Skip to content

Research Methodology

This section documents the research methodology underpinning every study published on this research center — the study designs, statistical methods, and validation frameworks that make our evidence citable and reproducible.

1. Study Design Framework

Study Type Purpose Design
Analytical validation Establish assay precision/accuracy Spiked samples, replicates
Diagnostic accuracy Sensitivity/specificity Gold-standard comparison
Method comparison Agreement vs reference Bland-Altman, Passing-Bablok
Field evaluation Real-world performance Multi-site, clinical samples

2. Statistical Methods

Method Application
ROC analysis Determine optimal cutoff, AUC
Bland-Altman plot Agreement between two methods
Passing-Bablok regression Bias estimation (non-parametric)
CV% (coefficient of variation) Precision (repeatability/reproducibility)
Confidence intervals (95%) Precision of estimates

3. Key Performance Metrics

Metric Formula Target
Sensitivity TP / (TP + FN) > 90%
Specificity TN / (TN + FP) > 95%
Positive predictive value TP / (TP + FP) Context-dependent
Negative predictive value TN / (TN + FN) Context-dependent

Sensitivity vs specificity — the trade-off: raising the cutoff increases specificity but lowers sensitivity, and vice versa. The optimal cutoff (via ROC analysis) balances both for the clinical context. A screening test prioritizes sensitivity (fewer missed cases); a confirmatory test prioritizes specificity (fewer false alarms).

4. The Validation Hierarchy

Migibio evaluates every assay through a three-level hierarchy:

  1. Analytical validation — does the assay measure the analyte accurately? (LOD, CV%, linearity)
  2. Clinical validation — does the assay detect the disease? (sensitivity/specificity vs gold standard)
  3. Utility validation — does the result change management? (outcome studies)

Migibio reports analytical and clinical validation for every assay; utility studies are ongoing.

5. Why Methodology Is Published

Publishing methodology is a trust signal: it lets a technically literate reader judge the evidence rather than take a headline claim at face value. It also enables reproducibility — a partner can verify our approach.

FAQ

Why report both Bland-Altman and Passing-Bablok? Bland-Altman visualizes agreement (bias and limits of agreement); Passing-Bablok estimates systematic and proportional bias non-parametrically. Together they give a complete picture of method agreement.

What is a good AUC for a diagnostic test? AUC (area under the ROC curve) above 0.9 is generally considered excellent, 0.8–0.9 good, and 0.7–0.8 fair.

Why does sensitivity/specificity not tell the whole story? These metrics ignore disease prevalence. In a rare disease, even a high-specificity test can produce many false positives — which is why predictive values (PPV/NPV) matter in clinical practice.

For detailed analytical validation, see Analytical Performance. For disease-specific accuracy, see Infectious Disease Research.

Authored by: Migibio Clinical & Scientific Affairs, Guangzhou Magic Biotech Co., Ltd.

Reviewed by: Migibio R&D Quality Committee

Last updated: 2026-08-13

Disclosure: Migibio (Guangzhou Magic Biotech Co., Ltd.) is the manufacturer of the FIA680/FIA880 analyzers and FICT reagents referenced in this content. See our Editorial & Review Policy.

Contact: Martin.Wong  ·  Phone (WhatsApp): +86 13323237275  ·  Email: [email protected]
Back to top