PAN-cODE: COVID-19 forecasting using conditional latent ODEs DOI

Ruian Shi,

Haoran Zhang, Quaid Morris

et al.

Journal of the American Medical Informatics Association, Journal Year: 2022, Volume and Issue: 29(12), P. 2089 - 2095

Published: Sept. 1, 2022

The coronavirus disease 2019 (COVID-19) pandemic has caused millions of deaths around the world and revealed need for data-driven models spread. Accurate caseload forecasting allows informed policy decisions on adoption non-pharmaceutical interventions (NPIs) to reduce transmission. Using COVID-19 as an example, we present Pandemic conditional Ordinary Differential Equation (PAN-cODE), a deep learning method forecast daily increases in infections deaths. By using latent variable model, PAN-cODE can generate alternative trajectories based alternate adoptions NPIs, allowing stakeholders make manner. also estimation regions that are unseen during model training. We demonstrate that, despite less detailed data having fully automated training, PAN-cODE's performance is comparable state-of-the-art methods 4-week-ahead 6-week-ahead forecasting. Finally, highlight ability realistic outcome select US regions.

Language: Английский

Deep learning for Covid-19 forecasting: State-of-the-art review. DOI
Firuz Kamalov, Khairan Rajab, Aswani Kumar Cherukuri

et al.

Neurocomputing, Journal Year: 2022, Volume and Issue: 511, P. 142 - 154

Published: Sept. 8, 2022

Language: Английский

Citations

38
An evaluation of prospective COVID-19 modelling studies in the USA: from data to science translation DOI Creative Commons

Kristen Nixon,

Sonia Jindal,

Felix Parker

et al.

The Lancet Digital Health, Journal Year: 2022, Volume and Issue: 4(10), P. e738 - e747

Published: Sept. 20, 2022

Infectious disease modelling can serve as a powerful tool for situational awareness and decision support policy makers. However, COVID-19 efforts faced many challenges, from poor data quality to changing human behaviour. To extract practical insight the large body of literature available, we provide narrative review with systematic approach that quantitatively assessed prospective, data-driven studies in USA. We analysed 136 papers, focused on aspects models are essential have documented forecasting window, methodology, prediction target, datasets used, geographical resolution each study. also found fraction papers did not evaluate performance (25%), express uncertainty (50%), or state limitations (36%). remedy some these identified gaps, recommend adoption EPIFORGE 2020 model reporting guidelines creating an information-sharing system is suitable fast-paced infectious outbreak science.

Language: Английский

Citations

37
A spatiotemporal machine learning approach to forecasting COVID-19 incidence at the county level in the USA DOI Open Access
Benjamín Lucas, Behzad Vahedi, Morteza Karimzadeh

et al.

International Journal of Data Science and Analytics, Journal Year: 2022, Volume and Issue: 15(3), P. 247 - 266

Published: Jan. 15, 2022

Language: Английский

Citations

35
Real-time mechanistic Bayesian forecasts of COVID-19 mortality DOI
Graham Gibson, Nicholas G Reich, Daniel Sheldon

et al.

The Annals of Applied Statistics, Journal Year: 2023, Volume and Issue: 17(3)

Published: Sept. 1, 2023

The COVID-19 pandemic emerged in late December 2019. In the first six months of global outbreak, U.S. reported more cases and deaths than any other country world. Effective modeling course can help assist with public health resource planning, intervention efforts, vaccine clinical trials. However, building applied forecasting models presents unique challenges during a pandemic. First, case data available to real time represent nonstationary fraction true incidence due changes diagnostic tests test-seeking behavior. Second, interventions varied across geography leading large transmissibility over We propose mechanistic Bayesian model that builds upon classic compartmental susceptible–exposed–infected–recovered (SEIR) operationalize time. This framework includes nonparametric varying transmission rates, death discrepancies testing reporting issues, joint observation likelihood on new counts deaths; it is implemented probabilistic programming language automate use reasoning for quantifying uncertainty forecasts. has been used submit forecasts Centers Disease Control through Forecast Hub under name MechBayes. examine performance relative baseline as well alternate submitted forecast hub. Additionally, we include an ablation test our extensions SEIR model. demonstrate significant gain both point scoring measures using MechBayes, when compared model, show MechBayes ranks one top two out nine which regularly duration pandemic, trailing only ensemble part.

Language: Английский

Citations

12
Machine learning for data-centric epidemic forecasting DOI
Alexander Rodríguez, Harshavardhan Kamarthi,

Pulak Agarwal

et al.

Nature Machine Intelligence, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 27, 2024

Language: Английский

Citations

4
Auditing the fairness of the US COVID-19 forecast hub’s case prediction models DOI Creative Commons
Saad Mohammad Abrar, Naman Awasthi, Daniel Smolyak

et al.

PLoS ONE, Journal Year: 2025, Volume and Issue: 20(4), P. e0319383 - e0319383

Published: April 22, 2025

The US COVID-19 Forecast Hub, a repository of forecasts from over 50 independent research groups, is used by the Centers for Disease Control and Prevention (CDC) their official communications. As such, Hub critical centralized resource to promote transparent decision making. While has provided valuable predictions focused on accuracy, there an opportunity evaluate model performance across social determinants such as race urbanization level that have been known play role in pandemic. In this paper, we carry out comprehensive fairness analysis show statistically significant diverse predictive determinants, with minority racial ethnic groups well less urbanized areas often associated higher prediction errors. We hope work will encourage modelers CDC report metrics together reflect potential harms models specific contexts.

Language: Английский

Citations

0
Comprehensive compartmental model and calibration algorithm for the study of clinical implications of the population-level spread of COVID-19: a study protocol DOI Creative Commons

Brandon Robinson,

Jodi D. Edwards, Tetyana Kendzerska

et al.

BMJ Open, Journal Year: 2022, Volume and Issue: 12(3), P. e052681 - e052681

Published: March 1, 2022

The complex dynamics of the coronavirus disease 2019 (COVID-19) pandemic has made obtaining reliable long-term forecasts progression difficult. Simple mechanistic models with deterministic parameters are useful for short-term predictions but have ultimately been unsuccessful in extrapolating trajectory because unmodelled and unrealistic level certainty that is assumed predictions.

Language: Английский

Citations

5
Estimating and forecasting the burden and spread of Colombia’s SARS-CoV2 first wave DOI Creative Commons

Jaime Cascante-Vega,

Juan Manuel Cordovez, Mauricio Santos‐Vega

et al.

Scientific Reports, Journal Year: 2022, Volume and Issue: 12(1)

Published: Aug. 9, 2022

Following the rapid dissemination of COVID-19 cases in Colombia 2020, large-scale non-pharmaceutical interventions (NPIs) were implemented as national emergencies most country's municipalities, starting with a lockdown on March 20th, 2020. Recently, approaches that combine movement data (measured number commuters between units), metapopulation models to describe disease dynamics subdividing population into Susceptible-Exposed-Asymptomatic-Infected-Recovered-Diseased and statistical inference algorithms have been pointed practical approach both nowcast forecast deaths. We used an iterated filtering (IF) framework estimate model transmission parameters using reported across 281 municipalities from late October locations more than 50 deaths Colombia. Since is high dimensional (6 state variables every municipality), those highly non-trivial, so we Ensemble-Adjustment-Kalman-Filter (EAKF) time variable system states parameters. Our results show model's ability capture characteristics outbreak country provide estimates epidemiological at level. Importantly, these could become base for planning future well evaluating impact NPIs effective reproduction ([Formula: see text]) critical parameters, such contact rate or reporting rate. However, our presents some inconsistency it overestimates Medellín. Nevertheless, demonstrates real-time, publicly available ensemble forecasts can short-term predictions Therefore, this be forecasting tool evaluate aid policymakers infectious management control.

Language: Английский

Citations

4
Chimeric forecasting: combining probabilistic predictions from computational models and human judgment DOI Creative Commons
Thomas McAndrew, Allison Codi,

Juan Cambeiro

et al.

BMC Infectious Diseases, Journal Year: 2022, Volume and Issue: 22(1)

Published: Nov. 10, 2022

Abstract Forecasts of the trajectory an infectious agent can help guide public health decision making. A traditional approach to forecasting fits a computational model structured data and generates predictive distribution. However, human judgment has access same as models plus experience, intuition, subjective data. We propose chimeric ensemble—a combination forecasts—as novel predicting agent. Each month from January, 2021 June, we asked two generalist crowds, using criteria COVID-19 Forecast Hub, submit distribution over incident cases deaths at US national level either or three weeks into future combined these forecasts with submitted Forecasthub ensemble. find ensemble compared including only improves predictions shows similar performance for deaths. is flexible, supportive tool promising results spread

Language: Английский

Citations

4
A data-driven semi-parametric model of SARS-CoV-2 transmission in the United States DOI Creative Commons
John M. Drake, Andreas Handel, Éric Marty

et al.

PLoS Computational Biology, Journal Year: 2023, Volume and Issue: 19(11), P. e1011610 - e1011610

Published: Nov. 8, 2023

To support decision-making and policy for managing epidemics of emerging pathogens, we present a model inference scenario analysis SARS-CoV-2 transmission in the USA. The stochastic SEIR-type includes compartments latent, asymptomatic, detected undetected symptomatic individuals, hospitalized cases, features realistic interval distributions presymptomatic periods, time varying rates case detection, diagnosis, mortality. accounts effects on human mobility using anonymized data collected from cellular devices, difficult to quantify environmental behavioral factors latent process. baseline rate is product metric obtained this fitted We fit incident death reports each state USA Washington D.C., likelihood Maximization by Iterated particle Filtering (MIF). Observations (daily reports) are modeled as arising negative binomial reporting estimate time-varying rate, parameters sigmoidal fraction cases that result death, extra-demographic process noise, two dispersion observation process, initial sizes classes. In retrospective covering March–December 2020, show how strength became decoupled across distinct phases pandemic. decoupling demonstrates need flexible, semi-parametric approaches modeling infectious disease dynamics real-time.

Language: Английский

Citations

2