Aptamer-Directed Assembly of NSET Microneedle Patch for Early Detection of Acute Kidney Injury Facilitated by Deep Learning-Enabled Target Signal Extraction

Neutrophil gelatinase-associated lipocalin (NGAL) is a promising early biomarker for acute kidney injury (AKI), yet current biosensors lack the capability for minimally invasive, in situ, real-time analysis of interstitial fluid (ISF)-a key medium suitable for reliable continuous monitoring. Herein, we propose an aptamer-directed assembly strategy for the preparation of a nanocomposite microneedle patch. Briefly, a DNA1-decorated microneedle is successively modified with DNA-modified carbon dot (CD-DNA2) and gold nanoparticle (AuNP-DNA3, NGAL aptamer) via complementary base pairing. This triggers nanoparticle-based surface energy transfer between a luminescent donor (CD) and an absorbing acceptor (AuNP), quenching CD fluorescence. When the nanocomposite microneedle patch is transdermally applied to AKI mice for ISF sampling, recognition of NGAL by its aptamer triggers dissociation of the DNA2-DNA3 duplex, leading to AuNP release and CD fluorescence restoration. The captured irregular fluorescence images of the microneedle microarray with background interference are processed using a U-Net-based deep learning model for precise identification and signal extraction. This model establishes a strong positive correlation between fluorescence recovery and NGAL concentration (R2 = 0.9936), significantly outperforming the direct analysis of raw images (R2 = 0.5668). Therefore, this detection strategy precisely reveals elevated NGAL levels (∼80.7 ng/mL) as early as day 1 after modeling, signaling AKI onset 2-3 days earlier than clinical serum creatinine measurements using a commercial handheld analyzer or detection kit. In addition, this nanocomposite microneedle sensor possesses other advantages like minimal invasion and rather fast ISF collection (