Accuracy-based equivalence evaluation using reference data for the clinical application of patient-based quality...

Patient-based real-time quality control (PBRTQC) is an effective quality control (QC) strategy using patient test results to monitor analytical performance. Most studies have proposed different PBRTQC models and tested their performance by developers themselves. However, there remains a need for validating the accuracy-based equivalence of the models/software before clinical application and for periodic review, which is also the requirement of testing and calibration laboratories competence accreditation criteria. To address this issue, we introduce the concept of digital metrology and algorithm traceability into PBRTQC to enable a standardized evaluation.

A reference data set was established as the digit-measuring instrument (DMI) and applied for metrological traceability, taking red blood cell (RBC) count as an example. RBC count data containing nine dimensions in each group were collected. The DMI was developed through digital twin technology, comprised of standardized twin data with measurement uncertainty (MU), serving as a standard to evaluate five PBRTQC models.

A hierarchical traceability chain was constructed, from the PBRTQC models/software output results tracing back to the DMI, and then to the original data generated by the instrument. The DMI included both in-control and out-of-control states, corresponding to the two situations of QC. The algorithm evaluation values Y of the five PBRTQC models were calculated, ranging from 0.638986 to 0.798970, with MU of (4.1-4.3) × 10^-5.

Establishing reference data as DMI and conducting algorithm traceability could provide an accuracy-based equivalence evaluation and offer evidence for laboratory users to select and deploy PBRTQC models/software in clinical settings.