나의 연구 마음 가짐 (스스로의 독백)
연구를 하는 것은 물론 그 자체로서 의미 가 있으나, 연구가 실생활에 어떻게 쓰이는 지를 모르는 채로 연구 한다면, 불가피하게 수학적 난이도 만 높이진다.
특히, 통계 연구는 기반 이론은 수학이므로, 얼마든지, 불가피하게 복잡하게 논문을 만들수 있다.
때로는 그런 수학적 난해함에 기반한 통계학이, 그것을 잘 이해하지 못 하는 사람들에게는 매우 우아하게 보일수 있다.
내가 추구하는 연구는 실생활 응용에 중심을 두지만, 기반 이론은 수학으로 하는 연구 이다.
하지만, 이런 것을 추구하는 나라고 할 지라도, 탑저널 혹은 탑학회지 출판을 목표로 하기 위해서는 수학적인 내용을 가미 시켜야 한다.
그렇다 할지라도, 그럼에도 불구하고, 내가 추구하는 연구는, 실제로 쓰일수 있는 통계 이다.
아무도 이 길을 가지 않는다. 스스로 강해져야만 갈수 있는 길이다.
나는 나 자신을 음지 에서 스스로 연구 하는 사람이라고 정의 하고 싶다. 그렇기 때문에 나에게는 수학은 곧 생명이다.
나는 내 연구의 모든 것을 알파에서 오메가 까지 스스로 해야 한다.
그리고 그 길에서 아무도 (효과적으로) 도와주지 않는다.
그렇기 때문에 내가 가는 길이 값진 길이다.
나는 혼자서 사냥 한다.
내가 메인으로 하는 연구는 반드시 탑저널 (Top 4 (JASA/JRSSB/AOS/Bka)) 혹은 탑학회 (Top 3 (ICML/NIPS/ICLR)) 만 이제 노린다.
최소 바이오통계 Top 3 (BIOM/BIOS/SIM) 정도는 되어야 한다.
지금의 내 논문 프로파일에서, 위의 논문이 아니라면 사실상 내 프로파일에 마이너스 이다. 그것을 알고, 또한 내가 가는 길에 아무도 도와주지 않음을 알기 때문에, 스스로 동기를 부여하고, 스스로 연구 한다.
이것이 이세윤 이다.
My research develops statistical methodology that connects theory, computation, and clinical practice, with the broader aim of improving decision-making in modern biomedical research. My recent work has appeared or been accepted in leading venues in statistics, biostatistics, machine learning, and artificial intelligence, including ICML 2026, Statistics in Medicine, Statistical Methods in Medical Research, and Electronic Journal of Statistics.
At a high level, my research focuses on the use of causal inference in clinical trials, real-world studies, and biomedical research. The foundational knowledge underlying my research is grounded in semiparametric efficiency theory and/or Bayesian analysis. I am familiar with a wide range of clinical endpoints, including continuous endpoints, binary endpoints, win-ratio endpoints, and survival endpoints. I actively aim to integrate recent developments in ML/AI into causal inference research, while also deeply studying the relevant convergence theory and practical utility of these methods. The latter, practical utility, typically requires regulatory knowledge, particularly familiarity with guidance from the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA).
During my Ph.D. training in Statistics at Texas A&M University, College Station, I studied statistical methods for complex innovative trial designs, with emphasis on adaptive designs, Bayesian clinical trial methodology, and regulatory considerations in confirmatory trials. I also gained early industry experience through Ph.D. summer internships at Novartis and Merck, where I was exposed to clinical trial design and statistical practice from a sponsor’s perspective.
After graduation, I continued working in the pharmaceutical industry through full-time roles at leading companies, including AbbVie, Johnson & Johnson, and Amgen. These experiences helped me understand practical challenges in clinical development, regulatory decision-making, and real-world biomedical evidence generation.
My academic training and industry experience allow me to approach scientific problems from two complementary perspectives: rigorous statistical theory and practical clinical utility. This perspective has shaped my research agenda, which focuses on developing statistically principled and practically useful methods for modern biomedical research.
Se Yoon Lee (2023) “A flexible dose-response modeling framework based on continuous toxicity outcomes in phase I cancer clinical trials,” BMC Trials Journal
Se Yoon Lee, Alain Munafo, Pascal Girard, and Kosalaram Goteti. (2022) “Optimization of dose selection using multiple surrogates of toxicity as a continuous variable in Phase I cancer trial,” Contemporary Clinical Trials Journal Github Slides
Se Yoon Lee and Bani K. Mallick. (2021) “Bayesian Hierarchical modeling: application towards production results in the Eagle Ford Shale of South Texas,” Sankhyā: The Indian Journal of Statistics, Series B Journal SharedIt Github
Kahkashan Afrin*, Ashif Iquebal*, Mostafa Karimi*, Allyson Larsen*, Se Yoon Lee*, and Bani K Mallick*. (2020) “Directionally Dependent MultiView Clustering Using Copula Model,” PLOS ONE (*: Equal contribution, authors are alphabetically ordered in the last name) Journal
Se Yoon Lee, Bowen Lei, and Bani K. Mallick. (2020) “Estimation of COVID19 spread curves integrating global data and borrowing information,” PLOS ONE Journal Github
Se Yoon Lee. (2026) “Asymptotic validity of Schoenfeld’s sample size formula for the Cox proportional hazards model via the Wald test approach”, Statistical Methods in Medical Research Journal
Se Yoon Lee. (2025) “A note on the sample size formula for a win ratio endpoint”, Statistics in Medicine Journal Github
My research in causal inference is grounded in semiparametric efficiency theory. I am particularly interested in this framework because it can be flexibly combined with modern machine-learning methods while still providing deep theoretical insight into estimation, efficiency, and valid inference. Another foundation of my causal inference research is Bayesian methodology, especially Bayesian approaches for borrowing information and incorporating prior or external evidence in a principled manner. Together, these two perspectives allow me to develop causal inference methods that are theoretically rigorous, computationally flexible, and practically useful for modern biomedical research.
Se Yoon Lee and Jae Kwang Kim (2026) “MEC: Machine-Learning-Assisted Generalized Entropy Calibration for Semi-Supervised Mean Estimation”, Forty-third International Conference on Machine Learning (Just Accepted) ICML 2026 Link Arkiv (6,352 accepted among 23,918 reviewed submissions; acceptance rate: 26.6%)
Se Yoon Lee, Peng Zhao, Debdeep Pati, Bani K.Mallick. (2024) “Tail-adaptive Bayesian Shrinkage”, Electronic Journal of Statistics Journal Github Arkiv
I view hybrid randomized controlled trials and synthetic control arm trials as important interdisciplinary research settings in which causal inference, machine learning, and artificial intelligence can be integrated to improve the efficiency of clinical evidence generation. When appropriately designed and implemented, these approaches have the potential to reduce the required sample size relative to traditional randomized controlled trials, lower development costs, and accelerate patient access to new therapies. At the same time, their use requires careful attention to causal assumptions, bias control, uncertainty quantification, and evolving regulatory guidance.
Motivated by these challenges, I seek to develop statistical methods that are both theoretically rigorous and practically relevant. My industry experience has helped me better understand the practical constraints and scientific questions that arise in real-world drug development, and it continues to inform the research problems I pursue. Ultimately, my research goal is to develop semiparametrically efficient estimators for clinically meaningful causal parameters under practically reasonable assumptions, while leveraging modern machine-learning and artificial-intelligence tools to improve efficiency without compromising the validity of statistical inference.
Se Yoon Lee and Jae Kwang Kim (2026) “MEC: Machine-Learning-Assisted Generalized Entropy Calibration for Semi-Supervised Mean Estimation”, Forty-third International Conference on Machine Learning (Just Accepted) ICML 2026 Link Arkiv (6,352 accepted among 23,918 reviewed submissions; acceptance rate: 26.6%)
Se Yoon Lee. (2025) “Power Priors and Type I Error Control: Constrained Borrowing of External Control Data”, Journal of Biopharmaceutical Statistics Journal
Se Yoon Lee (2024) “Eliciting the discount parameter in a power prior method on the basis of the type I error consideration,” Statistics in Biopharmaceutical Research Journal
Se Yoon Lee (2024) “Using Bayesian Statistics in Confirmatory Clinical Trials in the Regulatory Setting: A Tutorial Review,” BMC Medical Research Methodology Journal