Fairness-Aware Interpretable Modeling (FAIM) for Trustworthy Machine Learning in Healthcare

TL;DR

FAIM is an interpretable framework that enhances fairness in healthcare machine learning models by reducing biases while maintaining high performance, involving domain experts through an interactive interface.

Contribution

Introduces FAIM, a novel interpretable and interactive framework that improves fairness in healthcare ML models without performance loss, integrating clinical expertise.

Findings

FAIM reduces sex and race biases in hospital admission predictions.

FAIM models achieve high discriminatory performance.

FAIM outperforms common bias-mitigation methods.

Abstract

The escalating integration of machine learning in high-stakes fields such as healthcare raises substantial concerns about model fairness. We propose an interpretable framework - Fairness-Aware Interpretable Modeling (FAIM), to improve model fairness without compromising performance, featuring an interactive interface to identify a "fairer" model from a set of high-performing models and promoting the integration of data-driven evidence and clinical expertise to enhance contextualized fairness. We demonstrated FAIM's value in reducing sex and race biases by predicting hospital admission with two real-world databases, MIMIC-IV-ED and SGH-ED. We show that for both datasets, FAIM models not only exhibited satisfactory discriminatory performance but also significantly mitigated biases as measured by well-established fairness metrics, outperforming commonly used bias-mitigation methods. Our approach demonstrates the feasibility of improving fairness without sacrificing performance and provides an a modeling mode that invites domain experts to engage, fostering a multidisciplinary effort toward tailored AI fairness.