Clinical Metabolomics


Group Leader

Facts & Details: 

The timely and precise diagnosis of a patient’s disease state and prediction of future outcomes is of utmost importance to facilitate and improve both global as well as personalized health care. Metabolomics, the quantitative study of small organic compounds, so-called metabolites, present in a biological specimen, provides a footprint of the current state of an organisms. It is thus predestined to offer important insights into disease pathomechanisms as well as to enable powerful disease diagnosis and prognosis.
Mathematical models derived by state-of-the art machine learning algorithms make such disease outcome predictions possible. These machine learning algorithms autonomously combine outcome-relevant information, i.e. variables, into mathematical models, which can subsequently be used to make predictions for newly examined patients. These variables can comprise any available patient information, ranging from biomedical data, disease history, and demographic data, to large-scale metabolomics measurements.
Our group is particularly interested in the improved diagnosis and outcome prediction of widespread systemic diseases, e.g. chronic kidney disease, employing large-scale metabolomics data. Our bioinformatics methods cover a broad range of different machine learning algorithms ranging from multivariate Cox regression, classification, and scale-invariant zero-sum regression, to Gaussian and Mixed Graphical Models.
We are always interested in motivated students to conduct their bachelor, master, or MD thesis in our group. If you are interested, please send an e-mail with your research interests to h.zacharias [at]