Prashant Chandrasekar
  • Home
  • About Me
  • Research
  • Contact Me
  • Projects
CS6604: Quantitative Research on Network-based Clinical Trials
Team Members : Prashant Chandrasekar
​
​Link to Project artifacts
  1. Final Report
  2. Presentation slides (.pdf) (Waiting for IRB Approval)
  3. Software (.tar) (Waiting for IRB Approval)

Abstract

The Social Interactome (SI) is an ongoing research study of the Ad- diction Recovery Research Center at VT Carilion Research Institute and the Computer Science and Statistics departments at VT. The high-level aim of the study is to demonstrate through clinical trials involving our closed social-network system whether our proposed interventions aid people in their recovery from addictions. Much of the success of a social-network-based user study that analyzes the behavior of participants, depends on the study design and evaluation. The study design parameters are fine-tuned by instantiating clinical trials, analyzing the data and instantiating the next trial based on the results. So far, much of the research has focused on understanding participant engagement in aggregate forms. Through exploratory research, we aim to analyze the engagement graph to understand the behavior of participant in silo as well as in their friendship groups, to infer network-based characteristics such as homophily, social influence and contagion. The results will be a reinforcement to the efforts of hypothesis testing conducted by the SI team and will have a direct implication in the design of following clinical trials. 

Project Aim

The overall research is focused on properties that relate to participant engagement. Therefore, the goal of this research is to understand the following: 
  • Why did the participants in the homophily-based network engage more than participants who were randomly organized? Was this occurrence random (a coincidence)? If not, 
  • Can we observe and treat homophily as a latent variable and build a graphical model that explains how features of homophily and other factors such as monetary incentives, etc. effect engagement of participants to the website and each other. And finally,
  • Will this model apply to our observations to the engagement (or lack thereof) that we have observed in our other five clinical trails. If not, what else would we need to consider to make this model more generalizable to our setting.  

 Results via Tables and Figures (disclosed after IRB approval)

Powered by Create your own unique website with customizable templates.