Else Kröner Fresenius Center (EKFZ) for Digital Health researchers at TU Dresden, along with partners, have developed an autonomous AI agent designed to enhance clinical decision-making in the field of oncology.

This tool has the potential to support healthcare professionals by simplifying the analysis of complex medical data and assisting in tailored cancer treatment plans.

The autonomous AI agent, built upon the advanced language model GPT-4, integrates various digital tools, such as radiology report generation and medical image analysis.

It can predict genetic alterations from histopathology slides and search through medical databases, including PubMed and OncoKB.

The AI's knowledge base includes approximately 6,800 documents from established oncology guidelines and clinical resources, ensuring its decisions are informed by current medical knowledge.

Although the system has shown promise in simulated cases, further validation is necessary.

The research team's future efforts will concentrate on enhancing the AI's conversational abilities, incorporating human feedback, and maintaining data privacy by operating on local servers.

EKFZ for Digital Health at TU Dresden Clinical Artificial Intelligence professor and Dresden University Hospital Carl Gustav Carus oncologist Jakob Kather said: “To fully leverage the potential of AI agents in the future, it will be crucial to integrate them smoothly into routine clinical practice with minimal disruption.

“Challenges such as interoperability with existing systems, compliance with data privacy laws, the need for regulatory approval processes as medical devices, and ensuring accountability still need to be addressed.”

The researchers believe that AI agents similar to this one, with the relevant tools and data, could be adapted for various medical fields in the future.