[PolyPhy] Unveiling Medicine Patterns: 3D Clustering with Polyphy/Polyglot
- Mentors
- Oskar Elek, Ez_ra, Kiran Deol
- Organization
- UC OSPO
- Technologies
- python, jupyter
- Topics
- machine learning, data science, natural language processing
Polyglot, a web application revolutionizing data exploration, offers dynamic 3D visualizations. Imagine zooming and rotating through a complex dataset, where each data point is a unique element. Polyglot empowers researchers to delve deeper into intricate relationships hidden within their data. This project proposes leveraging Polyglot's strengths to unlock hidden connections in medicine. By creating custom medicine embeddings that capture the relationships between drugs based on their shared salts (beyond just active ingredients), we can explore a massive dataset of over 600,000 medicines. Visualizing these embeddings in Polyglot's 3D space could allow researchers to identify previously unknown connections between medicines based on their salt compositions, potentially leading to breakthroughs in drug discovery, pharmacology, and personalized medicine.