Modelling Science

Recent advances in the fields of scientometrics, network analysis and visualization open up new perspectives on the organization and the evolution of the science system. Most of the available evidence of semantic and social networks of science reveal a skew connectivity distribution and a core-periphery structure. From a dynamic point of view, a dense and stable core of general concepts and leading scientists obscure a highly instable fringe of new and emerging concepts and scholars. For technological foresight purposes, we need to identify new tools that allow to single out the pattern of growth and weak ties among peripherical concepts. In the filed of biomedical sciences, rare diseases represent a classical example of peripherical phenomena. According to European definition, a rare disease is a serious or lethal illness that affected less than one person over 2000 individuals. Despite rarity, five new rare diseases are discovered each week and according to recent estimates not less than 30 million European citizens
suffer from them. Health care institution have designed special policies to support patients and research in this field. Despite growing attention, in 8.284 cases out of 9.471 rare diseases (87 per cent) there is no information in publicly available dataset (Medline, USPTO, EPO, WIPO, EMEA; FDA, orpha.net, rarediseases.gov) and no unifying knowledge base.
In this study, in order to contribute to filling this gap, we develop a thesaurus of rare diseases and we extract from pubmed more than 4 million publications about rare diseases published by scientists working in the U.S., Japan, Germany, France, U.K. and Italy since 1984. We analyze the dynamics of semantic and social networks in this field by means of a newly developed software called DyNet searching for regularities in the genesis of R&D on newly identified diseases. We find that in some cases, new diseases speciate out from a diffused diseases while in other cases, they are
defined genetically or symptomatically through similarities and differences with widely known pathologies. To better understand the emergence of novel diseases, we complement this analysis to four in-depth case studies: Alzheimer, Gaucher, RATT, LEOPARD diseases. The evolution of Alzheimer and Gaucher diseases is seen in retrospect through archival primary and secondary data, from their identification as rare and new diseases to their recognition as widely diffused. RATT and LEOPARD diseases allow to focus on the recognition process of rarity through direct data collection.