Learning outcomes
The skills and methods taught in this course apply broadly across infectious disease epidemiology, from outbreak response to routine surveillance of endemic diseases. While examples often use outbreak scenarios for clarity, participants should consider how these approaches apply to their epidemiological contexts.
R and statistical concepts used
- understanding of common probability distributions used for epidemiological delays
- familiarity with the basics of Bayesian inference and MCMC sampling
- familiarity with using stan to estimate parameters of a probability distribution
- ability to interpret a posterior summary, including key convergence diagnostics
Delay distributions
- ability to estimate parameters of epidemiological delay distributions
- understanding of the ubiquity of delays in epidemiological data
Biases in delay distributions
- understanding of how censoring affects the estimation and interpretation of epidemiological delay distributions
- ability to estimate parameters of probability distributions from observed delays, taking into account censoring, using R
- understanding of right truncation in epidemiolgical data
- ability to estimate parameters of probability distributions from observed delays, taking into account truncation, in R
Using delay distributions to model the data generating process
- understanding of using delay distributions to model population-level data generating processes
- ability to use convolutions to combine count data with a distribution of individual probabilities, adjusting continuous probability distributions with discretisation
- understanding of the need to introduce additional uncertainty to account for the observation process at a population level
\(R_t\) estimation and the renewal equation
- understanding of the reproduction number and challenges in its estimation
- awareness of broad categories of methods for estimating the reproduction number, including estimation from population-level data
- understanding of the renewal equation as an epidemiological model
- familiarity with the generation time as a particular type of delay distributions
- ability to estimate static and time-varying reproduction numbers from time-series data in R
Nowcasting
- understanding of nowcasting as a particular right truncation problem
- Ability to perform a simple nowcast in R
- awareness of the breadth of methods to perform nowcasting
- \(R_t\) estimation as a nowcasting problem
Joint nowcasting with an unknown reporting delay
- understanding of the reporting triangle as a representation of partially-observed reporting data
- ability to jointly estimate a reporting delay distribution and a nowcast in R
- awareness of how nowcasting can be combined with \(R_t\) estimation in a single model
Forecasting concepts
- understanding of forecasting as an epidemiological problem, and its relationship with nowcasting and \(R_t\) estimation
- ability to use a forecasting model on an epidemiological time series in R
- familiarity with the forecasting paradigm of maximising sharpness subject to calibration
Forecasting models
- awareness of forecasting models as a spectrum from mechanistic to statistical
- ability to extend a renewal-equation forecasting model with additional mechanistic structure (e.g. susceptible depletion)
- ability to extend a forecasting model with additional statistical structure (e.g. autoregressive components)
Evaluating forecasts (and nowcasts)
- ability to visually assess forecasts and nowcasts
- familiarity with metrics for evaluating probabilistic forecasts and their properties
- understanding of proper scoring rules and the components of the weighted interval score (sharpness, over- and under-prediction)
- ability to assess calibration and bias using PIT histograms
- ability to score probabilistic forecasts in R
- ability to compare different models by their forecast scores
Ensemble models
- understanding of predictive ensembles and their properties
- ability to create a predictive ensemble of forecasts in R
- familiarity with weighted ensembles, including inverse-WIS weighting and quantile regression averaging