Quick Search for Similar Patient Situations

Developed an AI-powered system for retrieving similar patient cases using Large Language Models (LLM) and

embedding-based similarity search.

• Implemented GPT-4o for medical text summarization and text-embedding-ada-002 for similarity matching,

achieving high retrievalaccuracy.

• Designed an interactive UI with Gradio, ensuring seamless access for healthcare professionals.

• Optimized retrieval speed, achieving an average cosine similarity score above 0.85 with response times

under 2 seconds.