Covid-19: Comparison of Time Series Forecasting Models and Hybrid ARIMA-ANN

Priya, N. Hema; Harish, S; Natesan, Subramanian; Surendiran, B.

doi:10.1007/978-981-19-1122-4_59

Cited by 4 publications

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There is substantial literature on forecasting, particularly for influenza [4][5][6][7][8][9][10][11]. However, forecast models are usually assessed on whether they can detect increased activity associated with outbreaks or the accuracy of daily or weekly forecasts, not the accuracy of forecasting peak activity [12]. Model selection methods that minimise forecast errors or maximise the sensitivity and specificity of outbreak detection will not necessarily provide models that are optimised for forecasting the timing and intensity of seasonal peaks.…”

Section: Introductionmentioning

confidence: 99%

Machine learning forecasts for seasonal epidemic peaks: Lessons learnt from an atypical respiratory syncytial virus season

Morbey,

Todkill,

Watson

et al. 2023

PLoS ONE

View full text Add to dashboard Cite

Seasonal peaks in infectious disease incidence put pressures on health services. Therefore, early warning of the timing and magnitude of peak activity during seasonal epidemics can provide information for public health practitioners to take appropriate action. Whilst many infectious diseases have predictable seasonality, newly emerging diseases and the impact of public health interventions can result in unprecedented seasonal activity. We propose a Machine Learning process for generating short-term forecasts, where models are selected based on their ability to correctly forecast peaks in activity, and can be useful during atypical seasons. We have validated our forecasts using typical and atypical seasonal activity, using respiratory syncytial virus (RSV) activity during 2019–2021 as an example. During the winter of 2020/21 the usual winter peak in RSV activity in England did not occur but was ‘deferred’ until the Spring of 2021. We compare a range of Machine Learning regression models, with alternate models including different independent variables, e.g. with or without seasonality or trend variables. We show that the best-fitting model which minimises daily forecast errors is not the best model for forecasting peaks when the selection criterion is based on peak timing and magnitude. Furthermore, we show that best-fitting models for typical seasons contain different variables to those for atypical seasons. Specifically, including seasonality in models improves performance during typical seasons but worsens it for the atypical seasons.

show abstract

Section: Introductionmentioning

confidence: 99%

Machine learning forecasts for seasonal epidemic peaks: Lessons learnt from an atypical respiratory syncytial virus season

Morbey,

Todkill,

Watson

et al. 2023

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…However, forecast models are usually assessed on whether they can detect increased activity associated with outbreaks or the accuracy of daily or weekly forecasts, not the accuracy of forecasting peak activity (12). Model selection methods that minimise forecast errors or maximise the sensitivity and specificity of outbreak detection will not necessarily provide models that are optimised for forecasting the timing and intensity of seasonal peaks.…”

Section: Introductionmentioning

confidence: 99%

Machine learning forecasts for seasonal epidemic peaks: lessons learnt from an atypical respiratory syncytial virus season

Morbey¹,

Todkill²,

Watson³

et al. 2023

Preprint

View full text Add to dashboard Cite

Seasonal peaks in infectious disease incidence put pressures on health services. Therefore, early warning of the timing and magnitude of peak activity during seasonal epidemics can provide information for public health practitioners to take appropriate action. Whilst many infectious diseases have predictable seasonality, newly emerging diseases and the impact of public health interventions can result in unprecedented seasonal activity. We propose a machine learning process for generating short-term forecasts, where models are selected based on their ability to correctly forecast peaks in activity and can be useful during the aforementioned atypical seasonal activity, in contrast to traditional modelling. We have validated our forecasts using typical and atypical seasonal activity, using respiratory syncytial virus (RSV) activity during 2019-2021 as an example. During the winter of 2020/21 the usual winter peak in RSV activity in England did not occur but was deferred until the Spring of 2021. We compare a range of machine learning regression models, with alternate models including different independent variables, e.g. with or without seasonality or trend variables. We show that the best-fitting model which minimises daily forecast errors is not the best model for forecasting peaks when the selection criterion is based on peak timing and magnitude. Furthermore, we show that best-fitting models for typical seasons contain different variables to those for atypical black swan seasons. Specifically, including seasonality in models improves performance during typical seasons but worsens it for the atypical seasons. In conclusion, we have found that including seasonality in forecast models can result in overfitting, where the models are required to be used out-of-season or during atypical seasons.

show abstract

A glimpse into the future: modelling global prevalence of hypertension

Boateng,

Ampofo

2023

BMC Public Health

View full text Add to dashboard Cite

Background Hypertension is a major risk factor for cardiovascular diseases. Insights and foresights on trends of hypertension prevalence are crucial to informing health policymaking. We examined and projected the patterns of hypertension prevalence among sexes. Methods Using annual hypertension prevalence (18 + years) data sourced from WHO Global Health Observatory data repository from 1975 to 2015, Prophet models were developed to forecast the 2040 prevalence of hypertension in males, females, and both sexes. We used k-means clustering and self-organising maps to determine the clusters of hypertension prevalence concerning both sexes among 176 countries. Results Worldwide, Croatia is estimated to have the highest prevalence of hypertension in males by 2040, while that of females is in Niger. Among the world’s most populated countries, Pakistan and India are likely to increase by 7.7% and 4.0% respectively in both sexes. South-East Asia is projected to experience the largest hypertension prevalence in males, whereas Africa is estimated to have the highest prevalence of hypertension in females. Low-income countries are projected to have the highest prevalence of hypertension in both sexes. By 2040, the prevalence of hypertension worldwide is expected to be higher in the male population than in female. Globally, the prevalence of hypertension is projected to decrease from 22.1% in 2015 to 20.3% (20.2 – 20.4%) in 2040. We also identified three patterns of hypertension prevalence in 2040, cluster one countries are estimated to have the highest prevalence of hypertension in males (29.6%, 22.2 – 41.1%) and females (29.6%, 19.4 – 38.7%). Conclusion These findings emphasise the need for new and effective approaches toward the prevention and control of hypertension in Africa, South-East Asia, and Low-income countries.

show abstract

Covid-19: Comparison of Time Series Forecasting Models and Hybrid ARIMA-ANN

Cited by 4 publications

References 17 publications

Machine learning forecasts for seasonal epidemic peaks: Lessons learnt from an atypical respiratory syncytial virus season

Machine learning forecasts for seasonal epidemic peaks: Lessons learnt from an atypical respiratory syncytial virus season

Machine learning forecasts for seasonal epidemic peaks: lessons learnt from an atypical respiratory syncytial virus season

A glimpse into the future: modelling global prevalence of hypertension

Contact Info

Product

Resources

About