SwimRAG- AI-Powered Chatbot for Swimming Literature
Patrick McOsker
Lay Summary:
I created a web app for a RAG workflow powered by an LLM to answer queries about swimming injuries. The LLM references a curated database of over 200 scientific articles on swimming injuries to provide accurate and grounded information.
Abstract:
The vigorous training of swimmers and in particular the repetitive stress on the upper body musculature make them especially susceptible to injury. Naturally, there is a great deal of scientific literature detailing various swimming injuries and their respective preventative measures. Due to the volume of research on the topic, finding answers to specific questions can be tedious, particularly for clinicians and therapists who must balance the literature search with clinical duties. We introduce SwimRAG, a web application that allows for querying of a curated database of swimming injury literature. This is powered by a type of artificial intelligence known as Large Language Models (LLMs). LLMs have seen an increase in research over the past recent years due to their ability to perform a variety of tasks with accuracy that approaches human ability. Unfortunately, LLMs have been known to have certain imperfections such as presenting false or outdated information to the query. Retrieved Augmented Generation (RAG) can help with this by providing information to the model based on an external knowledge base, whether it be contextual or relevant information. With RAG, LLMs can provide more accurate answers to user queries.. We created SwimRAG, a RAG workflow that references a curated database of 200 scholarly articles about swimming injuries to provide contextual grounding to summarize the findings to answer any queries a researcher might have. We hope this tool will allow clinicians and sports medicine professionals to obtain information more quickly and provide quality care.
Q&A:
Bios: Patrick McOsker
Program Track: Advanced Research
GitHub Username:
Patrick123codered -Patrick McOsker
What was your favorite seminar? Why?
Linda Vahdat’s seminar on inhibiting cancer spread via copper. I thought it was a really interesting approach and I didn’t know copper was so integral to metastasis -Patrick McOsker
If you were to summarize your summer internship experience in one sentence, what would it be?
Challenging, eye-opening, but well supported -Patrick McOsker