Key features
- ~70% accuracy baseline for research & learning - Transparent, extensible codebase (RDKit, ML, GNN roadmap) - Safe "sandbox" for experimenting with drug metabolism prediction - Reproducible and community-driven Open source → research/teaching now, bridge to clinical-grade systems later.
Repo: https://github.com/Flamehaven/Arr-Medic-CYP3A4
Try the live demo here: https://huggingface.co/spaces/Flamehaven/arr-medic-cyp3a4-de... Read how it all began: "My First Attempt at a Medical AI with ELI5" on Medium: https://medium.com/p/ccc0edbd2bce
Feedback welcome: How do you think such tools could be responsibly used or improved?