Andreas Metzger

Andreas Metzger Presents at Dagstuhl Seminar on AI-Driven Process Adaptation

In his presentation, Prof. Metzger explored various AI-based approaches to optimize actionable recommendations for adapting business processes. The presented methods address the trade-off between prediction accuracy and earliness.

The seminar took place from May 4–9, 2025 at Schloss Dagstuhl – Leibniz Center for Informatics. More information about the seminar is available at: https://www.dagstuhl.de/25192

Abstract of the Talk „AI-Driven Process Execution and Adaptation“

Prescriptive business process monitoring aims to guide process managers on adapting processes to avoid negative outcomes. A key challenge is balancing prediction accuracy with earliness: earlier predictions allow more adaptation time, but are often less reliable. This talk explores existing methods for managing this trade-off and compares their performance by using real-world datasets. By evaluating their cost savings, this talk identifies factors influencing effectiveness and provides practical recommendations for selecting appropriate approaches. Based on these insights, the talk explores directions and opportunitiues for future research, involving emerging topics such as exaplainable AI and LLMs.

The slides are available here: https://adaptive-systems.org/images/documents/dagstuhl-presentation-25.pdf