Back to Search
Start Over
Even a good influenza forecasting model can benefit from internet-based nowcasts, but those benefits are limited
- Source :
- PLoS Computational Biology, PLoS Computational Biology, Vol 15, Iss 2, p e1006599 (2019)
- Publication Year :
- 2019
- Publisher :
- Public Library of Science (PLoS), 2019.
-
Abstract
- The ability to produce timely and accurate flu forecasts in the United States can significantly impact public health. Augmenting forecasts with internet data has shown promise for improving forecast accuracy and timeliness in controlled settings, but results in practice are less convincing, as models augmented with internet data have not consistently outperformed models without internet data. In this paper, we perform a controlled experiment, taking into account data backfill, to improve clarity on the benefits and limitations of augmenting an already good flu forecasting model with internet-based nowcasts. Our results show that a good flu forecasting model can benefit from the augmentation of internet-based nowcasts in practice for all considered public health-relevant forecasting targets. The degree of forecast improvement due to nowcasting, however, is uneven across forecasting targets, with short-term forecasting targets seeing the largest improvements and seasonal targets such as the peak timing and intensity seeing relatively marginal improvements. The uneven forecasting improvements across targets hold even when “perfect” nowcasts are used. These findings suggest that further improvements to flu forecasting, particularly seasonal targets, will need to derive from other, non-nowcasting approaches.<br />Author summary It has been demonstrated in retrospective settings that flu forecasting can be improved by augmenting forecasting models with internet-based nowcasts (i.e., one-week-ahead forecasts generated with internet data). The improvement of internet-assisted forecasting models, however, has not translated to overall improvement in flu forecasting of public health-relevant targets. This is partially due to the relatively simple forecasting models that nowcasts are augmenting and partially because nowcasts do not directly inform all public health-relevant targets, such as the start, the peak timing, and the peak intensity of the flu season (i.e., seasonal targets). This paper conducts an experiment to provide clarity on the benefits and limitations of augmenting flu forecasting models with internet-based nowcasts in practice. We show that 1) nowcasts can improve flu forecasting across all geographic regions, flu seasons, and public health-relevant targets, including seasonal targets, and 2) the benefits of nowcasting are uneven, as even in the perfect nowcasting setting, only marginal improvements to forecasts of seasonal targets are possible relative to short-term forecasts. This is because nowcasts provide direct information about short-term targets but only indirect information about seasonal targets. This paper supports previous findings that nowcasts can improve flu forecasting but clarifies the limitations of those improvements, both in terms of magnitude of possible improvement as well as which targets can be improved. The continued advancement of flu forecasting hinges on the development of model improvements separate from nowcasting.
- Subjects :
- 0301 basic medicine
Viral Diseases
Nowcasting
Computer science
Social Sciences
Geographical locations
Disease Outbreaks
law.invention
Mathematical and Statistical Techniques
0302 clinical medicine
Sociology
Internet based
law
Medicine and Health Sciences
Econometrics
Public and Occupational Health
Biology (General)
Computer Networks
Controlled experiment
Geography
Ecology
Statistical Models
Statistics
Online Encyclopedias
Infectious Diseases
Computational Theory and Mathematics
Modeling and Simulation
Physical Sciences
The Internet
Public Health
Seasons
Research Article
Computer and Information Sciences
QH301-705.5
Research and Analysis Methods
03 medical and health sciences
Cellular and Molecular Neuroscience
Influenza, Human
Genetics
Humans
Mass Media
Statistical Methods
Molecular Biology
Ecology, Evolution, Behavior and Systematics
Internet
business.industry
Statistical model
Influenza
Communications
United States
Regional Geography
030104 developmental biology
North America
Earth Sciences
CLARITY
Encyclopedias
People and places
business
Mathematics
030217 neurology & neurosurgery
Forecasting
Subjects
Details
- ISSN :
- 15537358
- Volume :
- 15
- Database :
- OpenAIRE
- Journal :
- PLOS Computational Biology
- Accession number :
- edsair.doi.dedup.....8c5854a4929ebf2e2d9888032e3130ba