Antonio Olivas-Martinez

Biostatistician and physician developing causal inference methods for complex settings in health research.

I develop statistical methods for causal inference in complex settings, with particular interest in semiparametric theory, unmeasured confounding, proxy variables, and inverse problems. My work is motivated by problems in biomedical and public health research, with an emphasis on transparent assumptions, flexible estimation, and tools that other researchers can use.

Methodological work

  • Source-Condition Analysis of Kernel Adversarial Estimators — with Andrea Rotnitzky, arXiv preprint (2025).
    Finite-sample analysis of RKHS-regularized adversarial estimators for ill-posed regression, comparing their stability and assumptions with existing methods.
    Preprint

  • Proximal Causal Inference for Modified Treatment Policies — with Peter B. Gilbert and Andrea Rotnitzky, arXiv preprint (2025).
    Extends proximal causal inference to continuous exposures and realistic treatment modifications in settings with unmeasured confounding, using negative controls and debiased machine learning. Motivated by vaccine immunobridging studies.
    Preprint · R implementation

  • A General Framework for Designing and Evaluating Active-Controlled Trials with Non-Inferiority Objectives — with Fei Gao and Holly Janes, Statistics in Medicine (2026).
    Unifies analytical methods and success criteria for active-controlled trials, enabling systematic comparison of type I error, power, robustness, and design trade-offs.
    Article · R implementation

Research software

  • proxi-mtp — R code and a simulated dataset demonstrating a doubly robust, cross-fitted estimator for proximal modified treatment policy effects.
  • ni-design — R functions for non-inferiority margins, event and sample-size calculations, unconditional power, and sensitivity analysis in active-controlled trial designs.

Selected collaborative projects

  • Immune Correlates and Vaccine Immunobridging: Statistical Innovations, Challenges, and Opportunities — Co-author, The Journal of Infectious Diseases (2025). Overview of recent statistical developments relevant to immunobridging, including variable importance prediction for correlates of risk, controlled risk causal analysis for correlates of protection, and vaccine efficacy transportability methods. DOI
  • Test Sensitivity in a Prospective Cancer Screening Program: A Critique of a Common Proxy Measure — Co-author, Statistical Methods in Medical Research (2023).
    Analyzed the relationship between empirical and true sensitivity in cancer screening programs, clarifying conditions under which proxy measures can misrepresent diagnostic performance.
    DOI
  • Sensitivity Measures in Studies of Cancer Early Detection Biomarkers — Co-author, Cancer Epidemiology, Biomarkers & Prevention (2025).
    Investigated how sensitivity estimates from different phases of biomarker development relate to true preclinical sensitivity, highlighting biases that can arise across study designs and screening intervals.
    DOI

For a broader publication list, see Google Scholar, ORCID, or my applied work.