Calibration of Low-Cost Particulate Matter Sensors for Exposure Assessment: A Critical Review

Low-cost particulate matter (PM) sensors make it possible to measure air pollution in homes, workplaces, and communities that are poorly represented by regulatory networks. Their readings, however, are sensitive to humidity, aerosol composition, concentration range, unit-to-unit variation, and aging. This review examines laboratory and field calibration evidence across environmental and occupational settings, emphasizing transferability and the consequences of calibration error for exposure assessment and health inference.

The review identified 66 peer-reviewed studies published from 2015 to 2025, including 32 laboratory evaluations and 34 field studies. The median reported calibrated coefficient of determination (R²) was 0.92 (n = 31) for laboratory studies and 0.72 (n = 29) for field studies; the corresponding median root mean square errors (RMSEs) were 2.6 µg/m³ (n = 7) and 12.3 µg/m³ (n = 9). These values summarize a heterogeneous literature rather than a common pooled effect. Studies differed in aerosol mixture, comparison instrument, concentration range, averaging interval, and validation design. Relative humidity and temperature were considered in most studies. Machine-learning and hybrid models appeared more often in field work and sometimes lowered error, although independent validation across time, place, and sensor units was uneven. Calibration is not a universal correction that can be applied once and carried unchanged into any deployment. Laboratory testing is well suited to defining response limits, particle-size sensitivity, and unit variability. Field co-location is needed to determine how the same device behaves under the sources, weather, concentrations, and time scales relevant to a health study. Credible use in epidemiology or public-health decision-making also requires independent validation, planned checks for drift, and clear reporting of calibration uncertainty.