Rapid, On-Site SARS-CoV‑2 Variant Detection and Differentiation Using GLAD-Pristine Silver Nanorod Arrays and Machine...

The rapid evolution of viruses like SARS-CoV-2 and its emerging variants requires advanced diagnostic techniques for effective pandemic management. This study introduces a machine learning (ML)-based surface-enhanced Raman scattering (SERS) methodology for the precise strains, substrains-based detection, and differentiation of SARS-CoV-2 in clinical nasopharyngeal swab samples. Pristine silver nanorod substrates fabricated using the glancing angle deposition method were used for the sensitive detection of the wildtype, kappa, delta, and omicron variants of SARS-CoV-2. Also, four different substrains of omicron strain (BA.1, BA.2, BA.5, and XBB) were detected and distinguished using the developed platform. A detection limit of around 100 pfu/mL was established for the 4 variants and 4 covariants of the COVID-19 virus. However, challenges arise in the clinical samples due to the subtle spectral variations between closely related variants of SARS-CoV-2. To address this, ML models were integrated with SERS data to discern intricate patterns, enhancing the differentiation capabilities. In this study, we employed two different classifiers, support vector machine (SVM) and bidirectional long short-term memory network (BiLSTM), for identifying the targeted variants from nasopharyngeal swabs of 122 positive patients, who were previously identified as the specific strain of SARS-CoV-2 through next-generation sequencing. The SVM classifier achieved an accuracy of 88.79% (95% CI: 83.18-94.39) and the BiLSTM model 85.98% (95% CI: 79.44-92.52) for variant classification on the validation set. Further, the models were validated on a blind test set, where an accuracy of 74.77% (95% CI: 67.29-83.18) and 70.09% (95% CI: 62.59-78.50) was achieved, respectively. Furthermore, the SVM classifier, trained for subvariant classification of omicrometer variants, obtained an accuracy of 95.83% (95% CI: 87.50-100.00) on the validation set. This integrated ML-SERS approach not only enhances detection efficacy but also provides on-site disease prediction ability, which will be immensely helpful for disease management.