# Causal Inference with Unstructured Outcomes

arXiv: https://arxiv.org/abs/2608.03085 · HTML: https://mlsanity.com/paper/2608.03085/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 51 references that are in the map.
- First posted on arXiv 4 Aug 2026.

## Builds on (most cited first, 10 of 10)

- [SONAR: Sentence-Level Multimodal and Language-Agnostic Representations](https://mlsanity.com/paper/2308.11466/) (2023)
- [Stochastic Hyperparameter Optimization through Hypernetworks](https://mlsanity.com/paper/1802.09419/) (2018)
- [DoubleMLDeep: Estimation of Causal Effects with Multimodal Data](https://mlsanity.com/paper/2402.01785/) (2024)
- [Causal Inference on Outcomes Learned from Text](https://mlsanity.com/paper/2503.00725/) (2025)
- [Leveraging Generative Artificial Intelligence for Causal Inference with Unstructured Data](https://mlsanity.com/paper/2507.03897/) (2025)
- [Integrating Unstructured Text into Causal Inference: Empirical Evidence from Real Data](https://mlsanity.com/paper/2602.14274/) (2026)
- [The Partial Testimony of Logs: Evaluation of Language Model Generation under Confounded Model Choice](https://mlsanity.com/paper/2605.01311/) (2026)
- [GenAI Powered Dynamic Causal Inference with Unstructured Data](https://mlsanity.com/paper/2605.07834/) (2026)
- [Identifying sparse treatment effects in high-dimensional outcome spaces](https://mlsanity.com/paper/2404.14644/) (2024)
- [Causal Inference with Unstructured Treatments](https://mlsanity.com/paper/2608.00657/) (2026)

## Abstract (arXiv)

Causal inference has traditionally centered on scalar outcomes: whether a patient recovers, how much a worker earns, or how many visits a website receives. Modern studies increasingly ask causal questions about outcomes with richer form, such as clinical notes, open-ended survey responses, and images. A hospital may want to know how an AI documentation tool changes the notes physicians write, or how a nurse training program alters what patients say in survey responses. For such outcomes, the usual average treatment effect is ill-defined: one cannot meaningfully subtract one text or image from another. To this end, we propose a causal query for unstructured outcomes. The key idea is to learn what features of the outcome are most causally affected by the treatment, which we call the maximally contrasting feature (MCF). To estimate the MCF, we learn a feature-scoring function that maps each outcome to a scalar and exposes the sharpest contrast between treated and control potential outcomes. We develop identification conditions and estimation algorithms for this query, and extend it to heterogeneous effects by allowing the feature-scoring function to depend on observed covariates. We also handle settings where both the treatment and the outcome are unstructured. Empirical studies on text and images show that the algorithm recovers salient aspects of an outcome changed by a treatment.