A Neural Network Based Early Warning System for Diarrheal Disease Burden in the Asia Pacific Region
(2023) 35th Annual Conference of the International Society for Environmental Epidemiology- Abstract
- BACKGROUND AND AIM: Warming temperature promotes bacterial growth while extreme precipitation enhances fecal-oral route of exposures to these pathogens, particularly in resource limited settings, increasing burden of diarrheal diseases. Given climate change related increases in frequency of extreme events, there is a pressing need to develop meaningful early warning system with adequate lead time (months instead of days) to inform public health preparedness. METHOD: Using diarrheal disease data from Nepal (2002-2014), Taiwan (2008-2019) and Vietnam (2000-2015), we trained shallow neural network time-series models, and applied it to predict diarrheal disease incidence for the last year of available data with a 12-month lead time. RESULTS:... (More)
- BACKGROUND AND AIM: Warming temperature promotes bacterial growth while extreme precipitation enhances fecal-oral route of exposures to these pathogens, particularly in resource limited settings, increasing burden of diarrheal diseases. Given climate change related increases in frequency of extreme events, there is a pressing need to develop meaningful early warning system with adequate lead time (months instead of days) to inform public health preparedness. METHOD: Using diarrheal disease data from Nepal (2002-2014), Taiwan (2008-2019) and Vietnam (2000-2015), we trained shallow neural network time-series models, and applied it to predict diarrheal disease incidence for the last year of available data with a 12-month lead time. RESULTS: Accuracy of the neural network model varied based on the type of input parameter as well as by administrative districts, ranging from 33% to 100%. Even in the absence of most recent disease data, the model performed reasonably well, providing results that can be useful to generate categorical probability of disease burden (Low, Medium, High) several months ahead of time and guide public health decision-making at a seasonal time scale. Our approach can be extended to include other climate sensitive diseases. CONCLUSIONS: Further work is needed to carry out prospective evaluation of such early warnings to estimate disease burden at a seasonal scale. This will build confidence in application of early warning systems to enhance public health adaptation to climate sensitive diseases at a seasonal scale. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/cc115914-5bcc-404f-88d0-4532a545327a
- author
- Sapkota, Amir
; Cano Rau, l Cruz
; He, Hao
; Dhimal, Megnath
; Thu Dang, Anh
; Zhang, Linus
LU
; Ma, Tianzhou ; Liang Xin, Zhong ; Gao, Chuansi LU and Wang, Yu Chun
- organization
- publishing date
- 2023-09-17
- type
- Contribution to conference
- publication status
- published
- subject
- conference name
- 35th Annual Conference of the International Society for Environmental Epidemiology
- conference location
- Kaohsiung, Taiwan
- conference dates
- 2023-09-17 - 2023-09-21
- DOI
- 10.1289/isee.2023.SA-070
- language
- English
- LU publication?
- yes
- id
- cc115914-5bcc-404f-88d0-4532a545327a
- date added to LUP
- 2025-02-11 11:20:43
- date last changed
- 2025-04-04 14:59:39
@misc{cc115914-5bcc-404f-88d0-4532a545327a, abstract = {{BACKGROUND AND AIM: Warming temperature promotes bacterial growth while extreme precipitation enhances fecal-oral route of exposures to these pathogens, particularly in resource limited settings, increasing burden of diarrheal diseases. Given climate change related increases in frequency of extreme events, there is a pressing need to develop meaningful early warning system with adequate lead time (months instead of days) to inform public health preparedness. METHOD: Using diarrheal disease data from Nepal (2002-2014), Taiwan (2008-2019) and Vietnam (2000-2015), we trained shallow neural network time-series models, and applied it to predict diarrheal disease incidence for the last year of available data with a 12-month lead time. RESULTS: Accuracy of the neural network model varied based on the type of input parameter as well as by administrative districts, ranging from 33% to 100%. Even in the absence of most recent disease data, the model performed reasonably well, providing results that can be useful to generate categorical probability of disease burden (Low, Medium, High) several months ahead of time and guide public health decision-making at a seasonal time scale. Our approach can be extended to include other climate sensitive diseases. CONCLUSIONS: Further work is needed to carry out prospective evaluation of such early warnings to estimate disease burden at a seasonal scale. This will build confidence in application of early warning systems to enhance public health adaptation to climate sensitive diseases at a seasonal scale.}}, author = {{Sapkota, Amir and Cano Rau, l Cruz and He, Hao and Dhimal, Megnath and Thu Dang, Anh and Zhang, Linus and Ma, Tianzhou and Liang Xin, Zhong and Gao, Chuansi and Wang, Yu Chun}}, language = {{eng}}, month = {{09}}, title = {{A Neural Network Based Early Warning System for Diarrheal Disease Burden in the Asia Pacific Region}}, url = {{http://dx.doi.org/10.1289/isee.2023.SA-070}}, doi = {{10.1289/isee.2023.SA-070}}, year = {{2023}}, }