Comparative evaluation of GUM and Monte Carlo simulation for uncertainty estimation in LC–MS/MS: impact of input correlations and application to urinary N-desethyl blonanserin

Abstract To improve the reliability of the liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis of biological samples, it is necessary to evaluate the measurement uncertainty (MU). Although the Guide to the Expression of Uncertainty in Measurement (GUM) is widely used as an international standard, it does not sufficiently reflect potential correlations between input variables and nonlinear factors, resulting in an overestimated uncertainty of an analytical method. In this study, GUM and the Monte Carlo method (MCM) were used to estimate the MU of urinary N -desethyl blonanserin (DBNS) concentration. LC–MS/MS was used to quantify urine DBNS levels based on a calibration curve. GUM overestimates the MU in sample preparation and calibration curve construction compared to MCM. In the case of sample preparation, GUM obtained the combined relative standard uncertainty (RSU) for each sample by taking the square root of the sum of RSU values, assuming a uniform error distribution. By contrast, MCM generated random numbers within the actual error range of each sample and added them to form a trapezoidal distribution as the sum of two uniform distributions. In the case of calibration curve construction, GUM used approximate values to estimate the uncertainty for the calibration curve fitted using least squares regression. However, the correlation coefficients between inputs have a considerable effect on the estimated MU. Therefore, the correlation coefficients between input variables in MCM mitigated the contribution of the input variables to the uncertainty budget, ultimately reducing uncertainty by 9.9%, and thereby improving the reliability of quantifying low‑concentration metabolites in biological samples. Despite its mathematical advantages, routine adoption of MCM in bioanalytical laboratories is often limited by programming barriers. To address this, we provide an adaptable, open–source R–script framework that simplifies MCM implementation in LC–MS/MS workflows. By facilitating robust uncertainty estimation, this toolkit enhances the reliability of trace–level bioanalytical measurements and extends the methodological utility of MCM to routine laboratory practice without requiring advanced computational expertise.