Small extracellular vesicles-mediated formation of CuPd nanoparticles with boosted nanozyme activity: Machine learning-assisted photoelectrochemical biosensing.

Small extracellular vesicles (sEVs) are promising biomarkers for early cancer diagnosis. Herein, a new machine learning-assisted photoelectrochemical (PEC) biosensor is developed for the highly sensitive detection of plasma sEVs, based on the terminal deoxynucleotidyl transferase (TdT) catalytic initiation and nanozyme activity enhancement of dual-identity sEVs. The glycosyl and phenolic hydroxyl groups on sEVs membrane proteins can act as substrates for TdT initiation DNA polymerization, and the generated polyT sequences are extended in situ on sEVs to mediate the formation of CuPd nanoparticles (sEVs-polyT-CuPd NPs). Then, the obtained sEVs-polyT-CuPd NPs are immobilized on the PEC platform via specific capture of EpCAM aptamer, and act as excellent nanozymes to catalyze the formation of insoluble precipitates, inhibiting the photocurrent. The constructed PEC biosensing platform has high sensitivity for sEVs assay with a low detection limit (10 particles/μL). Furthermore, the machine learning model is used for intelligent analysis of sEVs, and can discriminate between cancer patients and healthy individuals with 90.0% accuracy and 100% specificity. This work combines PEC biosensor with machine learning offers a novel strategy for sEVs analysis and cancer intelligent diagnosis.