基于医疗大模型的患者咨询问答安全及伦理管控  

Safety and Ethical Control of Patient Consultation and Question-answering Based on Medical Large Language Models

在线阅读下载全文

作  者:王天笑 于佳婕 钱琨 WANG Tianxiao;YU Jiajie;QIAN Kun(Zhongshan Hospital,Fudan University,Shanghai 200032,China)

机构地区:[1]复旦大学附属中山医院,上海市200032

出  处:《中国卫生信息管理杂志》2025年第2期202-207,共6页Chinese Journal of Health Informatics and Management

摘  要:目的面向患者医疗咨询领域,研究大模型技术应用的安全及伦理管控体系。方法通过分析大模型技术在患者医疗咨询问答领域面临的安全和伦理挑战,结合实际案例,设计基于大模型的患者咨询问答系统,评估该系统在实际环境中对安全和伦理的风险管控效果。结果构建了大模型在患者医疗咨询问答场景中的安全和伦理框架,基于此框架能够对可能出现的风险进行识别和管控。结论该体系通过对患者医疗咨询领域中出现的隐私保护、数据安全、透明性、伦理风险等核心问题进行具体分析和有效应对,可显著提高大模型在医疗知识问答场景下的可解释性和准确性,扩大其真实世界应用前景。Objective In the field of patients’medical consultations,conduct research on the safety and ethical control system for the application of large language model technology.Methods A patient consultation Q&A system based on large models is designed by analyzing the security and ethical challenges of largescale models in the field of patient medical consultation and Q&A and incorporating practical case studies.The system’s effectiveness in managing security and ethical risks in real-world environments is then evaluated.Result A security and ethical framework for the application of large-scale models in patient medical consultation and Q&A and its corresponding risk management and identification strategies were established.Conclusion This framework conducts specific analyses and takes effective countermeasures against core issues such as privacy protection,data security,transparency,and ethical risks that arise in the field of patients’medical consultations.As a result,it can significantly enhance the interpretability and accuracy of large language models in the context of medical knowledge Q&A scenarios,and broaden their application prospects in the real world.

关 键 词:大模型 安全 伦理 患者咨询 问答系统 

分 类 号:R197.323.2[医药卫生—卫生事业管理] R319[医药卫生—公共卫生与预防医学]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象