A Digital Physician Twin for Diagnostic Reasoning: Grounding Large Language Models in Clinical Measurements and Expert Knowledge

• Applicants: Privatdozent Holger A. Lindner, Ph.D.; Professor Dr. Stefan Riezler
• Subject Area: Medical Informatics and Medical Bioinformatics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 554382092

The goal of the proposed project is to build a digital physician twin that implements artificial intelligence support for clinical reasoning. From a health-care perspective, the purpose of a digital physician twin is to aid medical practitioners in clinical reasoning in an intelligible way that allows corrective feedback via human-system interactions. From a machine learning perspective, the challenge and promise of a such a system is to enable large language models to perform structured reasoning from clinical features documented in electronic health records and to extract potential novel rules of medical reasoning from data labeled by human experts.

• DFG Programme: Research Grants