Our group is studying metabolic diseases (e.g. type 2 diabetes and complications). We use statistical, bioinformatical and machine learning approaches to integrate metabolomics, transcriptomics, proteomics, epigenomics, and genomics data, as well as information from public databases to better understand the pathophysiological mechanisms. This helps us to identify candidate biomarkers and drug targets, with the final goal to translate our discoveries into clinical practice.
Intelligent management of type 2 diabetes and its complications
Despite a plethora of therapeutic approaches many people with diabetes fail to reach their treatment goals. Intelligent diabetes management is needed to enable precision biomedicine and to improve patient outcomes. Based on larger human cohort data (e.g. multi-level and multi-time points OMICs data with clinical variables), we aim to develop digital patient assessment tools and explore predictive modeling of type 2 diabetes and its complications using artificial intelligent approaches. For example, for each individual, using genetic data, we will create a genome-wide polygenic score to identify increased risk for type 2 diabetes and related complications (e.g. CVD and cancer). Additionally, using machine leaning approaches, we integrate epigenome, proteome and metabolome data, which reflect physiological, environmental as well as altered life style factors (physical activities, smoking, alcohol intake and medication). Moreover, we use reinforcement learning to optimize the performance of a personalized risk predictive model. Ultimately, we aim to translate our knowledge by combining our model in digital tools directly assessable by patients.