Paper short abstract:

'Digital Epidemiology' is the algorithmic analysis of social media data rather than of cases reported through medical and public institutions. This paper discusses the significant political and epistemological implications of this digital disruption of public health.

Paper long abstract:

'Digital Epidemiology' is the algorithmic analysis and digital visualisation of previously untapped data from social media, internet search terms and access-logs. This new kind of computational epidemiology is supposed to disrupt conventional epidemiological measurements of health and disease. Significance of medical data is no longer ultimately determined by a doctor's diagnosis but allocated to models, theories and semantic ontologies. This considerable expansion of the production, gathering and structuring of health-related data is purposefully placed outside of the traditional institutions of governmental health surveillance and usually to be found in independent research labs, start-up incubators or in the digital industry.

Using semi-structured interviews with digital epidemiologists and utilising a historical revision of formal epidemiology this paper will address two urgent questions the digital disruption of public health provokes:

First, a political question addresses the shifting accountability of a 'start-up epidemiology' in public health. While the promise of turning almost any data source into medically relevant information escapes and undermines the rationale of traditional government-led health surveillance, new constellations emerge, in which questions of austerity, resource scarcity and digital innovation align.

Second, an epistemological question focuses on the increasing significance of models and algorithms in the visualisation of health trends, disease transmission routes or treatment allocation guidelines. While the sheer volume of new empirical digital data creates a new optimism about the 'end of theory' in epidemiology, this paper follows the curious resilience and persistence of traditional concepts and models, applied in epidemiology for almost a hundred years (Morabia 2004).