{"id":"arxiv/2602.21423","url":"https://arxiv.org/abs/2602.21423","tags":["math.ST","stat.ME"],"score":0,"title":"Causal Inference with High-Dimensional Treatments","domain":"math.ST","source":"arxiv","authors":["Patrick Kramer","Edward H. Kennedy","Isaac M. Opper"],"reviews":0,"subject":"math.ST","summary":"In this work, we consider causal inference in various high-dimensional treatment settings, including for single multi-valued treatments and vector treatments with binary or continuous components, when the number of treatments can be comparable to or even larger than the number of observations. These settings bring unique challenges: first, the treatment effects of interest are represented by a high-dimensional vector rather than a scalar; second, positivity violations are often unavoidable; and third, estimation can be based on a smaller effective sample size. We first discuss fundamental…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-02-24","reactions":{}}