The abstract is read at the publisher; the summary is JClub's.
A site-specific regression using nearby 59T fixed-site PM2.5 predicts missing 6-h DustTrak observations in Bangkok with LOOCV RMSE 6.43 μg/m3 and cross-validated R2 0.971.
- Why it matters: Portable PM2.5 monitoring can be interrupted, risking loss of short-term data continuity. The study tests whether nearby fixed-station measurements can reconstruct isolated missing 6-h portable sessions without assuming direct interchangeability.
- What they did: Six-hour DustTrak sessions (n=24) were run in Bangkok during wet and dry seasons from August 2024 to February 2025, with 22 sessions matched to 59T. Models used 59T PM2.5 alone or with meteorological predictors, evaluated by leave-one-session-out cross-validation and a temporally structured hold-out subset.
- The result: DustTrak and 59T show very strong temporal correspondence (Spearman’s ρ=0.979, p<0.001) but direct substitution performs poorly (LOOCV RMSE 16.03 μg/m3). The 59T-only model reduces RMSE by 59.9% and remains strong in temporally separated validation (R2 0.926; RMSE 5.68 μg/m3).
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.