Domain Generalization and Adaptation in Intensive Care with Anchor Regression

TL;DR

This study explores causality-inspired methods, including anchor regression and anchor boosting, to improve predictive model robustness across diverse ICU datasets, addressing distribution shifts in clinical applications.

Contribution

It introduces anchor boosting, a novel tree-based nonlinear extension, and a framework to evaluate external data utility, advancing domain generalization and adaptation in ICU predictive modeling.

Findings

Anchor regularization improves out-of-distribution performance.

Methods are robust to violations of theoretical assumptions.

Identifies regimes where external data is most beneficial.

Abstract

The performance of predictive models in clinical settings often degrades when deployed in new hospitals due to distribution shifts. This paper presents a large-scale study of causality-inspired domain generalization on heterogeneous multi-center intensive care unit (ICU) data. We apply anchor regression and introduce anchor boosting, a novel, tree-based nonlinear extension, to a large dataset comprising 400,000 patients from nine distinct ICU databases. We find that anchor regularization yields improvements of out-of-distribution performance, particularly for the most dissimilar target domains. The methods appear robust to violations of theoretical assumptions, such as anchor exogeneity. Furthermore, we propose a novel conceptual framework to quantify the utility of large external data datasets. By evaluating performance as a function of available target-domain data, we identify three regimes: (i) a domain generalization regime, where only the external model should be used, (ii) a domain adaptation regime, where refitting the external model is optimal, and (iii) a data-rich regime, where external data provides no additional value.