1.湖南科技大学数学与计算科学学院,湖南省湘潭市雨湖区桃园路, 411201
2.中国中医科学院中医临床基础医学研究所
3.中国中医科学院西苑医院
4.北京中医药大学东直门医院
彭叶辉,pengyehui@hnust.edu.cn
杨伟,yangyxq@ruc.edu.cn
收稿:2021-03-15,
修回:2021-11-15,
纸质出版:2022-04-02
移动端阅览
向兴华,彭叶辉,杨伟等.顽固性高血压发生主要不良心血管事件患者的中医四诊信息可解释性研究——基于随机森林规则提取方法[J].中医杂志,2022,63(07):628-634.
XIANG Xinghua,PENG Yehui,YANG Wei,et al.Interpretability of Chinese Medicine Four Examinations Information of Major Adverse Cardiovascular Events in Resistant Hypertension: based on Random Forest Rule Extraction Method[J].Journal of Traditional Chinese Medicine,2022,63(07):628-634.
向兴华,彭叶辉,杨伟等.顽固性高血压发生主要不良心血管事件患者的中医四诊信息可解释性研究——基于随机森林规则提取方法[J].中医杂志,2022,63(07):628-634. DOI: 10.13288/j.11-2166/r.2022.07.006.
XIANG Xinghua,PENG Yehui,YANG Wei,et al.Interpretability of Chinese Medicine Four Examinations Information of Major Adverse Cardiovascular Events in Resistant Hypertension: based on Random Forest Rule Extraction Method[J].Journal of Traditional Chinese Medicine,2022,63(07):628-634. DOI: 10.13288/j.11-2166/r.2022.07.006.
目的
2
基于随机森林规则提取方法探讨顽固性高血压(RH)发生主要不良心血管事件(MACE)患者的临床信息分类规则及特征。
方法
2
采集住院治疗的2093例RH患者的电子病历信息,其中RH患者807例,RH+MACE患者1286例。收集患者的个人基本信息(包括性别、年龄、既往史、家族史、个人史、传染病史)、中医四诊信息、西医诊断信息数据。统计前期研究发现的与RH患者发生MACE密切相关的140个指标的分布情况,并基于密切相关指标运用随机森林规则提取方法分析RH患者发生MACE的影响特征;对数据进行结构化处理并基于随机森林训练构造多个去相关决策树,生成丰富的决策规则,再采用规则提取方法从规则中提取、度量、修剪、选择规则,利用序贯覆盖法筛选简化的分类规则集,并定义规则的可解释性度量。
结果
2
2093例患者中医四诊信息频次前5位的密切相关指标分别为弦脉1101次(52.6%)、苔腻947次(45.25%)、睡眠欠佳870次(41.57%)、面色少华784次(37.46%);患者基本信息及西医诊断信息频次前5位分别为心率正常1765次(84.33%)、既往高血压病1601次(76.49%)、年龄大于60岁1547次(73.91%)、舒张压正常1428次(68.23%)、高血压3级1151次(54.99%)。RH患者发生MACE影响特征的可解释性分析结果显示,RH+MACE且误差<0.05的规则有3条,规则1:咳嗽c('1') & 老舌c('1') & 头晕c('0') & 胸闷c('0');规则2:恶心c('0') & 颈动脉硬化c('1') & 弦脉c('1');规则3 :憋喘c('1')。3条规则分类准确率均大于95%(误差小于0.05),并具有一定的临床可解释性。分类规则集在准确性与可解释性方面的统计性能分别为灵敏度0.85、特异度0.73、精度0.83、受试者工作特征曲线下面积(AUC值)0.80以及可解释性值0.96。
结论
2
咳嗽、老舌、颈动脉硬化、脉弦、憋喘且无头晕、胸闷、恶心症状为RH患者发生MACE的核心临床信息分类要素。
Objective
2
To explore the classification rules and characteristics of clinical information in resistant hypertension (RH) patients with major adverse cardiovascular events (MACE).
Methods
2
We collected the electronic medical records of 2503 RH patients, among which 1286 cases had MACE. The basic personal information including gender, age, past history, family history, personal history, and history of infectious diseases, traditional Chinese medicine (TCM) four examinations information, and western medicine diagnosis information were collected. The previous study identified 140 indicators closely related to the occurrence of MACE in RH patients. The distribution of the closely related indicators in 2093 patients was analyzed, and the random forest rule extraction method was used to analyze the characteristics that influence the occurrence of MACE in RH patients. Multiple decorrelation decision trees were buit through structured data processing and random forest training, which can generate a wealth of decision rules. The rule extraction method was used to extract, measure, trim, and select rules. The sequential coverage method was used to filter the simplified classification rule set, and the interpretability measurement of the rules was defined.
Results
2
The top frequent five closely related indicators of the TCM four examinations information among 2093 patients were wiry pulse (1101, 52.6%), greasy coating (947, 45.25%), poor sleep (870, 41.57%), and pale complexion (784, 37.46%). The top five frequent information related to patients' basic information and western medicine diagnosis information were normal heart rate (1765, 84.33%), previous hypertension (1601, 76.49%), age older than 60 years old (1547, 73.91%), normal diastolic blood pressure (1428, 68.23%), and grade 3 hypertension (1151, 54.99%). The interpretability analysis of the influencing factors showed that were three rules for predicting RH+MACE with error less than 0.05, which were Rule Ⅰ: cough c('1') & old tongue c('1') & dizziness c('0') & chest tightness c('0'); Rule Ⅱ: Nausea c('0') & carotid arteriosclerosis c('1') & Xuanmai c('1'); and Rule Ⅲ: suffocation c('1'). The classification accuracy of the three rules has reached more than 95% (error less than 0.05), and has a certain clinical interpretability. The model evaluation index results of the rule set: sensitivity (0.85), specificity (0.73), accuracy (0.83), area under the receiver operating characteristic curve (AUC) value (0.80), and interpretability value (0.96).
Conclusion
2
Presentation of cough, old tongue, carotid arteriosclerosis, wiry pulse, wheezing, and with no symptoms of dizziness, chest tightness and nausea are taken as the core classification elements for predicting MACE in RH patients.
WHELTON PK , CAREY RM , ARONOW WS , et al . 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/ NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines [J]. J Americ College Cardiol , 2018 , 71 ( 19 ): e127 - e248 .
KASIAKOGIAS A , TSIOUFIS C , DIMITRIADIS K , et al . Dynamic resistant hypertension patterns as predictors of cardiovascular morbidity: a 4-year prospective study [J]. J Hypertens , 2014 , 32 ( 2 ): 415 - 422 .
刘大胜 , 向兴华 , 邢云飞 , 等 . 基于文本挖掘的顽固性高血压的治疗现状及“证-药-效”的规律研究 [J]. 世界科学技术-中医药现代化 , 2020 , 22 ( 3 ): 836 - 842 .
YANG J , LI Y , LIU Q , et al . Brief introduction of medical database and data mining technology in big data era [J]. J Evid Based Med , 2020 , 13 ( 1 ): 57 - 69 .
ZHANG J , KOWSARI K , HARRISON JH , et al . Patient 2Vec: a personalized interpretable deep representation of the longitudinal electronic health record [J]. IEEE Access , 2018 , 6 : 65333 - 65346 . doi: 10.1109/ACCESS. 2018. 2875677 http://dx.doi.org/10.1109/ACCESS.2018.2875677 .
STEELE AJ , DENAXAS SC , SHAH AD , et al . Machine learning models in electronic health records can outperform conventional survival models for predicting patient mortality in coronary artery disease [J]. Plos One , 2018 , 13 ( 8 ): e0202344 . doi: 10.1371/journal.pone.0202344 http://dx.doi.org/10.1371/journal.pone.0202344 .
刘大胜 . 真实世界下顽固性高血压病不良心血管事件的诊治规律研究 [D]. 北京 : 中国中医科学院 , 2020 .
FOKKEMA M . Fitting prediction rule ensembles with R package pre [J]. J Statist Software , 2020 , 92 ( 12 ): 1 - 30 .
DENG H . Interpreting tree ensembles with in trees [J]. Int J Data Science Analyt , 2018 , 7 ( 4 ): 277 - 287 .
HAILESILASSIE T . Rule extraction algorithm for deep neural networks: a review [J]. Int J Comput Science Inform Securit , 2016 , 14 ( 7 ): 376 - 381 .
FRIEDMAN JH , POPESCU BE . Predictive learning via rule ensembles [J]. Annal Appl Statist , 2008 , 2 ( 3 ): 916 - 954 .
中国高血压防治指南修订委员会 . 中国高血压防治指南2010 [J]. 中华心血管病杂志 , 2011 , 39 ( 7 ): 579 - 616 .
《中国高血压防治指南》修订委员会 . 中国高血压防治指南(2018年修订版) [J]. 中国心血管杂志 , 2019 , 24 ( 1 ): 24 - 56 .
常宝宝 . 自然语言分析与生成术语简介 [J]. 术语标准化与信息技术 , 2010 ( 4 ): 19 - 22 .
NGUYEN TT , NGUYEN MP , PHAM XC , et al . Heterogeneous classifier ensemble with fuzzy rule-based meta learner [J]. Informat Science , 2018 , 422 : 144 - 160 . doi: 10.1016/j.ins.2017.09.009 http://dx.doi.org/10.1016/j.ins.2017.09.009 .
DENG H , RUNGER G , TUV E , et al . CBC: an associative classifier with a small number of rules [J]. Decis Support System , 2014 , 59 : 163 - 170 . doi: 10.1016/j.dss.2013.11.004 http://dx.doi.org/10.1016/j.dss.2013.11.004 .
DENG H , RUNGER G . Gene selection with guided regularized random forest [J]. Pat Recognit , 2013 , 46 ( 12 ): 3483 - 3489 .
FÜRNKRANZ J . Separate-and-conquer rule learning [J]. Artific Intell Review , 1999 , 13 : 3 - 54 . doi: 10. 1023/A:1006524209794 http://dx.doi.org/10.1023/A:1006524209794 .
周志华 . 机器学习 [M]. 北京 : 清华大学出版社 , 2016 : 347 - 352 .
GORZAŁCZANY MB , RUDZIŃSKI F . A multi-objective genetic optimization for fast, fuzzy rule-based credit classification with balanced accuracy and interpretability [J]. Appl Soft Comput , 2016 , 40 : 206 - 220 . doi: 10.1016/j.asoc.2015.11.037 http://dx.doi.org/10.1016/j.asoc.2015.11.037 .
项莎特 , 瞿溢谦 , 叶含笑 . 多层学习联合建模方法设计在气阴两虚型咳嗽证候的辨证诊断中的应用 [J]. 中国卫生统计 , 2020 , 37 ( 6 ): 892 - 894 .
丁亮 , 章新友 , 刘莉萍 , 等 基于深度神经网络的原发性肝癌证型诊断分类预测模型 [J]. 世界科学技术-中医药现代化 , 2020 , 22 ( 12 ): 4185 - 4192 .
崔伟锋 , 刘萧萧 , 韩静旖 , 等 . 基于随机森林的原发性高血压心血管风险预后模型 [J]. 中国老年学杂志 , 2020 , 40 ( 4 ): 814 - 816 .
RUDIN C . Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead [J]. Nat Mach Intell , 2019 , 5 : 206 - 215 . doi: 10.1038/s42256-019-0048-x http://dx.doi.org/10.1038/s42256-019-0048-x .
0
浏览量
5
下载量
2
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010102007558号