Prediction accuracy of available early warning systems for air pollution across WHO South-East Asia region: a systematic review
- 1Indian Council of Medical Research, ICMR Headquarters , ,
- 2National Institute for Research in Environmental Health, Indian Council of Medical Research , ,
Abstract
Background: Ambient air pollution is a major environmental health risk across developing nations in the WHO South-East Asia Region (SEARO), contributing substantially to premature mortality and the global disease burden. While Air Quality Early Warning Systems (AQEWS) enable proactive public health interventions, their regional predictive accuracy remains uncharacterized. Despite their increasing importance, the predictive performance and methodological quality of AQEWS across SEARO have not been systematically evaluated. To address this gap, this study systematically evaluated the accuracy of available AQEWS across the region.
Methods: This systematic review (PROSPERO: CRD420251275901) searched major databases from inception to February 10, 2026, for observational modeling studies reporting quantitative AQEWS performance in SEARO. Reviewers extracted data and assessed bias using PROBAST+AI. Primary outcomes were RMSE, R ² , categorical accuracy, and forecast bias.
Results: Fourteen studies (India, Bangladesh, Thailand) from 839 records were synthesized; heterogeneity disallowed meta-analysis. Hybrid deep learning architectures consistently yielded the lowest RMSE (0.84–4.61 μg/m ³ ) and highest explained variance ( R ² up to 0.998) across the studies, particularly for PM 2.5 forecasting; however, evidence was heterogeneous and did not establish consistent superiority across pollutants, forecast horizons, validation settings, or operational AQEWS. Machine learning achieved up to 98.34% categorical accuracy. Chemical transport models demonstrated significant bias reduction (~70%) exclusively when integrated with satellite data assimilation. Most studies showed a high risk of bias due to inadequate external validation and algorithmic overfitting.
Conclusion: While hybrid and deep learning models demonstrated superior forecasting accuracy over traditional methods, widespread policy implementation across SEARO was hindered by poor methodological validation. This review provides a regional synthesis of AQEWS forecasting performance and identifies key methodological limitations that currently constrain their reliability for public health decision-making. Standardized reporting and rigorous external validation are urgently required to ensure these systems reliably inform regional public health emergency responses.
Systematic Review Registration: identifier: CRD 420251275901.
Related
- Hybrid stacked bidirectional deep learning with attention based air pollution prediction for environmental sustainability
- Machine learning-driven prediction of heavy metal pollution for aquatic health surveillance and disease control
- Input-data perturbation in groundwater level prediction: a controlled experiment on accuracy–reliability trade-offs among data augmentation and Bayesian-type uncertainty estimation methods
- Microplastic and Nanoplastic Pollution in Zooplankton: A Systematic Literature Review and Bibliometric Analysis of Ingestion, Ecotoxicological Effects, and Research Gaps
- Water quality indices and their impact on coastal pollution and tourist satisfaction: A systematic review
BibTeX
@article{mahawar2026prediction,
title = {{Prediction accuracy of available early warning systems for air pollution across WHO South-East Asia region: a systematic review}},
author = {Charu Mahawar and Sivakumar Sasidharan and Vikram Singh and Manoj Kumar and Neeraj Parmar and Sanghamitra Pati and Sumit Aggarwal},
journal = {Frontiers in public health},
year = {2026},
month = sep,
doi = {10.3389/fpubh.2026.1945362},
url = {https://europepmc.org/article/PMC/PMC13662169},
note = {CC BY},
}
RIS
TY - JOUR
TI - Prediction accuracy of available early warning systems for air pollution across WHO South-East Asia region: a systematic review
AU - Charu Mahawar
AU - Sivakumar Sasidharan
AU - Vikram Singh
AU - Manoj Kumar
AU - Neeraj Parmar
AU - Sanghamitra Pati
AU - Sumit Aggarwal
PY - 2026
DA - 2026/09/30
JO - Frontiers in public health
DO - 10.3389/fpubh.2026.1945362
UR - https://europepmc.org/article/PMC/PMC13662169
AB - Background: Ambient air pollution is a major environmental health risk across developing nations in the WHO South-East Asia Region (SEARO), contributing substantially to premature mortality and the global disease burden. While Air Quality Early Warning Systems (AQEWS) enable proactive public health interventions, their regional predictive accuracy remains uncharacterized. Despite their increasing importance, the predictive performance and methodological quality of AQEWS across SEARO have not been systematically evaluated. To address this gap, this study systematically evaluated the accuracy of available AQEWS across the region. Methods: This systematic review (PROSPERO: CRD420251275901) searched major databases from inception to February 10, 2026, for observational modeling studies reporting quantitative AQEWS performance in SEARO. Reviewers extracted data and assessed bias using PROBAST+AI. Primary outcomes were RMSE, R ² , categorical accuracy, and forecast bias. Results: Fourteen studies (India, Bangladesh, Thailand) from 839 records were synthesized; heterogeneity disallowed meta-analysis. Hybrid deep learning architectures consistently yielded the lowest RMSE (0.84–4.61 μg/m ³ ) and highest explained variance ( R ² up to 0.998) across the studies, particularly for PM 2.5 forecasting; however, evidence was heterogeneous and did not establish consistent superiority across pollutants, forecast horizons, validation settings, or operational AQEWS. Machine learning achieved up to 98.34% categorical accuracy. Chemical transport models demonstrated significant bias reduction (~70%) exclusively when integrated with satellite data assimilation. Most studies showed a high risk of bias due to inadequate external validation and algorithmic overfitting. Conclusion: While hybrid and deep learning models demonstrated superior forecasting accuracy over traditional methods, widespread policy implementation across SEARO was hindered by poor methodological validation. This review provides a regional synthesis of AQEWS forecasting performance and identifies key methodological limitations that currently constrain their reliability for public health decision-making. Standardized reporting and rigorous external validation are urgently required to ensure these systems reliably inform regional public health emergency responses. Systematic Review Registration: identifier: CRD 420251275901.
SN - 2296-2565
ER -