Suppr超能文献

高腺苷脱氨酶结核性胸腔积液预测模型的开发与验证

Development and validation of a predictive model for tuberculous pleural effusion with high adenosine deaminase.

作者信息

Lin Xuedan, Liu Yanchao, Chen Meiyun, Bao Anni, Yang Tianxing

机构信息

Sanmen People's Hospital, Taizhou, Zhejiang, China.

First Affiliated Hospital, Zhejiang University School of Medicine, Zhejiang, China.

出版信息

Sci Rep. 2025 Aug 30;15(1):31978. doi: 10.1038/s41598-025-17864-8.

Abstract

Tuberculous pleural effusion (TPE) can be effectively diagnosed using adenosine deaminase (ADA); however, high ADA levels in pleural effusion (PE) have also been shown to be linked to other diseases. In this study, we aimed to develop and validate a prediction model and differentiate TPE in patients with high ADA levels. This retrospective analysis of patients with ADA levels ≥ 25 IU/L was conducted at our healthcare institution between January 2017 and December 2023. After collecting and analyzing clinical and laboratory data, we developed predictive models using 31 indicators from serum and PE. The model's performance was assessed using the area under the receiver operating characteristics curve (AUC), accuracy, sensitivity, and specificity. Based on their significance in disease prediction, the top five variables were selected for use in constructing the prediction model. We externally validated the diagnostic model using a cohort from a different hospital. Among the 237 included patients with high ADA levels, 133 (56.1%) and 104 (43.9%) were diagnosed with TPE and non-TPE, respectively. The LightGBM model was superior to the other models, achieving an AUC of 0.926, with high accuracy, sensitivity, and specificity. The top five variables, including effusion lymphocyte percentage, effusion lactate dehydrogenase/ADA, age, effusion total protein, and peripheral blood platelet count, were essential for creating an accurate predictive model, which demonstrated strong performances on the training, test, internal validation, and external validation sets. The results were then validated with decision curve analysis and a calibration curve. This novel predictive model based on clinical and laboratory features in serum and PE showed strong diagnostic capability in detecting TPE in patients with high ADA levels.

摘要

结核性胸腔积液(TPE)可通过腺苷脱氨酶(ADA)进行有效诊断;然而,胸腔积液(PE)中ADA水平升高也与其他疾病有关。在本研究中,我们旨在开发并验证一种预测模型,以鉴别ADA水平升高患者中的TPE。我们对2017年1月至2023年12月在我们医疗机构中ADA水平≥25 IU/L的患者进行了回顾性分析。在收集并分析临床和实验室数据后,我们使用血清和PE中的31项指标开发了预测模型。使用受试者操作特征曲线(AUC)下面积、准确性、敏感性和特异性来评估模型的性能。根据它们在疾病预测中的重要性,选择前五个变量用于构建预测模型。我们使用来自另一家医院的队列对诊断模型进行了外部验证。在纳入的237例ADA水平升高的患者中,分别有133例(56.1%)和104例(43.9%)被诊断为TPE和非TPE。LightGBM模型优于其他模型,AUC为0.926,具有较高的准确性、敏感性和特异性。包括胸腔积液淋巴细胞百分比、胸腔积液乳酸脱氢酶/ADA、年龄、胸腔积液总蛋白和外周血血小板计数在内的前五个变量对于创建准确的预测模型至关重要,该模型在训练集、测试集、内部验证集和外部验证集上均表现出色。然后通过决策曲线分析和校准曲线对结果进行了验证。这种基于血清和PE临床及实验室特征的新型预测模型在检测ADA水平升高患者的TPE方面显示出强大的诊断能力。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/48e3/12398481/da93cb7aa923/41598_2025_17864_Fig1_HTML.jpg

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验