medRxiv [Preprint]. 2026 Mar 17:2026.03.13.26348337. doi: 10.64898/2026.03.13.26348337.
ABSTRACT
INTRODUCTION: Particulate Matter (PM 2.5 ) exposure contributes to the global disease burden, yet its monitoring remains sparse and uneven and is limited in many limited ground monitoring network settings. Road-traffic proxy indicators can provide indirect estimates of PM 2.5 where measurements are limited but require context-specific validation. We evaluated three PM 2.5 road-traffic related proxies:(I) population-Weighted Road Network Density (WRND), (ii) Euclidean (straight line) distance from highways (EH), and (iii) Euclidean distance from main roads (EM).
METHODS: We validated proxies using high-resolution outdoor filtered PM 2.5 personal exposure measurements collected over 1 year from 343 postpartum participants in The Gambia, Kenya, and Mozambique. Village-level spatial patterns for the PM 2.5 -proxy relationship were mapped using 5 km hexagonal aggregated tessellations. Proxy-PM 2.5 associations were assessed using Spearman correlation, and predictive utility was tested using country-specific and global Random Forest (RF) models (3-fold cross-validation), reporting R 2 , RMSE, and feature importance.
RESULTS: Spatial mapping showed heterogeneous proxy-PM 2.5 relationships across and within sites, with elevated PM 2.5 occurring in both low- and high-proxy contests. WRND-PM 2.5 correlations were weak overall and statistically significant only in Mozambique (r = 0.351; p = 0.005 ), with non-significant associations in Kenya (r = -0.041; p = 0.673 ) and The Gambia (r = -0.020; p = 0.909 ). EH-PM 2.5 correlations were positive in The Gambia (r = 0.335; p = 0.053 ) and Mozambique (r = 0.292; p = 0.020 ) but negative and significant in Kenya (r = -0.224; p = 0.018 ).Single-variable RF models performed poorly across all countries (R 2 < 0.45) and the Global model (R 2 =0.42). Combining proxies improved performance in Kenya (R 2 =0.52; RMSE=31.7µg/m 3 ) and Mozambique (R 2 =0.60; RMSE=8.9 µg/m 3 ), Global R 2 =0.46; RMSE=29.1 µg/m 3 ), although in The Gambia, the combined model (R 2 =0.53; RMSE=37.6 µg/m 3 ) did not exceed the best single-proxy model.
CONCLUSION: Road-network proxies provide context-dependent signals of personal PM 2.5 exposure, and predictive performance is strengthened when proxies are combined in a hybrid model.
PMID:41891017 | PMC:PMC13015645 | DOI:10.64898/2026.03.13.26348337
