Target specification bias, counterfactual prediction, and algorithmic fairness in healthcare

Eran Tal

Bias in applications of machine learning (ML) to healthcare is usually attributed to unrepresentative or incomplete data, or to underlying health disparities. This article identifi…