Integrative Medical Logic Engine ("Gem")
Gemini Custom GPT bridging Eastern and Western medicine
Top outcome
Reduced diagnostic analysis from an hour to minutes
Stack
Gemini Custom GPT (Gem) Framework · RAG (Medical Corpus) · Python · Advanced System Instructions
Overview
An integrative medical AI system built on Gemini's Custom GPT framework that bridges the gap between Traditional Chinese Medicine (TCM) and Western medical diagnostics. The system uses RAG over curated medical texts to provide evidence-based recommendations while maintaining strict translation tables that map TCM pattern concepts to Western clinical terms. Designed as a "collaborative consultant" that enhances practitioner decision-making without replacing clinical judgment.
The Problem
Synthesizing Eastern (TCM) and Western medicine diagnostics into a unified intake process, where terminology and reasoning paradigms fundamentally differ.
The Solution
Gemini Custom GPT framework with RAG for medical texts and translation tables that harmonize cross-tradition terminology.
Impact
- 1Reduced diagnostic analysis from an hour to minutes
- 2Harmonized terminology across medical traditions
- 3Eliminated clinician confusion via "Translation Tables"
Key Decisions
- Chose "Collaborative Agent" persona over "Prescriptive Agent" to maintain human-in-loop
- Built strict translation tables to ensure TCM/Western term alignment
- Implemented semantic validation layer (data-heavy vs. pattern-heavy reasoning)
Lessons Learned
Semantics matter: Western medicine is data-heavy; TCM is pattern-heavy. System instructions must reflect this.
Practitioners need confidence in AI insights; framing as "consultant" reduces bias risk
RAG quality directly impacts diagnostic accuracy
Screens
1 / 3