Introduction to forecasting

Forecasting & evaluation of infectious disease dynamics

Forecasting in infectious disease epidemiology

Metcalf & Lessler, Science, 2017

Modelling the future can help with …

… decision making

how many beds do we need?
where should we allocate vaccines?
where should we trial vaccines?

…situational awareness

nowcast: where are we now?
forecast: where are we heading?

Warning

But…

not all modelling is prediction
not all modelling of the future is forecasting

CDC use of influenza forecasts

CDC: About Flu Forecasting

Forecasting: relationship with \(R_t\) estimation

In infectious disease epidemiology, many relevant interventions and other changes affect the strength of transmission
Things that affect transmission don’t affect the predicted outcomes directly, but via \(R_t\)
In that sense, predicting infections comes down to predicting \(R_t\)
Commonly, forecast models assume no change in \(R_t\). Is this a good assumption?

Forecasting: relationship with nowcasting

Nowcast: we have some data from the dates we’re predicting
Forecast: we have no data from the dates we’re predicting (usually b/c they’re in the future)

Figure credit: Scenario Modeling Hub

Forecasts are often probabilistic

Because the future is uncertain, it is natural to express predictions in probabilities, e.g. there is a X% chance of exceeding Y hospital admissions.

Probabilistic epi forecasts

Figure credit: Evan Ray

The epi forecasting philosophical debate

Camp A

“It’s a machine learning problem! We can gather enough data to make good predictions, and we don’t really have to understand the underlying processes that well.”

Camp B

“We have clear theory that explains transmission of pathogens in a networked population. I don’t need to confront my model’s statements with data.”

What camp are you in?

Exhibit from Camp A:

Your Turn

Learn about and explore seasonal influenza-like illness data from US CDC.
Build a simple forecast model and understand how it is represented in R.

Return to the session