An Offline Deep Learning-Assisted Automated Paper-Based Microfluidic Platform

This paper reports an automated microfluidic paper-based analytical device (μPAD) platform featuring a highly integrated rotary valve with deep learning-assisted smartphone offline detection for early screening of Alzheimer's disease (AD). Unlike existing platforms, our platform utilizes deep learning-assisted smartphones to achieve offline detection, avoiding data transfer with the assistance of large-scale equipment and privacy leakage issues for the cloud. Meanwhile, we use a simple mechanical rotary valve to achieve complex enzyme-linked immunoassay (ELISA) detection of blood biomarker, β-amyloid peptide 1-42 (Aβ 1-42). In this paper, we performed 38 clinical serum samples (healthy: 19, unhealthy: 19; N=6), and the platform provided 98.4% mean average precision (mAP).