An acoustofluidic dynamic sieve enables sample-to-answer extracellular vesicle analysis for point-of-care liquid biopsy.

Liquid biopsy based on extracellular vesicle (EV) surface protein provides a non-invasive strategy for lung cancer detection and clinical stratification, yet its translation to point-of-care testing (POCT) is hindered by labor-intensive preprocessing and nonspecific background from contaminating proteins. Here, we report an acoustofluidic chip that works as an "acoustofluidic dynamic sieve," integrating on-chip staining, separation, enrichment, and detection into a sample-to-answer EV assay. Acoustic and flow fields synergistically generate localized vortices that selectively retain and concentrate EVs (>30 nm) while laminar flow removes smaller proteins. This on-chip purification ensures that fluorescence signals originate predominantly from captured EVs, achieving 125-fold signal amplification with high specificity. The platform completes multi-target surface protein analysis using less than 1.5 μL of plasma without off-chip purification, simplifying the workflow and reducing sample consumption. In clinical testing, 5-fold cross-validation distinguished lung cancer patients from healthy controls with 94.12% sensitivity, 93.33% specificity, and 93.75% accuracy. This acoustofluidic dynamic sieve strategy overcomes the specificity challenge in EV-based diagnostics and shows significant potential for clinical translation.