AI Driven Clinical Communication

On September 18th 2026, the AID-CC consortium met again in Graz in the brand-new radiology building on the Medical University campus. Projectmanagement-wise we reviewed the past six months of the project and coordinated our progress and next steps with the project plan. This time the focus of our discussions revolved around evaluating different aspects as well as components of the dialogue system, for instance,

– What data should be used (from a technical perspective — what is attributable to Multiple Sclerosis (MS) and what is not)?
– Which findings are interrelated? What information needs to be summarized in the “MS story,” and how is the distinction made regarding other clinical issues unrelated to MS (e.g., cardiovascular conditions, etc.)?
– What is the agent’s procedure regarding guidelines for handling special cases? Currently, the system does not prioritize whether information should be retrieved from the findings or the guidelines first. Should the agent also make inquiries in cases of uncertainty?
– How to weigh the patient’s self-report?

The next phase of the project includes (i) comparison of medical large language models applied to the on-site clinical MUG data, (ii) an extension of our hybrid AI dialogue system by a reasoning component (e.g. by integrating guidelines), (iii) preparation of quantitative as well as qualitative evaluation of the system (iv) intensify publication activities.