Machine Learning‐assisted Ultrasensitive SERS Immunoassays Across Wide Concentration Ranges Toward Clinical Ovarian Cancer Diagnosis
Shuang Lin, Minglang Dong, Chi Li, Xiang Lin, Cong Yan, Wen Xu, Zhouzhou Bao, Bin Dong
Abstract
Abstract The development of plasmonic devices featuring high‐density hotspots with wide linear response ranges has become critically essential for ultrasensitive carbohydrate antigen 125 (CA125) biomarker detection, enabling both early‐stage diagnosis and real‐time monitoring of ovarian cancer. In this study, an innovative nanophotonic platform is developed through assembling dual‐layers Au nanorods (NRs) within the SiO 2 microbowls for ultrasensitive detection of CA125 biomarkers. The interlaced nanorods inside the microbowl structure can fully utilize the light reflection and focus the energy of the light field, resulting in significantly amplified near‐field enhancements effect and exceptionally sensitive surface‐enhanced Raman scattering (SERS) performance with an analytical enhancement factor (AEF) of 1.09 × 10 8 . More importantly, the 3D nanogap of the NRIB configuration enables extensive contact with large CA125 protein molecules, facilitating the acquisition of their comprehensive SERS signal. As a proof of concept, a cost‐effective NRIB‐based SERS immunoassay platform is fabricated for sensitive CA125 detection across an ultra‐wide concentration range from 1 to 5000 U mL −1 through machine learning‐assisted analysis. The concentration analysis results of CA125 in serum samples from both healthy donors and ovarian cancer patients showed a strong correlation with clinical hospital tests. This hierarchical cavity‐enhanced platform, integrated with machine learning algorithms, is promising for early disease diagnosis and continuous patient monitoring.