tetano
Editor, Senior Moderator
ACS Omega
. 2022 Dec 8;7(50):47357-47371.
doi: 10.1021/acsomega.2c06786. eCollection 2022 Dec 20.
Robustness of FTIR-Based Ultrarapid COVID-19 Diagnosis Using PLS-DA
Sreejith Remanan Pushpa[SUP] 1 2 [/SUP], Rajeev Kumar Sukumaran[SUP] 3 2 [/SUP], Sivaraman Savithri[SUP] 1 2 [/SUP]
Affiliations
Abstract
The World Health Organization (WHO) declared the Omicron variant (B.1.1.529) of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the pathogen responsible for the Coronavirus disease 2019 (COVID-19) pandemic, as a variant of concern on 26 November 2021. By this time, 42% of the world's population had received at least one dose of the vaccine against COVID-19. As on 1 October 2022, only 68% of the world population got the first dose of the vaccine. Although the vaccination is incredibly protective against severe complications of the disease and death, the highly contagious Omicron variant, compared to the Delta variant (B.1.617.2), has led the whole world into more chaotic situations. Furthermore, the virus has a high mutation rate, and hence, the possibility of a new variant of concern in the future cannot be ruled out. To face such a challenging situation, paramount importance should be given to rapid diagnosis and isolation of the infected patient. Current diagnosis methods, including reverse transcription-polymerase chain reaction and rapid antigen tests, face significant burdens during a COVID-19 wave. However, studies reported ultrarapid, reagent-free, cost-efficient, and non-destructive diagnosis methods based on chemometrics for COVID-19 and COVID-19 severity diagnosis. These studies used a smaller sample cohort to construct the diagnosis model and failed to discuss the robustness of the model. The current study systematically evaluated the robustness of the diagnosis models trained using smaller (real and augmented spectra) and larger (augmented spectra) datasets. The Monte Carlo cross-validation and permutation test results suggest that diagnosis using models trained by larger datasets was accurate and statistically significant (Q [SUP]2[/SUP] > 99% and AUROC = 100%).
. 2022 Dec 8;7(50):47357-47371.
doi: 10.1021/acsomega.2c06786. eCollection 2022 Dec 20.
Robustness of FTIR-Based Ultrarapid COVID-19 Diagnosis Using PLS-DA
Sreejith Remanan Pushpa[SUP] 1 2 [/SUP], Rajeev Kumar Sukumaran[SUP] 3 2 [/SUP], Sivaraman Savithri[SUP] 1 2 [/SUP]
Affiliations
- PMID: 36570187
- PMCID: PMC9773962
- DOI: 10.1021/acsomega.2c06786
Abstract
The World Health Organization (WHO) declared the Omicron variant (B.1.1.529) of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the pathogen responsible for the Coronavirus disease 2019 (COVID-19) pandemic, as a variant of concern on 26 November 2021. By this time, 42% of the world's population had received at least one dose of the vaccine against COVID-19. As on 1 October 2022, only 68% of the world population got the first dose of the vaccine. Although the vaccination is incredibly protective against severe complications of the disease and death, the highly contagious Omicron variant, compared to the Delta variant (B.1.617.2), has led the whole world into more chaotic situations. Furthermore, the virus has a high mutation rate, and hence, the possibility of a new variant of concern in the future cannot be ruled out. To face such a challenging situation, paramount importance should be given to rapid diagnosis and isolation of the infected patient. Current diagnosis methods, including reverse transcription-polymerase chain reaction and rapid antigen tests, face significant burdens during a COVID-19 wave. However, studies reported ultrarapid, reagent-free, cost-efficient, and non-destructive diagnosis methods based on chemometrics for COVID-19 and COVID-19 severity diagnosis. These studies used a smaller sample cohort to construct the diagnosis model and failed to discuss the robustness of the model. The current study systematically evaluated the robustness of the diagnosis models trained using smaller (real and augmented spectra) and larger (augmented spectra) datasets. The Monte Carlo cross-validation and permutation test results suggest that diagnosis using models trained by larger datasets was accurate and statistically significant (Q [SUP]2[/SUP] > 99% and AUROC = 100%).