Whether it's the shortage of staff or the huge rush of patients: hospital emergency rooms are subject to great stress. In order to relieve the medical staff, four cooperation partners of the innovation network AIMECA, including the Hochschule Niederrhein, are developing an intelligent real-time planning system for emergency rooms over a period of about two years (duration until November 2025). The "NotPASS" project is being sponsored by the Central Innovation Program for SMEs (ZIM) with a total volume of around 875,000 euros.
The aim is to improve process planning in hospitals by applying AI in the form of reinforcement learning algorithms. The AI is intended to support medical staff in decision-making, so that treatment processes are improved and patient waiting times are shortened. In addition, the project partners are working on technical solutions that ensure the continuity and availability of patient data from the receipt of an emergency call to care and treatment in the hospital.
The Hochschule Niederrhein (HSNR) is participating in the project with two university-owned facilities. The competence centre "CC eHealth" identifies and prepares the relevant data for the AI and develops the system logic. System boundaries are prepared and user profiles are created, which are required for subsequent data input. The entire emergency process chain must be analyzed in detail so that proper planning can be created and data can be collected. This also involves the speed with which the planning system must react to changes and how they should ultimately be executed in concrete terms.
"I am particularly pleased about this project. Analyses of compliance with maximum times between patient arrival and doctor contact, show that these are not infrequently exceeded. The project results will lay the foundation courses for better coordination between patient needs and required resources in medicine and care. This minimizes risks without incurring additional expenses," says Professor Dr. Hubert Otten, project leader at "CC eHealth" and HSNR lecturer for technical systems, operational organization and logistics in healthcare facilities at the Faculty of Health Care.
HSNR's Institute for Modeling and High Performance Computing has to overcome some challenges on the way to the finished planning system. Here, a planning optimization based on reinforcement learning is developed in order to comprehend the complex system of the emergency room with its actors and technical devices in an all-encompassing way. The different fields of application as well as the methods suitable for them will be identified. The algorithms are to be continuously optimized in the project so that the results can be transmitted to a feedback system. To ensure that the system can be successfully applied in the future, the project partners are working with various hospitals that are testing the AI in practical applications. "If research succeeds in expanding machine learning methods to include probabilistic analysis, artificial intelligence will become more and more similar to human intelligence and thus more efficient and reliable in terms of forecasting," says Dr. Dirk Roos, Professor of Computer Simulation and Design Optimization at the Faculty of Mechanical and Process Engineering.
In addition to The Hochschule Niederrhein, two other partners are involved in "NotPASS": Health365 AC GmbH and cibX GmbH. All cooperation partners are part of the AIMECA innovation network, which bundles competencies from information and communication technologies as well as from scientific and medical disciplines.
