A primer on model selection using the Akaike Information Criterion

Portet, Stéphanie

doi:10.1016/j.idm.2019.12.010

Cited by 271 publications

(218 citation statements)

References 20 publications

Supporting

Mentioning

214

Contrasting

Unclassified

Order By: Relevance

“…When a large number of models are under consideration or the models are not nested, the model selection rules are different. We refer the reader to recent lecture notes (Portet, 2020) for an introduction to model selection.…”

Section: Methods Of Model Selection Using Akaike Information Criterionmentioning

confidence: 99%

“…In this section, we give a brief description of a model calibration method based on Bayesian inference and the method of model selection using Akaike Information Criterion (AIC). For more details the reader is referred to (Portet, 2020;Roda, 2020). Other model calibration procedures using nonlinear squares or more general maximum likelihood methods are not described here, and we refer the reader to (Rossi, 2018).…”

Section: Model Calibration and Model Selectionmentioning

confidence: 99%

See 1 more Smart Citation

Why is it difficult to accurately predict the COVID-19 epidemic?

Roda

Varughese

Han

et al. 2020

Infectious Disease Modelling

590

597

View full text Add to dashboard Cite

SIR and SEIR models Bayesian inference Model selection Nonidentifiability Quarantine Peak time of epidemic a b s t r a c tSince the COVID-19 outbreak in Wuhan City in December of 2019, numerous model predictions on the COVID-19 epidemics in Wuhan and other parts of China have been reported. These model predictions have shown a wide range of variations. In our study, we demonstrate that nonidentifiability in model calibrations using the confirmed-case data is the main reason for such wide variations. Using the Akaike Information Criterion (AIC) for model selection, we show that an SIR model performs much better than an SEIR model in representing the information contained in the confirmed-case data. This indicates that predictions using more complex models may not be more reliable compared to using a simpler model. We present our model predictions for the COVID-19 epidemic in Wuhan after the lockdown and quarantine of the city on January 23, 2020. We also report our results of modeling the impacts of the strict quarantine measures undertaken in the city after February 7 on the time course of the epidemic, and modeling the potential of a second outbreak after the return-to-work in the city.

show abstract

Section: Methods Of Model Selection Using Akaike Information Criterionmentioning

confidence: 99%

Section: Model Calibration and Model Selectionmentioning

confidence: 99%

Why is it difficult to accurately predict the COVID-19 epidemic?

Roda

Varughese

Han

et al. 2020

Infectious Disease Modelling

590

597

View full text Add to dashboard Cite

show abstract

“…The fits of the unconstrained and constrained models then were examined using Akaike's information criterion for small sample sizes (AICc) (Portet, 2020), where:…”

Section: Schild Methodsmentioning

confidence: 99%

Predicting the activation of the androgen receptor by mixtures of ligands using Generalized Concentration Addition

Schlezinger

Heiger-Bernays

Webster

2020

Preprint

View full text Add to dashboard Cite

Concentration/dose addition (CA) is widely used for compounds that act by similar mechanisms.But, CA cannot make predictions for mixtures of full and partial agonists for effect levels above that of the least efficacious component. As partial agonists are common, we developed Generalized Concentration Addition (GCA), which has been successfully applied to systems in which ligands compete for a single binding site. Here, we applied a pharmacodynamic model for a system with two binding sites, the androgen receptor (AR). AR acts according to the classic homodimer activation model: each cytoplasmic AR protein binds ligand, undergoes a conformational change that relieves inhibition of dimerization, and binds to DNA response elements as a dimer. We generated individual dose-response data for full (dihydroxytestosterone, BMS564929) and partial (TFM-4AS-1) agonists and a competitive antagonist (MDV3100) using reporter data generated in the MDA-kb2 cell line. We used the Schild method to estimate the binding affinity of AR for MDV3100. Data for individual compounds fit the AR pharmacodynamic model well. The partial agonist had agonistic effects at low effect levels and antagonistic effects at high levels, as predicted by pharmacological theory. The GCA model fit the empirical mixtures data-full/full agonist, full/partial agonist and full agonist/antagonist-as well or better than relative potency factors (a special case of CA) or effect summation. The ability of generalized concentration addition to predict the activity of mixtures of different types of androgen receptor ligands is important as a number of environmental compounds act as partial AR agonists or antagonists.

show abstract

“…We used the Ensemble Adjustment Kalman Filter (EAKF) to infer epidemiological parameters of the model based on the number of cases presenting symptoms per day in Henan province [21][22][23]. The EAKF is a data assimilation algorithm, which only needs hundreds of ensemble members to obtain good results, especially suitable for the estimation of high-dimensional parameters of the model [24,25], and has been successfully applied to epidemics such as cholera and influenza [22,25]. In this study,…”

Section: Estimation Of Parameters In the Modelmentioning

confidence: 99%

Epidemic Dynamics of COVID-19 Based on SEAIUHR Model Considering Asymptomatic Cases in Henan Province, China

Zhu

Chang

et al. 2020

Preprint

View full text Add to dashboard Cite

Background New coronavirus disease (COVID-19), an infectious disease caused by a type of novel coronavirus, has emerged in various countries since the end of 2019 and caused a global pandemic. Many infected people went undetected because their symptoms were mild or asymptomatic, but the proportion and infectivity of asymptomatic infections remained unknown. Therefore, in this paper, we analyzed the proportion and infectivity of asymptomatic cases, as we as the prevalence of COVID-19 in Henan province. Methods We constructed SEAIUHR model based on COVID-19 cases reported from 21 January to 26 February 2020 in Henan province to estimate the proportion and infectivity of asymptomatic cases, as we as the change of effective reproductive number, \({R}_{t}\). At the same time, we simulated the changes of cases in different scenarios by changing the time and intensity of the implementation of prevention and control measures. Results The proportion of asymptomatic cases among COVID-19 infected individuals was 42% and infectivity of asymptomatic cases was 10% of that symptomatic ones. The basic reproductive number\({R}_{0}\)=2.73, and \({R}_{t}\) dropped below 1 on 1 February under a series of measures. If measures were taken five days earlier, the number of cases would be reduced by 2/3, and after 5 days the number would more than triple. Conclusions In Henan Province, the COVID-19 epidemic spread rapidly in the early stage, and there were a large number of asymptomatic infected individuals with relatively low infectivity. However, the epidemic was quickly brought under control with national measures, and the earlier measures were implemented, the better.

show abstract

A primer on model selection using the Akaike Information Criterion

Cited by 271 publications

References 20 publications

Why is it difficult to accurately predict the COVID-19 epidemic?

Why is it difficult to accurately predict the COVID-19 epidemic?

Predicting the activation of the androgen receptor by mixtures of ligands using Generalized Concentration Addition

Epidemic Dynamics of COVID-19 Based on SEAIUHR Model Considering Asymptomatic Cases in Henan Province, China

Contact Info

Product

Resources

About