Smart Searching in Patient Records Using AI: A New KIMed Webinar Shows How

News | 24 April 2026

How to (Not) Build a RAG System – Lessons Learned from Developing an Intelligent Search System for Patient Records | KIMed Network Insights on May 21, 2026, at 5:00 p.m.

With its new webinar series “KIMed Network Insights – After-Work Seminar on AI and Medicine,” KIMed brings together interdisciplinary stakeholders working on the application of artificial intelligence (AI) in medicine. In short keynote presentations, network partners will present their projects, experiences, and research questions—ranging from the development of new AI solutions to their application in everyday clinical practice. A moderated discussion following each presentation will provide an opportunity for questions, exchange, and new perspectives.

The series kicks off with a webinar titled “How to (Not) Build a RAG System – Lessons Learned in Developing an Intelligent Search for Patient Records”  on May 21, 2026, from 5:00 to 6:00 p.m. Sarah Rockstroh and Dr. Robert Haase, research associates at the National AI Competence Center SCADS.AI Dresden/Leipzig, will share their experiences, challenges, and insights from the development and use of Retrieval-Augmented Generation (RAG) systems.

RAG is a method that specifically enhances the capabilities of large AI language models: Instead of relying exclusively on the knowledge stored during training, a RAG system can first search a defined knowledge base—such as a database—in response to a specific query and then incorporate the information found into the generated response. In clinical practice, AI systems could thus access specific, patient-related documents and process them to support medical decision-making.

This very principle underlies the “Intelligent Patient Record (iPA)” project, which will be presented in the webinar. Physicians are familiar with the problem: Finding relevant findings, previous reports, or medication histories quickly and reliably in extensive patient records takes valuable time—time that is often in short supply in everyday clinical practice. The iPA project aims to make this process more efficient through AI-supported retrieval methods. Instead of scrolling linearly through documents, physicians can search specifically for relevant content.

The webinar will explain how typical medical search strategies can be replicated using modern language models, and what needs to be considered in order to translate current research findings into clinical practice. The event is explicitly aimed at anyone interested in the topic, regardless of their prior technical knowledge.

Title: “How to (Not) Build a RAG System: Lessons Learned from Developing an Intelligent Search System for Patient Records”

Date and Time: May 21, 2026, 5:00–6:00 p.m.

Target Audience: Clinicians, physicians, clinician scientists, and anyone interested

Access link: https://uni-leipzig.zoom-x.de/j/61702883162?pwd=wjBYv3jgQR1CkawHnd6yxPkSTbMYqv.1