Forecasting the spread of COVID-19 based on policy, vaccination, and Omicron data

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: April 30, 2024

The COVID-19 pandemic caused by the novel SARS-COV-2 virus poses a great risk to world. During pandemic, observing and forecasting several important indicators of epidemic (like new confirmed cases, cases in intensive care unit, deaths for each day) helped prepare appropriate response (e.g., creating additional unit beds, implementing strict interventions). Various predictive models predictor variables have been used forecast these indicators. However, impact prediction on performance has not systematically well analyzed. Here, we compared using linear mixed model terms (mathematical, statistical, AI/machine learning models) (vaccination rate, stringency index, Omicron variant rate) seven selected countries with highest vaccination rates. We decided our best based Bayesian Information Criterion (BIC) analyzed significance predictor. Simple were preferred. selection use rate considered essential improving accuracies. For test data period before emergence, was most significant factor accuracy. after deciding models, ARIMA, lightGBM, TSGLM generally performed both periods. Linear country as random effect proven that choice determining accuracies highly vaccinated countries. Relatively simple fit either or data, produced results enhancing data.

Language: Английский

---

Model Diagnostics and Forecast Evaluation for Quantiles

Annual Review of Statistics and Its Application, Journal Year: 2022, Volume and Issue: 10(1), P. 597 - 621

Published: Nov. 1, 2022

Model diagnostics and forecast evaluation are closely related tasks, with the former concerning in-sample goodness (or lack) of fit latter addressing predictive performance out-of-sample. We review ubiquitous setting in which forecasts cast form quantiles or quantile-bounded prediction intervals. distinguish unconditional calibration, corresponds to classical coverage criteria, from stronger notion conditional as can be visualized quantile reliability diagrams. Consistent scoring functions—including, but not limited to, widely used asymmetricpiecewise linear score pinball loss—provide for comparative assessment ranking, link coefficient determination skill scores. illustrate use these tools on Engel's food expenditure data, Global Energy Forecasting Competition 2014, US COVID-19 Forecast Hub.

Language: Английский

---

The accuracy of forecasted hospital admission for respiratory tract infections in children aged 0–5 years for 2017/2023

Frontiers in Pediatrics, Journal Year: 2025, Volume and Issue: 12

Published: Jan. 6, 2025

Healthcare services are in need of tools that can help to ensure a sufficient capacity periods with high prevalence respiratory tract infections (RTIs). During the COVID-19 pandemic, we forecasted number hospital admissions for RTIs among children aged 0-5 years. Now, 2024, aim examine accuracy and usefulness our forecast models. We conducted retrospective analysis using data from 753,070 years, plotting observed monthly RTI admissions, including influenza coded RTI, syncytial virus (RSV) other upper lower January 1st, 2017, until May 31st, 2023. determined four different models, all based on assumptions regarding pattern transmission, computed ordinary least squares regression adjusting seasonal trends. compared vs. numbers between October 2021, 2023, metrics such as mean absolute error (MAE), percentage (MAPE) dynamic time warping (DTW). In most accurate prediction, assumed proportion who remained uninfected non-hospitalized during lockdown would be prone hospitalization subsequent season, resulting increased when measures were eased. this difference at peak hospitalizations requiring not support November 2021 2022 was 26 (394 420) 48 (1810 1762). scenarios similar transmission viruses is suppressed an extended period, simple model, assuming hospitalized following accurately admission numbers. These forecasts may useful planning activities hospitals.

Language: Английский

---

Time series analysis and prediction of the trends of COVID-19 epidemic in Singapore based on machine learning

Computer Methods and Programs in Biomedicine Update, Journal Year: 2025, Volume and Issue: unknown, P. 100190 - 100190

Published: April 1, 2025

Language: Английский

---

Understanding the leading indicators of hospital admissions from COVID-19 across successive waves in the UK

Epidemiology and Infection, Journal Year: 2023, Volume and Issue: 151

Published: Jan. 1, 2023

Following the end of universal testing in UK, hospital admissions are a key measure COVID-19 pandemic pressure. Understanding leading indicators at National Health Service (NHS) Trust, regional and national geographies help health services plan for ongoing pressures. We explored spatio-temporal relationships hospitalisations across SARS-CoV-2 waves England. This analysis includes an evaluation internet search volumes from Google Trends, NHS triage calls online queries, app, lateral flow devices (LFDs), ZOE app. Data sources were analysed their feasibility as using Granger causality, cross-correlation, dynamic time warping fine spatial scales. Trends triages consistently temporally led most locations, with lead times ranging 5 to 20 days, whereas inconsistent relationship was found LFD testing, which diminished resolution, showing cross-correlation leads between -7 7 days. The results indicate that novel surveillance can be used effectively understand expected healthcare burden within administrative areas though temporal heterogeneity these is determinant operational public utility.

Language: Английский

---

Applications of time series analysis in epidemiology: Literature review and our experience during COVID-19 pandemic

World Journal of Clinical Cases, Journal Year: 2023, Volume and Issue: 11(29), P. 6974 - 6983

Published: Oct. 13, 2023

Time series analysis is a valuable tool in epidemiology that complements the classical epidemiological models two different ways: Prediction and forecast. related to explaining past current data based on various internal external influences may or not have causative role. Forecasting an exploration of possible future values predictive ability model hypothesized and/or influences. The time approach has advantage being easier use (in cases more straightforward linear such as Auto-Regressive Integrated Moving Average). Still, it limited forecasting time, unlike Susceptible-Exposed-Infectious-Removed. Its applicability comes from its better accuracy for short-term prediction. In basic form, does assume much theoretical knowledge mechanisms spreading mutating pathogens reaction people regulatory structures (governments, companies, etc. ). Instead, estimates directly. allows testing hypotheses factors positively negatively contribute pandemic spread; be school closures, emerging variants, It can used mortality hospital risk estimation new cases, seroprevalence studies, assessing properties estimating excess relationship with pandemic.

Language: Английский

---

Quantifying the impact of hospital catchment area definitions on hospital admissions forecasts: COVID-19 in England, September 2020–April 2021

BMC Medicine, Journal Year: 2024, Volume and Issue: 22(1)

Published: April 17, 2024

Abstract Background Defining healthcare facility catchment areas is a key step in predicting future demand epidemic settings. Forecasts of hospitalisations can be informed by leading indicators measured at the community level. However, this relies on definition so-called or geographies whose populations make up patients admitted to given hospital, which are often not well-defined. Little work has been done quantify impact hospital area definitions forecasting. Methods We made forecasts local-level admissions using scaled convolution local cases (as defined area) and delay distribution. Hospital were derived from either simple heuristics (in people their nearest any nearby hospital) historical data (all emergency elective 2019, COVID-19 admissions), plus marginal baseline based distribution all admissions. evaluated predictive performance each weighted interval score considered how changed length horizon, date forecast was made, location. also change, if any, relative retrospective vs. real-time settings, different spatial scales. Results The choice affected accuracy admission forecasts. resulted most accurate both 7- 14-day horizon one top two best-performing across dates locations. “nearby” heuristic performed well, but less consistently than definition. baseline, did include information, lowest-ranked larger when case compared observed cases. All results consistent scales definitions. Conclusions Using context-specific improve Where available, carefully chosen sufficiently good substitute. There clear value understanding what drives patterns, further research needed understand where trends more heterogeneous.

Language: Английский

---

Real-time forecasting of COVID-19-related hospital strain in France using a non-Markovian mechanistic model

PLoS Computational Biology, Journal Year: 2024, Volume and Issue: 20(5), P. e1012124 - e1012124

Published: May 17, 2024

Projects such as the European Covid-19 Forecast Hub publish forecasts on national level for new deaths, cases, and hospital admissions, but not direct measurements of strain like critical care bed occupancy at sub-national level, which is particular interest to health professionals planning purposes. We present a French framework forecasting based non-Markovian compartmental model, its associated online visualisation tool retrospective evaluation real-time it provided from January December 2021 by comparing three baselines derived standard statistical methods (a naive auto-regression, an ensemble exponential smoothing ARIMA). In terms median absolute error unit two-week horizon, our model only outperformed baseline 4 out 14 geographical units underperformed compared 5 them 90% confidence ( n = 38). However, same week was never statistically any despite outperforming 10 times spanning 7 units. This implies modest utility longer horizons may justify application models in context hospital-strain surveillance future pandemics.

Language: Английский

---

An adaptive weight ensemble approach to forecast influenza activity in an irregular seasonality context

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Oct. 4, 2024

Language: Английский

---

Decision-Making Algorithm and Predictive Model to Assess the Impact of Infectious Disease Epidemics on the Healthcare System: The COVID-19 Case Study in Italy

IEEE Journal of Biomedical and Health Informatics, Journal Year: 2022, Volume and Issue: 26(8), P. 3661 - 3672

Published: May 11, 2022

To improve decision-making strategies and prediction based on epidemiological data, so far biased by highly-variable criteria, algorithms using unbiased morbidity parameters, i.e. Intensive Care Units (ICU) Ordinary Hospitalizations (OH), are proposed. ICU/OH acceleration velocities mathematically modeled available official data to derive two thresholds, alerting 30 % ICU 40 OH of COVID-19 daily occupancy settled the Italian Minister Health, as a case study. A predictive model is also proposed estimate in hospitals for each region, Susceptible-Infected-Recovered-Death (SIRD) epidemic further extend regional district. Computed validated models Italy after almost years pandemic, obtaining agreements with Presidential Decree regardless different trends waves. Therefore, algorithm resulted valuable tools, retrospectively, be tested prospectively sustainable curb impact COVID-19, or any other pandemic threats aggregate local healthcare systems.

Language: Английский

---