her2

Inspiration

Since I was diagnosed with Cancer in Fall of 2021, it has a journey from being anxious to being curious and appreciating it as an accelerated learning challenge. It has helped to connect with people having similar cancer and learn from each other. It has also been beneficial to study other similar cases on the web - Notably from Radical Remission.

What it does

Provides prompts to specify to personal cancer type and provide context .

How we built it

Using open ai LLM and text embeddings and giving a context and corpus of text to focus questions on that

Challenges we ran into

Working with embeddings running into technical errors. Understanding how the local context and larger search data in GPT could be connected.

Accomplishments that we're proud of

Getting it working partially with football data.

What we learned

Getting clear about scope of GPT usage, api usage, figuring out keys, embeddings, prompt engineering

What's next for her2

clear up errors, define the data better and have a large online dataset available. Figure out as how to give weightage for Context sensitive data from RadicalRemission.org and my FB Private HER2 feed, and Standard of Care at Stanford Cancer Center hospital, to see what are the main modalities in Standard of Care across the world, and what are alternative treatments, targeted therapies which worked for many patients after their Stage 4 cancer diagnosis.

Built With

• 3.9
• colab
• engineering
• openai
• prompt
• python
• tiktoken