Computer simulation approaches to evaluate the interaction between analytical performance characteristics and clinical (mis)classification: a complementary tool for setting indirect outcome-based analytical performance specifications

Simulation-based approaches for setting indirect outcome-based analytical performance specifications (APS) predominantly involve test repetition through analytical reruns or resampling. These methodologies assess the agreement between original and simulated measurement results, determining the APS corresponding to pre-established performance thresholds. For APS related to imprecision and bias, both analytical performance characteristics (APCs) are typically considered in simulations, whereas for APS regarding measurement uncertainty, bias is excluded in alignment with traceability standards. This paper introduces the "APS Simulator," a novel tool designed to complement the existing APS Calculator by simulating APS under various scenarios involving imprecision, bias, and measurement uncertainty. The APS Simulator facilitates simulations using distinct analytical rerun and resampling models, enabling laboratory professionals to explore a wide range of performance levels for their specific needs. While the APS Simulator provides valuable insights, significant challenges remain in the broader application of indirect outcome-based APS. These include incorporating sources of diagnostic uncertainty, setting appropriate thresholds for performance metrics, validating clinical decision limits, and accounting for population data characteristics. Addressing these limitations will be essential to enhancing the standardization and robustness of APS determination. The source code and desktop application for the APS Simulator are freely available at https://github.com/hikmetc/APS_Simulator, providing a user-friendly platform for researchers and clinicians to further explore these methodologies.