.png){kind=icon}
.png){kind=icon}
Return
High-contrast colorimetric NanoPCR enabled by dual-thiolated primers and machine learning for visual detection of hepatitis B virus DNA
.png){kind=icon}
OA
AI
P
H
N
N
L
T
P
N
N
L
M
A
T
DOI：10.1016/j.jsamd.2026.101218.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Point-of-care molecular diagnostics are frequently constrained by their reliance on sophisticated laboratory instrumentation. While a colorimetric assay based on gold nanoparticles (AuNPs) combined with polymerase chain reaction (nanoPCR) offers a visual alternative, its clinical translation is hindered by poor signal contrast and the double-stranded DNA's (dsDNA) inability to efficiently stabilize AuNPs. Here, we report a high-contrast, instrument-free colorimetric nanoPCR platform for hepatitis B virus (HBV) DNA detection that circumvents these biophysical constraints through an integration of urea-assisted chemical denaturation, dual-thiolated primer engineering, and a machine learning-based predictive framework. We demonstrate that 10 M urea efficiently dissociates dsDNA into single-stranded DNA (ssDNA) while maintaining AuNP stability. Simultaneously, dual-thiolated primers facilitate the formation of a dense, robust ssDNA corona on AuNP surfaces via strong Au-S covalent bonding. Systematic optimization of these interactions yields a pronounced and highly reproducible colorimetric response. To eliminate subjective visual bias, we employed a hybrid approach of image processing and machine learning (ML). Our findings reveal that the LAB (CIELAB) color space provides superior accuracy and consistency for colorimetric nanoPCR, with a Random Forest regression model exhibiting the best generalization performance. This ML-augmented analysis establishes a linear calibration between colorimetric data and viral load. Validation with 31 clinical samples showed total concordance with standard HBsAg assays. By removing the need for complex optics, this robust bio-computational methodology provides a scalable solution for decentralized molecular diagnostics in resource-limited settings.
Keywords:
Gold nanoparticles
nanoPCR
Colorimetric detection
Hepatitis B virus
Thiolated DNA
Urea denaturation
Plasmonic sensing
Machine learning
Journal
J
IF:
0
Papers:
152
Citations:
0
Organization
Cited Papers