From Social Networks to Machine Learning: When Epidemiology Is Going Digital

Can epidemiology, in its quantitative and qualitative dimensions, benefit from the formidable window offered by the Internet and social networks on the population's perception of public health issues and even emergencies? It was on this basis that the speaker at this seminar pioneered what might be called "digital epidemiology". It enables us to detect the emergence of an unusual situation, such as an epidemic, with extreme speed and sensitivity, thanks to the detection of temporal and geographical aggregates of targeted information searches. It also enables us to quickly assess the level of public support or distrust for a public health measure. This was the case for mistrust of vaccination against pandemic influenza A H1N1 in 2009. This approach has its strengths, weaknesses, biases and dangers of data interpretation, but it must undeniably find its place in modern epidemiology, whose anthropological dimension is increasingly asserting itself.