AccScience Publishing / EER / Online First / DOI: 10.36922/EER025370067
Submit to EER
Cite this article
11
Download
90
Views
Journal Browser
Volume | Year
Issue
Search
Forthcoming Issue
Current Issue
View All
News and Announcements
View All
ORIGINAL RESEARCH ARTICLE

Global PM2.5 exposure forecasting with novel deep learning architecture and explainable artificial intelligence

Syed Azeem Inam1* ,  Saddam Umer1 ,  Haider Rajput1
Show Less
1 Department of Artificial Intelligence and Mathematical Sciences, Faculty of Information Technology, Sindh Madressatul Islam University, Karachi, Pakistan
EER, 025370067 https://doi.org/10.36922/EER025370067
Received: 10 September 2025 | Revised: 30 November 2025 | Accepted: 11 December 2025 | Published online: 24 December 2025
© 2025 by the Author(s). This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International License ( https://creativecommons.org/licenses/by/4.0/ )
Download PDF
XML
Cite
Abstract

Particulate matter (PM) of fine size (≤2.5 μm) remains one of the most significant global environmental risk factors for early mortality and morbidity, and more than 90% of the global population currently lives in areas exceeding the World Health Organization 2021 guideline value of 5 μg/m3. This study introduces a temporally constrained transformer-based forecasting model to anticipate annual population-weighted PM2.5 exposure across 204 countries and territories between 1990 and 2020, aimed at supporting evidence-based air quality and climate policy development. The framework is based on a filtered dataset from the State of Global Air, comprising 6,323 country–year observations with harmonized exposure estimates and uncertainty bounds, allowing the model to capture long-range temporal variations and enduring heterogeneity among countries in exposure trends. A time-aware expanding-window cross-validation approach was strictly implemented to prevent information leakage and ensure realistic predictive conditions. Five-fold temporal validation demonstrates strong performance across geographical locations, with mean squared error ranging from 0.00043 to 0.00115, root mean squared error from 0.0207 to 0.0339 μg/m3, and mean absolute error from 0.0094 to 0.0193 μg/m3, with Nash–Sutcliffe efficiencies exceeding 0.95 on average. Continental-scale evaluation shows similar high accuracy in Europe and Oceania (root mean squared error <0.01 μg/m3; R2 > 0.98), while systematically higher errors are observed in Asia and Africa, which bear a higher pollution burden. The attention-weight inspection offers clear decompositions of temporal trends and country-specific patterns that drive predictions. The proposed framework is, therefore, a methodological and practical addition to transformer-based environmental forecasting and policy-relevant global health-risk assessment.

Keywords
PM2.5 exposure forecasting
Transformer architecture
Country embeddings
Global environmental health
Time-series deep learning
Explainable artificial intelligence
Funding
None.
Conflict of interest
The authors declare they have no competing interests.
References
  1. Inam SA. A review of artificial intelligence for predicting climate driven infectious disease outbreaks to enhance global health resilience. Discover Public Health. 2025;22(1):738. doi: 10.1186/s12982-025-01167-4

 

  1. Inam SA, Khan AA, Mazhar T, et al. PR-FCNN: A data-driven hybrid approach for predicting PM2.5 concentration. Discover Artif Intell. 2024;4(1):75. doi: 10.1007/s44163-024-00184-7

 

  1. Inam SA, Zaidi SMH, Khan AA, Ullah S. A neural network approach to carbon emission prediction in industrial and power sectors. Discover Appl Sci. 2025;7(6):640. doi: 10.1007/s42452-025-07257-x

 

  1. The World Bank. The Global Health Cost of PM2.5 Air Pollution: A Case for Action Beyond 2021. Washington, DC: The World Bank; 2022. doi: 10.1596/978-1-4648-1816-5

 

  1. Pai SJ, Carter TS, Heald CL, Kroll JH. Updated world health organization air quality guidelines highlight the importance of non-anthropogenic PM2.5. Environ Sci Technol Lett. 2022;9(6):501-506. doi: 10.1021/acs.estlett.2c00203

 

  1. Liu J, He C, Si Y, et al. Toward better and healthier air quality: Global PM2.5 and O3 pollution status and risk assessment based on the new WHO air quality guidelines for 2021. Glob Chall. 2024;8(4):2300258. doi: 10.1002/gch2.202300258

 

  1. Edlund KK, Killman F, Molnár P, Boman J, Stockfelt L, Wichmann J. Health risk assessment of PM2.5 and PM2.5- bound trace elements in Thohoyandou, South Africa. Int J Environ Res Public Health. 2021;18(3):1359. doi: 10.3390/ijerph18031359

 

  1. Elbarbary M, Oganesyan A, Honda T, et al. Systemic inflammation (c-reactive protein) in older Chinese adults is associated with long-term exposure to ambient air pollution. Int J Environ Res Public Health. 2021;18(6):3258. doi: 10.3390/ijerph18063258

 

  1. McDuffie EE, Martin RV, Spadaro JV, et al. Source sector and fuel contributions to ambient PM2.5 and attributable mortality across multiple spatial scales. Nat Commun. 2021;12(1):3594. doi: 10.1038/s41467-021-23853-y

 

  1. Ghosh R, Causey K, Burkart K, Wozniak S, Cohen A, Brauer M. Ambient and household PM2.5 pollution and adverse perinatal outcomes: A meta-regression and analysis of attributable global burden for 204 countries and territories. PLoS Med. 2021;18(9):e1003718. doi: 10.1371/journal.pmed.1003718

 

  1. Southerland VA, Anenberg SC, Harris M, et al. Assessing the distribution of air pollution health risks within cities: A neighborhood-scale analysis leveraging high-resolution data sets in the bay area, California. Environ Health Perspect. 2021;129(3):37006. doi: 10.1289/EHP7679

 

  1. Fang T, Di Y, Xu Y, et al. Temporal trends of particulate matter pollution and its health burden, 1990–2021, with projections to 2036: A systematic analysis for the global burden of disease study 2021. Front Public Health. 2025;13:1579716. doi: 10.3389/fpubh.2025.1579716

 

  1. Booth A, James P, McGough S, Solaiman E. Cross-regional deep learning for air quality forecasting: A comparative study of CO, NO2, O3, PM2.5, and PM10. Forecasting. 2025;7(4):66. doi: 10.3390/forecast7040066

 

  1. Benali AAE, Cafaro M, Epicoco I, Pulimeno M, Schioppa EJ. Just in Time Transformers. New York: Cornell University; 2024.

 

  1. Muthukumar P, Cocom E, Nagrecha K, et al. Predicting PM2.5 atmospheric air pollution using deep learning with meteorological data and ground-based observations and remote-sensing satellite big data. Air Qual Atmos Health. 2022;15(7):1221-1234. doi: 10.1007/s11869-021-01126-3

 

  1. Nedungadi V, Munir MA, Rußwurm M, et al. AirCast: Improving Air Pollution Forecasting through Multi-Variable Data Alignment. New York: Cornell University; 2025.

 

  1. Oldenburg V, Cardenas-Cartagena J, Valdenegro-Toro M. Forecasting Smog Clouds with Deep Learning. New York: Cornell University; 2024.

 

  1. Boogaard H, Pant P, Künzli N. Editorial: Science to foster the WHO air quality guideline values. Int J Public Health. 2025;69:1608249. doi: 10.3389/ijph.2024.1608249

 

  1. Tello-Leal E, Ramirez-Alcocer UM, Macías-Hernández BA, Hernandez-Resendiz JD. Evaluation of deep learning models for predicting the concentration of air pollutants in urban environments. Sustainability. 2024;16(16):7062. doi: 10.3390/su16167062

 

  1. Huang S, Xiao X, Guo H. A novel method for carbon emission forecasting based on EKC hypothesis and nonlinear multivariate grey model: Evidence from transportation sector. Environ Sci Pollut Res Int. 2022;29(40):60687-60711. doi: 10.1007/s11356-022-20120-5

 

  1. Ran Q, Bu F, Razzaq A, et al. When will China’s industrial carbon emissions peak? Evidence from machine learning. Environ Sci Pollut Res Int. 2023;30(20):57960-57974. doi: 10.1007/s11356-023-26333-6

 

  1. Costantini L, Laio F, Mariani MS, Ridolfi L, Sciarra C. Forecasting national CO2 emissions worldwide. Sci Rep. 2024;14(1):22438. doi: 10.1038/s41598-024-73060-0

 

  1. Lee JB, Lee JB, Koo YS, et al. Development of a Deep Neural Network for Predicting 6 Hour Average PM2.5 Concentrations up to Two Subsequent Days Using Various Training Data. [Preprint]; 2021. doi: 10.5194/gmd-2021-356

 

  1. Damkliang K, Chumnaul J. Deep learning and statistical approaches for area-based PM2.5 forecasting in Hat Yai, Thailand. J Big Data. 2025;12(1):36. doi: 10.1186/s40537-025-01079-9

 

  1. Utku A, Can U, Alpsülün M, et al. Advancing air quality monitoring: Deep learning-based CNN-RNN hybrid model for PM2.5 forecasting. Atmosphere. 2025;16(9):1003. doi: 10.3390/atmos16091003

 

  1. Zhang Y, Zhang W, Wu B, Yi W. Regional PM2.5 pollution forecasting using a hybrid model based on multi-scales feature fusion and deep learning algorithms. PLoS One. 2025;20(10):e0333489. doi: 10.1371/journal.pone.0333489

 

  1. Yeo I, Choi Y, Lops Y, Sayeed A. Efficient PM2.5 forecasting using geographical correlation based on integrated deep learning algorithms. Neural Comput Appl. 2021;33(22):15073-15089. doi: 10.1007/s00521-021-06082-8

 

  1. Samal KKR, Babu KS, Das SK. Multi-directional temporal convolutional artificial neural network for PM2.5 forecasting with missing values: A deep learning approach. Urban Clim. 2021;36:100800. doi: 10.1016/j.uclim.2021.100800

 

  1. Potavijit C, Pattarapanitchai P, Nontapa C. A hybrid deep learning model for forecasting PM2.5 concentrations in Northern Thailand from satellite images. Glovento J Integr Stud. 2025;1(1):6. doi: 10.63665/gjis.v1.6

 

  1. Wu Y, Wang X, Wang M, Liu X, Zhu S. Time-series forecasting of PM2.5 and PM10 concentrations based on the integration of surveillance images. Sensors (Basel). 2024;25(1):95. doi: 10.3390/s25010095

 

  1. Dairi A, Harrou F, Sun Y. Efficient Deep Learning-Driven Approach for PM2.5 Forecasting at Different Locations in Spain. In: 2021 IEEE 3rd Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS). IEEE; 2021. p. 173-178. doi: 10.1109/ECBIOS51820.2021.9510462

 

  1. Vignesh PP, Jiang JH, Kishore P. Predicting PM2.5 concentrations across USA using machine learning. Earth Space Sci. 2023;10(10):1-15. doi: 10.1029/2023ea002911
Next article in this issue
Share
Back to top
Explora: Environment and Resource, Electronic ISSN: 3060-9046 Published by AccScience Publishing
We use cookies on our website to ensure you get the best experience.
Read more about our cookies here
Accept