Personalized Treatment Effect Estimation from Unstructured Data
Henri Arno, Thomas Demeester · 2025
Read Personalized Treatment Effect Estimation from Unstructured Data by Henri Arno, Thomas Demeester (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-20993.txt.
Description
Henri Arno and Thomas Demeester tackle a practical problem in learning which patients benefit from which treatment when the evidence sits in messy medical records. Records mix tables with free-text notes, and standard causal methods assume neat data; applied naively to learned representations of text, they can produce biased benefit estimates. Clinicians therefore fall back on risk scores, which find the sickest patients rather than the ones a treatment helps most.
Their method uses expert annotations during training to keep the confounding information honest, then, once trained, predicts benefit from the raw multimodal record alone. Across synthetic, semi-synthetic, and real record datasets, it beat both risk-based and representation-based rivals. The authors offer the approach as a workable bridge: causal policy learning that respects the text in the chart, aimed at helping doctors aim treatments where they do the most good.
Catalog record
| CREATOR | Henri Arno, Thomas Demeester |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-20993