摘要
目的探索冠心病病变全程的证候要素分布与组合规律。方法基于冠心病证候、证候要素诊断的行业标准,运用改良动手深度学习(Transformer)算法构建冠心病证候要素诊断系统,对比系统诊断结果与4位临床医师基于治疗指南的诊断结果,获知系统证候要素诊断准确性;借助冠心病证候要素诊断系统对中国中医科学院广安门医院的1221例冠心病住院病例进行证候要素诊断及不同分期证候要素分布、组合规律研究。结果通过临界病变、冠心病心绞痛未行介入手术、冠心病心绞痛介入术后12周内、冠心病心绞痛介入术后1年以上并伴心功能降低(NYHA分级为II级或III级)患者病例各20例的人、机对比,冠心病证候要素诊断系统平均准确度可达96.46±8.957%。冠心病证候要素诊断系统对证候要素判断结果提示,冠心病冠脉临界病变阶段以气滞、血瘀的实性证候要素及组合为主;冠心病心绞痛未行介入手术阶段存在血瘀、气虚、痰浊、气滞等实性证候要素为主的分布及其组合形式;冠心病介入术后12周内的患者可呈现阴虚、气虚、气滞、血瘀等证候要素分布及其组合形式;冠心病介入术后1年以上伴心功能不全的患者多表现为气虚、痰浊、阳虚等虚性证候要素为主的证候要素分布及组合形式。结论冠心病病变全程的证候要素分布及组合形式存在一定时间性的演变规律,可出现由实性证候要素向虚实夹杂的证候要素分布及组合形式演变,最终发展成虚性为主的证候要素分布及组合模式,而基于深度学习算法的冠心病证候要素辨证系统为证候要素诊断的客观化提供途径。
Objective To explore the distribution and combination of syndrome elements in the whole course of coronary artery disease(CAD).Methods The modified Transformer algorithm model was built based on the standard of CAD syndrome and syndrome elements diagnosis.The diagnostic results of the system were compared with the diagnostic results of four clinicians based on treatment guidelines to obtain the diagnostic accuracy of the system’s syndrome elements;the diagnosis of syndrome elements and the distribution and combination patterns of syndrome elements at different stages were studied in 1,221 inpatient cases of coronary heart disease at Guang’anmen Hospital of the China Academy of Chinese Medical Sciences with the help of the syndrome element diagnosis system.Results The average accuracy of the diagnostic system for coronary artery disease evidence elements was 96.46±8.957%in 20 cases each of patients with critical lesions,coronary angina without intervention,within 12 weeks after coronary angina intervention,and more than 1 year after coronary angina intervention with reduced cardiac function(NYHA class II or III).The results of the diagnosis system of coronary artery disease syndrome elements suggested that the solid syndrome elements and combinations of qi stagnation and blood stasis were predominant in the stage of coronary artery critical lesions;the distribution of excessive syndrome elements such as blood stasis,qi deficiency,phlegm and qi stagnation;patients within12 weeks after coronary artery intervention could present yin deficiency,qi deficiency,qi stagnation and blood stagnation.Patients with cardiac insufficiency more than 1 year after coronary intervention mostly showed the distribution and combination of deficiency elements such as qi deficiency,phlegm stagnation and yang deficiency.Conclusion The distribution and combination of syndrome elements of coronary artery disease throughout the whole course of the disease follow a certain temporal evolution pattern,which can evolve from the excessive syndrome elements to the distribution and combination of mixed deficiency and excessive syndrome elements,and finally develop into the distribution and combination of deficiency-oriented syndrome elements.
作者
李洪峥
王阶
张振鹏
郭雨晨
杜强
高嘉良
董艳
李剑楠
李谦一
Li Hongzheng;Wang Jie;Zhan Zhenpeng;Guo Yuchen;Du Qiang;Gao Jialiang;Dong Yan;Li Jiannan;Li Qianyi(Guang'anmen Hospital,China Academy of Chinese Medical Sciences,Beijing 100053,China;Graduate School of Beijing University of Chinese Medicine,Beijing 100029,China;Department of Automation,Tsinghua University,Beijing 100084,China;Beijing Xiaobai Century Network Technical Co.Ltd,Beijing 100084,China)
出处
《世界科学技术-中医药现代化》
CSCD
北大核心
2021年第9期3086-3094,共9页
Modernization of Traditional Chinese Medicine and Materia Medica-World Science and Technology
基金
国家中医药管理局传承与创新“百千万”人才工程(020100401):岐黄学者,负责人:王阶
国家中医药管理局循证能力建设项目(60103):冠心病病证结合人工智能诊疗的循证能力建设研究,负责人:王阶
关键词
冠心病
证候要素
分布规律
组合规律
人工智能
深度学习
Coronary artery disease
Syndrome elements
Laws of distribution
Laws of combination
Artificial intelligence
Deep learning