中医人工智能辨证研究现状与发展  

Research Status and Development of Artificial Intelligence Syndrome Differentiation in Traditional Chinese Medicine

作　　者：李连新 杨璠 朱兆鑫 杨珊珊 杨汐 王雪颖 陈菊 Li Lianxin;Yang Fan;Zhu Zhaoxin;Yang Shanshan;Yang Xi;Wang Xueying;Chen Ju(College of Clinical Medicine,Chengdu University of Traditional Chinese Medicine,Chengdu 611137,China;College of Medical Information Engineering,Chengdu University of Traditional Chinese Medicine,Chengdu 611137,China)

机构地区：[1]成都中医药大学临床医学院,成都611137 [2]成都中医药大学医学信息工程学院,成都611137

出　　处：《世界科学技术-中医药现代化》2021年第11期4268-4276,共9页Modernization of Traditional Chinese Medicine and Materia Medica-World Science and Technology

基　　金：国家自然科学基金委员会青年科学基金项目(81804222):基于深度学习的“因位性势”四维中医辨证论治算法模型设计,负责人:陈菊;国家科学技术部国家重点研发计划中医药现代化重大专项(2018YFC1704104):便携设备健康服务云平台构建及示范应用研究,负责人:温川飙;四川省卫生健康委员会普及应用项目(20PJ168):基于自动编码器和RNN结合的深度学习方法对肺系疾病进行临床诊断的应用研究,负责人:陈菊。

摘　　要：目的 总结中医人工智能辨证领域的研究历史与现状、应用状况、所面临的挑战,为中医辨证智能发展提供帮助。方法 检索中国知网(China National Knowledge Infrastructure,CNKI)总库1980年-2020年人工智能与中医辨证相关文献,筛选出301篇,以时间为线索,从专家系统、机器学习、深度学习和知识图谱,对中医辨证算法研究进行分类;以地域为线索,结合文献计量学可视化分析,梳理出中医辨证人工智能地域发展态势。结果 通过三个阶段中医辨证算法的比较,机器学习的神经网络、决策树和贝叶斯等算法模型辨证准确率均值在90%以上,适用于中医辨证,未来深度学习与知识图谱在中医辨证方面极具潜力。结论 中医辨证信息化迅猛发展并处于持续深入的状态,但中医文本数据的处理和结构化工作滞后,缺乏评价系统来评判人工智能算法在辨证上的效果,需予以重视。Objective To summarize the research history, current situation, application and challenges in the field of TCM artificial intelligence syndrome differentiation, so as to provide help for the development of TCM dialectical intelligence. Methods A total of 301 articles were selected from CNKI database from 1980 to 2020. The research on TCM syndrome differentiation algorithm was classified from expert system, machine learning, deep learning and knowledge map with time as the clue. The regional development trend of artificial intelligence in TCM syndrome differentiation was sorted out based on geographical clues and bibliometric visualization analysis.Results Through the comparison of the three stages of TCM dialectical algorithm, the average dialectical accuracy of neural network, decision tree and Bayesian algorithm model of machine learning was more than 90%, which was suitable for TCM syndrome differentiation. In the future, deep learning and knowledge mapping had great potential in TCM syndrome differentiation.Conclusion The rapid development of TCM dialectical informatization is in a continuous and in-depth state, but the work of TCM diagnosis objectification and standardization lags behind, and there is a lack of evaluation system to evaluate the effect of artificial intelligence algorithm on syndrome differentiation, which needs to be paid attention to.

关 键 词：人工智能 中医 辨证 机器学习 

分 类 号：R2-03[医药卫生—中医学]