Refining Epidemiological Forecasts with Simple Scoring Rules

TL;DR

This paper introduces a method to improve COVID-19 epidemiological forecasts by applying simple scoring rules to tune model hyperparameters, enhancing forecast accuracy and consistency with observed data.

Contribution

It presents a novel approach to refine infectious disease model forecasts using scoring rules to optimize hyperparameters for better accuracy.

Findings

Improved forecast accuracy through hyperparameter tuning.

Enhanced consistency of models with observed data.

Demonstrated effectiveness on multisource COVID-19 data.

Abstract

Estimates from infectious disease models have constituted a significant part of the scientific evidence used to inform the response to the COVID-19 pandemic in the UK. These estimates can vary strikingly in their bias and variability. Epidemiological forecasts should be consistent with the observations that eventually materialise. We use simple scoring rules to refine the forecasts of a novel statistical model for multisource COVID-19 surveillance data by tuning its smoothness hyperparameter.