How social networks predict epidemics
In class, we have been talking about epidemics and its spread in a network. However, we don’t know whether we can detect the epidemic before the CDC learns about its nature. This is what Nicholas Christakis, a sociologist and physician, studied through the help of social networks, defined as the group of connections people make in real life, not online. With the ordinary S-curve as the rate of the epidemic, Christakis inquires how we can rapidly identify an epidemic occurring.
In his research, Christakis used a network of 105 people with relationships, and utilized the structure of the network to identify “sensors”, people who have a lot of connections and are central in the structure. Information or germs would be contracted by these central sensors and their connections more quickly than a group of random people. In his research, the epidemic was identified in the sensor group 40 minutes quicker than the random group.
Christakis also realized the problem that mapping social networks and sensors is not easy, but they exploited the fact that friends nominated by random people have a higher degree and are more central, and doing this for every node gets you closer to the “party host”, the more popular person who can act as a sensor. The friends of the sensor, therefore, can be followed to see if their sensor and themselves are infected, and if so, then an epidemic is occurring. Christakis concluded with the fact that epidemics take root and affect the central population firstly, and with that insight we can detect the epidemic far earlier and improve human well-being.
Nicholas Christakis: How social networks predict epidemics