Introduction to nowcast
This package is about nowcasting mortality surveillance data.
The goal of nowcast is to predict the present. This is also called delay adjustment or correction, as registration delay can underestimate the real number of outcomes of interest. This is particularly important in an outbreak.
Nowcast corrections are usually compared with a baseline. If the corrected mortality deviates from the baseline by much, there is reason to look into possible outbreaks.
These are the planned content for this package.
1. mortality reporting data simulation
Purpose: generate realistic datasets for mortality reporting. We do not have the possibility to share and distribute real-life mortality data from the official registry. However, we simulate it based on our knowledge of the real data.
The data simulator can be used for two geographic levels:
- county level
- national level
We allow the following specifications in the basic version:
- reporting delay pattern
- pseudo-mortality counts based on historical data
More advanced simulator will allow user to specify
- population size and mortality rate
- excess mortality due to disease (e.g. influenza)
3. nowcast delay adjustment
Models (Bayesian regression framework)
- negative binomial
- negative binomial with random effects (each county)
Assessment methods are included.