因果中介分析常用於社會科學、醫學及流行病學等領域，用於探討暴露因子是如何透過感興趣的中介因子去影響到結果變量。然而在進行多重中介因子分析時，中介因子間的模型和結構通常十分複雜，傳統方法需要每個中介因子的模型皆正確假設，否則容易得到錯誤的結論。VanderWeele and Vansteelandt (2014) 針對多重中介因果分析提出了有序分解法並發展出逆機率權重（Inverse probability Weighting approach for sequential effect decomposition, SEQ-w）估計量。SEQ-w 估計量有效地放寬中介因子模型假設，在多個複雜的中介因子模型時具有穩健性。然而，SEQ-w 估計量在因果中介路徑的解釋尚未明瞭，同時，SEQ-w 估計量只局限於連續型反應變數。因此，本研究嘗試完備 SEQ-w 估計量相關性質。在因果解釋層次，本研究透過反事實模型確立了 SEQ-w 估計量在路徑特定效應上的分類，並證明出與 Tai et al. (2022a) 提出的完全交互作用分解法具有相同的機制解釋。在資料應用廣度層次，本文也將 SEQ-w 估計量拓展至多階段模型 (Multi-stage model) 以及存活分析。最後將以上的方法程式化進行模擬分析，並應用於癌症基因組圖譜（TCGA）資料，以探討肺腺癌及其潛在致癌基因組間的因果機制。

Causal mediation analysis is widely used in fields such as social sciences, medicine, and epidemiology to investigate how exposure influence outcome through mediators of interest. However, in causal multi-mediation analysis, the models and structures of these mediators are often highly complex. Traditional methods require correct specification of each mediator model, otherwise, erroneous conclusions may result. VanderWeele and Vansteelandt (2014) proposed a inverse probability weighting approach for sequential effect decomposition (SEQw) for multiple mediator causal analysis. SEQ-w effectively relaxes the assumptions of mediator models, demonstrating robustness in scenarios with complex mediator models. Nonetheless, the causal interpretation of SEQ-w pathways remains unclear, and SEQ-w is limited to continuous response variables. Therefore, this study aims to further develop the properties of SEQ-w. At the level of causal interpretation, we establish the classification of path-specific effects through counterfactual models, proving that SEQ-w shares the same mechanistic explanation as the fully mediated interaction effect decomposition proposed by Tai et al. (2022a). At the level of data application, this paper extends the SEQ to survival analysis and multi-stage models. Finally, we implement and simulate these methods, applying them to The Cancer Genome Atlas (TCGA) data to explore the causal mechanisms between lung adenocarcinoma and its potential oncogenic genomes.

Jeffrey M Albert, Jang Ik Cho, Yiying Liu, and Suchitra Nelson. Generalized causal mediation and path analysis: Extensions and practical considerations. Statistical Methods in Medical Research, 28(6):1793–1807, 2019.
Reuben M Baron and David A Kenny. The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51(6):1173, 1986.
Ralf Bender, Thomas Augustin, and Maria Blettner. Generating survival times to simulate cox proportional hazards models. Statistics in medicine, 24(11):1713–1723, 2005.
Jiajia Chen, Rong Wu, Ying Xuan, Min Jiang, and Yuecan Zeng. Bioinformatics analysis and experimental validation of ttk as a biomarker for prognosis in non-small cell lung cancer. Bioscience reports, 40(10):BSR20202711, 2020.
Frank C Detterbeck, Daniel J Boffa, Anthony W Kim, and Lynn T Tanoue. The eighth edition lung cancer stage classification. Chest, 151(1):193–203, 2017.
Yen-Tsung Huang and Hwai-I Yang. Causal mediation analysis of survival outcome with multiple mediators. Epidemiology, 28(3):370–378, 2017.
Minoru Kanehisa and Susumu Goto. Kegg: kyoto encyclopedia of genes and genomes. Nucleic acids research, 28(1):27–30, 2000.
Maria Teresa Landi, Tatiana Dracheva, Melissa Rotunno, Jonine D Figueroa, Huaitian Liu, Abhijit Dasgupta, Felecia E Mann, Junya Fukuoka, Megan Hames, Andrew W Bergen, 63 et al. Gene expression signature of cigarette smoking and its role in lung adenocarcinoma development and survival. PloS one, 3(2):e1651, 2008.
Bo Li, Xiaojuan Gu, Hanbing Zhang, and Hao Xiong. Comprehensive analysis of the prognostic value and immune implications of the ttk gene in lung adenocarcinoma: a meta-analysis and bioinformatics analysis. Animal Cells and Systems, 26(3):108–118, 2022.
Judea Pearl. Direct and indirect effects. In Oren Breese and Daphne Koller, editors, Proceedings of the Seventeenth Conference on Uncertainty and Artificial Intelligence, pages 411–420, San Francisco, CA, 2001. Morgan Kaufmann.
James Robins. A new approach to causal inference in mortality studies with a sustained exposure period—application to control of the healthy worker survivor effect. Mathematical modelling, 7(9-12):1393–1512, 1986.
James M Robins and Sander Greenland. Identifiability and exchangeability for direct and indirect effects. Epidemiology, 3(2):143–155, 1992.
Donald B Rubin. Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of educational Psychology, 66(5):688, 1974.
Jonathan M Snowden, Sherri Rose, and Kathleen M Mortimer. Implementation of gcomputation on a simulated data set: demonstration of a causal inference technique. American journal of epidemiology, 173(7):731–738, 2011.
An-Shun Tai and Sheng-Hsuan Lin. Integrated multiple mediation analysis: A robustness-specificity trade-off in causal structure. Statistics in Medicine, 40(21):4541–4567, 2021.
An-Shun Tai and Sheng-Hsuan Lin. Multiply robust estimation of natural indirect effects with multiple ordered mediators. Statistics in Medicine, 43(4):656–673, 2024.
An-Shun Tai, Le-Hsuan Liao, and Sheng-Hsuan Lin. On the conventional definition of path-specific effects: Fully mediated interaction with multiple ordered mediators. Epidemiology, 33(6):817–827, 2022a.
An-Shun Tai, Pei-Hsuan Lin, Yen-Tsung Huang, and Sheng-Hsuan Lin. Path-specific effects in the presence of a survival outcome and causally ordered multiple mediators with application to genomic data. Statistical Methods in Medical Research, 31(10):1916–1933, 2022b.
Dustin Tingley, Teppei Yamamoto, Kentaro Hirose, Luke Keele, and Kosuke Imai. Mediation: R package for causal mediation analysis. 2014.
Tyler VanderWeele and Stijn Vansteelandt. Mediation analysis with multiple mediators. Epidemiologic methods, 2(1):95–115, 2014.
Tyler J VanderWeele and Stijn Vansteelandt. Conceptual issues concerning mediation, interventions and composition. Statistics and its Interface, 2(4):457–468, 2009.
Tyler J VanderWeele, Stijn Vansteelandt, and James M Robins. Effect decomposition in the presence of an exposure-induced mediator-outcome confounder. Epidemiology, 25(2): 300–306, 2014.
Qingzhao Yu and Bin Li. mma: an r package for mediation analysis with multiple mediators. Journal of Open Research Software, 5(1):11–11, 2017.