An Alloy Platform of Dual‐Fingerprints for High‐Performance Stroke Screening
Weikang Shu, M. Zhang, Chaoqi Zhang, Rongxin Li, Congcong Pei, Yu Zeng, Liang Zhao, Jing Zhao, Jingjing Wan
Abstract
Abstract A multi‐modal serum profiling platform holds promise for precision diagnosis of diseases. Still, advanced tools are in demand to deliver the multi‐modal serum profiling. Herein, a bimodal spectrometric protocol is designed for stoke serum profiling using an alloy platform, by integrating label‐free surface‐enhanced Raman spectroscopy (SERS) and laser desorption/ionization mass spectrometry (LDI‐MS). The PdAu@Au concave cube with a wide localized surface plasmonic resonance (LSPR) range simultaneously enhances the signals from SERS and LDI‐MS, enabling high‐throughput co‐detection of vibrational and metabolic fingerprints of 0.1 µL serum in 2 min with simple pretreatment. Further, a dual‐fingerprints screening model of stroke is constructed, by adaptive machine learning with a programming nonlinear fitting model. The area under the curve are 0.949 (0.917–0.977, 95% confidence interval (CI)) and 0.911 (0.812–0.984, 95% CI), in the discovery and validation cohorts, respectively. Finally, five metabolites are identified that correlated to SERS signals and mapped the relevant pathways. This study features high performance in terms of throughput, speed, sample volume, and accuracy, providing new insight into the construction of multiplexed characterization platforms for precision diagnostic.