Maule Geran, Alomari Ahmad, Rayyan Abdallah, Aghahowa Ogbeide, Khraisat Mohammad, Javier Luis
Department of Clinical Sciences, University of Central Florida College of Medicine, Orlando, Florida, USA.
Department of Graduate Medical Education, HCA Florida North Florida Hospital, Gainesville, Florida, USA.
Can Respir J. 2025 Aug 6;2025:2882255. doi: 10.1155/carj/2882255. eCollection 2025.
The detection and classification of pleural effusion present significant challenges in clinical practice, often contributing to delayed diagnoses and suboptimal patient outcomes. Recent advancements in artificial intelligence (AI) and machine learning (ML) techniques hold substantial promise for enhancing the accuracy and efficiency of pleural effusion diagnostics. This paper reviews the current landscape of AI applications in pleural effusion detection, synthesizing findings across diverse studies to illustrate the transformative potential of these technologies. We examine various ML models, including deep learning and ensemble methods, that leverage clinical, laboratory, and imaging data to improve diagnostic performance. Notably, models such as Light Gradient Boosting Machine (LGB) and XGBoost have achieved accuracy levels up to 96% and high AUC values (e.g., AUC = 0.883 for pleural effusion differentiation). This overview highlights the importance of integrating diverse diagnostic parameters to enhance pleural effusion diagnostic accuracy and outlines future research directions essential for optimizing patient management and outcomes.
胸腔积液的检测和分类在临床实践中面临重大挑战，常常导致诊断延迟和患者预后不佳。人工智能（AI）和机器学习（ML）技术的最新进展为提高胸腔积液诊断的准确性和效率带来了巨大希望。本文综述了AI在胸腔积液检测中的应用现状，综合不同研究的结果以阐明这些技术的变革潜力。我们研究了各种ML模型，包括深度学习和集成方法，这些模型利用临床、实验室和影像数据来提高诊断性能。值得注意的是，诸如轻梯度提升机（LGB）和XGBoost等模型已实现高达96%的准确率和较高的AUC值（例如，胸腔积液鉴别诊断的AUC = 0.883）。本综述强调了整合多种诊断参数以提高胸腔积液诊断准确性的重要性，并概述了优化患者管理和预后所需的未来研究方向。
