Medical co-pilot: First AI-supported medical device approved to assist clinical decision-making

AIQ and VDE are driving forward the development and approval of AI-supported medical devices. As a medical device based on a large language model (LLM), the Prof. Valmed software application supports medical professionals in making diagnoses and selecting suitable therapeutic approaches. AIQ and VDE provided expertise in technology and regulation. Innovative AI approaches in healthcare require consideration of regulatory and technical requirements.

Function and benefits of the software

To support medical professionals in making diagnoses and therapy recommendations, the application uses a system based on a large language model (LLM) to access around 2.5 million validated medical documents from sources such as guidelines, regulatory texts, PubMed and Cochrane. The aim is to provide evidence-based care and relieve the burden on medical staff in their daily work.

Technical architecture and role of AIQ

AIQ’s development and testing approach uses Retrieval-Augmented Generation (RAG), which enables the LLM to search specific medical sources in real time and integrate relevant content. This structure aims for accuracy, transparency and trust – key requirements in medicine. Technology validation, safety testing and regulatory requirements were addressed jointly. Certification according to DIN EN ISO 13485:2021 confirms compliance with quality standards for medical devices.

Regulatory strategy and partners

An early focus on regulatory requirements and the implementation of a strategic ‘compliance-by-design’ process were crucial. VDE and AIQ provided support in setting up a quality management system and creating complete technical documentation as a prerequisite for CE certification. Thanks to the structured approach, the project was able to obtain certification in a very short period of time.

Significance of the development

This is one of the first LLM-based applications for clinical decision support with successful CE certification as a Class IIb medical device according to MDR. This represents an important step towards the safe and compliant use of AI in everyday clinical practice.