Evaluating short-term forecasting of COVID-19 cases among different epidemiological models under a Bayesian framework

Li, Qiwei; Bedi, Tejasv; Lehmann, Christoph U.; Xiao, Guanghua; Xie, Yang

doi:10.1093/gigascience/giab009

Cited by 11 publications

(15 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…During the COVID-19 pandemic, the use of phenomenological models for short-term prediction of the epidemic curve and/or derivation of characteristics of the epidemic wave has become very popular [12,25,28,29]. In this paper, we illustrate how simple phenomenological models such as the exponential model, logistic growth model and Richards model can be used to fit the daily number of new hospitalisations from epidemic outbreaks in a Bayesian framework.…”

Section: Discussionmentioning

confidence: 99%

“…While the RMSE is commonly used to assess the predictive performance, it has the disadvantage that it depends on the size of the epidemic, i.e. larger cases will tend to result in larger RMSE [25]. The MAPE and sMAPE are independent from the scale of the epidemic.…”

Section: Forecasting New Hospitalizations At Different Phases In the Epidemicmentioning

confidence: 99%

“…The proposed modeling procedure provides insights into ongoing outbreaks, and uses the available information at different phases of the epidemic. The proposed method facilitates real-time public health responses when faced with infectious disease outbreaks such as covid-19.Several alternative phenomenological models exist for the proposed phases of an epidemic outbreak.Chowell et al introduced the generalized logistic and generalized Richards growth model, which includes an additional deceleration of growth parameter and which allows for sub-exponential growth[3].Li et al show that also other alternative growth curve models can be used for short-term forecasting of COVID-19 cases and that these models can be cast into a Bayesian framework, including the generalized logistic and generalized Richards model, the von Bertalanffy model, Gompertz model and the generalized growth curve model[25].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Multivariate phenomenological models for real-time short-term forecasts of hospital capacity for COVID-19 in Belgium from March to June 2020

et al. 2021

View full text Add to dashboard Cite

Phenomenological models are popular for describing the epidemic curve. We present how they can be used at different phases in the epidemic, by modeling the daily number of new hospitalisations (or cases). As real-time prediction of the hospital capacity is important, a joint model of the new hospitalisations, number of patients in hospital and in intensive care unit is proposed. This model allows estimating the length of stay in hospital and ICU, even if no (or limited) individual level information on length of stay is available. Estimation is done in a Bayesian framework. In this framework, real-time alarms, defined as the probability of exceeding hospital capacity, can be easily derived. The methods are illustrated using data from the covid-19 pandemic in March-Jun 2020 in Belgium, but are widely applicable.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Forecasting New Hospitalizations At Different Phases In the Epidemicmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Multivariate phenomenological models for real-time short-term forecasts of hospital capacity for COVID-19 in Belgium from March to June 2020

et al. 2021

View full text Add to dashboard Cite

show abstract

“…(ii) It is a pure learning algorithm and model parameters do not provide any information of epidemiological interest [36].…”

Section: Types Of Modelingmentioning

confidence: 99%

Modeling and Simulation: A Study on Predicting the Outbreak of COVID-19 in Saudi Arabia

Msmali

Meetei

Mechai

et al. 2021

Discrete Dynamics in Nature and Society

View full text Add to dashboard Cite

The novel coronavirus disease (COVID-19) has resulted in an ongoing pandemic affecting the health system and economy of more than 200 countries worldwide. Mathematical models are used to predict the biological and epidemiological tendencies of an epidemic and to develop methods for controlling it. In this work, we use a mathematical model perspective to study the role of behavior change in slowing the spread of COVID-19 in Saudi Arabia. The real-time updated data from March 2, 2020, to January 8, 2021, were collected from the Saudi Ministry of Health, aiming to provide dynamic behaviors of the epidemic in Saudi Arabia. During this period, 363,692 people were infected, resulting in 6293 deaths, with a mortality rate of 1.73%. There was a weak positive relationship between the spread of infection and mortality R 2 = 0.459 . We used the susceptible-exposed-infection-recovered (SEIR) model, a logistic growth model, with a special focus on the exposed, infected, and recovered individuals to simulate the final phase of the outbreak. The results indicate that social distancing, hygienic conditions, and travel limitations are crucial measures to prevent further spread of the epidemic.

show abstract

“…Some of them, such as the Richards curve, have already been used in previous situations of this nature (Hsi et al [12], Wang et al [13]). Concerning COVID-19 specifically, mention must be made of Català et al [14], where the authors employed the Gompertz curve; and of Li et al [15], a comparative study performed using the Richards, logistic, Von Bertalanffy, and Gompertz curves, among others. However, such curves present a fairly rigid behavior, so it is necessary to resort to more flexible models.…”

Section: Introductionmentioning

confidence: 99%

Two Multi-Sigmoidal Diffusion Models for the Study of the Evolution of the COVID-19 Pandemic

et al. 2021

View full text Add to dashboard Cite

A proposal is made to employ stochastic models, based on diffusion processes, to represent the evolution of the SARS-CoV-2 virus pandemic. Specifically, two diffusion processes are proposed whose mean functions obey multi-sigmoidal Gompertz and Weibull-type patterns. Both are constructed by introducing polynomial functions in the ordinary differential equations that originate the classical Gompertz and Weibull curves. The estimation of the parameters is approached by maximum likelihood. Various associated problems are analyzed, such as the determination of initial solutions for the necessary numerical methods in practical cases, as well as Bayesian methods to determine the degree of the polynomial. Additionally, strategies are suggested to determine the best model to fit specific data. A practical case is developed from data originating from several Spanish regions during the first two waves of the COVID-19 pandemic. The determination of the inflection time instants, which correspond to the peaks of infection and deaths, is given special attention. To deal with this particular issue, point estimation as well as first-passage times have been considered.

show abstract

Evaluating short-term forecasting of COVID-19 cases among different epidemiological models under a Bayesian framework

Cited by 11 publications

References 56 publications

Multivariate phenomenological models for real-time short-term forecasts of hospital capacity for COVID-19 in Belgium from March to June 2020

Multivariate phenomenological models for real-time short-term forecasts of hospital capacity for COVID-19 in Belgium from March to June 2020

Modeling and Simulation: A Study on Predicting the Outbreak of COVID-19 in Saudi Arabia

Two Multi-Sigmoidal Diffusion Models for the Study of the Evolution of the COVID-19 Pandemic

Contact Info

Product

Resources

About