Li, Xuan
(2015)
Statistical analysis of infectious disease data on networks.
Master's Thesis, University of Pittsburgh.
(Unpublished)
Abstract
Purpose
Infectious disease modeling has a long history in helping researchers to understand the complex spread pattern of infectious disease. Social contact networks and agent-based models can be used to conceptualize social contact pattern and spread process of infectious disease. The goal of this research is to investigate the relationship between network measurements and individual infection risk using statistical analysis.
Public Health significance
This research will help in gaining a better understanding of the important factors of infection risk in a population. Identification of central people may be used to inform building an efficient surveillance and prevention program.
Methods
Three social contact network models were used in this thesis, Erdos-Renyi network, Barabasi-Albert network and Jefferson County contact network using FRED platform. We simulated mild and severe epidemic outbreaks on them and calculated infection risk and infection speed of each individual. Network measurements, degree, betweenness centrality, closeness centrality, eigenvector centrality, PageRank, and clustering coefficient were measured on the ability to identify groups of different infection risk level and infection speed. Random Forest and variable importance were used to estimate the most important factors in predicting infection risk
Results
For Barabasi-Albert and Erdos-Renyi networks, centrality measurements are critical factors in identifying infection risk. Degree is the most important factor in Barabasi-Albert network while closeness and degree are the most important in the mild outbreak and severe outbreak respectively in the Erdos-Renyi network. Results of Jefferson County contact network in FRED find out the importance of location sizes. The highly clustered structure of location-based model makes betweenness centrality and clustering coefficient important in predicting infection risk.
Conclusion
Different network structures and characteristics of the disease will influence the importance of network measurements. Network structures also influence the correlations between network measurements. Random forest is a powerful tool for classifying infection risk. Centrality network measurements may help in identifying high infection risk people.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
28 September 2015 |
| Date Type: |
Publication |
| Defense Date: |
29 June 2015 |
| Approval Date: |
28 September 2015 |
| Submission Date: |
24 July 2015 |
| Access Restriction: |
3 year -- Restrict access to University of Pittsburgh for a period of 3 years. |
| Number of Pages: |
80 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
School of Public Health > Biostatistics |
| Degree: |
MS - Master of Science |
| Thesis Type: |
Master's Thesis |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
Social Contact Network; Random Forest; Infectious Disease; Agent-Based Model;FRED; |
| Date Deposited: |
28 Sep 2015 18:32 |
| Last Modified: |
01 Sep 2018 05:15 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/25754 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[img]](http://d-scholarship.pitt.edu/style/images/fileicons/application_pdf.png)
![[feed]](/images/feed-icon-32x32.png){kind=icon}