Transforming electrophysiology workflows with natural language processing and agentic artificial intelligence
Akshar Patel, Stanley Joseph, Caryl F. Bailey, Ashish Sakharpe, Mallikarjuna Devarapalli · Heart Rhythm O2 · 2025
This article explores how Natural Language Processing (NLP) models and agentic AI can streamline workflows in electrophysiology (EP). It discusses fine-tuning models such as BioBERT for EP-specific tasks, Named Entity Recognition for identifying key terms, real-time guideline updates using web scraping, and the integration of these components into a unified agentic AI workflow. The Hugging Face Transformers library and its pipeline() function are leveraged for various NLP tasks, including summarization, text generation, and translation, to automate literature reviews, guideline monitoring, and report generation.