Research Experience
Robust Sensitivity Analysis for Quantiles of Individual Treatment Effects in Paired Observational Studies
June 2021 - Present, advised by Prof. Xinran Li, University of Chicago
Extended conventional sensitivity analysis to a robust sensitivity analysis framework for quantiles of individual effects, incorporating quantiles of hidden biases in matched studies.
Implemented the algorithm in R and developed R packages to support this methodology.
Sensitivity Analysis for Quantiles of Hidden Biases in Matched Observational Studies
June 2020 - June 2021, advised by Prof. Xinran Li, University of Chicago
Extended conventional sensitivity analysis to address general quantiles of hidden biases across all matched sets, offering more robust insights compared to focusing solely on the maximum bias.
Demonstrate through simulations and implications that the proposed sensitivity analysis for all quantiles of hidden biases is simultaneously valid and and enhances the conventional sensitivity analysis.
Developed and released an R package for the proposed methodology.
Recommendation System on Traditional Chinese Medicine Big Data
June 2017 - June 2018, advised by Prof. Xiaohua Zhou, Peking University
Implemented recommendation systems in Python using neighborhood-based and model-based collaborative filtering, along with deep learning techniques such as item embedding, feedforward networks, and autoencoders.
Built a hybrid recommendation system combining DeepFM and neighborhood-based models, tested and validated on real-world data.
Publications
Sensitivity Analysis for Quantiles of Hidden Biases in Matched Observational Studies.
Dongxiao Wu, Xinran Li. Journal of the American Statistical Association. [arXiv] [Journal] [R package]
Robust Sensitivity Analysis for Quantiles of Individual Treatment Effects in Paired Observational Studies.
Dongxiao Wu, Xinran Li. To be submitted.