Yulin Zhao PP
Keywords: clinical trial, information retrieval, question-answering, large language model, information extraction, answer generation, code generation, ranking, clinicaltrial.gov, ChatGPT, gpt-3.5-turbo, NewBing, BioGPT
Abstract: Large language models have revolutionized natural language processing tasks with their impressive performance. In this paper, we introduce TrialGPT, a novel language model designed to augment clinical trial information retrieval and question-answering tasks. TrialGPT incorporates a search function that leverages the ClinicalTrials.gov API documentation to generate search expressions and retrieve trial information. Additionally, it features a question-answering function that utilizes information extraction and answer generation modules to generate NLP answers based on extracted trial information. We present the architecture and implementation details of TrialGPT, along with a dataset created for evaluating its performance.
Submission Number: 7
Loading