Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting
In this paper we investigate how various discretization schemes could be incorporated in regularization algorithms for stable parameter estimation and forecasting in epidemiology. Specifically, we compare parametric and nonparametric discretization tools in terms of their impact on the accuracy of r...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
AIMS Press
2019-04-01
|
| Series: | Mathematical Biosciences and Engineering |
| Subjects: | |
| Online Access: | https://www.aimspress.com/article/10.3934/mbe.2019182?viewType=HTML |
| _version_ | 1818288357514412032 |
|---|---|
| author | Aurelie Akossi Gerardo Chowell-Puente Alexandra Smirnova |
| author_facet | Aurelie Akossi Gerardo Chowell-Puente Alexandra Smirnova |
| author_sort | Aurelie Akossi |
| collection | DOAJ |
| description | In this paper we investigate how various discretization schemes could be incorporated in regularization algorithms for stable parameter estimation and forecasting in epidemiology. Specifically, we compare parametric and nonparametric discretization tools in terms of their impact on the accuracy of recovered disease parameters as well as their impact on future projections of new incidence cases. Both synthetic and real data for 1918 ``Spanish Flu" pandemic in San Francisco are considered. The discrete approximation of a time dependent transmission rate is combined with the Levenberg-Marquardt algorithm used to solve the nonlinear least squares problem aimed at fitting the model to limited incidence data for an unfolding outbreak. Our simulation study highlights the crucial role of a priori information at the early stage of an epidemic in mitigating the lack of stability in over-parameterized models with insufficient data. Fortunately, our results suggest that a balanced combination of problem-oriented regularization techniques is one way in which scientists can still draw useful conclusions about system parameters and in turn generate reliable forecasts that policy makers could use to guide control interventions. |
| first_indexed | 2024-12-13T01:55:06Z |
| format | Article |
| id | doaj.art-3271a988b7b944d7b46de37e807b34dc |
| institution | Directory Open Access Journal |
| issn | 1551-0018 |
| language | English |
| last_indexed | 2024-12-13T01:55:06Z |
| publishDate | 2019-04-01 |
| publisher | AIMS Press |
| record_format | Article |
| series | Mathematical Biosciences and Engineering |
| spelling | doaj.art-3271a988b7b944d7b46de37e807b34dc2022-12-22T00:03:24ZengAIMS PressMathematical Biosciences and Engineering1551-00182019-04-011653674369310.3934/mbe.2019182Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecastingAurelie Akossi0Gerardo Chowell-Puente1Alexandra Smirnova21. Department of Mathematics and Statistics, Georgia State University, Atlanta, USA2. Department of Population Health Sciences, Georgia State University, Atlanta, USA1. Department of Mathematics and Statistics, Georgia State University, Atlanta, USAIn this paper we investigate how various discretization schemes could be incorporated in regularization algorithms for stable parameter estimation and forecasting in epidemiology. Specifically, we compare parametric and nonparametric discretization tools in terms of their impact on the accuracy of recovered disease parameters as well as their impact on future projections of new incidence cases. Both synthetic and real data for 1918 ``Spanish Flu" pandemic in San Francisco are considered. The discrete approximation of a time dependent transmission rate is combined with the Levenberg-Marquardt algorithm used to solve the nonlinear least squares problem aimed at fitting the model to limited incidence data for an unfolding outbreak. Our simulation study highlights the crucial role of a priori information at the early stage of an epidemic in mitigating the lack of stability in over-parameterized models with insufficient data. Fortunately, our results suggest that a balanced combination of problem-oriented regularization techniques is one way in which scientists can still draw useful conclusions about system parameters and in turn generate reliable forecasts that policy makers could use to guide control interventions.https://www.aimspress.com/article/10.3934/mbe.2019182?viewType=HTMLdisease forecastingparameter estimationregularization |
| spellingShingle | Aurelie Akossi Gerardo Chowell-Puente Alexandra Smirnova Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting Mathematical Biosciences and Engineering disease forecasting parameter estimation regularization |
| title | Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting |
| title_full | Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting |
| title_fullStr | Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting |
| title_full_unstemmed | Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting |
| title_short | Numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting |
| title_sort | numerical study of discretization algorithms for stable estimation of disease parameters and epidemic forecasting |
| topic | disease forecasting parameter estimation regularization |
| url | https://www.aimspress.com/article/10.3934/mbe.2019182?viewType=HTML |
| work_keys_str_mv | AT aurelieakossi numericalstudyofdiscretizationalgorithmsforstableestimationofdiseaseparametersandepidemicforecasting AT gerardochowellpuente numericalstudyofdiscretizationalgorithmsforstableestimationofdiseaseparametersandepidemicforecasting AT alexandrasmirnova numericalstudyofdiscretizationalgorithmsforstableestimationofdiseaseparametersandepidemicforecasting |