AutoQUEST: A Chain-of-Thought Pipeline for Automated Question Generation and Validation in MAUDE Research

The process of formulating research questions using the Manufacturer and User Facility Device Experience (MAUDE) database is often complicated by the challenges of data preprocessing and analysis. To meet the challenges, AutoQUEST, a Python-based prompt pipeline that capitalizes on large language models (LLMs) and Chain-of-Thought (CoT) has been proposed to facilitate the automation of question formulation. In five distinct test cases, AutoQUEST yielded an accuracy rate of 100% in generating valid research questions and attained query execution success rates ranging from 75% to 100%. This innovative CoT pipeline facilitates the research question formulating process, reduces technical barriers in data extraction and transformation, and enhances the efficacy of patient safety research concerning medical devices.