Deep Learning-Assisted G4 Nanowire-Enhanced Carbon Dot Biosensor for Exosomal LncRNA Artificial Intelligence Diagnosis.

Exosomal long noncoding RNAs (lncRNA) have significant potential as a biomarker for early cancer diagnosis. Accurate and sensitive detection of this abnormal expression remains challenging. Herein, we develop an innovative dual-mode photoelectrochemical/ratiometric fluorescence (PEC/RFL) biosensor based on G-quadruplex nanowire (G4 NW)-enhanced copper/cerium codoped carbon dots (CuCe-CDs) for exosomal lncRNA assays, using deep learning for artificial intelligence diagnosis of exosomal lncRNA. Target exosomal lncRNA triggers three-dimensional bipedal DNAzyme walkers to form numerous CuCe-CD-labeled G4 NWs (G4-CD NWs) with the joint help of K and bifunctional Mg. Since G4 NWs not only load abundant CuCe-CDs but also enhance the nanozyme activity and FL intensity of CuCe-CDs, a dual-mode PEC/RFL biosensor for the highly sensitive and accurate assay of exosomal lncRNA is built. Moreover, deep learning is utilized to excavate the potential pattern hidden in data of our established method and successfully distinguishes lung cancer in people with a diagnostic accuracy of 95.7%. We think that the developed biosensor combined with deep learning offers an innovative tool for exosomal lncRNA assays and early intelligent diagnosis of cancer.