FAISS-Based Semantic Search over Cleaned Clinical Trial Corpus using Sentence Transformers

This study presents a semantic search framework for clinical trials using sentence-transformer embeddings and FAISS vector indexing. By embedding cleaned metadata from thousands of trials, the model supports real-time retrieval of topically relevant studies. Each query yields interpretable outputs, including word clouds and PrettyTables of trial summaries. The system bridges raw clinical data and meaningful insights, helping researchers navigate vast biomedical corpora. Our lightweight architecture enables efficient querying without sacrificing interpretability, facilitating adoption in academic, regulatory, and industrial settings.