Career

Department of Medical Statistics

Topics for Theses

We offer a range of topics for theses in the areas mentioned above. However, we also welcome your own suggestions. The following topics are currently offered by our working and service groups.

Working group Statistical Methods for Clinical Trials:

• "Agent-based modelling of therapeutic interventions: How can we explore complex patient-therapist interactions and their effects in childhood obesity therapy with agent based models (ABM)?" (joint project with Professor Kerstin Wiegand​​​​​​​)

Working group Clinical Epidemiology and Health Economics:

• "Validation and recalibration of kidney transplantation risk prediction models using data from the German transplantation registry"
• "Examination whether the implementation of a fixed effect in beta-binomial regression models preserves the principle of concurrent control in blocked designs such as meta-analyses"
• "Cross-Validation of real-world/routine health care data and data collected for clinical trials from the University Medical Center Göttingen"

Working group Computational Statistics:

• "Comparing Causal Forests with GAMLSS for Heterogeneous Treatment Effect Evaluation"
• "Treatment effects on a bivariate outcome using distributional copula regression models"
• "Incorporating Variability into Treatment Optimization Algorithms"
• "Variable selection for causal inference in observational data: propositions for GAMLSS"

Scientific Core Facility Medical Biometry and Statistical Bioinformatics:

• "Design aspects of federated learning" - Federated learning refers to an approach of learning on distributed data that does not require the merging of data. Such approaches are particularly attractive in a medical context due to the high data protection requirements. As part of the German Health Centre for Cardiovascular Diseases, this project aims to investigate various design aspects of federated learning as applied to CT and ECG data in cardiac patients.
• "Implementation and Testing of an Analysis Pipeline for Single-Cell Data Analysis" - Single-Cell Experiments are new Techniques with an increased use to tackle questions in molecular biology. While the analysis steps are already implemented, the increasing demand suggests to implement a fully-automatized analysis pipeline. The project would be supervised in collaboration with the Translational multiomics and precision cardiology group.
• "Prediction of Cas13 RNA silencing efficency from gRNA primary sequence" - Cas13 is a protein that degrades specific RNA molecules in dependence of a gRNA sequence. The Efficency of gRNA sequences remains elusive and its prediction is subject to ongoing research. The Translational multiomics and precision cardiology group collected data that can serve as training data for machine learning algorithms to tackle that problem.
• "Multimodal single-cell based Up-stream Analysis" - Multimodal single-cell experiments allow to gather information about gene status and its regulation simultanously in a single-cell resolution. These data bares the potential to identify potential triggers for a specific gene expression change in a specific cell context. The development and extension of an integrative analysis method remains a challenge.
• "Identification of genes that drive epigenetic changes" -  In a previous project we used epigenetic features to predict gene activation. From this project the idea was born to reverse the task and implement an interpretable machine learning model in order to identify which genes steer the relevant epigenetic changes. 

If you are interested in writing a thesis, please contact the head of the relevant working or service group directly.