# Filipe Burmester

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Filipe Burmester](https://cerebralvalley.ai/u/filipeburmester)
- **GitHub:** https://github.com/fburmester/pharmacyp
- **Demo video:** https://youtu.be/-fyEVTyoO9s
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/202

Modern medicine is prescribing more drugs to the same patient than ever before. ~42% of U.S. adults aged 65+ take five
   or more prescriptions, and adverse drug reactions are a top-six cause of preventable death — the JAMA estimate is
  ~106,000 fatal ADRs per year in U.S. hospitals alone. Two failure modes drive most of that harm: the cytochrome P450
  enzymes (which metabolize ~75% of clinical drugs and vary genetically between patients), and drug–drug interactions
  that compound silently across a regimen.

  Pharmacyp is a clinical decision support tool that catches both. A clinician enters a patient's drug list and CYP
  genotype; Pharmacyp returns a structured risk report through three layers: a deterministic rule engine over CPIC
  pharmacogenomic guidelines and curated DDI data (DDInter 2.0), two graph-neural-network microservices that predict
  interactions and CYP behavior from chemical structure when no curated data exists, and a Claude Opus 4.7
  explainability layer that annotates every finding with a citation back to the underlying clinical literature. Every
  output is labeled by source — RULE-BASED, ML-PREDICTED, or INSUFFICIENT DATA — so uncertainty is never hidden behind a
   confident tone.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/202. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt