论智能中医辨证可解释研究的挑战和研究思路  

Challenges and Ideas in the Interpretability of Intelligent Syndrome Differentiation of Traditional Chinese Medicine

作　　者：张婷婷[1] 王亚强[2] 温川飙[1] 叶莹[1] ZHANG Tingting;WANG Yaqiang;WEN Chuanbiao;YE Ying(Chengdu University of Traditional Chinese Medicine,Chengdu,610075;Chengdu University o fin formation Technology)

机构地区：[1]成都中医药大学,四川省成都市610075 [2]成都信息工程大学

出　　处：《中医杂志》2022年第6期506-510,共5页Journal of Traditional Chinese Medicine

基　　金：国家自然科学基金(61801058);中国博士后科学基金(2021M693785);成都中医药大学“杏林学者”学科人才科研提升计划(BSH2020031);成都中医药大学校教育教学改革基金(XJG2019-11982)。

摘　　要：智能中医辨证的研究已有数十年的历史,但由于不理解机器做出决策的原理,决策者难以信任机器的辨证结果,导致至今智能辨证的临床应用仍十分有限。通过分析中医智能辨证的发展,认为其局限主要包括有限的可用数据、语料库的缺乏、评价方法的局限,且中医辨证本身的解释难度大和非规范化等因素,致使中医辨证可解释研究的开展面临挑战。提出智能辨证可解释研究具有必要性和可行性,未来的可解释研究需要多学科合作,充分利用已有的解释资源,构建辨证的解释体系和大型语料库,以支持对可解释智能中医辨证的探索。Although it has been decades since the intelligent syndrome differentiation was produced in traditional Chinese medicine(TCM),its clinical application has been limited due to the difficulty in understanding how machines make decisions and thereafter the distrust of decision makers in the machine results. The development of intelligent TCM syndrome differentiation is mainly restricted by limited available data,lack of corpus,insufficient evaluation methods,as well as the challenges in interpretability and standardization of TCM syndrome differentiation.Therefore,it is necessary and feasible to conduct researches to explore the interpretability of intelligent TCM syndrome differentiation by integrating multi-disciplines,making full use of existing interpretability-related resources,and building syndrome-differentiated interpretable system and large corpora.

关 键 词：智能辨证 可解释 人工智能 自然语言处理 

分 类 号：R241[医药卫生—中医诊断学]