Machine learning-advanced hydrogel-based transcription-coupled positive-feedback CRISPR/Cas13a analysis for novel microRNA signatures in differential diagnosis of non-small cell lung cancer

MicroRNAs (miRNAs) hold significant potential as biomarkers for the precise diagnosis of non-small cell lung cancer (NSCLC). However, miRNAs remain underused due to their low abundance, high heterogeneity, and complex matrix interference in liquid biopsies, as well as the requirement for specialized techniques. Herein, we devised a hydrogel-based transcription-coupled positive-feedback CRISPR/Cas13a (TCPFC) enhanced electrochemiluminescent (ECL) analyzer for advanced analysis of plasma miRNA signatures via machine learning (ML). Initially, three NSCLC-associated miRNA signatures (miR-203b, miR-450b, and miR-642a) were identified from plasma miRNA datasets and validated using RT-qPCR. An Au@ACZ-SA-PEG hydrogel emitter was engineered to deliver a robust ECL output on a glassy carbon electrode. Additionally, the TCPFC strategy utilized transcription-coupled positive-feedback CRISPR/Cas13a to achieve cascade signal amplification. Concurrently, collateral cleavage eliminated dopamine quenchers, thereby restoring the ECL signal ("OFF-ON") for readout and achieving attomolar-level detection. The integration of ML algorithms with the hydrogel-based TCPFC-ECL platform yielded differential diagnosis accuracies of 100.00% (train, n = 110) and 92.73% (test, n = 110), effectively distinguishing healthy controls (HC) from patients with stages I/II and III/IV NSCLC. Consequently, this biosensing platform demonstrates considerable promise as a practical tool for the precise diagnosis of NSCLC.