Dilution & Regression Guidelines
Actual Dilution Calculation
The reference plasma used in the study has its own factor activity level, which may differ from exactly 100%. The Reference Factor Activity (RAF) — obtained from the reference plasma package insert — corrects for this lot-to-lot variation. The platform computes the Actual Dilution automatically:
Actual Dilution (%)
Actual Dilution = Theoretical Dilution × RAF / 100
Where Theoretical Dilution is the intended mixing percentage (e.g., 35%) and RAF is the Reference Factor Activity of the specific reference plasma lot in use (e.g., 108%). For example: 35% × 108 / 100 = 37.8% ≈ 38% actual dilution. This adjustment ensures that sensitivity estimates reflect true factor activity, not the nominal mixing ratio.
Actual dilutions are used for all calculations and graph axes. Theoretical dilutions are preserved in the data table for traceability.
Regression Analysis
Once clotting times are collected at each dilution level, the platform fits a regression curve to model the relationship between factor activity (actual dilution %) on the x-axis and mean clotting time (seconds) on the y-axis. Two regression types are supported:
Polynomial Regression
Y = a₀ + a₁X + a₂X² + a₃X³ (degree 1, 2, or 3)
Fits a polynomial curve to the dilution-clotting time relationship. First degree (linear) is appropriate when the response is approximately linear. Second degree (quadratic) is the most common choice and handles the typical curvilinear response well. Third degree (cubic) provides additional flexibility for complex response curves. The degree should be selected based on how well the curve fits the observed data.
Logarithmic Regression
Y = a + b × ln(X)
Fits a logarithmic curve, which is often appropriate for coagulation assays where the clotting time response flattens at higher factor activity levels. The logarithmic model can produce a better fit in assays where the dose-response relationship is inherently non-linear across the tested range.
The regression is fit using only data points within the configured Regression Lower Limit and Regression Upper Limit (expressed as actual dilution percentages). Dilutions outside this range are still plotted on the graph in gray for reference but are excluded from the curve fit. Restricting the regression to the 20%–75% actual dilution range is recommended, as very low dilutions can distort the curve near the clinically relevant ULNR crossover region.
The quality of the regression fit is assessed using the Standard Error of the Estimate (SEE) and R². Both values are displayed on the chart. A high R² (typically >0.95) and a low SEE indicate a reliable curve fit and a more precise Calculated Sensitivity estimate.