fortiss played a central role in the ZNAflow project. Scientists from the fields of Requirements Engineering and Machine Learning systematically recorded the requirements of clinical practice and mapped them in a specially developed artifact model. This model realistically depicts the complex processes and dependencies in an emergency room, thus creating the basis for a practical assistance system.
Building on this, fortiss developed a prototype for an AI-based prediction model. It processes historical ZNA data, urgency levels, weather and calendar information, and other influencing factors to predict overload situations up to three hours in advance. At the same time, the system transparently displays the most important influencing factors so that staff can understand the predictions.
Greater safety and relief for hospital staff
Another focus of fortiss' work was the investigation of the ethical, legal, and social aspects (ELSI) of using AI in emergency rooms. The findings were incorporated into a practical guide that provides hospitals and developers with guidance on the responsible use of data-based systems.
The assistance system visualizes the predicted workload on an intuitive dashboard. This enables medical teams to identify critical situations at an early stage and respond in a targeted manner – for example, by adjusting duty rosters, keeping diagnostic capacities free, or transferring patients in good time. This significantly increases the reliability of care and confidence in decision-making, while reducing the burden on staff. “ZNAflow enables us to evaluate the performance of AI-based assistance systems in the emergency room in a practical setting. It is particularly exciting to see how data-based predictions can support medical processes and at the same time serve as a scientific basis for further optimizations,” emphasizes Florian Angermeir, project manager at fortiss.
Testing planned in other emergency departments
The completion of ZNAflow marks an important step toward the digitization and optimization of clinical processes. In the future, the system will be tested in other central emergency departments (ZNA) to validate the quality of the forecasts and further improve the application.
In addition to fortiss, AGAPLESION gAG, AGAPLESION Evangelische Krankenhaus Mittelhessen, DOCYET GmbH, and the Technical University of Munich as the network coordinator played a key role in the success of ZNAflow.
