Diagnostic tests are the starting point for all clinical decision-making and are therefore responsible for the quality of care available to a patient. As a consequence, high-quality diagnostic studies are of great importance in order to provide the most appropriate diagnostic test for a given individual in a given diagnostic setting.

However, statistical methods for diagnostic studies are not as well developed as those for therapeutic studies. A crucial gap is that repeated observations per person, which lead to so-called clustered data, are not sufficiently taken into account in the evaluation process when using the classic measures of diagnostic accuracy such as sensitivity, specificity and the area under the ROC curve.

The central aim of this project is to provide a comprehensive concept for planning and analysing diagnostic studies with spatially or temporally clustered data and to implement this concept in research practice. To this end, we will first systematically review the methods currently available in the field of diagnostic studies with clustered data and compare them using systematic simulation studies.

We will then extend the currently developed estimation framework for diagnostic studies to the case of temporally and spatially clustered data. Based on the estimation concept, we will further develop methodological approaches to obtain valid estimates and systematically compare them with existing methods.

In addition, methods for predictive values and sample size planning will be developed. Finally, we will provide an implementation of the new framework as a guideline and an R-package to enable translation into research practice. All this will be done in close collaboration with experts in the field of methodology and application of diagnostic tests.

Duration: 2024 - 2027

Dr. Nicole Rübsamen (Universitätsklinikum Münster)

• Prof. Dr. Vanessa Didelez (BIPS, Bremen)
• Prof. Dr. Larissa Fabritz (Universitätsklinikum Hamburg-Eppendorf)
• Prof. Dr. Tim Hahn (Universitätsklinikum Münster)
• PD Dr. Thomas Jack (Medizinische Hochschule Hannover)
• Dr. Katharina Kramer (Universität Augsburg)
• Prof. Dr. Antje Wulff (Universität Oldenburg)