An MLI-Guided Framework for Subgroup-Aware Modeling in Electronic Health Records (AdaptHetero)

TL;DR

AdaptHetero is a novel framework that uses machine learning interpretation to identify and improve subgroup-specific modeling in electronic health records, enhancing predictive accuracy and clinical relevance.

Contribution

It introduces a new MLI-guided approach that transforms interpretability insights into actionable subgroup-aware modeling strategies for EHR data.

Findings

Improves predictive performance up to 174.39% in certain subpopulations

Uncovers clinically meaningful subgroup characteristics

Enhances robustness and equity in clinical models

Abstract

Machine learning interpretation (MLI) has primarily been leveraged to foster clinician trust and extract insights from electronic health records (EHRs), rather than to guide subgroup-specific, operationalizable modeling strategies. To bridge this gap, we propose AdaptHetero, a novel MLI-driven framework that transforms interpretability insights into actionable guidance for tailoring model training and evaluation across subpopulations. Evaluated on three large-scale EHR datasets -- GOSSIS-1-eICU, WiDS, and MIMIC-IV -- AdaptHetero consistently uncovers heterogeneous model behaviors in predicting ICU mortality, in-hospital death, and hidden hypoxemia. Integrating SHAP-based interpretation with unsupervised clustering, AdaptHetero identifies clinically meaningful, subgroup-specific characteristics, improving predictive performance across many subpopulations (with gains up to 174.39 percent) while proactively flagging potential risks in others. These results highlight the framework's promise for more robust, equitable, and context-aware clinical deployment.