On-Site Precise Diagnosis of COVID-19 Using an Attention-Based PLS-1D-CNN Model With Biomolecular Infrared Signatures
- 1 August 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (4) , 2982-2997
Abstract
During the early stage of respiratory virus outbreaks, such as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the efficient utilize of limited nasopharyngeal swabs for rapid and accurate screening is crucial for public health. In this study, we present a methodology that integrates attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR) with the adaptive iteratively reweighted penalized least squares (airPLS) preprocessing algorithm and a channel-wise attention-based partial least squares one-dimensional convolutional neural network (PLS-1D-CNN) model, enabling accurate screening of infected individuals within 10 minutes. To evaluate the quality of spectral signals obtained under different experimental protocols, a biomolecular importance (BMI) evaluation method was introduced to quantify the contribution of virus related biomolecules during feature extraction. Two cohorts of nasopharyngeal swab samples, comprising 126 and 112 samples from suspected SARS-CoV-2 Omicron variant cases, were collected at Beijing Youan Hospital for verification. For the ATR-FTIR spectra in Cohort 2, which exhibited higher BMI values, the airPLS algorithm was applied for baseline correction, followed by classification using the channel-wise attention based PLS-1D-CNN model. The proposed diagnostic system achieved a screening sensitivity of 96.24% and a specificity of 97.14%, outperforming previously reported methods. To further examine the generalizability and robustness of the proposed approach, an additional dataset (Cohort 3) was collected and validated under varying experimental conditions. The system maintained a screening sensitivity of approximately 96.00%, and the consistent results across different settings confirm the effectiveness and reliability of the proposed diagnostic method.Keywords
Funding Information
- National Key R&D Program of China (2022YFE0102300)
- National Medical Research Council (MOH-000927)
- A*Star (A2090b0144)
- National Research Foundation (NRF-CRP29-2022-0003)
- National Natural Science Foundation of China (62203307)
This publication has 35 references indexed in Scilit:
- A comprehensive diagnostic platform leveraging voice-control feature for rapid SARS-CoV-2 detection using reverse transcription loop-mediated isothermal amplificationSensors and Actuators B: Chemical, 2025
- On the feasibility of Vis–NIR spectroscopy and machine learning for real time SARS-CoV-2 detectionSpectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2023
- A novel random multi-subspace based ReliefF for feature selectionKnowledge-Based Systems, 2022
- Effect of variable selection algorithms on model performance for predicting moisture content in biological materials using spectral dataAnalytica Chimica Acta, 2022
- Noninvasive Diagnostic for COVID-19 from Saliva Biofluid via FTIR Spectroscopy and Multivariate AnalysisAnalytical Chemistry, 2022
- Infrared Based Saliva Screening Test for COVID‐19Angewandte Chemie, 2021
- Accuracy of a nucleocapsid protein antigen rapid test in the diagnosis of SARS-CoV-2 infectionClinical Microbiology and Infection, 2021
- Ultrarapid On-Site Detection of SARS-CoV-2 Infection Using Simple ATR-FTIR Spectroscopy and an Analysis Algorithm: High Sensitivity and SpecificityAnalytical Chemistry, 2021
- Fast Screening and Primary Diagnosis of COVID-19 by ATR–FT-IRAnalytical Chemistry, 2021
- Structure, Function, and Antigenicity of the SARS-CoV-2 Spike GlycoproteinCell, 2020