Statistical methods and software for validation studies on new in vitro toxicity assays.

Area of Science:

• Toxicology
• Assay Development
• Biostatistics

Background:

• New in vitro assay methods require rigorous validation.
• Validation ensures reliability against established knowledge or reference assays.
• Binary outcome assays are common in toxicological testing.

Purpose of the Study:

• To provide a framework for validating new in vitro assay methods.
• To detail the calculation of predictive values and confidence intervals for binary assays.
• To illustrate these methods with toxicological examples using R software.

Main Methods:

• Utilizing 2x2 tables to display binary assay outcomes.
• Calculating positive and negative predictive values based on sensitivity, specificity, and prevalence.
• Employing statistical methods for confidence interval computation.
• Demonstrating sample size calculations for toxicity assays.

Main Results:

• The study presents a clear methodology for assay validation.
• R code is provided for calculating confidence intervals and performing sample size calculations.
• The application is illustrated using two toxicological case studies.

Conclusions:

• The described validation approach enhances the reliability of new in vitro assays.
• Accessible R tools facilitate the statistical analysis of assay validation data.
• This work supports the robust implementation of novel toxicological testing methods.