Statistical Modeling of Primary Data Unit

The unit "Statistical Modeling of Primary Data" is the contact for all questions related to the statistical-methodological aspects of planning, conducting, evaluating and publishing epidemiological studies. The aim is the biometrically sound conception of projects, the adaptation or development of suitable analytical methods, and the statistical training of staff. This applies both to BIPS-internal research projects and to studies that are located outside BIPS and conducted together with the unit. In addition, the so-called federated data analysis is gaining in importance, in which analysis codes are transmitted to the data while maintaining data protection and data security and no personal data need to be transferred on the part of the data-holding organizations.
However, the unit does not see its actual task in the routine preparation of standard analyses, but explicitly in the adaptation and further development of innovative analysis procedures. For example, together with three other European study centers, approaches are being developed to improve the classification of objectively measured exercise and eating behavior with the help of machine learning methods. In another project, artificial intelligence methods are also being used to derive optimal diagnostic parameters (so-called biomarkers) by enriching cancer registry data with diagnostic data from pathology, which could then form the basis for more effective treatment methods. Improving data quality is also the goal of a DFG-funded project in which tools are being developed to assess the data quality of epidemiological studies as standardized as possible. It is also intended to help strengthen the basis for the FAIR criteria of findability, accessibility, interoperability and reusability. In this way, the project ties in seamlessly with the goals of NFDI4Health, an interdisciplinary research consortium in which the unit is also involved. One of the aims of this initiative is to develop a centralized catalog of health research-relevant study data with user-friendly data access management and to improve the linking of research-relevant data sources in Germany, taking into account the special data protection requirements of personal health data. In this context, the unit is developing an interactive online evaluation tool for the analysis of existing primary data and software for the use of generalized additive models for location, scale and shape parameters (GAMLSS), taking into account the problem of distributed data.

Staff

Buck, Christoph, Dr.
Tel.: +49 (0)421 218-56944
Fax: +49 (0)421 218-56941
buck(at)leibniz-bips.de

Intemann, Timm, Dr.
Tel.: +49 (0)421 218-56984
Fax: +49 (0)421 218-56921
intemann(at)leibniz-bips.de

Mändle, Andreas, Dr.
Tel.: +49 (0)421 218-56794
maendle(at)leibniz-bips.de

Peters, Manuela, Dr.
Tel.: +49 (0)421 218-56924
Fax: +49 (0)421 218-56941
mpeters(at)leibniz-bips.de

Pohlabeln, Hermann, Dr.
Tel.: +49 (0)421 218-56947
Fax: +49 (0)421 218-56941
pohlabeln(at)leibniz-bips.de

Swenne, Annika
Tel.: +49 (0)421 218-56943
Fax: +49 (0)421 218-56941
swenne(at)leibniz-bips.de