---
title: "Proxy label"
description: "A training target that stands in for the construct a decision is meant to capture, such as health-care cost standing in for health need."
canonical: https://aigovernanceengineer.com/glossary/proxy-label
author: "Jorge García Aibar"
license: "CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)"
doi: https://doi.org/10.5281/zenodo.22956197
version: "0.5.0"
updated: 2026-09-24
---

# Proxy label

A training target that stands in for the construct a decision is meant to capture, such as health-care cost standing in for health need [1]. When the proxy is shaped by unequal treatment, a model can be accurate on the proxy and biased on the construct.

- Developed in: [Case: a health risk score with a proxy label](https://aigovernanceengineer.com/cases/health-risk-score-proxy)
- Chapters: [ch. 16, Fairness & XAI](https://aigovernanceengineer.com/bok/fairness-and-explainability)
- Contrast with: [Proxy variable](https://aigovernanceengineer.com/glossary/proxy-variable)
- In the glossary chapter: https://aigovernanceengineer.com/bok/glossary#t-proxy-label

## Sources

[1] "Dissecting racial bias in an algorithm used to manage the health of populations" (Obermeyer, Powers, Vogeli and Mullainathan; Science 366(6464):447-453; cost as a proxy for need). Science. 2019-10-25. https://doi.org/10.1126/science.aax2342 (verified: primary)