KOL discovery using social network analysis
Understanding social network analysis
Social networks consist of nodes, which represent actors in the social network, and edges connecting the nodes. Each edge stands for a kind of relationship between the nodes it connects
[Newman, 2000]. Clearly, there may be many different kinds of relationships, ranging from personal acquaintance to membership in the same association.
Social networks are being studied in social sciences for more than the past 50 years, variations on this general concept are being applied in various industries like social media, movies, operations, etc. to identify key actors or clusters (communities) of experts [Catanese et al., 2011, Ding and Yilmaz, 2010, Kima et al., 2011]. In the pharmaceutical industry, social network analysis is being used to identify KOLs. The broad and flexible concept of social networks lends itself to represent the deep and diverse data landscape of the pharmaceutical industry.
Despite being known for a long time, the methods of social network analysis are applicable to large networks only through recent advances in technology, and are able to uncover insights which are not feasible through human effort. For example, people working a lifetime in a scientific area might recognize the top 10, 20 or even 50 leading researchers for their community. However, not only will the answers be inconsistent across multiple researchers due to subjective biases, but also the fast pace of research will impose constant change on the collaboration landscape.
Automated social network analysis in contrast not only allows a researcher to identify key HCPs, but compare and rank them over time. Also, social network analysis of HCP networks avoids the biases introduced by subjective views by analyzing large quantities of objective data. In fact, social network analysis relies on social principles. For example, experts in a field prefer to work with other experts [Newman and Park, 2003].
A social network mapping all such collaborations, therefore, provides rich contextual information [Newman, 2004]. By graphing these collaborative relationships over multiple years, a very clear picture of the research community in a specific therapeutic area emerges. The patterns in this social network are not a result of random processes. Rather, the key is that the people in the research community make active choices. The aggregation of these conscious individual decisions is what creates network patterns which can be analyzed and utilized.