Forecasting Based on Surveillance Data

TL;DR

This paper reviews statistical methods for evaluating epidemic forecasts, demonstrating their application through case studies on influenza and norovirus surveillance data, and provides accessible tools for practitioners.

Contribution

It introduces a comprehensive overview of forecast evaluation techniques specifically for epidemic count data, with practical case studies and open-source code.

Findings

Effective evaluation methods for epidemic forecasts are demonstrated.

Case studies illustrate application to influenza and norovirus data.

Tools and code are made publicly available for reproducibility.

Abstract

Forecasting the future course of epidemics has always been one of the main goals of epidemic modelling. This chapter reviews statistical methods to quantify the accuracy of epidemic forecasts. We distinguish point and probabilistic forecasts and describe different methods to evaluate and compare the predictive performance across models. Two case studies demonstrate how to apply the different techniques to uni- and multivariate forecasts. We focus on forecasting count time series from routine public health surveillance: weekly counts of influenza-like illness in Switzerland, and age-stratified counts of norovirus gastroenteritis in Berlin, Germany. Data and code for all analyses are available in a supplementary R package.