呼吸道感染的无创筛查方法

Introduction: Globally, respiratory infections pose significant challenges to the public health system and have the potential to cause severe illness, leading to acute respiratory failure. Despite high sensitivity and specificity, current diagnostic procedures are often invasive and stressful for children. Therefore, advancing non-invasive methods is crucial for improving the tolerability for children. By utilizing electronic nose technology, specifically Cyranose 320 (Sensigent, USA), this study investigates whether individuals with confirmed respiratory infections can be reliably differentiated from healthy controls based on volatile organic compounds (VOCs) and to determine the optimal sample type for diagnosis.

引言： 在全球范围内，呼吸道感染对公共卫生系统构成重大挑战，并有可能导致严重疾病，导致急性呼吸衰竭。尽管敏感性和特异性很高，但目前的诊断程序对儿童来说往往是侵入性和压力性的。因此，推进非侵入性方法对于提高儿童的耐受性至关重要。通过利用电子鼻技术，特别是Cyranose 320（Sensigent，美国），本研究调查了是否可以根据挥发性有机化合物（VOC）将确诊的呼吸道感染患者与健康对照组可靠地区分开来，并确定诊断的最佳样本类型。 


Materials & Methods: Using Cyranose 320, 2710 samples from 202 participants comprised children (n=103, positive=44, healthy controls=59, ages 0-17 years; median age: 3 years, 11 months) and adults (n=99, positive=32, healthy controls=37, ages 18 years and older; median age: 28 years, 3 months) were analyzed to differentiate bacterial, viral, and co-infections across a panel of different pathogens including SARS-CoV-2, Influenza A, RSV, Adenovirus, Rhinovirus, and key bacteria such as H. influenzae (including type B), M. catarrhalis, and S. pneumoniae, alongside a sample-centric analysis. Breath samples were collected before and after water intake, along with saliva and sweat samples. The methodological approach employed Linear Discriminant Analysis (LDA) on pre-processed (using standardization and feature selection via PCA) sensor data, validated through a rigorous pairwise leave-one-out-cross-validation framework. The diagnostic performance was determined using sensitivity, specificity, PPV, NPV and Mahalanobis distances (MD).

材料和方法： 使用Cyranose 320，对来自202名参与者的2710个样本进行了分析，包括儿童（n=103，阳性=44，健康对照=59，年龄0-17岁；中位年龄：3岁11个月）和成人（n=99，阳性=32，健康对照=37，年龄18岁及以上；中位岁：28岁3个月），以区分不同病原体的细菌、病毒和合并感染，包括严重急性呼吸系统综合征冠状病毒2型、甲型流感、呼吸道合胞病毒、腺病毒、鼻病毒和关键细菌，如流感嗜血杆菌（包括B型）、卡他支原体和肺炎链球菌，以及以样本为中心的分析。在饮水前后收集呼吸样本，以及唾液和汗液样本。该方法采用线性判别分析（LDA）对预处理（通过PCA进行标准化和特征选择）的传感器数据进行分析，并通过严格的成对留一交叉验证框架进行验证。诊断性能由灵敏度、特异性、PPV、NPV和马氏距离（MD）确定。 


Results: The principal findings demonstrate that the VOC analysis yielded high diagnostic performance compared to PCR-confirmed diagnosis when differentiating bacterial, viral and co-infections, with sensitivity ranging from 75% to 85%, specificity over 90%, and MD>2.0 from healthy controls. However, its ability to differentiate specific pathogens is still limited. Crucially, the study also establishes that simple water-intake before breath samples significantly improves the accuracy of breath analysis by up to 90%. Furthermore, saliva samples demonstrated a clear discriminative ability over sweat samples.
结果： 主要发现表明，与PCR确认的诊断相比，VOC分析在区分细菌、病毒和合并感染时具有较高的诊断性能，灵敏度为75%至85%，特异性超过90%，与健康对照组相比，MD>2.0。然而，它区分特定病原体的能力仍然有限。至关重要的是，该研究还确定，在呼吸样本之前简单的饮水可以将呼吸分析的准确性提高90%。此外，唾液样本显示出对汗液样本的明显辨别能力。 

Conclusion: In conclusion, VOC analysis using Cyranose 320 has the potential to differentiate several respiratory infections from healthy controls. Breath samples after water intake demonstrated superior diagnostic performance, while saliva samples served as a complementary diagnostic source. While establishing this method as a reliable tool requires further multicenter validation and protocol optimization, this study provides a scalable framework for clinical integration, enabling cost-effective, non-invasive, and efficient management of respiratory infections.

结论： 总之，使用Cyranose 320进行VOC分析有可能将几种呼吸道感染与健康对照区分开来。饮水后的呼吸样本显示出卓越的诊断性能，而唾液样本则可作为补充诊断来源。虽然将这种方法确立为可靠的工具需要进一步的多中心验证和方案优化，但这项研究为临床整合提供了一个可扩展的框架，实现了成本效益高、非侵入性和有效的呼吸道感染管理。