Spurious Correlations and Beyond: Understanding and Mitigating Shortcut Learning in SDOH Extraction with Large Language Models
This work addresses reliability issues in LLMs for healthcare analytics, which is crucial for accurate downstream applications, though it is incremental as it builds on existing methods for mitigating biases in specific domains.
The study tackled the problem of spurious correlations and shortcut learning in large language models (LLMs) for extracting social determinants of health (SDOH) from clinical text, specifically focusing on drug status extraction, and found that mentions of alcohol or smoking led to false positives and gender disparities in performance, with mitigation strategies like prompt engineering reducing these errors.
Social determinants of health (SDOH) extraction from clinical text is critical for downstream healthcare analytics. Although large language models (LLMs) have shown promise, they may rely on superficial cues leading to spurious predictions. Using the MIMIC portion of the SHAC (Social History Annotation Corpus) dataset and focusing on drug status extraction as a case study, we demonstrate that mentions of alcohol or smoking can falsely induce models to predict current/past drug use where none is present, while also uncovering concerning gender disparities in model performance. We further evaluate mitigation strategies - such as prompt engineering and chain-of-thought reasoning - to reduce these false positives, providing insights into enhancing LLM reliability in health domains.