This group introduces the application of statistical and machine learning techniques to forecasting infectious diseases epidemics. The rationale is to use information on past incidence data in order to forecast future epidemics in terms of magnitude, timing and duration. This can be with or without of covariates such as demographic, immunological or climatic data. Such so-called “black-box” methods do not aim at understanding the details of diseases transmission but simply aim at forecasting future epidemics as accurately as possible by whatever means it takes. Such forecasting are crucial for deploying efficient infectious diseases prevention and control measures.
- Bias-variance trade-off
- Training/testing data
- Tree regression
- Ensemble methods
- ARIMA models
Hannah received a bachelors in mathematics from Oxford University (2007), followed by an MSc in Epidemiology (2009) and a PhD in Infectious Disease Dynamics (2013) from Imperial College London. She then spent 2 years working as a postdoctoral researcher in the Department of Epidemiology at Johns Hopkins School of Public Health. She has recently joined the Oxford University Clinical Research Unit (OUCRU) in Ho Chi Minh city, Vietnam as a postdoctoral researcher. Hannah’s work focused on the integration of models and data to better understand processes governing infectious disease dynamics. Hannah’s work to date has mainly focused on dengue, modelling both the within and between host dynamics to understand the interaction virus and immunity and how they lead to observed biological and epidemiological patterns. She has also taken part in dengue prediction work in Thailand and is currently also working on sero-epidemiology studies of dengue and other viral infections.
Matthew Graham completed an undergraduate degree in mathematics at the University of Oxford, UK in 2008, before spending a year in industry working on optimizing the supply chain for a large international retail company. He undertook a master’s degree and PhD at the University of Warwick, UK from 2009-2015 before taking up a post-doctoral research position at Johns Hopkins. His work has focused on the role of contact networks in disease transmission and control and measles disease dynamics including analyzing and forecasting an outbreak in Guinea. His current and future work is on predicting influenza in tropical climates. He currently resides in Ho Chi Minh City, Vietnam.
Michael Johansson completed a PhD at the Johns Hopkins Bloomberg School of Public Health in 2008. He is currently a Biologist at the Centers for Disease Control and Prevention Dengue Branch and a Visiting Scientist at the Harvard TH Chan School of Public Health Center for Communicable Disease Dynamics. He uses statistical and mathematical modeling to investigate infectious disease dynamics and identify ways to improve surveillance, prevention, and control. He also leads the CDC Epidemic Prediction Initiative, the CDC Zika Response Modeling Team, is a Deputy Editor at PLoS Neglected Tropical Diseases, and contributes to various other efforts to advance infectious disease forecasting.
Alex Perkins is an Assistant Professor in the Department of Biological Sciences and Eck Institute for Global Health at the University of Notre Dame. He completed a BA in Computational Ecology at the University of Tennessee and a PhD in Population Biology at the University of California, Davis. He received postdoctoral training through the NIH Fogarty International Center's Research and Policy for Infectious Disease Dynamics (RAPIDD) Program. Research in his lab uses a variety of mathematical, computational, and statistical techniques to answer questions about the dynamics and control of mosquito-borne diseases. Ongoing projects range from using mechanistic models of spatiotemporal transmission dynamics to estimate past transmission patterns and forecast future transmission, using simulation models to inform design and analysis of trials for vaccines and vector control products, and inferring transmission networks in near-elimination malaria settings based on a combination of epidemiological and genetic data.