Predicting mosquito-borne disease outbreaks using poisson and negative binomial models: A comparative study.
Uložené v:
| Názov: | Predicting mosquito-borne disease outbreaks using poisson and negative binomial models: A comparative study. |
|---|---|
| Autori: | Al-Manji A; Department of Family Medicine and Public Health, College of Medicine and Health Sciences, Sultan Qaboos University, Muscat, Oman. Electronic address: s132154@student.squ.edu.om., Al Wahaibi A; Ministry of Health, Muscat, Oman. Electronic address: adilwahaibi@gmail.com., Al-Azri M; Department of Family Medicine and Public Health, College of Medicine and Health Sciences, Sultan Qaboos University, Muscat, Oman. Electronic address: mhalazri@squ.edu.om., Chan MF; Department of Family Medicine and Public Health, College of Medicine and Health Sciences, Sultan Qaboos University, Muscat, Oman. Electronic address: moonf@squ.edu.om. |
| Zdroj: | Journal of infection and public health [J Infect Public Health] 2025 Nov; Vol. 18 (11), pp. 102906. Date of Electronic Publication: 2025 Jul 18. |
| Spôsob vydávania: | Journal Article; Comparative Study |
| Jazyk: | English |
| Informácie o časopise: | Publisher: Elsevier Country of Publication: England NLM ID: 101487384 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1876-035X (Electronic) Linking ISSN: 18760341 NLM ISO Abbreviation: J Infect Public Health Subsets: MEDLINE |
| Imprint Name(s): | Original Publication: Oxford : Elsevier, 2008- |
| Výrazy zo slovníka MeSH: | Disease Outbreaks* , Dengue*/epidemiology , Vector Borne Diseases*/epidemiology, Humans ; Retrospective Studies ; Animals ; Oman/epidemiology ; Mosquito Vectors/virology ; Poisson Distribution ; Bayes Theorem ; Models, Statistical ; Forecasting/methods ; Temperature ; Mosquito-Borne Diseases |
| Abstrakt: | Competing Interests: Declaration of Competing Interest The authors declare no financial or personal conflicts of interest that could have influenced the findings of this study, “Predicting Mosquito-Borne Disease Outbreaks Using Poisson and Negative Binomial Models: A Comparative Study”. Background: Dengue fever, a major mosquito-borne disease (MBD), continues to impose a growing global burden fueled by urbanization, climate change, and increased human mobility. Accurate predictive models are crucial for early detection and outbreak mitigation. This study aimed to develop and compare hierarchical models, with and without lagged predictors, for forecasting dengue cases in Oman. Methods: A retrospective analysis was conducted using weekly data from 2020 to 2024 across multiple districts. Predictors included climate variables (temperature, humidity, wind, rainfall), mosquito surveillance indicators (trap positivity, mosquito density), and population demographics. Four hierarchical Bayesian models were developed: Poisson without lag, Poisson with lag, Negative Binomial without lag, and Negative Binomial with lag. Models incorporated fixed effects and random intercepts for epidemiological week, district, governorate, year, and seasonal components. Model performance was evaluated through convergence diagnostics, Mean Squared Error (MSE), Area Under the Curve (AUC), confusion matrices, and Leave-One-Out Information Criterion (LOOIC). Results: All models demonstrated excellent convergence and fit the historical weekly data (2020-2024) accurately. The Negative Binomial model with lagged variables performed best, achieving the highest AUC (0.881, 95 % CI: 0.858-0.902), the lowest LOOIC (3234.6 ± 109.4), and the smallest MSE. Mosquito trap positivity was consistently the strongest predictor, while wind speed showed a moderate positive effect and temperature showed significant delayed negative effects. Rainfall, humidity, and population size were not significant predictors. Importantly, short-term forecasts for the first weeks of 2025 closely matched the observed case counts, confirming that the models' prediction metrics reflected both retrospective fit and real-world forecasting performance. Conclusions: Incorporating delayed climatic and entomological factors using a Negative Binomial hierarchical framework significantly enhanced dengue outbreak prediction in Oman. The findings support the integration of lagged predictors and hierarchical modeling into early warning systems for mosquito-borne diseases, facilitating timely public health interventions and improved outbreak preparedness. (Copyright © 2025 The Author(s). Published by Elsevier Ltd.. All rights reserved.) |
| Contributed Indexing: | Keywords: Climate; Dengue; Hierarchical Bayesian models; Mosquito-borne diseases; Negative Binomial regression; Oman; Poisson regression; Vector surveillance |
| Entry Date(s): | Date Created: 20250803 Date Completed: 20251010 Latest Revision: 20251010 |
| Update Code: | 20251011 |
| DOI: | 10.1016/j.jiph.2025.102906 |
| PMID: | 40753686 |
| Databáza: | MEDLINE |
Full Text Finder 
Nájsť tento článok vo Web of Science | Abstrakt: | Competing Interests: Declaration of Competing Interest The authors declare no financial or personal conflicts of interest that could have influenced the findings of this study, “Predicting Mosquito-Borne Disease Outbreaks Using Poisson and Negative Binomial Models: A Comparative Study”.<br />Background: Dengue fever, a major mosquito-borne disease (MBD), continues to impose a growing global burden fueled by urbanization, climate change, and increased human mobility. Accurate predictive models are crucial for early detection and outbreak mitigation. This study aimed to develop and compare hierarchical models, with and without lagged predictors, for forecasting dengue cases in Oman.<br />Methods: A retrospective analysis was conducted using weekly data from 2020 to 2024 across multiple districts. Predictors included climate variables (temperature, humidity, wind, rainfall), mosquito surveillance indicators (trap positivity, mosquito density), and population demographics. Four hierarchical Bayesian models were developed: Poisson without lag, Poisson with lag, Negative Binomial without lag, and Negative Binomial with lag. Models incorporated fixed effects and random intercepts for epidemiological week, district, governorate, year, and seasonal components. Model performance was evaluated through convergence diagnostics, Mean Squared Error (MSE), Area Under the Curve (AUC), confusion matrices, and Leave-One-Out Information Criterion (LOOIC).<br />Results: All models demonstrated excellent convergence and fit the historical weekly data (2020-2024) accurately. The Negative Binomial model with lagged variables performed best, achieving the highest AUC (0.881, 95 % CI: 0.858-0.902), the lowest LOOIC (3234.6 ± 109.4), and the smallest MSE. Mosquito trap positivity was consistently the strongest predictor, while wind speed showed a moderate positive effect and temperature showed significant delayed negative effects. Rainfall, humidity, and population size were not significant predictors. Importantly, short-term forecasts for the first weeks of 2025 closely matched the observed case counts, confirming that the models' prediction metrics reflected both retrospective fit and real-world forecasting performance.<br />Conclusions: Incorporating delayed climatic and entomological factors using a Negative Binomial hierarchical framework significantly enhanced dengue outbreak prediction in Oman. The findings support the integration of lagged predictors and hierarchical modeling into early warning systems for mosquito-borne diseases, facilitating timely public health interventions and improved outbreak preparedness.<br /> (Copyright © 2025 The Author(s). Published by Elsevier Ltd.. All rights reserved.) |
|---|---|
| ISSN: | 1876-035X |
| DOI: | 10.1016/j.jiph.2025.102906 |