3D-Printed Microfluidic-Integrated SERS Salivary Biosensor Utilizing Fe@Ag/Carbon Nanofibers for Advanced Machine Learning-Driven Noninvasive, Label-Free Mass Screening of Lung Cancer

Developing an advanced, noninvasive diagnostic tool for early-stage lung cancer screening is critical to enable prediagnosis and improve patient outcomes. Herein, a microfluidic surface-enhanced Raman scattering (SERS) biosensor platform fabricated using three-dimensional (3D) printing has been developed to facilitate noninvasive, point-of-care lung cancer diagnostics by integrating carbon nanofibers (CNFs) anchored with Iron@Silver (Fe@Ag) core–shell nanoparticles. The evaluation of the SERS performance using Rhodamine 6G demonstrated a significant surface enhancement factor (EF) of approximately 10 7, with an ultralow detection limit down to 10 –12 M, affirming its superior sensitivity. The plasmonic hotspot exhibited an electric field intensity enhancement factor (| E | 2 /| E 0 | 2 ) exceeding 600, induced at the vicinity of Fe@Ag nanoparticles, which was identified as a major factor contributing to the superior performance of the developed sensors. The integration of Fe@Ag/CNFs-based SERS substrates into the microfluidic platform addresses challenges associated with xerostomia, a common condition in lung cancer patients that limits saliva production, by enabling effective analysis of low-volume saliva samples with improved reproducibility and statistical robustness, enabling high-throughput, real-time detection of lung cancer biomarkers from saliva samples. To improve diagnostic precision, machine learning techniques have been utilized to distinguish the salivary SERS profiles of individuals with lung cancer ( n = 44) from those of healthy controls ( n = 45). To improve diagnostic accuracy, machine learning techniques were utilized to distinguish the salivary SERS signals of lung cancer groups ( n = 44) from those of healthy individuals ( n = 45). Principal Component Analysis has been used to reduce the data dimensionality, followed by application of a support vector machine classifier, achieving a classification accuracy of 94%, with a sensitivity of 93.5% and a specificity of 88%. Combining SERS with machine learning techniques underscores the promise of this microfluidic biosensor as a noninvasive and reliable tool for the early detection of lung cancer, paving the way for more accurate and efficient clinical diagnostics.