Hybrid AI System for Delirium Risk Prediction – Supporting Nursing Practice
In surgical and internal-medicine hospital wards, up to 50 % of patients may develop delirium, an acute but usually reversible disturbance of cognition and attention. Delirium is associated with complications, longer hospital stays, and poorer health outcomes. It also places a considerable burden on nursing staff, as caring for affected patients is often complex and time-consuming.
However, delirium is often preventable when patients at risk are identified early and appropriate preventive measures are implemented.
The KIDELIR research project aims to support this early detection. In collaboration with nursing professionals, the research team is developing a hybrid artificial intelligence (AI) decision-support system that estimates each patient’s individual risk of developing delirium. Part of the team are our BrainLinks-BrainTools members Prof. Tonio Ball, Prof. Joschka Bödecker, Prof. Christoph Maurer und Jun.-Prof. Philippe Kellmeyer. By integrating data from multiple clinical sources, the system is designed to identify patterns associated with increased delirium risk and to support clinical decision-making in everyday care.
The goal is to provide practical support for healthcare professionals by:
- analysing routinely collected patient data to estimate an individual delirium risk score
- alerting nursing staff when a patient’s risk exceeds a defined threshold
- recommending evidence-based preventive measures tailored to individuals at increased risk
Recognition at the “Ideas for Impact” Health Prize
The project was recognised by the Bosch Health Campus as part of the Ideas for Impact Health Prize, awarded on behalf of the Robert Bosch Stiftung.
The Ideas for Impact prize honours innovative healthcare concepts in Germany that have the potential to sustainably improve the healthcare system, particularly with regard to the health and quality of life of older people. It highlights projects that use digital technologies and artificial intelligence to strengthen prevention, care delivery, and patient support.
In the 2026 edition, KIDELIR was selected as one of five finalists out of around 100 projects reviewed by the foundation. Representatives of the project consortium were invited to the official award ceremony on 26 February 2026 in Berlin.
Congratulations to our BrainLinks-BrainTools researchers and all project partners on this recognition!