Background
When you get blood test results, they typically show whether each value is within or outside the normal range. But the normal range is broad — it covers 95% of the population. Being “normal” doesn’t mean being optimal. CRP below 10 is normal, but above 3 increases cardiovascular risk. Vitamin D above 50 is normal, but below 75 is suboptimal according to recent research.
I wanted a tool that compares my blood tests with what the research actually says — not just standard reference intervals.
What I built
AI-Lægens Bord is a set of Claude Code Skills that together do one thing: fetch your health data and cross-reference it with current medical research.
Tech stack
- sundhed.dk skill fetches blood tests, medication, vaccinations and records via Playwright and MitID
- PubMed and medRxiv skills search medical research — both peer-reviewed and preprints
- Lab review skill compares each biomarker against evidence-based optimal intervals
- SQLite stores everything locally — no cloud, no server, no sharing
AI agents run in parallel: 4-8 agents simultaneously research different biomarkers against PubMed and medRxiv, identify patterns across markers, and compile an action plan with prioritized recommendations and expected impact.
Results
All data stays on your own machine. The entire project is open source on GitHub, built with zero dependencies beyond Node.js built-ins. No npm packages, no cloud services, no tracking.
I use it myself. It’s not medical advice — it’s a tool that makes the research accessible to anyone willing to read it.
