What statistical methods are used for bioassay analysis and relative potency determination?

Bioassay statistical analysis extracts potency information from dose-response data: 1) Parallel line analysis - classic approach assuming parallel dose-response curves for sample and reference; linear portion of log-dose vs response analyzed. 2) Four-parameter logistic (4PL) regression - fits sigmoid curves: Response = Bottom + (Top-Bottom)/(1 + 10^((LogEC50-LogDose)*HillSlope)). Relative potency calculated from EC50 shift. 3) Parallelism assessment - evaluate if curves are parallel (same slope, shape); f-test or equivalence testing; non-parallel curves indicate quality issues. 4) Confidence intervals - 95% CI for potency typically required within 80-125% for lot release. 5) Weighted regression - appropriate when variance is non-constant. 6) Combination indices - for assays with multiple responses. Regulatory expectations: USP chapters 1032-1034 guidance, EDQM biostatistics, and CombiStats software. Key: assay qualification establishes acceptable precision; system suitability ensures each run is valid.