Multi-Modal Exploration of Medical Data with Agentic AI
Generative AI and Large Language Models (LLMs) have become integral to modern data systems both in academia and industry. However, a critical limitation of LLMs is their tendency to produce non-factual outputs, commonly referred to as hallucinations. In this talk, recent results of leveraging agentic AI technologies for multi-modal data exploration in medicine will be presented, where datasets comprise heterogeneous modalities, including structured numerical data, unstructured text, and visual content. Kurt Stockinger, professor at the ZHAW School of Engineering, will demonstrate how agentic AI systems can be used to query such multi-modal datasets in natural language and thus provide the medical personnel with a powerful tool to gain new insights into their patient data in an intuitive way.