CANCELLED  TL18: Limit of Quantitation by Mean Inverse Model Approach
*KuangLin He, Fujirebio Diagnostics, Inc. Keywords: Immunoassay, Limit of Quantitation, Limit of Detection, Modeling Conventionally, Immunoassay limit of detection (LoD) and limit of quantitation (LoQ) were evaluated based on distribution of measured concentration. The LoD and LoQ were low and look like more sensitive, but often lower than linear range. In CLSI EP17A2, there is another modeling/regression option. There is a concentration mean polynomial model, Std = B0` + B1`*Mean + B2'*Mean**2, for LoD and a concentration mean power function model, %CV = B0*Mean**B1, for LoQ, where LoQ is general within linear range. A model relative standard deviation (%RSD) or model CV is 12.5%. AIC=95.86. If the LoD equation is divided by Mean at both sides, Std/Mean = B0'/Mean + B1' +B2'*Mean is deduced. From this equation, we proposed mean inverse model approach for LoQ, %CV=B0+B1/Mean+B2*Mean or %CV=B0+B1/Mean, if B2 pvalue is statistical insignificant. Allowable %CV=10% and previous equation to determine LoQ. The %RSD of mean inverse model is 5.7% using same data of power function model and AIC=96.28. While Total error is TotErr=Bias+2Std, Bias+2Std =B0` + B1`*Mean. (Bias+2Std)/Mean= B0'/Mean + B1', if divided by Mean both sides. PctTE=PctBias+2CV. PctBias + 2CV = 5% + 2*10%. Allowable PctTEa=25% and PctTE = B0 + B1/Mean to determine LoQ. The %RSD of mean inverse model is much smaller and more mathematical meaningful than the power function model.

Key Dates

June 3, 2014  September 7, 2015
Online Registration 
June 3, 2015  August 15, 2015
Housing 
July 31  August 17, 2015
Invited Abstract Editing 
August 10, 2015
Short Course materials due from Instructors 
August 26, 2015
Advanced Registration Deadline 
September 7, 2015
Cancellation Deadline 
September 16  18, 2015
Marriott Wardman Park, Washington, DC