2015年美國總統歐巴馬在國情咨文中提出“精準醫療計畫”，希望對於不同體質人群實施不同療法，達到精準醫療的目標。中醫很早就研究了“體質”這一概念，已經形成了一套比較完善的體系。有不少相關研究成果得到了數據的支持，但是這些結果只能說明中醫體質與某些疾病之間存在關聯性，並不知道他們之間是否存在因果關係。在本研究中，，我們使用孟德爾隨機化（Mendelian randomization）來驗證中醫體質與疾病是否有因果關係。孟德爾隨機化利用了基因是隨機從親代遺傳給自帶的事實，因此用基因當做衡量體質的指標，不受其他外部因素的影響。

此研究流程如下，第一步，對台灣人體資料庫（Taiwan Biobank）裡中醫部分問卷用傳統中醫分類與因子分析（Factor analysis）兩種方法來確定體質指標。第二步，用全基因體關聯性分析（Genome Wide Association Study）來找出決定體質的基因位點（SNP）。第三步，用孟德爾隨機化中的兩階段最小平方法（2 stage least square）來研究體質與疾病之間是否存在因果關係。第四步，用MR-Egger法檢驗選取的基因位點是否違反了孟德爾隨機化的假設前提。

本研究結果發現42組傳統中醫體質與疾病之間存在因果關係，其中氣喘與痰濕體質的因果關係、抑鬱症與血虛體質的因果關係可以在中醫理論中得到印證。這說明我們的研究具有一定程度的可靠性。但是因子分析的結果顯示，人們並不能通過這些題目區分出中醫體質，一些題目之間的區別難以分辨。因此如何用問卷分辨人們的中醫體質仍然是值得研究的方向。

In 2015, US President Obama proposed the "Precision Medicine Initiative" in his State of the Union address, hoping to implement different therapies for people of different physiques to achieve the goal of precision medicine. Traditional Chinese medicine has long studied the concept of "Traditional Chinese Medicine constitution" and has formed a relatively complete system. Many relevant research results are supported by data, but these results can only show that there is a correlation between the constitution of TCM and certain diseases, and it is not known whether there is a causal relationship between them. In this study, we used Mendelian randomization to verify whether there is a causal relationship between TCM constitution and diseases. Mendelian randomization takes advantage of the fact that genes are randomly inherited from their parents. Therefore, genes are used as indicators to measure TCM constitution and are not affected by other external factors.

The research process is as follows. In the first step, the traditional Chinese medicine classification and factor analysis are used to determine the TCM constitution index on the part of the questionnaire in the Taiwan Biobank. The second step is to use Genome Wide Association Study to find the SNPs that determines the TCM constitution. The third step is to use the 2 stage least square method in Mendelian randomization to study whether there is a causal relationship between TCM constitution and disease. The fourth step is to use the MR-Egger method to test whether the selected SNPs violate the assumptions of Mendelian randomization.

The results of this study found that there are 42 groups of causal relationship between traditional Chinese medicine constitutions and diseases. Among them, the causal relationship between asthma and “痰濕”constitution, and the causal relationship between depression and “血虛” constitution can be confirmed in the theory of Chinese medicine. This shows that our research has a certain degree of reliability. However, the results of factor analysis show that people cannot distinguish the constitution of Chinese medicine through these topics, and the differences between some topics are difficult to distinguish. Therefore, how to use the questionnaire to distinguish people's TCM physique is still a direction worthy of research.

Introduction 1
Method 2
Data 2
Exposure 2
Total score 2
Factor score 5
Outcome 6
Observational association 7
Genetic variant 7
Mendelian randomization 8
Two stage least square（2SLS） 8
Known confounder 9
Power 10
MR-Egger method 10

Results 11
Data 11
Exposure: Factor score 11
Exposure: Total score 15
Outcome 18
Genetic variants 19
工具變數(instrumental variable) 19
Association and Causal Inference 20
因子得分與家族疾病史的分析 20
中医体质总得分与家族疾病病史的分析 24
Power 25

Discussion 25
Reference 26

1. 匡调元, 两纲八要辨体质新论 中医药学刊, 2003(01): p. 108-110.
2. 王琦, 9种基本中医体质类型的分类及其诊断表述依据 北京中医药大学学报, 2005(04): p. 1-8.
3. 蘇奕彰, 中醫體質量表及分型指標之建立（2-2）. 中醫藥年報, 2008. 第26 期(5): p. 65-152.
4. Angrist, J.D., G.W. Imbens, and D.B. Rubin, Identification of Causal Effects Using Instrumental Variables. Journal of the American Statistical Association, 1996. 91(434): p. 444-455.
5. Bowden, J., G. Davey Smith, and S. Burgess, Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International Journal of Epidemiology, 2015. 44(2): p. 512-525.
6. Stock, J.H. and M. Yogo, Testing for Weak Instruments in Linear IV Regression. Nber Technical Working Papers, 2002. 14(1): p. 80-108.
7. Douglas Staiger, J.H.S., Instrumental Variables Regression with Weak Instruments. Econometrica, 1994. 65(3): p. 557-586.
8. 向珍蛹, et al., 膝骨关节炎中医证型分布的流行病学研究 上海中医药杂志, 2012. 46(12): p. 5-8.
9. 祝谌予, et al., 糖尿病证治 中医杂志, 1986(06): p. 10-12.
10. 李亚, et al., 支气管哮喘缓解期中医证素分布规律的文献分析. 时珍国医国药, 2010. 21(09): p. 2341-2342.
11. 郭蓉娟, et al., 抑郁症中医证候要素相关分析. 中医杂志, 2008(09): p. 828-829+847