Transcript Big Data for Disease Control Interdisciplinary approaches to data

Big Data for Disease Control
Interdisciplinary approaches to data linkage and management
Shona Jane Lee
Investigating Networks of Zoonosis Innovation
Centre for African Studies
• Intro to ‘Big Data’ in the context of infectious
disease control
• Achieving interdisciplinarity and data linkage across
disciplines and sectors - a One Health approach
• African Trypanosomiasis (AT) and the challenge of
data management for Neglected Zoonotic Diseases
• My research – tracing technological innovations and
data linkage and management for AT surveillance
and control in rural Uganda
What is ‘Big Data’?
• ‘Big Data’ generally understood in terms of its
‘volume, variety and velocity’, however size isn’t
everything…
• More about the methods of analysis, and
fundamentally networked nature of the data in
question – its ‘relationality’ to other data
• Value derived from patterns and connections that
can be drawn between different sources, and the
fresh insights this can reveal through novel methods
of analysis
Big Data for
Global Health & Development
•
UN’s High Level Panel 2013: calls for a ‘Data revolution’ to
achieve Post-2015 MDG targets – highlights importance of good
data for effective policy
•
Measuring impact - Allows comparisons to be made (for example
between prevalence rates of disease prior to and after intervention
campaigns).
•
Warning! Data may be plentiful but is often:
•
•
•
•
Unstructured
Collected with varying methods and units of measurement
Inaccessible
Analysed differently between groups or individuals
Can lead to misleading interpretations…
An opportunity for disease control?
• Challenges for infectious disease surveillance and control
(particularly zoonoses) surround the complex forces:
• Bringing together large datasets collected by multiple
stakeholders and disciplines with various objectives and ways
of collecting/managing data (vector control, livestock, public
health, pharma, parasitology, epidemiology, social science
etc.).
• Challenges associated with poor data collection and
management in developing countries (which predominantly
shoulder the burden of such diseases)
• Poor indicators and disputed methods of analysis among
analysts
One Health
‘One Health’ paradigm
•
Interdependency of human,
animal and ecosystem
health
•
“addressing NZDs requires
collaborative, crosssectoral efforts of human
and animal health systems
and a multidisciplinary
approach that considers the
complexities of the eco
systems where humans and
animals coexist” - WHO
Okello, et al. 2014
One Health
• One Health built on a premise of collaboration,
interdisciplinarity and linkage between sectors
• Big Data works on the premise of linking large data sets from
different sources and drawing patterns from these links
BUT
• Beyond being openly available, data needs to be aggregated,
standardised, compatible with and comparable to other data,
and analysed within the context of other data.
• Challenge: collaborative networks between veterinary, medical
and environmental disciplines remain compartmentalised, and
constrained by the inflexibility of its actors
African Trypanosomiasis
a.k.a ‘Sleeping Sickness’ –
Human African Trypanosomiasis
• Two strains:
• T.b. gambiense – chronic, transmitted between humans
via tsetse fly bite, mainly in West and Central Africa
• T.b. rhodesiense – acute form, zoonotic – transmitted
from animals (usually domestic livestock) to humans
via tsetse fly bite, limted to East Africa
Uganda the only country where both
strains are endemic
Mapping
•
•
•
•
•
Terrain
Vector distribution & density
Prevalence
Infrastructure
Demographic Factors
Modelling
Modelling
Modelling
Modelling
Challenges
Modelling
Challenges
• Modelling
Challenge
•
How to translate this complex evidence base to others in your
field?
•
How do you translate it to those outwith your own field? 
Necessary for an integrated approach
•
How do you translate it into actionable Policy??!
•
Sometimes the data requires streamlining, simplifying, or
analysing in different ways in order to produce something
manageable and achievable for policy – balancing act for analysts.
•
“you put garbage in, you get garbage out”…
•
Need high quality, up to date data
Dealing with Data in the
Absence of Active Screening
mHealth for One Health?
“provide access to further professional veterinary assistance in local areas, improve herd
management and the ability to share data with key stakeholders” – Cojengo, 2014
Case Study:
HAT control in Uganda
HAT poses several challenges to data collection/management and therefore
policy for control:
•
Different strains, requires different approaches and collaboration of more
stakeholders (veterinary, parasitology, medicine, pharma, vector control
etc.)
•
Poor livestock data (in terms of numbers, movement and disease
prevalence)
•
Poor human prevalence data owing to passive surveillance approach, poor
education among health staff on SS and under-reporting
•
Scant data on tsetse fly population density and distribution
• Poor data management and ICT infrastructure
Case Study:
HAT control in Uganda
My study will examine several questions, the following of which relate
particularly to innovative approaches to data management:
•
How are technological innovations in disease surveillance adopted and
used by practitioners?
• Ethnography of technology acceptance among veterinary health workers trialling
Cojengo’s Vet Africa tool
• Mapping and network analysis of data linkage across health network
• Assess feasibility of introducing mHealth approach into this system to improve data
linkage and management to enhance passive surveillance and better improve efforts
to map and model disease risk
•
How successful has the One Health framework been in facilitating
interdisciplinary approaches and taking advantage of big data for
Trypanosomiasis control?
• SNA of data and resourcing pooling between disciplines and sectors
Records from Bugiri Hospital Sleeping Sickness treatment centre
(Berrang-Ford et al., 2006)
Questions?