A Bioinspired Three‐Dimensional High‐Curvature Nano‐Interface Integrated Microfluidic Chip for Small Extracellular Vesicles Enrichment and Machine Learning‐Assisted Prostate Cancer Precision Diagnosis

Le Wang, Yizhong Liang, Manan Sulaiman, Jiaqi Du, Ming Jiang, Zhihua Wang, Xu Yu, Li Xu

ABSTRACT The efficient and unbiased isolation of small extracellular vesicles (sEVs) from complex biological fluids remains a major obstacle for clinical diagnostics. Here, we report a bioinspired microfluidic chip that integrates a three‐dimensional high‐curvature TiO 2 nano‐interface (3D Hic‐TiO 2 ) with a biotin‐modified artificial insertion peptide (BAIP) modification for rapid enrichment of sEV. Bowl‐shaped TiO 2 nanospheres fabricated via electrospray provide topological nanotraps that match the size and curvature of sEVs, enabling efficient size‐selective capture. Coupled with BAIP‐mediated membrane affinity and herringbone‐induced chaotic mixing, the BAIP‐TiO 2 ‐Chip achieved >90% capture efficiency within 5 min. Redox‐responsive BAIP variants enabled mild release of intact sEVs for downstream analysis. When the BAIP‐TiO 2 ‐Chip was applied to plasma samples from clinical prostate cancer (PCa) patients, mass spectrometry‐based proteomic profiling revealed 110 differentially expressed sEV‐associated proteins, including candidates involved in immune regulation and cell adhesion. In parallel, simultaneous quantification of PSA and PSMA mRNAs in sEVs could be achieved. Assisted by machine learning, a boosted decision tree model achieved 80% diagnostic accuracy in distinguishing PCa from benign conditions and healthy donors. This work presents a versatile platform for sEV isolation, enabling both transcriptomic mRNA analysis and proteomic profiling, and provides new molecular insights into PCa for improved early diagnosis.