Home > Events > Tuesday Talk - Shikang Liu

Tuesday Talk - Shikang Liu

Start: 1/23/2018 at 3:30PM
End: 1/23/2018 at 4:30PM
Location: 258 Fitzpatrick
Add to calendar:
iCal vCal

Shikang Liu

 What do social networks tell us about our health-related behaviors?"

Social communication data related to phone calls, SMS, Facebook interactions, etc. have become popular in studies of social networks. These studies have covered various aspects, encompassing the evolution of network structures, the relationship between different communication types, and the relationship between social networks and personality traits. A popular research direction has been to understand the impact of social networks on health. For example, the spread of obesity or smoking in social networks has been studied through quantitative analysis. The initial studies of the relationship between social networks and health dealt with analyses of static network data, while recent efforts have exploited the wealth of dynamic network data that has become available with the popularity of mobile devices such as cell phones. For instance, by clustering students that have similar evolving social network positions, and by comparing the resulting network-based clusters with the users' trait-based clusters, a number of network-trait relationships were detected and validated. With the popularity of highly capable and inexpensive wearable sensors, a wealth of dynamic user trait data can be collected easily. In particular, wearable sensors that collect health-related user traits have become very popular, such as FitBit devices. Hence, by integrating analyses of dynamic social network data with analyses of dynamic health-related trait data, one can study the link between social networks and health more accurately. To our knowledge, no one has done this yet.  We address this gap as follows. 

Our analysis consists of the following major steps: 1) How to construct a dynamic network from raw temporal user communication (e.g., SMS) events? 2) What does global network analysis reveal in terms of the network structure (e.g., degree distribution)? 3) What does local network analysis reveal in terms of the relationship between  users' evolving social network positions and evolving health-related behaviors?

We study longitudinal data from the NetHealth study in which the smartphone usage and health-related behaviors of around 700 college students are monitored for over two years. We systematically construct and analyze the dynamic social network. In addition, we identify users whose evolving social network positions correlate with his/her health-related behaviors.