools Identification of critical data gaps
Assessment of utility of potential VC interventions Assembly of
regional vector ecology profiles Local environments
Location-specific stratifications and data Intervention profiles,
incl. efficacy and resistance Interventions Malaria parasite
locations, rates Epidemiology Vector species ecology profiles and
ranges Entomology 1 Regulations, policies, financing Policies and
regulations 23 4 Second-wave input Supply, demand and financing
assessment Second-wave output
Slide 17
Analytical framework will capture four key types of data
Primary data components Secondary components (used to expand and/or
refine framework) Key sources for data Aggregate vector species
information Entomology 1 List of reproductively isolated vector
groups Vector ecology profiles (biting, resting, breeding sites,
sugar meal source) Vector presence coordinates Expert-derived
vector ranges Emergence of new species Mating and swarm behavior
Species genomic data Malaria Atlas Project (MAP) Disease Vector
Database Swiss Tropical Institute / MARA Walter Reed Biosystematics
Unit VectorBase / Anobase Consolidate multiple location- based
variables Local Environments 2 Political map Precipitation Human
density estimates Climate Topography Local resistance to active
ingredients Availability of alternative interventions (e.g., drugs,
vaccines) Climate change impact Human development impact Urban,
rural, agriculture stratifications Cost constraints
Infrastructure/accessibility Socio-political obstructions Relevant
cultural mores Use patterns for alt. interventions WHO MAP CIA
Factbook Koppen-Geiger Climate Classification SEDAC (GRUMP) Map
against malaria outbreak data (location, rate) Epidemiology Overlay
intervention profiles, including efficacy info. Interventions 34
Parasite rates and coordinates Expert-derived epidemiological
ranges Impact of human migration patterns Actual disease burden
Human and vector host resistance Malaria Atlas Project (MAP) WHO
Swiss Tropical Institute CDC Classified list of interventions 1
Efficacy and effectiveness Compliance Cost Impact of educational
efforts Ecological influences on intervention efficacy WHO Croplife
IVM evidence committee STI Vestergaard-Frandsen Academic literature
Expert input 1. Interventions to be classified by control paradigm,
target vector age, active ingredient(s), number of active
ingredients, safety, development status and robustness against
pyrethroid- resistant vectors WHOAFPMBANVR
Slide 18
Framework inputsIntermediate outputsEnd-user tools
Interventions Epidemiology Local Environments Integrated
epidemiological & vector species datasets / maps Vector species
datasets / maps Vector locations Location-specific boundaries &
data Stratification map Epidemiological map Intervention
effectiveness Integrated epidemiological & entomological
datasets / maps Profiles of current interventions Comprehensive
vector ecologies WHO, Academic lit., STI, Expert input MAP, GRUMP
WHO, Academic lit., Vestergaard- Frandsen MAP, DVD, Academic lit.,
Expert ranges MAP, WRBU, DVD, STI MAP, WHO, STI MAP Parasite
epidemiology MAP, Academic lit., Expert input Multiple data sets to
be combined and integrated Intervention utility map Data gaps
Intervention gap assessment Regional Vector Ecology Profiles MAP
Parasite rates and coordinates Expert-derived epidem. ranges Vector
ecology profiles MAP, WRBU, STI List of reproduct. isolated groups
DVD, MAP., STI Vector presence coordinates DVD, MAP, Academic lit.
Expert-derived vector ranges MAP, Expert input MAP, Academic lit.
MAP, Expert input List of interventions WHO, STI, Expert input,
academic literature Precipitation Political map Hum. population
NASA; MAP GRUMP MAP Local resistance to AIs Academic lit.,
Vestergaard-Frandsen, Altern. interven. WHO, Academic lit. Climate
NASA; MAP Topography MAP Intervention efficacy WHO, STI, academic
literature Expert input Entomology Searchable database and vector
or location- specific datasets Visual maps Searchable database and
vector or location- specific datasets Visual maps Searchable
database and vector or location- specific datasets Visual maps
MAP
Slide 19
What are we doing? VCDN consortia member Develop the cyber
infrastructure, applications and tools to enable broad-based
sharing of Malaria data and models; simulation and analysis to
drive positive and predictive outcomes Components: cloud-based
large scale data integration, collaborative tools, extraction/
modeling/analytic tools, visualization, GIS- mapping, search,
simulation and modeling
Slide 20
Challenges Data: integrity, formats, ontologies, currency and
curation, security.not to mention the politics of data
Collaboration: data owners dont always play nice Technology +
policy = impacts We are in unchartered territory.
Slide 21
Where do we go from here? UNSD NGO/IGO WHO UN/Glob al Pulse GF
at scale Public Health Extreme Scale Informati on Exhaust Data
Global view for Health