Keywords: deep learning, time series, dimensionality reduction, clustering
Hospitals collect and store data on their patients with the intent to facilitate better treatment. These data include the doses and types of medication administered, the medical procedures applied, laboratory results and other patient-related information, e.g. ECG data.
If properly utilized, this information could be used to find hidden patterns and enable precision medicine: What kind of medication would help this patient best? What intervention can be used for patients especially at risk, and who are they? Which research questions should be addressed in new clinical studies?
This technology has the potential of improving care for patients and lowering operating costs for hospitals at the same time. However, utilizing this information is a challenging problem. The data is very high-dimensional (there are many attributes), it is temporally complex (there are some regions with no or little data and some regions with dense data) and is sparse (the same data was not collected for each patient).
In my work, I address these challenges using deep-learning-based dimensionality reduction and clustering techniques.
|2018 - present||Doctoral candidate, Joint Research Center for Computational Biomedicine, RWTH Aachen University, Germany|
|2014 - 2018||Master of Science, Computer Science at RWTH Aachen University, Germany. Master thesis|
|2011 - 2014||Bachelor of Science, Computer Science at RWTH Aachen University, Germany. Bachelor thesis|