Last week, the German government announced – together with the WHO – the opening of a new Hub for Pandemic and Epidemic Intelligence in Berlin. Funded with 100 million USD, the hub promises to protect and prepare in ways which global institutions have failed through the Covid-19 pandemic of 2019 to 2021. At the heart of this initiative sits data science. Much of the hub’s vision appears focused on data linkage to enable an ambitious program of surveillance, prediction and anticipation.There is little to learn at this stage about the exact form of the Berlin hub, even less to say about what its scientific profile will eventually look like and what we might expect from its director, Dr Chikwe Ihekweazu. Tedros is cited in the press release as envisioning the hub to leverage “innovations in data science for public health surveillance and response,” while Angela Merkel hopes that the hub will help the world to be “better prepared for future epidemics and pandemics.” The hub’s official mission statement also points to the collection of data, which will be “vital to generating signals and insights on disease emergence, evolution and impact”, seeking to develop state-of-the-art analytic tools and design predictive models, to be shared with the world. From the outset, the hub’s vision presumes, like others before it, that the challenges of covid-19 have been merely technological in nature. But do we really need another ‘disruption’ of epidemiology?
Sure, recent developments in data science and AI hold the potential to improve aspects of epidemiological data collection and analysis, but epidemiology has not and will never be a purely numeric exercise. Epidemics are, after all, not governed solely by mechanical forces. While the HUB is not the only global epidemiological initiative emerging at this normalisation phase of the pandemic, it still comes as a surprise that such a narrowly conceived data project presents itself as the WHO’s potential means of resolution, and the culmination of presumed lessons learned from this pandemic.
Or how to keep Global Health stuck in a dated biosecurity model that is unable & unwilling to approach epidemics as complex social & ecological phenomena. Time for a paradigm shift – Time to think beyond "data" https://t.co/GsEDYPo6aD
— Visual Plague/ Global War Against the Rat 😷 (@visualplague) September 1, 2021
A commentary in Lancet Microbe was already raising concerns in early July, with Kevin Bardosch in particular emphasising an obvious mismatch between the WHO’s mission statements and the apparent lack of local expertise and social science in this refresh of epidemic intelligence:
“Moreover, numeric methods to obtain epidemiological data are just one part of countering a pandemic, and qualitative research must also be included, argued Voss. According to public health researcher Kevin Bardosh (University of Washington, Seattle, WA, USA), WHO has always claimed that “everything is about community engagement”, but its health emergency programme had only one or two social scientists. “The COVID-19 pandemic has completely challenged the field of epidemic social science and I think the hub would be uniquely placed to facilitate a serious discussion about how social sciences can best contribute to epidemic [or] pandemic intelligence”, Bardosh told The Lancet Microbe.” (The Lancet Microbe)
The relationship between data science and epidemiology is a key concern of our Epidemy project. Data science (or: Data Science) has been around for a while as an integral element of epidemiological reasoning, and long before the SARS-Cov-2 virus emerged. Indeed, many of the principles and values of contemporary data science have been developed with dedicated consideration of epidemics, contagious phenomena and medical problems. Thinking of epidemiology as a data-driven science is as old as considering the field of statistics to be a tool for governance and colonial control. However, with the newfound scientific status of a data science in the world of big data, machine learning and artificial intelligence, the very old struggle to define the scientific nature of epidemiological reasoning reignited. In 2019, a paper in Nature Communications had already proposed the reallocation of epidemiological modelling and data analytics into a new field of “outbreak science.” The apparently blunt tools of the social sciences and historical scholarship had no place in this shiny new endeavour of resolving epidemiological challenges through the proposition of modelling and mathematics. But, perhaps more importantly, epidemiological data science and modelling were pillars of the very biosecurity model that had failed devastatingly to predict and prepare for Covid-19 in 2019.
How, then, is the resurgence of epidemiological intelligence, repackaged as ‘data science’ or ‘outbreak science’ justified? In the opening ceremony of the Berlin hub, multiple speakers made reference to Berlin’s history of epidemiology and cited Rudolf Virchow as a pioneering figure and as a towering legacy for the new hub. There is much to say (and to complain) about the crude appropriation of Virchow as the father of data-driven approaches to epidemiology. I was left wondering how Virchow’s vital contributions to social medicine stand to be reflected in this hub. Perhaps this will be the place where a systematic consideration of the social determinants of health is no longer kept at a distance from regressive models of infection dynamics. Perhaps, with any luck, the hub will mend the long-standing schism between social and formal epidemiology, between card carrying epidemiologists and modellers. The hub might even be able finally to enrich the impoverished skeletons of “the social” as delivered by the behavioural sciences, and return to local expertise, longstanding scholarship of medical anthropologists, sociologists, historians and philosophers to understand what epidemics mean and what they are ‘on the ground’. Might this be the place where that other contribution of Virchow’s to the history of epidemiology is taken seriously and where politics are redefined as “medicine on a grand scale”? If the hub wants to address these pressing challenges – and it might be well placed to begin this work – it would need to shed the image of providing technical solutions to non-technical problems as soon as possible, and it must invest substantially in transparency as well as in epistemological diversity.
The announcement of the hub certainly raises important questions for the history of epidemiological reasoning. But rather than dig out Virchow and forget about his opposition to germ theory, those involved in the hub would be well-advised to re-examine the history of the specific tradition to which the hub is nominally dedicated: epidemiological intelligence. This vexing concept emerged out of the Second World War, pioneered in the US by Alexander Langmuir and driven by the threat of biological warfare. But it was largely conceived as an epidemiological training program, giving credence and personnel to the newly-formed Centers for Disease Control (CDC). And yes, while data gathering and analysis played a part in the intelligence provided by the service, Langmuir saw its fundamental purpose in the training of rounded epidemiologists, comfortable in field research and well-acquainted to the political “intricacies of the Public Health Service.”
The history of epidemiological intelligence is of course broader than these snippets can do justice. There is still much to be done to understand the specific historical contexts under which part of epidemiology sought to align itself with secret services rather than with academic endeavours. It’s great to see new research developing on these questions. It might be helpful in the meantime to dedicate more thought to the history and present of intelligence services at large. Over recent weeks one phrase has been reiterated in the news to defend the failure of Western services predicting the power grab of the Taliban: “Intelligence is not a science.” Rather, so the chorus of commentators argued here in the UK, what intelligence services do is gather information on the ground and try to make sense of it on the basis of what is known and what needs to be known. Far from assuming that intelligence services should be a model for the production of better epidemiology, one wonders which direction the hub will choose and what kind of intelligence its data-driven science will produce, with and for whom.
Featured image: Angela Merkel and Tedros Adhanom Ghebreyesus opening the WHO hub. Credit: German Federal Government/Denzel