LLM 기반 생성 AI를 활용한 임상시험
Clinical Trials Utilizing LLM-Based Generative AI
2
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 연구는 대규모 언어 모델(LLM) 기반의 Private LLM을 활용하여 의료기기 임상시험 분야에적용하여 업무 효율성과 전문성 향상을 탐구한다. Private LLM 시스템은 임상 데이터를 기반으로정교하고 정확한 답변을 제공하며, 의사결정 지원, 임상 전문가 활동 보조, 새로운 콘텐츠 생성, 문제 해결 등 다양한 응용 분야에서 활용 가능성을 보여준다. 연구는 다음 네 가지 주요 단계로구성된다. 첫째, 의료기기 임상시험에 특화된 데이터를 수집하고 이를 전처리하여 학습 가능한 형식으로 정리한다. 둘째, LLaMA와 같은 오픈소스 LLM 모델을 기반으로 PEFT(LoRA) 및 RAG 기법을 적용하여 특정 임상 도메인에 맞는 맞춤형 Private LLM 질의응답 시스템을 구축한다. 셋째, 구축된 시스템을 활용하여 전문가 수준의 질의응답 기능을 실현하고, 임상시험 운영 중 발생하는복잡한 질문과 문제를 해결한다. 마지막으로, 시스템의 성능을 평가하여 임상시험 운영과 의료기기 개발의 효율성과 신뢰성을 높이기 위한 방향성을 제안한다. 연구 결과, Private LLM 시스템은기존의 방법론 대비 업무 자동화와 정밀한 의사결정 지원에서 탁월한 성능을 보였다. 특히, 도메인 전문가의 질문에 대한 정확한 답변을 제공하고, 새로운 임상 기준 및 인사이트를 생성할 수있는 능력은 의료기기 임상시험 운영의 혁신적 도구로 자리잡을 가능성을 보여준다. 이를 통해정밀 의료, 임상시험 관리 자동화, 그리고 도메인 지식 기반의 질의응답 시스템에서 Private LLM 의 실질적 활용 가능성을 확인하였다.
Abstract
This study explores the improvement of work efficiency and expertise by applying Private LLM based on Large Language Model (LLM) to the field of clinical trials in medical devices. The Private LLM system provides sophisticated and accurate answers based on clinical data and shows its potential for use in various applications such as decision support, clinical expert activity assistance, new content generation, and problem solving. The study consists of the following four main steps. First, data specific to clinical trials of medical devices are collected, preprocessed, and organized into a learnable format. Second, based on open-source LLM models such as LaMA, PEFT (LoRA) and RAG techniques are applied to build a customized private LLM Q&A system for a specific clinical domain. Third, it realizes expert-level Q&A function by utilizing the established system and solves complex questions and problems that arise during clinical trial operation. Finally, by evaluating the performance of the system, we propose a direction to increase the efficiency and reliability of clinical trial operation and medical device development. As a result of the study, the Private LLM system has outperformed the existing methodology in supporting task automation and precise decision making. In particular, the ability to provide accurate answers to questions from domain experts and to generate new clinical criteria and insights shows the potential to become an innovative tool in clinical trial operation of medical devices. This confirmed the practical applicability of Private LLM in precision medical care, automation of clinical trial management, and a Q&A system based on domain knowledge.
참고문헌 (20)
- "Using AI to Improve Patient Outcomes in Clinical Trials"[URL]
- "AlphaFold: AI for Protein Structure Prediction"[URL]
- Chen, Q., "Keep up with the latest coronavirus research", Nature, 579(7798), pp.193~ (2020)[DOI]
- Radford, A., "Language Models are Unsupervised Multitask Learners", OpenAI Blog, 1(8), pp.9~ (2019)
- Johnson, A. E. W., "MIMIC-III, a freely accessible critical care database", Scientific Data, 3(1), pp.1~9 (2016)[DOI]
- Vaswani, A., "Attention is all you need", Advances in Neural Information Processing Systems, pp.5998~6008 (2017)
- Raffel, C., "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer", Journal of Machine Learning Research, 21(140), pp.1~67 (2020)[DOI]
- Kai He, "A Survey of Large Language Models for Healthcare : from Data, Technology, and Applications to Accountability and Ethics", Journal of LaTeX Class Files, 14(8), pp.1~8 (2021)
- "Prompting Engineering Guide"[URL]
- LLaMA : Open and Efficient Foundation Language Models, "Hugo Touvron", Meta AI Research, ~ (2023)
- Jason Wei, "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models", Google Research, Brain Team, ~ (2022)
- Xie, T., "A Survey of Large Language Models for Healthcare: From Data, Technology, and Applications to Accountability and Ethics", ~ (2023)
- He, K., "A Survey of Large Language Models for Healthcare : from Data, Technology, and Applications to Accountability and Ethics", IEEE Transactions on Neural Networks and Learning Systems (2024)
- Liu, L., "A Survey on Medical Large Language Models : Technology, Application, Trustworthiness, and Future Directions", Journal of LaTeX Class Files, 14(8), pp.1~ (2024)
- Yuan, M., "Large Language Models Illuminate a Progressive Pathway to Artificial Healthcare Assistant: A Review", ~ (2024)
- Mumtaz, U., "LLMs-Healthcare:Current Applications and Challenges of Large Language Models in various Medical Specialties", ~ (2024)