ClinStructor: AI-Powered Structuring of Unstructured Clinical Texts

2025-11-14

AI Summary

Clinical notes contain rich, high-value contextual information but their unstructured nature leads to model susceptibility to gender/racial biases, poor generalization across electronic health record (EHR) systems, and limited interpretability. To address this, we propose a large language model (LLM)-based framework that automatically converts raw clinical text into task-oriented question-answer (QA) pairs—a structured, semantically preserved intermediate representation. Evaluated on ICU mortality prediction, our method incurs only a modest 2–3 percentage-point AUC drop while substantially improving model transparency, fairness, and cross-EHR generalizability. Our key contribution is the first use of LLM-driven QA-style structuring as a unified interface for enhancing interpretability and mitigating bias—simultaneously preserving predictive performance, enabling fine-grained controllability, and ensuring deployment robustness.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Clinical notes contain valuable, context-rich information, but their unstructured format introduces several challenges, including unintended biases (e.g., gender or racial bias), and poor generalization across clinical settings (e.g., models trained on one EHR system may perform poorly on another due to format differences) and poor interpretability. To address these issues, we present ClinStructor, a pipeline that leverages large language models (LLMs) to convert clinical free-text into structured, task-specific question-answer pairs prior to predictive modeling. Our method substantially enhances transparency and controllability and only leads to a modest reduction in predictive performance (a 2-3% drop in AUC), compared to direct fine-tuning, on the ICU mortality prediction task. ClinStructor lays a strong foundation for building reliable, interpretable, and generalizable machine learning models in clinical environments.

Problem

• Addressing unstructured clinical text challenges like bias and poor generalization
• Converting clinical free-text into structured question-answer pairs using LLMs
• Enhancing transparency and interpretability while maintaining predictive performance

Innovation

• Converts clinical free-text into structured question-answer pairs
• Leverages large language models for clinical text structuring
• Enhances transparency and controllability in predictive modeling