Lina Zheng

Low-cost particulate matter (PM) sensors have transformed air-quality monitoring by enabling high-resolution, real-time measurements across ambient, mobile, indoor, and workplace environments. However, their affordability and portability come at the cost of reduced precision and susceptibility to environmental factors such as humidity, temperature, and sensor drift. This review synthesizes recent advances in machine learning–based and physics-informed calibration methods that correct low-cost sensor biases and extend their applicability to regulatory and occupational hygiene contexts. We first examine real-world deployments—ranging from community networks and vehicle-mounted platforms to smart-manufacturing floors and mining operations—highlighting how data-driven corrections improve spatial coverage and temporal responsiveness. Next, we critically evaluate calibration strategies: purely data-driven algorithms, purely physics-based models, and hybrid physics-informed machine learning frameworks. We compare their calibration accuracy, transferability, computational requirements, and interpretability, noting that while non-linear models often achieve sub-10 µg/m³ errors, they face challenges in embedded deployment, dependence on representative training data, and region-specific performance. Overall, no single calibration strategy is universally optimal, with machine-learning-based methods excelling in predictive performance, physics-based methods in interpretability and physical consistency, and hybrid approaches in balancing accuracy, robustness, and transferability. Finally, we discuss persistent barriers—sensor fouling, data governance, model generalization, and hardware constraints—and propose emerging solutions such as self-cleaning optics, federated learning, and knowledge-distilled on-device calibration. By mapping the current state of the art and outlining future research directions, this review aims to guide the integration of low-cost PM sensors into robust, scalable air-quality management and occupational health systems.

The health of miners working in underground mines is greatly threatened by their exposure to respirable crystalline silica (RCS) found in coal dust. While coal dust includes RCS, it is important to know the specific filtration efficiency for RCS particles compared to coal dust to improve protective measures. This study compares the filtration efficiency of reduced graphene oxide (rGO) composite-incorporated Polyacrylonitrile (PAN) nanofibrous filters at four different concentrations of rGO (0, 2, 6, and 10 wt%) for RCS particles and coal dust. The filters were made using the electrospinning method. They were tested separately with RCS particles and coal dust with particle sizes from 0.3 to 10 μm, within the respirable range, at various concentrations. Filtration effectiveness was evaluated using a particle counter. A dry dust aerosol generator, model 3410U, operating at 1–6 bar, was used to disperse particles. The results show that the rGO-PAN filters are almost equally effective in removing RCS particles and coal dust across all tested flow rates: 30, 50, and 80 L per minute (L min^–1). Moreover, at a flow rate of 50 L min^–1, the 6 wt% rGO-PAN filters showed greater efficiency for both types of dust. This indicates their strong potential for targeting both silica and coal dust particles. This study provides useful insights for creating effective filtration systems to lower health risks linked to RCS exposure in underground mining settings.

This review summarizes current evidence on low-cost particulate matter (PM) sensors for indoor and occupational environments and proposes a framework that links performance evaluation, calibration, and uncertainty to decision-making.

Aspergillus is more challenging to control than other fungi due to its ability to produce and spread spores that can readily colonize diverse environments. Invasive infection results in severe complications and a terrible mortality of 85.2%, where misdiagnosis can result in critical delays in receiving optimal treatment. The diagnosis relies on laboratory tests, where an accurate and rapid diagnostic method is essential for early detection of invasive Aspergillosis. Surface-enhanced Raman scattering (SERS) is an effective technique for fungal detection. However, its application in Aspergillus detection has been limited due to the low selectivity associated with existing substrates. Based on the mechanism of action of the antifungal drug amphotericin B (AMB), which tightly binds to ergosterol on the cell wall of Aspergillus, our work introduces high-affinity polydopamine (PDA) as a capture carrier, utilizing magnetic nanoparticles (MNP) as magnetic attraction core. Through an in situ polymerization method, PDA was successfully coated onto MNP. The high adhesiveness of PDA enabled the coupling of AMB, enabling the successful preparation of a high-affinity, high-capture efficiency nanocatcher (MNP@PDA-AMB). This approach allowed for in vitro diagnosis (IVD) of Aspergillus with SERS detection using silver nanowire (AgNWs) substrate, demonstrating that the brief capture time (8 min) and entire detection process only takes 20 min. Excitingly, this work also established an enhanced limit of detection (LOD) with 10-10² cells/mL and the requisite reproducibility (RSD < 10%). The proposed method represents a breakthrough in the rapid detection of Aspergillus, offering broad adaptability and significant potential for clinical applications.