Ultrasensitive\nDetection of Blood-Based Alzheimer’s\nDisease Biomarkers: A Comprehensive SERS-Immunoassay Platform Enhanced\nby Machine Learning

Accurate and early disease detection is crucial for improving\npatient\ncare, but traditional diagnostic methods often fail to identify diseases\nin their early stages, leading to delayed treatment outcomes. Early\ndiagnosis using blood derivatives as a source for biomarkers is particularly\nimportant for managing Alzheimer’s disease (AD). This study\nintroduces a novel approach for the precise and ultrasensitive detection\nof multiple core AD biomarkers (Aβ<sub>40</sub>, Aβ<sub>42</sub>, p-tau, and t-tau) using surface-enhanced Raman spectroscopy\n(SERS) combined with machine-learning algorithms. Our method employs\nan antibody-immobilized aluminum SERS substrate, which offers high\nprecision, sensitivity, and accuracy. The platform achieves an impressive\ndetection limit in the attomolar (aM) range and spans a wide dynamic\nrange from aM to micromolar (μM) concentrations. This ultrasensitive\nand specific SERS immunoassay platform shows promise for identifying\nmild cognitive impairment (MCI), a potential precursor to AD, from\nblood plasma. Machine-learning algorithms applied to the spectral\ndata enhance the differentiation of MCI from AD and healthy controls,\nyielding excellent sensitivity and specificity. Our integrated SERS-machine-learning\napproach, with its interpretability, advances AD research and underscores\nthe effectiveness of a cost-efficient, easy-to-prepare Al-SERS substrate\nfor clinical AD detection.