From ELIZA to RAG - Information Aggregation as the Foundation of Modern Medicine

This working paper addresses the critical information overload in modern medicine, particularly in oncology and general practice. It challenges the imprecise term "Artificial Intelligence" and proposes "probabilistic real-time information aggregation" as a more accurate scientific framework. Drawing a historical arc from ELIZA (1966) to modern Retrieval-Augmented Generation (RAG) , the paper outlines how the convergence of structured mass data, mathematical cybernetics, and GPU parallelism enables valid and reliable clinical decision support. The author introduces the 5-R Rule (Right documentation, patient, date, context, and time) as a quality framework for healthcare informatics. The paper further describes the pilot deployment of the J-Med Practice Assistant (codename: BitCarrot), a JavaFX-based overlay that integrates RAG, Neo4J graph databases, and local GDPR-compliant LLM processing to assist physicians in real-time.