Social Networks and Infectious Disease...

Dr. James Holland Jones of Stanford gave a presentation yesterday at my school and I dragged my girlfriend to it.  Her eyes glazed over after the first couple minutes because she's seriously not into social networks or disease, but it was fascinating.

What he presented on was a combination of simulation and modeling work and an interesting empirical study looking at social structure within a school.  Primarily, his team applies graph theory as a tool to understand social networks as a basis for looking at disease transmission. 

Different types of behaviors influence the effective form of social networks in a way that influences disease transmission.  For example, the social mores connected to sexual behavior often promote a dendritic or long-chain network while other sorts of behavior can create clustering.

The differences in these patterns influence the transmission of disease, especially when one compares the degree of modularity.  When a population is tightly clustered with most connections being between people in the same cluster vs. people outside their cluster, disease transmission measured in total infected at a particular time looks "lumpy" as the infection passes quickly inside clusters and slower between them.  When the proportion of internal connections to out-of-cluster connections is higher, the infection spreads more quickly and is smoother with a higher portion of the overall population getting infected.

The theoretical background also looked at individual person/node traits like degree (number of connections from one node to others) and the potential effect of relationship/connection cost. 

The empirical study primarily looked at whether people follow one or more of the theoretical models in forming social groups.  The study used "motes" or small sensors worn on the front of the body that could detect the signals of other motes at very short, conversational ranges.  Essentially, because the mass of the human body blocks the signals, whenever the motes "ping" every 20 seconds and detect another mote, you essentially have a face-to-face interaction.  For disease transmission, these are the sort of encounters allowing for flu transmission.

After listening to this presentation and one on a study looking at social networking and health  by one of my instructors a few semesters ago, Dr. Matt Newman, at University of Texas at Austin using a timed recorders, it makes me wonder if this sort of technology might be useful for sexual behavior inventories.  Obviously, ethical issues might prevent actually recording sexual behavior as it happened on a large scale, but being able to compare self-reports and mote-derived data might be useful. 

Additionally, using the methodology to understand stress response modification through social interaction could be interesting and useful, but might require a "ruggedized" type of mote with a low maintenance and annoyance footprint for the participant.