Introduction and course overview

From an epidemiological line list to informing decisions in real-time

Aim of the workshop

Data collected in an outbreak, or in routine surveillance, is used to answer questions like

  • What is the number of cases now (nowcasting)
  • Are infections rising or falling and by how much (\(R_t\) estimation)
  • What does this mean for the near future (forecasting)

This workshop focuses on the last of these, and on a question that matters just as much once a forecast has been made.

  • What is a forecast, and how do we visualise and interpret one?
  • Is the forecast any good (evaluation)?
  • How do we combine forecasts from several models (ensembles)?

To make sense of forecasts we need to understand the epidemiological processes that create the data we typically have available, and the challenges of doing this analysis in real time rather than retrospectively.

Let’s look at infectious disease surveillance data from the perspective of an individual infection. There are two types of processes happening:

  • upwards, from an individual infection through to being recorded in surveillance data; and

  • outwards, from each infection spreading to cause new infections in the population.

Both of these processes involve time delays which makes analysing data in real time especially tricky.

A forecast is a probabilistic statement about what this data will do next. Making one is only half the job. Much of this workshop is about the other half, telling whether a forecast is any good by confronting it with what actually happened, and using that to compare and combine models.

Why this workshop?

  • Forecasts are common in outbreak response and disease surveillance, and increasingly feed into decisions
  • It is easy to produce a forecast and much harder to know whether it is any good
  • Combining forecasts across models often improves predictions, but only if we can measure that improvement

Approach

Each session in the workshop

  • builds on the previous one so that by the end you have a route from making a forecast to evaluating it and combining it with others;
  • starts with a short introductory talk;
  • mainly consists of interactive content that you work through in R;
  • marks optional and additional material clearly (boxes labelled optional, plus a “Going further” section at the end of each session).

We work throughout from forecasts we made earlier, so you can focus on interpreting, evaluating and combining them without waiting for models to fit.

For those attending in person the workshop also has multiple instructors ready to answer questions, pauses at several points to discuss observations together, and ends with a wrap-up and discussion.

What we cover

  • forecasting concepts and visualisation
  • the spectrum of forecasting models
  • forecast evaluation
  • forecast ensembles

Let’s get started!