Routine petroleum product testing evaluates each parameter independently against its specification limit under a single-parameter compliance paradigm that implicitly assumes parameter independence. This study challenges that assumption, proposing that multi-parameter correlation structures carry diagnostic signals invisible to single-parameter review. An Isolation Forest (IF) model was fit to 10,120 full-test-panel samples (gasoline: 5432; diesel: 4688) from an ISO/IEC 17025-accredited testing laboratory in Guangdong, China (2018–2026), retaining 3,279 gasoline samples across 32 parameters and 3534 diesel samples across 22 parameters. Five alternative anomaly detection methods—MCD Mahalanobis distance, Local Outlier Factor (LOF), One-Class SVM (OC-SVM), PCA reconstruction error, and a shallow autoencoder—were compared at equivalent contamination settings (1.0
Blank matrix reference materials (RMs) are representative of a typical test matrix, but contain levels of target measurands below method limits of detection (LOD). In the absence of target measurands, conventional approaches used in the characterization of RMs are not applicable. Non-target analysis (NTA) using liquid chromatography–high-resolution mass spectrometry (LC–HRMS) enables collection of data on samples without prior information on the chemical constituents present. In this work, NTA was used to characterize CRM-Cyano-A, a blank cyanobacterial matrix (Aphanizomenon sp.) dietary supplement RM developed in support of cyanotoxin analysis methods. To provide extensive profile coverage, reversed-phase and hydrophilic interaction LC methods were used, and HRMS data were analyzed using metabolomics software. Information on thousands of chemical species was summarized and initially scrutinized to show that cyanobacterial toxins were not present above method LODs. The NTA data were then used to characterize the RM. Homogeneity was verified using data acquired on CRM units selected across the production series, examining compounds with varying polarities. Isochronous stability studies revealed changes in a broad range of matrix components, shedding light on the natural instability of the biological matrix, while facilitating selection of conditions for reference material transport and long-term storage. The NTA concept demonstrated is broadly applicable in biological and environmental RM development.
In order to ensure comparability of uncertainties for serum glucose quantified by ID-LC-MS/MS, a comprehensive uncertainty evaluation method was established, especially associated with weighing. A set of class E2 weights were weighed under repeatability conditions, then the indication error, standard deviation and uncertainty of each weight were combined to the uncertainty of repeatability ( u_rep.rel ). The relationship between the indications and certified mass of all weights was established by linear least square fitting, then the uncertainties of slope and temperature coefficient were combined to the uncertainty of linearity ( u_L.rel ). Test load at least Max/3 was used to evaluate the uncertainty of eccentricity ( u_ecci.rel ). Uncertainties of the densities for weights, samples and air were combined to the uncertainty of buoyancy correction factor( u_B_u.rel ). The contribution of u_rep.rel , u_L.rel , u_ecci.rel and u_B_u,rel to the combined uncertainty of weighing ( u_c(m_s).rel ) were then calculated. At low weighing mass (≤112.7 mg), u_rep.rel is the maximum uncertainty contribution, whereas u_B_u.rel becomes the major contribution for higher mass (>20 g). When weighing object is 1.9497 g, both u_rep.rel and u_B_u.rel are significant uncertainty contributors. Other contributions which are less than 10 u_c(m_s).rel were 0.0652 and 0.0515 u_c(m_s).rel based on ID-LC-MS/MS, only u_rep.rel and/or u_B_u.rel should be taken into account.
In proficiency testing (PT), a result outside the satisfactory range may indicate a performance concern without identifying the likely source of the deviation. This study examined whether prior differentiation of sample roles could build additional diagnostic information into PT returns. A matrix-free, pretreatment-free survey sample (SS) was distributed together with a conventional matrix-based, pretreatment-dependent assessment sample (AS) and conventional interference samples, forming an asymmetric sample design. Within the same laboratory return, the paired AS and SS results were treated as proxies for different portions of the analytical workflow. Across 147 paired AS–SS datasets, two diagnostic quantities used for interpretation, R_SS and the survey-referenced contrast ( SRC ), were derived. R_SS reflected deviation associated with the pretreatment-free portion of the analytical workflow, whereas SRC reflected additional deviation associated with the AS-specific portion of the analytical workflow. Representative cases illustrated how these quantities could help narrow the likely source domains of deviations and guide targeted review. The study shows that prior differentiation of sample roles can increase the yield of source-oriented diagnostic information from PT data and support more focused follow-up review and corrective action without replacing established PT performance assessment.
In this study, we systematically evaluated the contribution of storage time to measurement uncertainty for six pharmacopoeial certified reference materials (CRMs) over 180 days under in-use conditions. The experimental protocol—involving repeated sampling from a single storage bottle—was designed to simulate routine laboratory practice; consequently, the quantified uncertainty component reflects the combined effect of chemical degradation and repeated-opening-associated physical changes (e.g., localized evaporation), rather than chemical stability alone. We also compared three uncertainty estimation methods: trend analysis, the range method, and the pooled standard deviation method. We prepared the working solutions according to Chinese Pharmacopoeia specifications and stored them at 2–4 °C in amber glass bottles with PTFE-lined caps. We repeatedly sampled them over time to simulate routine laboratory practice. This experimental design evaluates the in-use stability of the MRC (measurement reference material), which differs fundamentally from the inventory stability protocol recommended by ISO 33405:2024, in which individual sealed vials are opened only once and variability from repeated sampling is eliminated. We identified storage time as the dominant contributor to total measurement uncertainty for all five tested compounds, accounting for 60.0–99.6
