Evaluating the Impact of Social Determinants on Health Prediction in the Intensive Care Unit

TL;DR

This study links community-level social determinants of health with EHR data to evaluate their impact on ICU health prediction models, revealing limited overall performance gains but benefits for fairness and bias auditing.

Contribution

It introduces an integrated EHR-SDOH database and systematically assesses the influence of SDOH features on health prediction and bias detection in ICU settings.

Findings

SDOH features do not improve overall model accuracy for general populations.

SDOH features enhance fairness in data-limited subpopulations.

SDOH features are useful for auditing algorithmic biases beyond traditional protected attributes.

Abstract

Social determinants of health (SDOH) -- the conditions in which people live, grow, and age -- play a crucial role in a person's health and well-being. There is a large, compelling body of evidence in population health studies showing that a wide range of SDOH is strongly correlated with health outcomes. Yet, a majority of the risk prediction models based on electronic health records (EHR) do not incorporate a comprehensive set of SDOH features as they are often noisy or simply unavailable. Our work links a publicly available EHR database, MIMIC-IV, to well-documented SDOH features. We investigate the impact of such features on common EHR prediction tasks across different patient populations. We find that community-level SDOH features do not improve model performance for a general patient population, but can improve data-limited model fairness for specific subpopulations. We also demonstrate that SDOH features are vital for conducting thorough audits of algorithmic biases beyond protective attributes. We hope the new integrated EHR-SDOH database will enable studies on the relationship between community health and individual outcomes and provide new benchmarks to study algorithmic biases beyond race, gender, and age.