Ultrasensitive electrochemical detection of clinically relevant genetic mutations via nanoparticle surface localized amplification and machine learning.

Genetic mutation detection is crucial for early disease diagnosis, prognosis assessment, and the implementation of precision medicine, as it enables the identification of key biomarkers that directly influence treatment decisions and patient outcomes. Random Forest classifier can help select critical mutations by prioritizing informative genetic mutations for target-recognition probe design. We develop an ultrasensitive and rapid genetic mutation detection platform integrated with the nanoparticle surface localized amplification (nSLAM) system and machine learning that shows superior sensitivity and shorter detection time compared to conventional qPCR. We carry out a proof-of-concept study using nucleic acids of SARS-CoV-2, which can be detected at concentrations as low as 0.13 aM (0.08 copies/μL) and the cycle threshold was only 8.38 cycles via advanced nSLAM. We validate the mutation detection system by identifying single nucleotide polymorphisms (SNPs), insertions, and deletions using various strains of SARS-CoV-2, including the wild-type, Alpha, Beta, Delta, and Omicron variants. This approach enables discrimination between closely related and clinically significant genetic mutations, thereby supporting clinical decisions from prevention to treatment.