What is the clinical evidence from this study?

Study design: Other. Population: Symptomatic severe aortic stenosis (n=782). Intervention: CardioTRAP (RAG framework).

What question did this study set out to answer?

The goal is to create a framework that improves data analysis and knowledge extraction in cardiology using advanced technologies.

March 26, 2026Open Access

Design of a RAG framework for cardiology EHR analysis

ADAnnamaria DefilippoMagna Graecia University GCGiovanni CaninoMagna Graecia University NPNicola ProcopioInstitute of Software

Key Result

CardioTRAP, a novel Retrieval-Augmented Generation framework combining dense and sparse retrieval, was developed to extract actionable insights from cardiology electronic health records.

Key Points

The goal is to create a framework that improves data analysis and knowledge extraction in cardiology using advanced technologies.
Developed CardioTRAP architecture for managing biomedical data specific to cardiology.
Incorporated advanced indexing techniques for efficient storage and retrieval.
Employed deep learning models to generate clinically relevant insights.
Used a hybrid approach combining supervised and unsupervised learning methods.
Achieved high accuracy and scalability in predictive analytics.
Enhanced identification of critical clinical features via benchmarks and metrics.
Facilitated patient risk stratification and novel biomarker discovery.

Structured PICO

Population

782 consecutive patients with symptomatic severe aortic stenosis who underwent Transcatheter Aortic Valve Replacement (TAVR)

Intervention

CardioTRAP (a hybrid dense-sparse Retrieval-Augmented Generation framework)

CardioTRAP provides a specialized Retrieval-Augmented Generation architecture to efficiently extract insights from unstructured cardiology electronic health records.

Limitations

Pre-trained embedding models often lack access to domain-specific vocabulary
Large Language Models have an inclination to generate factually incorrect or biased content (hallucinations)

Abstract

Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) systems have significantly advanced data modelling capabilities and improved opportunities for extracting knowledge from vast and heterogeneous biomedical datasets. Recent research has increasingly focused on integrating LLMs with custom-designed RAGs to create systems capable of handling complex biomedical challenges, with a growing demand for more reliable and precise prediction mechanisms in health-related contexts. This study introduces CardioTRAP, an architecture specifically designed to manage biomedical data, with a primary focus on cardiology. The system employs advanced indexing techniques to enable efficient storage and retrieval by integrating deep learning models that generate contextual and clinically relevant insights. By adopting a hybrid approach that combines supervised and unsupervised learning methods, CardioTRAP ensures both high accuracy and scalability, supporting predictive analytics, patient risk stratification, and the discovery of novel biomarkers. Benchmarks and practical applications, evaluated through state-of-the-art metrics, underscore its ability to enhance the identification of critical clinical features. Finally, CardioTRAP demonstrates how the integration of data management and RAG systems can serve as a bridge between biomedical research and clinical practice.

KI fragen

Bookmark

View Full Paper