Involved Working Groups
- Early detection of sepsis through analysis of complex patterns: Many of the parameters necessary for sepsis diagnosis are available in the bedside electronic patient data management system (PDMS). Automated filtering and analysis according to defined criteria or by artificial intelligence, could be helpful means to monitor parameter changes and detect complex critical correlations. This would allow a diagnosis to be made more quickly and an appropriate therapy to be started sooner. The current analysis includes about 25,000 patients who were treated in the surgical and internal intensive care units. Among them are about 2000 patients with severe sepsis and septic shock, which should be detected as early as possible through the use of neural networks and XGBoost.
Luz C, Vollmer M, Decruyenaere J, Nijsten M, Glasner C, Sinha B (2020)
Machine learning in infection management using routine electronic health records: tools, techniques, and reporting of future technologies
Clinical Microbiology and Infection, 26(10), 1291-1299; DOI:10.1016/j.cmi.2020.02.003
Vollmer M, Luz CF, Sodmann P, Sinha B , Kuhn S-O (2019)
Time-specific Metalearners for the Early Prediction of Sepsis
Computing in Cardiology (CinC), 2019; Vol 46; ISSN: 2325-887X; DOI:10.22489/CinC.2019.029