e-Labs: enabling collaboration between health and social research

e-Labs are secure, on-line environments that connect data, research methods and investigators to enable timely, comprehensive insights. The University of Manchester is developing a range of prototype e-Labs, including the ESRC-supported Obesity e-Lab (ObE) (www.obesityelab.org.uk).

What is obesity e-Lab (ObE)?
ObE is working to find more effective ways to share expertise and data between social and health sciences. It uses obesity research in academia and the public health service as a case study of slow, fragmented research that could be accelerated and expanded using e-Labs. An initial finding is that the Health Surveys for England (HSE) are under-used, largely because of the difficulty in extracting data relevant to specific questions. Similar problems have been addressed in the life sciences through workflow-sharing technologies, and ObE is therefore collaborating with the myExperiment project (www.myexperiment.org) to produce a prototype e-Lab that simplifies access to HSE and similar datasets.

What are research objects?
The basic currency of an e-Lab is the research object: a package of ingredients for reproducing an answer to a question and sharing it. A research object might contain an extract of data, the database query used to extract the data, statistical scripts (for data cleaning, deriving variables and modelling), reports, slides, and so on. The extent to which a research object is shared with others is the choice of the user, much like the trust model in networks like Facebook.

Why is e-Lab important for social researchers?
Social researchers form a particularly diverse, multi-disciplinary community: different researchers differ also in their degree of familiarity with data sources, their expertise with research methods, and the knowledge underpinning their interpretation of findings. This is a strength when researchers work together, leading to richer insights, but a weakness when they work in isolation.

For example, in the systems map for the UK Foresight Obesity project (www.shiftn.com/Obesity/Full-Map.html), the organising principle is energy balance. However, evidence on social determinants of obesity tends to be isolated to over-eating (energy in) or physical activity (energy out), when the key outcome is energy balance. Given easier access to data, methods, and a diverse community of researchers via e-Labs, the true complexity of problems like obesity might better be reflected in research, which in turn leads to outcomes more relevant to policy.

A typical use case for an e-Lab?
Say a social researcher wants to explore relationships between socio-economic status, gender, obesity, and geographic location using HSE data. By querying the e-Lab, she finds that a health researcher has asked a similar question and shared their analysis as a research object: the health researcher has commented on the potential bias of occupation on Body Mass Index (BMI) reflecting muscle rather than fat mass. The social researcher clones this research object, extends the analysis and adds detailed metadata on the strengths and weaknesses of educational attainment vs. household income as a relevant socio-economic measure. Thus by sharing this research object, the social researcher has lowered the barrier for the health researcher to use and understand existing research and data.

Organisational value
e-Labs can also increase organisational memory: for example, a new recruit would inherit a set of research objects showing how information was arrived at, rather than a set of reports that provide only a rough notion of the data and methods behind the information. In addition, a new postholder could quickly tap into a community of social researchers linked to their research objects. As the uses of data are more explicit when they are set in research objects, organisations could improve information governance and could apply more appropriate security models around sensitive data and information. In addition, with easier sharing and reproducibility of data uses, learning and professional development opportunities would be easier to identify.