Rapid and noninvasive artificial intelligence-assisted diagnostic method for oral squamous cell carcinoma

Oral squamous cell carcinoma (OSCC) remains the most common head and neck malignancy, for which early detection is critical yet challenging with current invasive methods. This study aimed to establish a comprehensive diagnostic framework for OSCC by integrating proton transfer reaction-time-of-flight mass spectrometry (PTR-TOF-MS) breath analysis and metagenomic sequencing with artificial intelligence (AI). Exhaled breath and saliva samples were collected from participants in a discovery cohort (n = 222) and an external validation cohort (n = 83). Samples were analyzed using PTR-TOF-MS and metagenomic sequencing, and multimodal diagnostic models were constructed and trained on the discovery cohort data. We identified OSCC-specific biomarkers, including methanethiol and Fusobacterium nucleatum, and developed an interactive online platform ( https://bio.futurecnn.com/ ) enabling real-time predictions and biomarker interpretability. The AI-driven diagnostic model achieved excellent accuracy (ROC-AUC: 0.92) in distinguishing OSCC patients from healthy controls in the external set. This approach offers a practical, noninvasive solution for OSCC screening and establishes an adaptable framework for other breath-based diagnostics.