基于真实世界和大数据技术的顽固性高血压痰瘀互结证精准方药方案的方法研究

It is an inevitable trend of Chinese medicine development into precision medicine to change the unfavorable status of its discipline. The project group previously found that the real-world study and big data technology can resolve the problem of lacking evidence, refinement and quantification in individualization of pattern differentiation and treatment. If the hypothesis be proved, the key problem in Chinese medicine precision methods can be resolved. Accordingly, the project take resistant hypertension (RH) as an example to study the methods of forming precise formula & herbs and therapeutic schedule for a RH patient with Tan Yu Hu Jie pattern on the basis of Tan Yu Bian Zhi Fang had good effects. Firstly, to construct the objective function of occurence rate of a RH patient with Tan Yu Hu Jie pattern who will develop into cardiovascular disease and construct several regression models of the objective function by generalized boosted model (GBM). Secondly, to linear combine the regression models by boosting algorithm of GBM, and determine the risk factors and occurence rate of the RH patient with Tan Yu Hu Jie pattern who will develop into cardiovascular disease. Therefore, the formula & herbs and the therapeutic schedule of the RH patient with Tan Yu Hu Jie pattern can be optimized and his/her precise formula & herbs and therapeutic schedule can be formed. Finally, the precision methods are tested in the real-world study with large sample to define its key technology and applicable scope, and provide the new discovery methodology for industry-specific technology of forming precise formula & herbs and precise therapeutic schedule in pattern differentiation and treatment.

中医走精准发展路是改变自身弱势学科地位的必然趋势，项目组初步发现，真实世界与大数据技术能解决辨证论治存在的客观证据与精细化、量化不足难题，假说此发现被证实，就解决了中医精准化方法的关键问题。据此，本项目以顽固性高血压(Resistant Hypertension, RH)为范例，在既往肯定痰瘀辨治RH有效的基础上，开展RH痰瘀互结证精准方药方案的方法研究。首先，采用大数据技术广义增强模型(Generalized Boosted Model, GBM)，建立RH痰瘀互结证个体发生心血管病概率的目标函数，构建目标函数的多个单棵回归树模型。其次，采用GBM中的Boosting算法，确定RH痰瘀互结证个体发生心血管的危险因素和概率，优化并形成精准治疗方药和治疗方案。最后，对精准化方法进行大数据真实世界临床验证，初步确定其关键技术和适用范围，为解决辨证论治中方药方案精准的共性技术提供方法学上新发现。

高概念、大数据时代，创新意识与原创思维得到高度重视，海量医疗数据的产生对医学研究提供了巨大便利。中医药学以天人相应、形神一体的思维模式与真实世界注重临床实际情况的理念不谋而合。将真实世界研究方法有效地融入到中医药的诊治规律及精准方药的探索中，使中医原创思维与形象思维、具象思维、逻辑思维相结合，是中医药学发展的重要思路。顽固性高血压（Resistant hypertension, RH）作为临床中常见、高发、难治性疾病，病情复杂、病程长，易引发心、脑、肾等多脏器的并发症，尤其容易引发主要不良心血管事件（Major adverse cardiovascular events，MACE）。目前中西医均未给出有效的治疗方案，使其严重威胁人类健康。本项目在既往肯定痰瘀辨治RH有效的基础上，在真实世界下研究RH痰瘀互结证的诊治规律及其与MACE的影响因素，克服辨证论治存在的客观证据欠缺与治疗方药缺乏精细的不足，逐步理清“病-症-证-药（方）”之间的关系，开展RH痰瘀互结证精准方药方案的方法研究。本项目以真实世界的HIS系统中的RH痰瘀互结证患者作为研究对象，采用大数据技术广义增强模型(Generalized Boosted Model,GBM)和XGBoost算法，建立RH痰瘀互结证个体发生心血管病概率的目标函数，构建目标函数的多个单棵回归树模型。采用GBM和XGBoost中的Boosting算法，确定RH痰瘀互结证个体发生心血管的危险因素和概率。通过描述性与推断性统计及关联规则的方法，寻求“症-证”、“症-药”的内在关联，探索了RH“病-证”、“证-药”的规律，初步形成顽固性高血压痰瘀互结证的精准方药、精准方案，从而证实中医辨证论治中精准方药方案的形成方法具有可行性。