Online Program Home
  My Program

Abstract Details

Activity Number: 471 - Statistical Methods and Challenges in Diagnostic Medicine
Type: Contributed
Date/Time: Wednesday, August 2, 2017 : 8:30 AM to 10:20 AM
Sponsor: Section on Medical Devices and Diagnostics
Abstract #324183 View Presentation
Title: Probability of Detection of Target Nucleic Acid Sequence with PCR Assay in Molecular Diagnostics: Math Model Derivation, Validation and Applications
Author(s): Jeffrey Vaks*
Companies: Roche Molecular Diagnostics
Keywords: Real time PCR-based assays ; Analytical sensitivity ; Binomial/Poisson modelling of PCR process ; Concentrations at specified detection rates: C5, C50 and C95
Abstract:

Probability of detection as function of target nucleic acid sequence concentration and minimum number of copies detectable determines important performance characteristic, analytical sensitivity, of polymerase chain reaction assay. Only math model for the case of single copy detectable was published elsewhere. General math model for probability of detection with any minimum number of copies detectable was derived and validated with ?2 goodness-of-fit test using data collected on several types of instruments for limit of detection evaluations of several viral and microbial assays. The p-values of 84 ?2 goodness of fit tests on such data ranged from 0.050 to 0.999, with ?2 test on combined data having 157 degrees of freedom and p-value of ~1 successfully validating the math model. Examples of applications provided are: (1) estimation of sample concentration with confidence bounds from observed detection rate, (2) estimation of concentrations corresponding to 5%, 50% and 95% detection rates with confidence bounds, (3) probability of detection vs. concentration and minimum number of copies detectable curves corresponding to several values of limit of detection with single copy detectable, and (4) bounds for concentrations corresponding to 5% and 95% detection rates, recommended as precision characteristic of qualitative assays in CLSI guideline EP12-A2.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association