Efficient nonparametric analysis of diagnostic and prognostic biomarkers

• Applicants: Professor Dr. Marc Ditzhaus; Professor Dr. Frank Konietschke
• Subject Area: Epidemiology and Medical Biometry/Statistics
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 527029865

Diagnostic trials to evaluate biomarker reliabilities have become an important area in clinical as well as in pre-clinical and translational research. Existing statistical methods mostly rely on total areas under ROC curves and thus take the entire range of true and false positives into account. Partial areas under the ROC curves (and other accuracy measures, e.g. Youden-indices) are more informative because they focus on clinically relevant segments. Statistical inference methods for the analysis of these measures in factorial diagnostic trials have not been developed and implemented yet. This project aims to close these gaps to improve biometric practice. The proposed research has been motivated by various real trials and statistical consulting.

• DFG Programme: Research Grants