Validation in Statistics and Machine Learning - Abstract
Within the pharmaceutical industry, the requirements for validation are mainly governed by regulatory guidances. The guidelines primarily deal with formal parts of the regulatory procedure. Validation traditionally plays an important role in the submission process. As the industry is increasingly investing in diagnostic techniques towards more targeted medical approaches, new aspects of validation arise. In particular, statistical approaches towards the verification of results found in biomarker or pharmacogenomic studies are of interest. This presentation highlights the different areas of validation in the drug development process, and differentiates the requirements, specifically in the area of new statistical methodology and with respect to the clinical utility of the respective outcomes.