Natural Language Processing Applications in Cardiology: A Narrative Review

2025-10-19

AI Summary

Cardiovascular diseases (CVDs) exhibit complex etiologies and heterogeneous data sources, necessitating deep knowledge extraction from unstructured textual sources—including patient narratives, electronic health records (EHRs), and scientific literature. This study presents a systematic review of natural language processing (NLP) applications in cardiology from 2014 to 2025. We construct, for the first time, a comprehensive taxonomy encompassing technological evolution, task categories, and data modalities, grounded in rigorous analysis of 265 high-quality studies. Our synthesis maps methodological pathways and limitations across text mining, information extraction, and machine learning/deep learning models. Key findings reveal a paradigm shift over the past decade—from rule-based systems toward end-to-end learning—and an expansion of application scope from isolated diagnostic tasks to longitudinal risk prediction and personalized intervention. The work establishes a theoretical framework and practical guidelines to advance NLP-driven precision diagnosis and management of cardiovascular and cerebrovascular diseases.

Technology Category

• Natural Language Processing: Information Extraction
• Cognitive Modeling & Cognitive Systems: Applications
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Cardiovascular disease has become increasingly prevalent in modern society and has a significant effect on global health and well-being. Heart-related conditions are intricate, multifaceted disorders, which may be influenced by a combination of genetic predispositions, lifestyle choices, and various socioeconomic and clinical factors. Information regarding these potentially complex interrelationships is dispersed among diverse types of textual data, which include patient narratives, medical records, and scientific literature, among others. Natural language processing (NLP) techniques have increasingly been adopted as a powerful means to analyse and make sense of this vast amount of unstructured data. This, in turn, can allow healthcare professionals to gain deeper insights into the cardiology field, which has the potential to revolutionize current approaches to the diagnosis, treatment, and prevention of cardiac problems. This review provides a detailed overview of NLP research in cardiology between 2014 and 2025. We queried six literature databases to find articles describing the application of NLP techniques in the context of a range of different cardiovascular diseases. Following a rigorous screening process, we identified a total of 265 relevant articles. We analysed each article from multiple dimensions, i.e., NLP paradigm types, cardiology-related task types, cardiovascular disease types, and data source types. Our analysis reveals considerable diversity within each of these dimensions, thus demonstrating the considerable breadth of NLP research within the field. We also perform a temporal analysis, which illustrates the evolution and changing trends in NLP methods employed over the last decade that we cover. To our knowledge, the review constitutes the most comprehensive overview of NLP research in cardiology to date.

Problem

• Analyzing unstructured text data for cardiovascular disease insights
• Applying NLP techniques to diverse cardiology tasks and diseases
• Reviewing NLP research evolution in cardiology over a decade

Innovation

• NLP techniques analyze unstructured cardiology text data
• Review covers NLP paradigms and cardiovascular disease types
• Temporal analysis shows evolution of NLP methods