Zihang Chen, Liang Liu, Qianmin Su, Gaoyi Cheng, Jihan Huang, Ying Li
School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, ShangHai, China.
Institute of Clinical Science, Zhongshan Hospital, Fudan University, ShangHai, China.
Sci Rep. 2025 Jul 28;15(1):27398. doi: 10.1038/s41598-025-11876-0.
The recruitment of participants for clinical trials has traditionally been a passive and challenging process, leading to difficulties in acquiring a sufficient number of qualified participants in a timely manner. This issue has impeded advancements in medical research. However, recent years have seen the evolution of knowledge graphs and the introduction of large language models (LLMs), providing innovative approaches for the pre-screening and recruitment phases of clinical trials. These developments promise enhanced recruitment efficiency and increased participant involvement. To ensure the safety and efficacy of clinical trials, it is crucial to establish precise inclusion and exclusion criteria for participant selection. This paper introduces a method to optimize the pre-recruitment stage by utilizing these criteria in conjunction with the cutting-edge capabilities of knowledge graphs and LLMs. The enhanced strategy includes the automated generation of questionnaires, algorithmic evaluation of eligibility, supplemental query-response functions, and a broader participant screening reach. The application of this framework yielded a detailed clinical trial recruitment questionnaire that accurately encompasses all necessary criteria. Its JSON output is noteworthy for its precision and reliability, achieving an impressive 90% accuracy rate in summarizing patient responses. Additionally, the questionnaire's ancillary question-and-answer feature complies with stringent legal and ethical standards, meeting the requirements for practical deployment. This study validates the practicality and technological soundness of the presented approach. Utilizing this framework is expected to enhance the efficiency of trial recruitment and the level of patient participation.
传统上，临床试验参与者的招募是一个被动且具有挑战性的过程，导致难以及时招募到足够数量的合格参与者。这个问题阻碍了医学研究的进展。然而，近年来知识图谱不断发展，大语言模型（LLMs）也被引入，为临床试验的预筛选和招募阶段提供了创新方法。这些进展有望提高招募效率并增加参与者的参与度。为确保临床试验的安全性和有效性，为参与者选择建立精确的纳入和排除标准至关重要。本文介绍了一种方法，通过将这些标准与知识图谱和大语言模型的前沿能力相结合来优化预招募阶段。增强后的策略包括自动生成问卷、算法评估资格、补充查询-响应功能以及更广泛的参与者筛选范围。该框架的应用产生了一份详细的临床试验招募问卷，准确涵盖了所有必要标准。其JSON输出因其精确性和可靠性而值得注意，在总结患者回答时准确率达到了令人印象深刻的90%。此外，问卷的辅助问答功能符合严格的法律和道德标准，满足实际部署的要求。本研究验证了所提出方法的实用性和技术合理性。预计利用这个框架将提高试验招募效率和患者参与水平。
