Algorithms Using Local Graph Features to Predict Epidemics

          @misc{ scivideos_22874,
            doi = {},
            url = {https://old.simons.berkeley.edu/talks/algorithms-using-local-graph-features-predict-epidemics-0},
            author = {},
            keywords = {},
            language = {en},
            title = {Algorithms Using Local Graph Features to Predict Epidemics},
            publisher = {The Simons Institute for the Theory of Computing},
            year = {2022},
            month = {oct},
            note = {22874 see, \url{https://scivideos.org/simons-institute/22874}}
          }
          

Abstract

Abstract People's interaction networks play a critical role in epidemics. However, precise mapping of the network structure is often expensive or even impossible. I will show that it is unnecessary to map the entire network. Instead, contact tracing a few samples from the population is enough to estimate an outbreak's likelihood and size. More precisely, I start by studying a simple epidemic model where one node is initially infected, and an infected node transmits the disease to its neighbors independently with probability p. In this model, I will present a nonparametric estimator on the likelihood of an outbreak based on local graph features and give theoretical guarantees on the estimator's accuracy for a large class of networks. Finally, I will extend the result to the general SIR model with random recovery time: Local graph features are enough to predict the time evolution of epidemics on a large class of networks.