Comparing trained and untrained probabilistic ensemble forecasts of COVID-19 cases and deaths in the United States

Collaborative forecasting of influenza-like illness in Italy: The Influcast experience

Collaborative hubs that integrate multiple teams to generate ensemble projections and forecasts for shared targets are now regarded as state-of-the-art in epidemic predictive modeling. In this paper, we introduce Influcast, Italy's first epidemic forecasting hub for influenza-like illness. During the 2023/2024 winter season, Influcast provided 20 rounds of forecasts, involving five teams and eight models to predict influenza-like illness incidence up to four weeks in advance at the national and regional administrative level. The individual forecasts were synthesized into an ensemble and benchmarked against a baseline model. Across all models, the ensemble most frequently ranks among the top performers at the national level considering different metrics and forecasting rounds. Additionally, the ensemble outperforms the baseline and most individual models across all regions. Despite a decline in absolute performance over longer horizons, the ensemble model outperformed the baseline in all considered horizons. These findings show the importance of multimodel forecasting hubs in producing reliable short-term influenza-like illnesses forecasts that can inform public health preparedness and mitigation strategies.

Nowcasting reported covid-19 hospitalizations using de-identified, aggregated medical insurance claims data

We propose, implement, and evaluate a method for nowcasting the daily number of new COVID-19 hospitalizations, at the level of individual US states, based on de-identified, aggregated medical insurance claims data. Our analysis proceeds under a hypothetical scenario in which, during the Delta wave, states only report data on the first day of each month, and on this day, report COVID-19 hospitalization counts for each day in the previous month. In this hypothetical scenario (just as in reality), medical insurance claims data continues to be available daily. At the beginning of each month, we train a regression model, using all data available thus far, to predict hospitalization counts from medical insurance claims. We then use this model to nowcast the (unseen) values of COVID-19 hospitalization counts from medical insurance claims, at each day in the following month. Our analysis uses properly-versioned data, which would have been available in real-time at the time predictions are produced (instead of using data that would have only been available in hindsight). In spite of the difficulties inherent to real-time estimation (e.g., latency and backfill) and the complex dynamics behind COVID-19 hospitalizations themselves, we find altogether that medical insurance claims can be an accurate predictor of hospitalization reports, with mean absolute errors typically around 0.4 hospitalizations per 100,000 people, i.e., proportion of variance explained around 75%. Perhaps more importantly, we find that nowcasts made using medical insurance claims are able to qualitatively capture the dynamics (upswings and downswings) of hospitalization waves, which are key features that inform public health decision-making.

The Future of HIV: Challenges in meeting the 2030 Ending the HIV Epidemic in the U.S. (EHE) reduction goal

Objectives

To predict the burden of HIV in the United States (US) nationally and by region, transmission type, and race/ethnicity through 2030.

Methods

Using publicly available data from the CDC NCHHSTP AtlasPlus dashboard, we generated 11-year prospective forecasts of incident HIV diagnoses nationally and by region (South, non-South), race/ethnicity (White, Hispanic/Latino, Black/African American), and transmission type (Injection-Drug Use, Male-to-Male Sexual Contact (MMSC), and Heterosexual Contact (HSC)). We employed weighted (W) and unweighted (UW) n-sub-epidemic ensemble models, calibrated using 12 years of historical data (2008-2019), and forecasted trends for 2020-2030. We compared results to identify persistent, concerning trends across models.

Results

We projected substantial decreases in incident HIV diagnoses nationally (W: 27.9%, UW: 21.9%), and in the South (W:18.0%, UW: 9.2%) and non-South (W: 21.2%, UW: 19.5%) from 2019 to 2030. However, concerning non-decreasing trends were observed nationally in key sub-populations during this period: Hispanic/Latino persons (W: 1.4%, UW: 2.6%), Hispanic/Latino MMSC (W: 9.0%, UW: 9.9%), people who inject drugs (PWID) (W: 25.6%, UW: 9.2%), and White PWID (W: 3.5%, UW: 44.9%). The rising trends among Hispanic/Latino MMSC and overall PWID were consistent across the South and non-South regions.

Conclusions

Although the forecasted national-level decrease in the number of incident HIV diagnoses is encouraging, the US is unlikely to achieve the Ending the HIV Epidemic in the U.S. goal of a 90% reduction in HIV incidence by 2030. Additionally, the observed increases among specific subpopulations highlight the importance of a targeted and equitable approach to effectively combat HIV in the US.

Flusion: Integrating multiple data sources for accurate influenza predictions

Over the last ten years, the US Centers for Disease Control and Prevention (CDC) has organized an annual influenza forecasting challenge with the motivation that accurate probabilistic forecasts could improve situational awareness and yield more effective public health actions. Starting with the 2021/22 influenza season, the forecasting targets for this challenge have been based on hospital admissions reported in the CDC's National Healthcare Safety Network (NHSN) surveillance system. Reporting of influenza hospital admissions through NHSN began within the last few years, and as such only a limited amount of historical data are available for this target signal. To produce forecasts in the presence of limited data for the target surveillance system, we augmented these data with two signals that have a longer historical record: 1) ILI+, which estimates the proportion of outpatient doctor visits where the patient has influenza; and 2) rates of laboratory-confirmed influenza hospitalizations at a selected set of healthcare facilities. Our model, Flusion, is an ensemble model that combines two machine learning models using gradient boosting for quantile regression based on different feature sets with a Bayesian autoregressive model. The gradient boosting models were trained on all three data signals, while the autoregressive model was trained on only data for the target surveillance signal, NHSN admissions; all three models were trained jointly on data for multiple locations. In each week of the influenza season, these models produced quantiles of a predictive distribution of influenza hospital admissions in each state for the current week and the following three weeks; the ensemble prediction was computed by averaging these quantile predictions. Flusion emerged as the top-performing model in the CDC's influenza prediction challenge for the 2023/24 season. In this article we investigate the factors contributing to Flusion's success, and we find that its strong performance was primarily driven by the use of a gradient boosting model that was trained jointly on data from multiple surveillance signals and multiple locations. These results indicate the value of sharing information across multiple locations and surveillance signals, especially when doing so adds to the pool of available training data.

When are predictions useful? A new method for evaluating epidemic forecasts

BACKGROUND

COVID-19 will not be the last pandemic of the twenty-first century. To better prepare for the next one, it is essential that we make honest appraisals of the utility of different responses to COVID. In this paper, we focus specifically on epidemiologic forecasting. Characterizing forecast efficacy over the history of the pandemic is challenging, especially given its significant spatial, temporal, and contextual variability. In this light, we introduce the Weighted Contextual Interval Score (WCIS), a new method for retrospective interval forecast evaluation.

METHODS

The central tenet of the WCIS is a direct incorporation of contextual utility into the evaluation. This necessitates a specific characterization of forecast efficacy depending on the use case for predictions, accomplished via defining a utility threshold parameter. This idea is generalized to probabilistic interval-form forecasts, which are the preferred prediction format for epidemiological modeling, as an extension of the existing Weighted Interval Score (WIS).

RESULTS

We apply the WCIS to two forecasting scenarios: facility-level hospitalizations for a single state, and state-level hospitalizations for the whole of the United States. We observe that an appropriately parameterized application of the WCIS captures both the relative quality and the overall frequency of useful forecasts. Since the WCIS represents the utility of predictions using contextual normalization, it is easily comparable across highly variable pandemic scenarios while remaining intuitively representative of the in-situ quality of individual forecasts.

CONCLUSIONS

The WCIS provides a pragmatic utility-based characterization of probabilistic predictions. This method is expressly intended to enable practitioners and policymakers who may not have expertise in forecasting but are nevertheless essential partners in epidemic response to use and provide insightful analysis of predictions. We note that the WCIS is intended specifically for retrospective forecast evaluation and should not be used as a minimized penalty in a competitive context as it lacks statistical propriety. Code and data used for our analysis are available at https://github.com/maximilian-marshall/wcis .

Optimizing Disease Outbreak Forecast Ensembles

On the basis of historical influenza and COVID-19 forecasts, we found that more than 3 forecast models are needed to ensure robust ensemble accuracy. Additional models can improve ensemble performance, but with diminishing accuracy returns. This understanding will assist with the design of current and future collaborative infectious disease forecasting efforts.

hubEnsembles: Ensembling Methods in R

Combining predictions from multiple models into an ensemble is a widely used practice across many fields with demonstrated performance benefits. The R package hubEnsembles provides a flexible framework for ensembling various types of predictions, including point estimates and probabilistic predictions. A range of common methods for generating ensembles are supported, including weighted averages, quantile averages, and linear pools. The hubEnsembles package fits within a broader framework of open-source software and data tools called the "hubverse", which facilitates the development and management of collaborative modelling exercises.

Human judgement forecasting of COVID-19 in the UK

Background

In the past, two studies found ensembles of human judgement forecasts of COVID-19 to show predictive performance comparable to ensembles of computational models, at least when predicting case incidences. We present a follow-up to a study conducted in Germany and Poland and investigate a novel joint approach to combine human judgement and epidemiological modelling.

Methods

From May 24th to August 16th 2021, we elicited weekly one to four week ahead forecasts of cases and deaths from COVID-19 in the UK from a crowd of human forecasters. A median ensemble of all forecasts was submitted to the European Forecast Hub. Participants could use two distinct interfaces: in one, forecasters submitted a predictive distribution directly, in the other forecasters instead submitted a forecast of the effective reproduction number R t. This was then used to forecast cases and deaths using simulation methods from the EpiNow2 R package. Forecasts were scored using the weighted interval score on the original forecasts, as well as after applying the natural logarithm to both forecasts and observations.

Results

The ensemble of human forecasters overall performed comparably to the official European Forecast Hub ensemble on both cases and deaths, although results were sensitive to changes in details of the evaluation. R t forecasts performed comparably to direct forecasts on cases, but worse on deaths. Self-identified "experts" tended to be better calibrated than "non-experts" for cases, but not for deaths.

Conclusions

Human judgement forecasts and computational models can produce forecasts of similar quality for infectious disease such as COVID-19. The results of forecast evaluations can change depending on what metrics are chosen and judgement on what does or doesn't constitute a "good" forecast is dependent on the forecast consumer. Combinations of human and computational forecasts hold potential but present real-world challenges that need to be solved.

Evaluation of the US COVID-19 Scenario Modeling Hub for informing pandemic response under uncertainty

Our ability to forecast epidemics far into the future is constrained by the many complexities of disease systems. Realistic longer-term projections may, however, be possible under well-defined scenarios that specify the future state of critical epidemic drivers. Since December 2020, the U.S. COVID-19 Scenario Modeling Hub (SMH) has convened multiple modeling teams to make months ahead projections of SARS-CoV-2 burden, totaling nearly 1.8 million national and state-level projections. Here, we find SMH performance varied widely as a function of both scenario validity and model calibration. We show scenarios remained close to reality for 22 weeks on average before the arrival of unanticipated SARS-CoV-2 variants invalidated key assumptions. An ensemble of participating models that preserved variation between models (using the linear opinion pool method) was consistently more reliable than any single model in periods of valid scenario assumptions, while projection interval coverage was near target levels. SMH projections were used to guide pandemic response, illustrating the value of collaborative hubs for longer-term scenario projections.

Comparison of combination methods to create calibrated ensemble forecasts for seasonal influenza in the U.S

The characteristics of influenza seasons vary substantially from year to year, posing challenges for public health preparation and response. Influenza forecasting is used to inform seasonal outbreak response, which can in turn potentially reduce the impact of an epidemic. The United States Centers for Disease Control and Prevention, in collaboration with external researchers, has run an annual prospective influenza forecasting exercise, known as the FluSight challenge. Uniting theoretical results from the forecasting literature with domain-specific forecasts from influenza outbreaks, we applied parametric forecast combination methods that simultaneously optimize model weights and calibrate the ensemble via a beta transformation and made adjustments to the methods to reduce their complexity. We used the beta-transformed linear pool, the finite beta mixture model, and their equal weight adaptations to produce ensemble forecasts retrospectively for the 2016/2017, 2017/2018, and 2018/2019 influenza seasons in the U.S. We compared their performance to methods that were used in the FluSight challenge to produce the FluSight Network ensemble, namely the equally weighted linear pool and the linear pool. Ensemble forecasts produced from methods with a beta transformation were shown to outperform those from the equally weighted linear pool and the linear pool for all week-ahead targets across in the test seasons based on average log scores. We observed improvements in overall accuracy despite the beta-transformed linear pool or beta mixture methods' modest under-prediction across all targets and seasons. Combination techniques that explicitly adjust for known calibration issues in linear pooling should be considered to improve probabilistic scores in outbreak settings.

Adaptive metrics for an evolving pandemic: A dynamic approach to area-level COVID-19 risk designations

Throughout the COVID-19 pandemic, policymakers have proposed risk metrics, such as the CDC Community Levels, to guide local and state decision-making. However, risk metrics have not reliably predicted key outcomes and have often lacked transparency in terms of prioritization of false-positive versus false-negative signals. They have also struggled to maintain relevance over time due to slow and infrequent updates addressing new variants and shifts in vaccine- and infection-induced immunity. We make two contributions to address these weaknesses. We first present a framework to evaluate predictive accuracy based on policy targets related to severe disease and mortality, allowing for explicit preferences toward false-negative versus false-positive signals. This approach allows policymakers to optimize metrics for specific preferences and interventions. Second, we propose a method to update risk thresholds in real time. We show that this adaptive approach to designating areas as "high risk" improves performance over static metrics in predicting 3-wk-ahead mortality and intensive care usage at both state and county levels. We also demonstrate that with our approach, using only new hospital admissions to predict 3-wk-ahead mortality and intensive care usage has performed consistently as well as metrics that also include cases and inpatient bed usage. Our results highlight that a key challenge for COVID-19 risk prediction is the changing relationship between indicators and outcomes of policy interest. Adaptive metrics therefore have a unique advantage in a rapidly evolving pandemic context.

Informing pandemic response in the face of uncertainty. An evaluation of the U.S. COVID-19 Scenario Modeling Hub

Our ability to forecast epidemics more than a few weeks into the future is constrained by the complexity of disease systems, our limited ability to measure the current state of an epidemic, and uncertainties in how human action will affect transmission. Realistic longer-term projections (spanning more than a few weeks) may, however, be possible under defined scenarios that specify the future state of critical epidemic drivers, with the additional benefit that such scenarios can be used to anticipate the comparative effect of control measures. Since December 2020, the U.S. COVID-19 Scenario Modeling Hub (SMH) has convened multiple modeling teams to make 6-month ahead projections of the number of SARS-CoV-2 cases, hospitalizations and deaths. The SMH released nearly 1.8 million national and state-level projections between February 2021 and November 2022. SMH performance varied widely as a function of both scenario validity and model calibration. Scenario assumptions were periodically invalidated by the arrival of unanticipated SARS-CoV-2 variants, but SMH still provided projections on average 22 weeks before changes in assumptions (such as virus transmissibility) invalidated scenarios and their corresponding projections. During these periods, before emergence of a novel variant, a linear opinion pool ensemble of contributed models was consistently more reliable than any single model, and projection interval coverage was near target levels for the most plausible scenarios (e.g., 79% coverage for 95% projection interval). SMH projections were used operationally to guide planning and policy at different stages of the pandemic, illustrating the value of the hub approach for long-term scenario projections.

Predictive performance of multi-model ensemble forecasts of COVID-19 across European nations

Background

Short-term forecasts of infectious disease burden can contribute to situational awareness and aid capacity planning. Based on best practice in other fields and recent insights in infectious disease epidemiology, one can maximise the predictive performance of such forecasts if multiple models are combined into an ensemble. Here, we report on the performance of ensembles in predicting COVID-19 cases and deaths across Europe between 08 March 2021 and 07 March 2022.

Methods

We used open-source tools to develop a public European COVID-19 Forecast Hub. We invited groups globally to contribute weekly forecasts for COVID-19 cases and deaths reported by a standardised source for 32 countries over the next 1-4 weeks. Teams submitted forecasts from March 2021 using standardised quantiles of the predictive distribution. Each week we created an ensemble forecast, where each predictive quantile was calculated as the equally-weighted average (initially the mean and then from 26th July the median) of all individual models' predictive quantiles. We measured the performance of each model using the relative Weighted Interval Score (WIS), comparing models' forecast accuracy relative to all other models. We retrospectively explored alternative methods for ensemble forecasts, including weighted averages based on models' past predictive performance.

Results

Over 52 weeks, we collected forecasts from 48 unique models. We evaluated 29 models' forecast scores in comparison to the ensemble model. We found a weekly ensemble had a consistently strong performance across countries over time. Across all horizons and locations, the ensemble performed better on relative WIS than 83% of participating models' forecasts of incident cases (with a total N=886 predictions from 23 unique models), and 91% of participating models' forecasts of deaths (N=763 predictions from 20 models). Across a 1-4 week time horizon, ensemble performance declined with longer forecast periods when forecasting cases, but remained stable over 4 weeks for incident death forecasts. In every forecast across 32 countries, the ensemble outperformed most contributing models when forecasting either cases or deaths, frequently outperforming all of its individual component models. Among several choices of ensemble methods we found that the most influential and best choice was to use a median average of models instead of using the mean, regardless of methods of weighting component forecast models.

Conclusions

Our results support the use of combining forecasts from individual models into an ensemble in order to improve predictive performance across epidemiological targets and populations during infectious disease epidemics. Our findings further suggest that median ensemble methods yield better predictive performance more than ones based on means. Our findings also highlight that forecast consumers should place more weight on incident death forecasts than incident case forecasts at forecast horizons greater than 2 weeks.

Funding

AA, BH, BL, LWa, MMa, PP, SV funded by National Institutes of Health (NIH) Grant 1R01GM109718, NSF BIG DATA Grant IIS-1633028, NSF Grant No.: OAC-1916805, NSF Expeditions in Computing Grant CCF-1918656, CCF-1917819, NSF RAPID CNS-2028004, NSF RAPID OAC-2027541, US Centers for Disease Control and Prevention 75D30119C05935, a grant from Google, University of Virginia Strategic Investment Fund award number SIF160, Defense Threat Reduction Agency (DTRA) under Contract No. HDTRA1-19-D-0007, and respectively Virginia Dept of Health Grant VDH-21-501-0141, VDH-21-501-0143, VDH-21-501-0147, VDH-21-501-0145, VDH-21-501-0146, VDH-21-501-0142, VDH-21-501-0148. AF, AMa, GL funded by SMIGE - Modelli statistici inferenziali per governare l'epidemia, FISR 2020-Covid-19 I Fase, FISR2020IP-00156, Codice Progetto: PRJ-0695. AM, BK, FD, FR, JK, JN, JZ, KN, MG, MR, MS, RB funded by Ministry of Science and Higher Education of Poland with grant 28/WFSN/2021 to the University of Warsaw. BRe, CPe, JLAz funded by Ministerio de Sanidad/ISCIII. BT, PG funded by PERISCOPE European H2020 project, contract number 101016233. CP, DL, EA, MC, SA funded by European Commission - Directorate-General for Communications Networks, Content and Technology through the contract LC-01485746, and Ministerio de Ciencia, Innovacion y Universidades and FEDER, with the project PGC2018-095456-B-I00. DE., MGu funded by Spanish Ministry of Health / REACT-UE (FEDER). DO, GF, IMi, LC funded by Laboratory Directed Research and Development program of Los Alamos National Laboratory (LANL) under project number 20200700ER. DS, ELR, GG, NGR, NW, YW funded by National Institutes of General Medical Sciences (R35GM119582; the content is solely the responsibility of the authors and does not necessarily represent the official views of NIGMS or the National Institutes of Health). FB, FP funded by InPresa, Lombardy Region, Italy. HG, KS funded by European Centre for Disease Prevention and Control. IV funded by Agencia de Qualitat i Avaluacio Sanitaries de Catalunya (AQuAS) through contract 2021-021OE. JDe, SMo, VP funded by Netzwerk Universitatsmedizin (NUM) project egePan (01KX2021). JPB, SH, TH funded by Federal Ministry of Education and Research (BMBF; grant 05M18SIA). KH, MSc, YKh funded by Project SaxoCOV, funded by the German Free State of Saxony. Presentation of data, model results and simulations also funded by the NFDI4Health Task Force COVID-19 (https://www.nfdi4health.de/task-force-covid-19-2) within the framework of a DFG-project (LO-342/17-1). LP, VE funded by Mathematical and Statistical modelling project (MUNI/A/1615/2020), Online platform for real-time monitoring, analysis and management of epidemic situations (MUNI/11/02202001/2020); VE also supported by RECETOX research infrastructure (Ministry of Education, Youth and Sports of the Czech Republic: LM2018121), the CETOCOEN EXCELLENCE (CZ.02.1.01/0.0/0.0/17-043/0009632), RECETOX RI project (CZ.02.1.01/0.0/0.0/16-013/0001761). NIB funded by Health Protection Research Unit (grant code NIHR200908). SAb, SF funded by Wellcome Trust (210758/Z/18/Z).

Adaptive metrics for an evolving pandemic A dynamic approach to area-level COVID-19 risk designations

Throughout the COVID-19 pandemic, policymakers have proposed risk metrics, such as the CDC Community Levels, to guide local and state decision-making. However, risk metrics have not reliably predicted key outcomes and often lack transparency in terms of prioritization of false positive versus false negative signals. They have also struggled to maintain relevance over time due to slow and infrequent updates addressing new variants and shifts in vaccine- and infection-induced immunity. We make two contributions to address these weaknesses of risk metrics. We first present a framework to evaluate predictive accuracy based on policy targets related to severe disease and mortality, allowing for explicit preferences toward false negative versus false positive signals. This approach allows policymakers to optimize metrics for specific preferences and interventions. Second, we propose a novel method to update risk thresholds in real-time. We show that this adaptive approach to designating areas as "high risk" improves performance over static metrics in predicting 3-week-ahead mortality and intensive care usage at both state and county levels. We also demonstrate that with our approach, using only new hospital admissions to predict 3-week-ahead mortality and intensive care usage has performed consistently as well as metrics that also include cases and inpatient bed usage. Our results highlight that a key challenge for COVID-19 risk prediction is the changing relationship between indicators and outcomes of policy interest. Adaptive metrics therefore have a unique advantage in a rapidly evolving pandemic context.

Significance Statement

In the rapidly-evolving COVID-19 pandemic, public health risk metrics often become less relevant over time. Risk metrics are designed to predict future severe disease and mortality based on currently-available surveillance data, such as cases and hospitalizations. However, the relationship between cases, hospitalizations, and mortality has varied considerably over the course of the pandemic, in the context of new variants and shifts in vaccine- and infection-induced immunity. We propose an adaptive approach that regularly updates metrics based on the relationship between surveillance inputs and future outcomes of policy interest. Our method captures changing pandemic dynamics, requires only hospitalization input data, and outperforms static risk metrics in predicting high-risk states and counties.

Using digital traces to build prospective and real-time county-level early warning systems to anticipate COVID-19 outbreaks in the United States

Coronavirus disease 2019 (COVID-19) continues to affect the world, and the design of strategies to curb disease outbreaks requires close monitoring of their trajectories. We present machine learning methods that leverage internet-based digital traces to anticipate sharp increases in COVID-19 activity in U.S. counties. In a complementary direction to the efforts led by the Centers for Disease Control and Prevention (CDC), our models are designed to detect the time when an uptrend in COVID-19 activity will occur. Motivated by the need for finer spatial resolution epidemiological insights, we build upon previous efforts conceived at the state level. Our methods-tested in an out-of-sample manner, as events were unfolding, in 97 counties representative of multiple population sizes across the United States-frequently anticipated increases in COVID-19 activity 1 to 6 weeks before local outbreaks, defined when the effective reproduction number Rt becomes larger than 1 for a period of 2 weeks.

Context-dependent representation of within- and between-model uncertainty: aggregating probabilistic predictions in infectious disease epidemiology

Probabilistic predictions support public health planning and decision making, especially in infectious disease emergencies. Aggregating outputs from multiple models yields more robust predictions of outcomes and associated uncertainty. While the selection of an aggregation method can be guided by retrospective performance evaluations, this is not always possible. For example, if predictions are conditional on assumptions about how the future will unfold (e.g. possible interventions), these assumptions may never materialize, precluding any direct comparison between predictions and observations. Here, we summarize literature on aggregating probabilistic predictions, illustrate various methods for infectious disease predictions via simulation, and present a strategy for choosing an aggregation method when empirical validation cannot be used. We focus on the linear opinion pool (LOP) and Vincent average, common methods that make different assumptions about between-prediction uncertainty. We contend that assumptions of the aggregation method should align with a hypothesis about how uncertainty is expressed within and between predictions from different sources. The LOP assumes that between-prediction uncertainty is meaningful and should be retained, while the Vincent average assumes that between-prediction uncertainty is akin to sampling error and should not be preserved. We provide an R package for implementation. Given the rising importance of multi-model infectious disease hubs, our work provides useful guidance on aggregation and a deeper understanding of the benefits and risks of different approaches.

National and subnational short-term forecasting of COVID-19 in Germany and Poland during early 2021

BACKGROUND

During the COVID-19 pandemic there has been a strong interest in forecasts of the short-term development of epidemiological indicators to inform decision makers. In this study we evaluate probabilistic real-time predictions of confirmed cases and deaths from COVID-19 in Germany and Poland for the period from January through April 2021.

METHODS

We evaluate probabilistic real-time predictions of confirmed cases and deaths from COVID-19 in Germany and Poland. These were issued by 15 different forecasting models, run by independent research teams. Moreover, we study the performance of combined ensemble forecasts. Evaluation of probabilistic forecasts is based on proper scoring rules, along with interval coverage proportions to assess calibration. The presented work is part of a pre-registered evaluation study.

RESULTS

We find that many, though not all, models outperform a simple baseline model up to four weeks ahead for the considered targets. Ensemble methods show very good relative performance. The addressed time period is characterized by rather stable non-pharmaceutical interventions in both countries, making short-term predictions more straightforward than in previous periods. However, major trend changes in reported cases, like the rebound in cases due to the rise of the B.1.1.7 (Alpha) variant in March 2021, prove challenging to predict.

CONCLUSIONS

Multi-model approaches can help to improve the performance of epidemiological forecasts. However, while death numbers can be predicted with some success based on current case and hospitalization data, predictability of case numbers remains low beyond quite short time horizons. Additional data sources including sequencing and mobility data, which were not extensively used in the present study, may help to improve performance.

The United States COVID-19 Forecast Hub dataset

Academic researchers, government agencies, industry groups, and individuals have produced forecasts at an unprecedented scale during the COVID-19 pandemic. To leverage these forecasts, the United States Centers for Disease Control and Prevention (CDC) partnered with an academic research lab at the University of Massachusetts Amherst to create the US COVID-19 Forecast Hub. Launched in April 2020, the Forecast Hub is a dataset with point and probabilistic forecasts of incident cases, incident hospitalizations, incident deaths, and cumulative deaths due to COVID-19 at county, state, and national, levels in the United States. Included forecasts represent a variety of modeling approaches, data sources, and assumptions regarding the spread of COVID-19. The goal of this dataset is to establish a standardized and comparable set of short-term forecasts from modeling teams. These data can be used to develop ensemble models, communicate forecasts to the public, create visualizations, compare models, and inform policies regarding COVID-19 mitigation. These open-source data are available via download from GitHub, through an online API, and through R packages.

The United States COVID-19 Forecast Hub dataset

Academic researchers, government agencies, industry groups, and individuals have produced forecasts at an unprecedented scale during the COVID-19 pandemic. To leverage these forecasts, the United States Centers for Disease Control and Prevention (CDC) partnered with an academic research lab at the University of Massachusetts Amherst to create the US COVID-19 Forecast Hub. Launched in April 2020, the Forecast Hub is a dataset with point and probabilistic forecasts of incident cases, incident hospitalizations, incident deaths, and cumulative deaths due to COVID-19 at county, state, and national, levels in the United States. Included forecasts represent a variety of modeling approaches, data sources, and assumptions regarding the spread of COVID-19. The goal of this dataset is to establish a standardized and comparable set of short-term forecasts from modeling teams. These data can be used to develop ensemble models, communicate forecasts to the public, create visualizations, compare models, and inform policies regarding COVID-19 mitigation. These open-source data are available via download from GitHub, through an online API, and through R packages.