After a two year research stay in the United States at Argonne National Laboratory, Mirko Hahn joined our group in Magdeburg at the end of June to finish his PhD Thesis about efficient solution methods for optimization problems with integers and partial differential equations.
Dr. Florian Kehrle finished his PhD thesis on Inverse Simulation for Cardiac Arrhythmia. There are several dozen different types of cardiac arrhythmia. While most of them can be easily differentiated by cardiologists, the discrimination between atrial fibrillation (AFib) and regular atrial arrhythmias including atrial flutter (AFlu) and focal atrial tachycardia poses a diagnostic challenge. The usually available data, a surface electrocardiogram (ECG), looks very similar in both cases, to laymen, physicians, and computerized algorithms alike. Florian developed a comprehensive approach for clinical decision support. It starts with automatic data generation with a smartphone. The mobile app HEAT can extract R wave times from ECG pictures and from beep recordings of a heart rate monitor. The data is sent via a secure connection to our computation solver where the optimization problem is solved. Results are directly exported as PDF files and sent back to the app. To solve the optimization problem Florian developed a tailored Branch and Bound algorithm.
Felix Bernhardt successfully defended his Master's Thesis on Comparison of machine learning, domain-specific and hybrid models for discrimination of cardiac arrhythmias in which he complemented the work of Florian Kehrle by applying machine learning algorithms. Starting in January 2019 he is a PhD student in our working group and will continue working on the intersection of machine learning and optimization applied to medical applications.
In 2018 the members of the MATHOPT group joined several conferences and presented new research results. The conferences with its participating group members are listed below.