Comparing causal parameters with many treatments and positivity violations.

A McClean¹, Y Li¹, S Bae, M McAdams DeMarco², I Díaz³, W Wu⁴

¹Artificial Intelligence Department, Ataraxis AI, 1239 Broadway, Suite 1502, New York, New York 10001, U.S.A.

²Department of Surgery, New York University, 1 Park Avenue, New York, New York 10016, U.S.A. yiting.li@nyulangone.org sunjae.bae@nyulangone.org mara.mcadamsdemarco@nyulangone.org.

³Division of Biostatistics, New York University, 180 Madison Avenue, New York, New York 10016, U.S.A ivan.diaz@nyu.edu.

⁴Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, 2024 E Monument Street, Baltimore, Maryland 21205, U.S.A wenbowu@jhu.edu.

PubMed PMC DOI PDF

Abstract

Comparing outcomes across treatments is essential in medicine and public policy. To do so, researchers typically estimate a set of parameters, possibly counterfactual, each targeting adifferent treatment. Treatment-specific means are commonly used, but their identification requires a positivity assumption: every subject has a nonzero probability of receiving each treatment. This assumption is ofte…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free Account Sign In

Want unlimited research access? See Pro plans

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.