LoQ Calculation Guidelines

Power Model

CV% = a × (Concentration)^b

A two-parameter power model relating CV% to concentration. The parameter 'a' scales the overall level of imprecision, and 'b' (typically negative) captures how CV% decreases as concentration rises. This model captures the typical inverse relationship observed in quantitative measurement procedures — where results at low concentrations are proportionally more variable than results at high concentrations.

Log-Linear Transformation

log(CV%) = log(a) + b × log(Concentration)

The power model is linearized by taking the logarithm of both sides. Ordinary least squares regression is then applied in log space to estimate the parameters 'a' and 'b'. This approach ensures that the fit is not disproportionately influenced by high-concentration levels where absolute CV% differences are smaller.

LoQ

LOQ = (Threshold / a)^(1 / b)

The LoQ is solved analytically by setting CV% equal to the acceptance threshold and inverting the power model. The threshold is the maximum acceptable CV% configured by the user. The resulting concentration is the lowest point at which the fitted precision model predicts CV% to be at or below the threshold.

Fitted CV%

Fitted CV% = a × (Concentration)^b

For each concentration level, the fitted CV% is back-calculated from the power model parameters. This is the expected CV% based on the model, as opposed to the measured CV% observed from the replicate data. The 95% CI for the fitted CV% reflects the uncertainty in the regression estimate.

Overall LoQ

Overall LoQ = max(LoQ_lot1, LoQ_lot2, …, LoQ_lotN)

When multiple reagent lots are evaluated, the LoQ is calculated independently for each lot using that lot's data. The maximum LoQ across all lots is reported as the overall procedure LoQ — a conservative, worst-case estimate that accounts for lot-to-lot variation in precision performance at low concentrations.