Abstract:
The current state of the concept of safety of agricultural raw materials and products has undergone significant changes in recent decades due to the active harmonization of national requirements with international standards. One of the requirements is the need to take into account
measurement uncertainty. In the case of measuring biosafety parameters, this procedure faces challenges, including inherent analyte and matrix variability. Most of the modern approaches proposed by international rules are focused on stable systems: calibration or chemical analysis, which have a number of advantages and disadvantages, and can significantly affect the quality of the calculated interval. Uncertainty has significant changes over time, so it is necessary to periodically reassess its interval. The purpose of the work is to develop a model approach to the assessment of uncertainty, which will allow taking into account
as many sources of influence on the result of the analysis of biological safety indicators as possible. In the process of stability assessment, no contradictory values were obtained, indicating the influence of special factors on the results. The integrated compliance limits are not violated. Data normality is established. The results of the uncertainty assessment demonstrate the similarity of the intervals obtained by the hybrid method to the intervals calculated by the estimation based on the chain of CRM measurements. The hybrid method reconciles the result of the assessment with the dynamics observed in the error recognized by the results of interlaboratory comparisons. The hybrid method of uncertainty
estimation forms limits of suitability almost twice narrower than those obtained from the results of
interlaboratory comparison. The least acceptable method is to use the arithmetic mean. The obtained results showed an insufficiently narrow uncertainty interval, which is 5 times narrower than the results of using
the hybrid method. The combination of standardized score charts and Shewhart control charts with the integration of additional limits of fit to a set of control limits holds great promise for providing objective evidence of the suitability of retrospective data for estimating measurement uncertainty. Combined maps
make it possible to make reasonable claims about the statistical controllability, stability and compliance of the measurement procedure in a specific period based on the analysis of variability and average values of measurements. The combination of retrospective analysis of the sequence of routine CPM measurement results with confirmed stability allows consideration of the effects of laboratory-specific variability in reproducibility conditions, temporal variability, and error contributions assessed using proficiency testing. The hybrid method of uncertainty assessment provides a reasonable characterization of laboratory uncertainty based on evidence and creates the possibility of balanced control of the probability of α and β
errors in the implementation of conclusions regarding the compliance of agricultural raw materials with modern national and international standards of biological safety requirements.