Machine Learning in Laboratory Medicine
The medical laboratory of the future should not only provide accurate measurements, but also support clinicians using data-driven diagnostic recommendations. The aim is to perform algorithm-based interpretations of laboratory results and thereby facilitate an integrative, step-by-step diagnostic approach.
To reach this goal, our working group analyzes laboratory data with non-linear, multiparametric machine learning methods in order to predict clinically relevant outcome events. High ethical standards as well as the European General Data Protection Regulation (GDPR/DSGVO) are respected.
We are currently working on various projects addressing retrospective and prospective data analysis: as an example, we are trying to identify data constellations that allow a more exact and earlier prediction of heparin-induced thrombocytopenia in intensive care patients.
The above projects are carried out in cooperation with the group of William Martin from the “Institute of Molecular Evolution” at the Heinrich Heine University in Duesseldorf as well as Georg Dorffner and Alexander Tolios from the "Institute of Artificial Intelligence and Decision Support” and the “Department of Blood Group Serology and Transfusion Medicine" at the Medical University of Vienna.