Machine Learning-Enabled Diagnosis of Viral Respiratory Infections from Exhaled Volatile Organic Compound Analysis

Background: The sensitivity and specificity of current breath biomarkers are often inadequate for effective requisite sensitivity for early-stage detection, thereby ignore early-stage treatment in patient. Methods: In this study, we developed a screening model for viral respiratory infections based on the combination of portable GC-MS and an artificial intelligence (AI) model. This platform employs neural network-assisted algorithms to enhance the specificity and sensitivity of the model. Subsequently, we applied this platform to analyze 200 viral respiratory infections and normal exhaled samples. Results: The diagnostic signatures, including 1-nonanethiol, 2-butanone, generated by the model effectively discriminated viral respiratory infections patients from normal controls with high sensitivity (85%), specificity (85%), and accuracy (AUC = 0.85). Furthermore, propionaldehyde, amylaldehyde, generated by the model effectively discriminated COVID-19 from influenza A patients with sensitivity (75%), specificity (83.3%), and accuracy (AUC = 0.80). Data from UKBiobank indicated that In the volatile metabolite profiles exhaled by patients with viral respiratory infections, some characteristic components are related to the metabolic products of the host's fatty acid β-oxidation pathway. Conclusion: This study introduces a diagnostic model capable of identifying novel and feasible breath biomarkers for early-stage viral respiratory infections detection. The promising results position the platform as an efficient noninvasive screening test for clinical applications, offering potential advancements in early detection for viral respiratory infections patients.