TY - GEN
T1 - Modeling individual-level infection dynamics using social network information
AU - Tuarob, Suppawong
AU - Tucker, Conrad S.
AU - Salathe, Marcel
AU - Ram, Nilam
N1 - Publisher Copyright:
© 2015 ACM.
PY - 2015/10/17
Y1 - 2015/10/17
N2 - Epidemic monitoring systems engaged in accurate discovery of infected individuals enable better understanding of the dynamics of epidemics and thus may promote effective disease mitigation or prevention. Currently, infection discovery systems require either physical participation of potential patients or provision of information from hospitals and health-care services. While social media has emerged as an increasingly important knowledge source that reflects multiple real world events, there is only a small literature examining how social media information can be incorporated into computational epidemic models. In this paper, we demonstrate how social media information can be incorporated into and improve upon traditional techniques used to model the dynamics of infectious diseases. Using flu infection histories and social network data collected from 264 students in a college community, we identify social network signals that can aid identification of infected individuals. Extending the traditional SIRS model, we introduce and illustrate the efficacy of an Online-Interaction-Aware Susceptible-Infected-Recovered-Susceptible (OIA-SIRS) model based on four social network signals for modeling infection dynamics. Empirical evaluations of our case study, flu infection within a college community, reveal that the OIA-SIRS model is more accurate than the traditional model, and also closely tracks the real-world infection rates as reported by CDC ILINet and Google Flu Trend.
AB - Epidemic monitoring systems engaged in accurate discovery of infected individuals enable better understanding of the dynamics of epidemics and thus may promote effective disease mitigation or prevention. Currently, infection discovery systems require either physical participation of potential patients or provision of information from hospitals and health-care services. While social media has emerged as an increasingly important knowledge source that reflects multiple real world events, there is only a small literature examining how social media information can be incorporated into computational epidemic models. In this paper, we demonstrate how social media information can be incorporated into and improve upon traditional techniques used to model the dynamics of infectious diseases. Using flu infection histories and social network data collected from 264 students in a college community, we identify social network signals that can aid identification of infected individuals. Extending the traditional SIRS model, we introduce and illustrate the efficacy of an Online-Interaction-Aware Susceptible-Infected-Recovered-Susceptible (OIA-SIRS) model based on four social network signals for modeling infection dynamics. Empirical evaluations of our case study, flu infection within a college community, reveal that the OIA-SIRS model is more accurate than the traditional model, and also closely tracks the real-world infection rates as reported by CDC ILINet and Google Flu Trend.
UR - http://www.scopus.com/inward/record.url?scp=84959286892&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84959286892&partnerID=8YFLogxK
U2 - 10.1145/2806416.2806575
DO - 10.1145/2806416.2806575
M3 - Conference contribution
AN - SCOPUS:84959286892
T3 - International Conference on Information and Knowledge Management, Proceedings
SP - 1501
EP - 1510
BT - CIKM 2015 - Proceedings of the 24th ACM International Conference on Information and Knowledge Management
PB - Association for Computing Machinery
T2 - 24th ACM International Conference on Information and Knowledge Management, CIKM 2015
Y2 - 19 October 2015 through 23 October 2015
ER -