Bland-Altman & MDP Guidelines
Bland-Altman Analysis
Two Bland-Altman plots are generated for each domain (PT and INR):
Absolute Difference Plot
Y-axis: (Y − X) per sample | X-axis: mean of X and Y
Plots the absolute difference between methods (test minus reference) against the average of both methods. The mean bias line shows the average systematic difference across the range. The allowable systematic error bounds (derived from TEa and bias allocation) are shown as horizontal reference lines.
Percentage Difference Plot
Y-axis: (Y − X) / X × 100 | X-axis: mean of X and Y
Plots the relative (percentage) difference between methods against the average. Useful for identifying whether bias is proportional (increases with concentration) or constant. Points scattered uniformly around zero indicate no systematic trend; a funnel shape indicates concentration-dependent differences.
The "Other Statistics" panel below the plots reports the mean absolute bias, mean percentage bias, and the bias allocation percentage — the proportion of the configured TEa assigned to systematic error (bias) for this study.
Medical Decision Points (MDPs)
When MDPs are enabled, the module estimates the predicted bias at specific clinically important PT and INR concentrations using the selected regression model. MDPs are configured independently for PT (in seconds) and INR:
Predicted Bias at MDP
B_c = a + (b − 1) × X_c
Where X_c is the medical decision point concentration (reference method value), b is the regression slope, and a is the regression intercept. This formula computes the predicted difference between the test and reference method at that specific concentration.
95% CI for Predicted Bias
CI = B_c ± t(α/2, N−2) × SE(B_c)
A confidence interval around the predicted bias at each MDP. A narrow CI indicates a precise bias estimate; a wide CI means the bias at that decision level is uncertain and requires more data or a wider measurement range to characterize reliably.
The MDP table shows: Reference Method value, Predicted Test Method value, Bias, and 95% CI for Bias — separately for PT and INR. For PT, typical decision points include the upper limit of normal (e.g., 13 s) and values in the therapeutic anticoagulation range (e.g., 20–30 s). For INR, key decision points include the therapeutic target range (2.0, 2.5, 3.0) and the supratherapeutic threshold (4.0–5.0).