You are here: Home Knowledge Discovery Projects

Selected project overview

Distributed Computing Under Data Protection Constraints

 

Individual patient data are highly confidential data meaning that data transfer might be restricted even in the context of research in joint projects.

Within the context of the projects MIRACUM and GESA, we develop specialized statistical methods to analyze these data without disclosing individual information jointly. Specifically, only aggregated, non-disclosive summary statistics are shared between the project partners. Our methods are built on the framework DataSHIELD.

Principal Investigator: Harald Binder

Contributor: Daniela Zöller, Stefan Lenz

distributed-computing-vert-small.png

MIRACUM - Medical Informatics in Research And Care in University Medicine

 

The MIRACUM consortium, an association of initially 8 university hospitals (Erlangen, Frankfurt, Freiburg, Gießen, Mainz, Magdeburg, Mannheim, and Marburg), is a project funded by the medical informatics initiative of the Federal Ministry of Education and Research (BMBF). The aim of the MIRACUM consortium is to build up a research infrastructure that can be used to jointly analyze decentralized data from treatment and research. This is demonstrated by three so-called "use cases."

Use Case 2 („From Data to Knowledge”), headed by Harald Binder, searches for patterns/subgroups in patient data. It consists of two parts. In the first part, clinical data and laboratory data of asthma/COPD patients are examined. The second part deals with clinical and genetic data (gene methylation) of patients with brain tumors.

Principal Investigator: Harald Binder

Contributor: Stefan Lenz

miracum

SoftMeta – Software for Meta-Analysis

 

The aim of this project is the implementation and dissemination of statistical methods for meta-analysis in the publicly available free software R. Starting point of this project are the R packages meta, metasens, netmeta, and diagmeta which have been developed on GitHub. Please feel free to visit our project page on ResearchGate for the latest updates as well as the companion website to our book "Meta-Analysis with R."

Principal Investigator: Guido Schwarzer

Contributor: Gerta Rücker

meta_cover

Modeling of ROC Curves in Meta-Analysis of Diagnostic Test Accuracy Studies and Network Meta-Analysis

 

This DFG-project works on two active research areas of evidence synthesis in medicine, meta-analysis of diagnostic test accuracy studies and network meta-analysis. We developed a new approach to meta-analysis of diagnostic test accuracy studies to combine full ROC curves and implemented this in a new R package diagmeta. Furthermore, we continuously expand our R package netmeta for network meta-analysis, for example with methods to disentangle the effects of combination therapies. The long-term objective is to combine both areas to network meta-analysis of diagnostic test accuracy studies.

Principal Investigator: Gerta Rücker

Contributor: Guido Schwarzer, Susanne Steinhauser (for diagnosis)

SROC

DynaMORE - Dynamic MOdeling of REsilience

 

The EU (HORIZON 2020) project DynaMORE exploits advanced mathematical modeling for the promotion of mental health and well-being.

As one of twelve consortium participants, we integrate multimodal and high-dimensional data (e.g., fMRI and behavioral data at baseline as well as ecological monitoring data from smartphones and wristbands) in our models and contribute statistical techniques for dynamic prediction of resilience. For both objectives, we also apply deep learning techniques and compare them to traditional approaches.

Principal Investigator: Harald Binder

Contributor: Göran Köber

dynamore

GESA - Gender-Sensitive Analyses of Psychological Health

 

In large cross-sectional studies, a considerable difference in the prevalence of mental disorders between men and women has been observed. Sex- and gender-sensitive analysis, as planned for the BMBF project GESA (FKZ 01GL1718A), can be used to uncover the causes of the observed differences.

We combine three large-scale population-based cohorts (SHIP / GHS / KORA) to enable a longitudinal analysis. The aim is to establish hypotheses for causal relationships and models.

Principal Investigator: Harald Binder

Contributor: Daniela Zöller

gesa-placeholder.png

MeInBio - BioInMe: Exploration of Spatio-Temporal Dynamics of Gene Regulation Using High-Throughput and High-Resolution Methods

 

MeInBio addresses fundamental open questions with regard to transcriptional control in transitory developmental or adaptation states. The research training group employs high-throughput sequencing approaches and high-resolution technology by downscaling existing methods to small cell numbers and single cells.

The design of single cell RNA-seq experiments, particularly sample size calculation, is a critical step when addressing biological hypotheses because it ensures adequate statistical power. We are assessing distributional assumptions of single cell gene expression measurements and use deep generative models to uncover complex patterns of interest. These patterns are used for sample size and design investigations concerning differential expression analysis, cell clusters and continuous processes in cells.

Principal Investigator: Harald Binder

Contributor: Martin Treppner

meInBio

AgeGain - Transfer of Cognitive Training Gains in Cognitively Healthy Aging: Mechanisms and Modulators

 

AgeGain is a multicentric, randomized controlled trial that investigates the mechanisms of transfer learning to identify measures for maintaining mental performance in old age. The study centers are the University Hospital Cologne, the German Sport University Cologne, and the University Medical Center in Mainz and Rostock. Participants are volunteers of age 60-85 without major physical or mental illnesses. One part of them receives special cognitive training, and a subgroup of those gets a fitness training in addition to that. In the course of the study, a variety of measurements is taken, including assessments of the physical and social activity, brain scans (fMRI and PET), and physical performance tests.

Our part of the project is to provide statistical support, especially concerning the challenges of analyzing such a multimodal and high-dimensional dataset.


Principal Investigator: Harald Binder

Contributor: Stefan Lenz, Göran Köber

gesa-placeholder.png

 

Other projects

 

For a list of all current projects please see here...

For a list of finished projects please see here...