Inspiration
In 2023, there were 65 million deaths worldwide. Out of which 20 millions died due to heart-related illnesses or disease. How can we apply GenerativeAI + Agentic RAG to regenerate human heart?
What it does
We have built CytoRAG, a Multi-modality copilot for Cell Biologists that integrates text and microscopic images to recommend, predict and generate innovative ways to regenerate cell tissues.
How we built it
- Build a data ingest pipeline that parse, dedup, clean the source data for ingestion to pinecone
- Integrate with llamaIndex workflow, LlamaCloud
- Evaluate with Arize Phoenix to study the vector space of the data
- Deploy with Vessl
Challenges we ran into
- Build a data ingest pipeline that parse, dedup, clean the source data for ingestion to pinecone
- Integrate with llama workflow
- Evaluate with Arize Phoenix to study the vector space of the data
- Deploy with Vessel
- Ensuring the right package version of llama-index, arize-phoenix :)
Accomplishments that we're proud of
Real-World Applications*: Human Biology remains one of the most complex subject for Computational AI. Heart tissue generation will significantly transform the lives of many who are suffering heart-related disease or illness.
Integration & Interoperability*
Transparency & Ethics: We've implemented additional measures (HarmCategory.HARM_CATEGORY) to address harmful content, including hate speech, dangerous content, sexually explicit material, and harassment. This enhanced transparency and ethical approach aims to create a safer online environment for all users.
Research & Development: To contribute to cutting-edge research in Cell Biology using RAG and agentic systems, our application will process up to 10 millions medical literatures published on https://pubmed.ncbi.nlm.nih.gov/. The complexity of human biology poses a significant challenge for computational AI. However, breakthroughs in heart tissue generation hold the potential to revolutionize the lives of those affected by heart-related diseases. By contributing to research and development in RAG and agentic systems, we can advance the capabilities of AI and explore new possibilities in this field.
What we learned
What's next for Regenerating Human Heart
In this Hackathon, we have ingest a small subset of the 10 millions medical literatures published on https://pubmed.ncbi.nlm.nih.gov/. We intend to continue after the hackathon Beyond medical literature and microscopic images, we intend to ingest additional modality of data including: Genomics (DNA data), Transcriptonics such as single-cell RNA sequence data, Proteomics, Metabolomics (metabolite data such as concentration of Oxygen, Calcium) and 10+ modalities. It is an monumental challenge but if we solve it, it will solve tissue regeneration, and transform the lives of 20 million people suffering heart-related illnesses.
Built With
- amazon-web-services
- llama-cloud
- llama-index
- llama-workflow
- pinecone
- python
- vessl
Log in or sign up for Devpost to join the conversation.