Machine learning-assisted SERS-based dual-aptamer biosensor for ultrasensitive clinical screening of breast cancer.

Breast cancer remains a global health challenge, with an increasing number of cases necessitating innovative approaches to streamline patient management prior to treatment. In this study, we present a comprehensive aptamer-involved surface-enhanced Raman spectroscopy (aptamer-SERS)-based protocol specifically designed for large-scale clinical screening of the circulating protein MUC1 overexpressed in the majority of breast cancer cases. Central to our approach was a "sandwich" assay, where MUC1 was anchored between aptamer-functionalized bimetallic core-shell nanoparticles (NPs) and magnetic nanobeads. The biosensor was applied to monitor MUC1 in serum samples through SERS signals from the reporter molecule 4-ATP, exhibiting a strong linear correlation across a wide dynamic range and the lower the limit of detection (LOD) of 2.96 fg/mL. To support clinical decision-making, the protocol was integrated with a machine learning (ML)model for the classification of SERS signal patterns, demonstrating performance metrics of 96% accuracy and 93.7% specificity when applied to diverse serum samples. This integration enabled robust, non-invasive preclinical screening that informs therapeutic regulation and patient monitoring long before clinical symptoms present. By establishing a scalable framework for continuous monitoring of MUC1 levels across large populations, our study offered a forward-thinking tool, namely, dual-aptamer-SERS biosensor, that may optimize individualized treatment planning and improve overall clinical management strategies.