Automatic studies of epidemics have up to now concentrated on the analysis and prediction of current outbreaks. In contrast, the RussianFlu project brings together researchers from Computer Science and History to examine a historic influenza epidemic, using Digital Humanities methods. The project especially looks at the worldwide spread of the epidemic and the perception towards it in the US and Germany/Austria, facilitated by the close collaboration of researchers at Virginia Tech and the L3S funded by the NEH-DFG initiative on Digital Humanities. It is of relevance to humanities scholars seeking new ways to understand popular and scientific perceptions of disease, epidemiologists studying the spread of infectious epidemics in global contexts and data analysts seeking to track and predict the spread of information about disease outbreaks.
This project examines US and German medical discussion and popular reporting during the Russian influenza epidemic, from its outbreak in late 1889 through the successive waves that lasted through 1893. The project will use historical materials to develop, apply, and evaluate new methods for computational epidemiology through applications such as word and term distribution analysis, fact extraction, sentiment analysis, network analysis and data visualization. The Russian Influenza is an especially appropriate case study for an approach that integrates the digital humanities and computational analysis. With the establishment of the global telegraph network, for the first time in world history, news about a disease could spread across long distances faster than the disease itself, which was limited by the speed of human travel. At this same time, medical discoveries were transforming both scholarly and public opinion about disease origins, transmission, and prevention. In this context of relative international calm, at least among the great powers, transnational communication, particularly between the United States and Germany, was facilitated by both the increased speed of electronic communications and a shared perception of the advantages of sharing scholarly insights. Finally, the Russian influenza is an excellent case study because although it had a relatively low mortality rate, it spread quickly and infected high proportions of the population in each region it reached, thus allowing for mapping of the spread of disease using popular and medical reporting.
Challenges and Highlights
In RussianFlu, the following main research questions have been identified:
- How does the tone of reporting during a disease outbreak change in relation to variables such as proximity to reporting location, number of cases, categories of victims, and accumulating deaths? We will use up-do-date NLP methods in fact extraction and sentiment analysis and adapt them to historic medical and newspaper texts to compare factual reporting to sentiment.
- How did newspapers and medical journals contribute to the narrative of the Russian flu, including therecognition of an outbreak, involvement of medical experts, attention to celebrity victims, the effort to shape public opinion, scope of opinions, and the response of authorities? As part of this research question, we are going to build social networks for the scientific communities working on the Russian flu and explore the influence of these communities on popular reporting.
- How accurate were predictions about the scope, impact, and significance of the Russian flu at distinct stages, by comparison to epidemiological data reported during and after the outbreak? The data sources include both popular newspapers and medical journals from the United States, Germany, Austria, Switzerland, and Russia, in both English and German languages. Digitized newspapers allow for tracking of the disease as it spreads, as well as evidence of the ways that expert medical knowledge was disseminated to public audiences. Digitized medical journals make it possible for computational methods to be applied to detailed reports about disease symptoms, public health responses, and transmission patterns. This project is thus unique among digital humanities projects by bringing together two distinctive approaches: first, the integration of popular newspaper reporting and expert medical analysis of the same disease outbreak, and second, developing analytical tools for source materials in two languages (English and German) to illustrate the nature of the transnational medical dialogue that also engaged with popular reporting on a global scale.