What is it about?
This study explores the use of large language models (LLMs) for improving the risk prediction performance based on Electronic Health Records (EHRs), demonstrating notable performance enhancements and advantages like adaptability and few-shot learning. Yet, caution is warranted due to LLM prompt sensitivity and persistent safety concerns.
Featured Image
Photo by National Cancer Institute on Unsplash
Why is it important?
The study offers a comparative assessment of current risk prediction methodologies and underscores the advantages of integrating LLMs in Electronic Health Records (EHR) representation. When used effectively, the knowledge retained by pre-trained LLMs hold significant potential for enhancing the performance of downstream tasks.
Perspectives
Large language models (LLMs) are being widely adopted for different healthcare applications. This research investigates how LLMs can be effectively used to represent the structured data within Electronic Health Records (EHRs) for different risk prediction tasks. Modeling the past medical histories using LLMs have significant benefits, with some potential concerns that need to be studied further.
Angeela Acharya
George Mason University
Read the Original
This page is a summary of: Clinical risk prediction using language models: benefits and considerations, Journal of the American Medical Informatics Association, February 2024, Oxford University Press (OUP),
DOI: 10.1093/jamia/ocae030.
You can read the full text:
Resources
Contributors
The following have contributed to this page
