Jadwal Sholat

Memuat jadwal sholatโ€ฆ

Kedokteran & Kesehatan editorial

Open AccessOA2023

Scaling Clinical Trial Matching Using Large Language Models: A Case Study in Oncology

This study explores the use of large language models (LLMs) to scale clinical trial matching in oncology, demonstrating that LLMs like GPT-4 can structure eligibility criteria and extract complex matching logic, outperforming baselines but still facing challenges in accuracy and context limitations.
C. Wong; Sheng Zhang; Yu Gu; C. Moung; Jacob Abel; N. Usuyama; R. Weerasinghe; B. Piening; Tristan Naumann; C. Bifulco; Hoifung Poonยท Machine Learning in Health Careยท 2023ยท DOI 10.48550/arXiv.2308.02180

The core problem

Clinical trial matching is a critical process in healthcare delivery and discovery, but it is hindered by overwhelming unstructured data and unscalable manual processing. This paper presents a systematic study on scaling clinical trial matching using large language models (LLMs), focusing on oncology. The study is grounded in a clinical trial matching system currently in test deployment at a large U.S. health network. The authors aim to evaluate the potential of LLMs to automate and improve the efficiency of matching patients to clinical trials.

Innovation

The study leverages cutting-edge LLMs, such as GPT-4, to structure elaborate eligibility criteria of clinical trials and extract complex matching logic, including nested AND/OR/NOT conditions. The system is designed to process unstructured clinical trial data and patient information from longitudinal medical records. The authors conduct a systematic evaluation of LLM performance against prior strong baselines, assessing the models' ability to handle the complexities of clinical trial matching in oncology.
Introduction
Clinical trial matching is a critical process in healthcare delivery and discovery, but it is hindered by overwhelming unstructured data and unscalable manual processing. This paper presents a systematic study on scaling clinical trial matching using large language models (LLMs), focusing on oncology. The study is grounded in a clinical trial matching system currently in test deployment at a large U.S. health network. The authors aim to evaluate the potential of LLMs to automate and improve the efficiency of matching patients to clinical trials.

Why it matters

The study reveals significant growth areas for applying LLMs to end-to-end clinical trial matching. Key challenges include context limitations and accuracy, especially in structuring patient information from longitudinal medical records. The authors discuss the implications of these findings for the deployment of LLM-based systems in clinical settings, emphasizing the need for further research to address these limitations and improve the reliability of automated matching.

Who should read this

Pembaca Research Digest dan praktisi rumpun ini.

Opening member contentโ€ฆ