Refining epidemiological forecasts with simple scoring rules

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 materialize. 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. This article is part of the theme issue ‘Technical challenges of modelling real-life epidemics and examples of overcoming these’.

Combining models to generate a consensus effective reproduction numberRfor the COVID-19 epidemic status in England

“…University of Liverpool Model [37] Bayesian statistical model that comprises a deterministic compartmental transmission model governed by a system of ODEs and a stochastic observation model. Fits to deaths, hospital admissions, and symptomatic report data from NHS 111 online.…”

Inferring community transmission of SARS-CoV-2 in the United Kingdom using the ONS COVID-19 Infection Survey

“…Estimation of these quantities by public health officials in the United Kingdom (UK) has relied on an ensemble of models which encompass a range of data sources and assumptions ( Park et al., 2023 ). These parameters are traditionally estimated using case data as a proxy for the infection incidence curve ( Cori, Ferguson, Fraser, & Cauchemez, 2013 ; Parag, 2021 ), but methods have also been developed to estimate these parameters from other sources, such as hospitalisations ( Moore, Rosato, & Maskell, 2022 ) and genomic data ( Vöhringer et al., 2021 ). The four nations of the UK (England, Scotland, Wales and Northern Ireland, herein ordered by population size) were recognised globally as having comprehensive SARS-CoV-2 testing surveillance systems ( Clarke, Beaney, & Majeed, 2022 ; Dean, 2022 ; Tapper, 2022 ), comprising of widescale community testing ( Coronavirus (COVID-19) in the UK, 2023 ), nationwide surveys of infection ( Coronavirus (COVID-19) Infection Survey, UK, 2022 ; Real-Time Assessment of Community Transmission (REACT) Study, 2022 ), genomic data ( COVID–19 Genomic Surveillance, 2023 ; The COVID-19 Genomics UK (COG-UK) consortium, 2020 ), and wastewater surveillance ( Morvan et al., 2022 ), but these were largely scaled down from their peak capacity as part of the transition from an “emergency” to “endemic” state in the first half of 2022 ( Coronavirus (COVID-19): Test and Protect - Transition Plan, 2022 ; COVID-19 Response: Living with COVID-19, 2022 ; COVID-19 Test, Trace and Protect Transition Plan, 2022 ; Together for a Safer Future: Wales’ Long-Term Covid-19 Transition from Pandemic to Endemic, 2022 ).…”

Inferring community transmission of SARS-CoV-2 in the United Kingdom using the ONS COVID-19 Infection Survey

“…1–4 Estimation of these quantities by public health officials in the United Kingdom (UK) has relied on an ensemble of models which encompass a range of data sources and assumptions. 5 These parameters are traditionally estimated using case data as a proxy for the infection incidence curve, 6,7 but methods have also been developed to estimate these parameters from other sources, such as hospitalisations 8 and genomic data. 9 The four nations of the UK (England, Scotland, Wales and Northern Ireland, herein ordered by population size) were recognised globally as having comprehensive SARS-CoV-2 testing surveillance systems, 10–12 comprising of widescale community testing, 13 nationwide surveys of infection, 14,15 genomic data, 16,17 and wastewater surveillance, 18 but these were largely scaled down from their peak capacity as part of the transition from an “emergency” to “endemic” state in the first half of 2022.…”