在資料驅動的決策領域中，精確評估個體處方效果（ITE）是資源最佳化的核心。然而，實務上常面臨「小樣本」與「處置分配不平衡」雙重挑戰，傳統因果機器學習模型的點估計往往過度自信，難以提供可靠的不確定性量化與風險保證。 為此，本研究提出結合多任務高斯過程（MTGP）與共形推論（Conformal Inference）的因果推論框架。
首先以 MTGP 為基礎，透過線性共區域化模型共享兩組特徵資訊，克服樣本失衡造成的估計偏差。其次，將共形推論整合至 ITE 預測，並評估非精確與精確巢狀方法（Inexact / Exact Nested Method）。此機制利用傾向分數加權校準誤差，不僅能處理共變數偏移 （Covariate Shift），更提供無需假設資料分布的覆蓋率保證。
模擬結果顯示，本框架在條件風險值（CVaR）資源配置的表現上，顯著優於 Xlearner 與因果森林等基準模型。ACTG175 臨床資料實證亦指出，相較於易產生無效區間的傳統模型，MTGP-inexact 方法即使在嚴苛的資料分布偏移下，仍能精準維持目標覆蓋率並有效收斂預測區間寬度。總結而言，本研究成功平衡了統計安全性與決策實用性，為高風險與小樣本環境下的 ITE 評估提供了強健的理論與實證基礎。

In data-driven decision-making, accurately evaluating the Individual Treatment Effect (ITE) is core to optimal resource allocation. However, practical challenges such as small sample sizes and imbalanced treatment allocations often cause traditional causal machine learning models to yield overconfident point estimates, failing to provide reliable uncertainty quantification and risk guarantees. To address this, this study proposes a causal inference framework combining Multi-Task Gaussian Processes (MTGP) and Conformal Inference. First, MTGP leverages a Linear Model of Coregionalization to share feature information across groups, mitigating estimation biases caused by data imbalance. Second, Conformal Inference is integrated into ITE predictions to evaluate the Inexact and Exact Nested Methods. By employing propensity score weighting to calibrate errors, this mechanism effectively handles covariate shifts and provides distribution-free coverage guarantees. Simulation results demonstrate that this framework significantly outperforms baselines like the X-learner and causal forest in Conditional Value at Risk (CVaR) resource allocation tasks. Furthermore, empirical analysis using the ACTG175 clinical dataset confirms that, unlike uncalibrated traditional models, the MTGP-inexact method precisely maintains target coverage and effectively narrows prediction intervals even under severe covariate shifts. In conclusion, this study successfully strikes a balance between statistical safety and practical utility, offering a robust foundation for ITE estimation in high-risk, small-sample environments.

Alaa, A. M., & van der Schaar, M. (2017). Bayesian inference of individualized treatment effects using multi-task gaussian processes. In Advances in neural information processing systems (nips) (Vol. 30, pp. 3427–3435). Curran Associates, Inc.
Álvarez, M. A., Rosasco, L., & Lawrence, N. D. (2012). Kernels for vector-valued functions: A review. Foundations and Trends® in Machine Learning, 4(3), 195– 266. doi: 10.1561/2200000036
Angelopoulos, A. N., & Bates, S. (2021). A gentle introduction to conformal prediction and distribution-free uncertainty quantification. arXiv preprint arXiv:2107.07511. doi: 10.48550/arXiv.2107.07511
Brand, J. E., Zhou, X., & Xie, Y. (2023). Recent developments in causal inference and machine learning. Annual Review of Sociology, 49, 81–110. doi: 10.1146/ annurev-soc-030420-015345
Champiat, S., Dercle, L., Ammari, S., Massard, C., Hollebecque, A., et al. (2017). Hyperprogressive disease is a new pattern of progression in cancer patients treated by anti-pd-1/pd-l1. Clinical Cancer Research, 23(8), 1920–1928. doi: 10.1158/ 1078-0432.CCR-16-1741
Chou, Y. C., Chuang, H. H., & Huang, K. (2026). Individualized treatment allocation under outcome risk and limited data (Working Paper). College of Commerce, National Chengchi University.
Desai, R. J., Glynn, R. J., Solomon, S. D., Claggett, B., Wang, S. V., & Vaduganathan, M. (2024). Individualized treatment effect prediction with machine learning—salient considerations. NEJM Evidence, 3(4), EVIDoa2300041. doi: 10.1056/ EVIDoa2300041 Guo, R., Cheng, L., Li, J., Hahn, P. R., & Liu, H. (2020). A survey of learning causality with data: Problems and methods. ACM Computing Surveys (CSUR), 53(4), 1–37. doi: 10.1145/3397269
Gutierrez, P., & Gérardy, J.-Y. (2017). Causal inference and uplift modelling: A review of the literature. In Proceedings of the 3rd international conference on predictive applications and apis (pp. 1–13).
Hammer, S. M., Katzenstein, D. A., Hughes, M. D., Gundacker, H., Schooley, R. T., Haubrich, R. H., . . . Merigan, T. C. (1996). A trial comparing nucleoside monotherapy with combination therapy in hiv-infected adults with cd4 cell counts from 200 to 500 per cubic millimeter. New England Journal of Medicine, 335(15), 1081– 1090. doi: 10.1056/NEJM199610103351501
Heckman, J. J., Smith, J., & Clements, N. (1997). Making the most out of programme evaluations and social experiments: Accounting for heterogeneity in programme impacts. The Review of Economic Studies, 64(4), 487–535. doi: 10.2307/2971729 Hitsch, G. J., Misra, S., & Zhang, W. W. (2024). Heterogeneous treatment effects and optimal targeting policy evaluation. Quantitative Marketing and Economics, 22, 115–168. doi: 10.1007/s11129-023-09278-5
Holland, P. W. (1986). Statistics and causal inference. Journal of the American Statistical Association, 81(396), 945–960. doi: 10.2307/2289064
Künzel, S. R., Sekhon, J. S., Bickel, P. J., & Yu, B. (2019). Meta-learners for estimating heterogeneous treatment effects using machine learning. Proceedings of the National Academy of Sciences, 116(10), 4156–4165. doi: 10.1073/pnas.1804597116 Lei, L., & Candès, E. J. (2021). Conformal inference of counterfactuals and individual treatment effects. Journal of the Royal Statistical Society Series B: Statistical Methodology, 83(5), 911–938. doi: 10.1111/rssb.12445
McFowland III, E., Somanchi, S., & Neill, D. B. (2023). Efficient discovery of heterogeneous quantile treatment effects in randomized experiments via anomalous pattern detection.
Pearl, J. (1995). Causal diagrams for empirical research. Biometrika, 82(4), 669–688. doi: 10.1093/biomet/82.4.669
Pion, A., & Vazquez, E. (2024). Gaussian process interpolation with conformal prediction: methods and comparative analysis. arXiv preprint arXiv:2407.08271. doi: 10.48550/arXiv.2407.08271
Qi, Z., Pang, J.-S., & Liu, Y. (2023). On robustness of individualized decision rules. Journal of the American Statistical Association, 118(543), 2143–2157. doi: 10 .1080/01621459.2022.2038180
Rasmussen, C. E., & Williams, C. K. (2005). Gaussian processes for machine learning. MIT Press. doi: 10.7551/mitpress/3206.001.0001
Rockafellar, R. T., & Uryasev, S. (2000). Optimization of conditional value-at-risk. Journal of risk, 2, 21–41. doi: 10.21314/JOR.2000.038
Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55. doi: 10.1093/ biomet/70.1.41
Rubin, D. B. (1974). Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of Educational Psychology, 66(5), 688–701. doi: 10.1037/h0037350
Schulz, E., Speekenbrink, M., & Krause, A. (2018). A tutorial on gaussian process regression with a focus on exploration-exploitation scenarios. Journal of Mathematical Psychology, 85, 1–16. doi: 10.1016/j.jmp.2018.03.001
Shalit, U., Johansson, F. D., & Sontag, D. (2017). Estimating individual treatment effect: generalization bounds and algorithms. In International conference on machine learning (icml) (pp. 3076–3085).
Shimodaira, H. (2000). Improving predictive inference under covariate shift by weighting the log-likelihood. Journal of statistical planning and inference, 90(2), 227–244. doi: 10.1016/S0378-3758(00)00115-4
Tibshirani, R. J., Barber, R. F., Candès, E., & Ramdas, A. (2019). Conformal prediction under covariate shift. In Advances in neural information processing systems (neurips) (Vol. 32).
Wager, S., & Athey, S. (2018). Estimation and inference of heterogeneous treatment effects using random forests. Journal of the American Statistical Association, 113(523), 1228–1242. doi: 10.1080/01621459.2017.1319839
Wang, L., Zhou, Y., Song, R., & Sherwood, B. (2018). Quantile-optimal treatment regimes. Journal of the American Statistical Association, 113(523), 1243–1254. doi: 10.1080/01621459.2017.1330204
Yao, L., Chu, Z., Li, S., Li, Y., Gao, J., & Zhang, A. (2021). A survey on causal inference. ACM Transactions on Knowledge Discovery from Data (TKDD), 15(5), 1–46. doi: 10.1145/3444944
Yoon, J., Jordon, J., & van der Schaar, M. (2018). Ganite: Estimation of individualized treatment effects using generative adversarial nets. In International conference on learning representations (iclr).