Sub-Micro Optical Spectroscopy Bioassay System for COVID-19 Neutralizing Antibody Immunity Response

Ciao-Ming Tsai,Wei-Huai Chiu,Wei-Yi Kong,Chitsung Hong,Cheng-Hao Ko,Weileun Fang

DOI: https://doi.org/10.1109/jsen.2024.3424557

IF: 4.3

2024-01-01

IEEE Sensors Journal

Abstract:This study introduced an innovative system for detecting neutralizing antibodies from serum samples, using optical spectroscopy. This system utilizes a compact optical chip to capture spectral data from lateral flow immunoassay (LFIA). Researchers analyzed 265 serum samples sourced from MacKay Memorial Hospital in Taiwan. The presence of neutralizing antibodies was confirmed using the LFIA test kit. For validation, the study compared the results with those from the FDA-approved neutralizing antibody ELISA kit, considered the gold standard. Raw light intensity data were converted into reflectance data using a data processing procedure that prominently featured the Savitzky-Golay filter (S-G filter) for data refinement. To predict neutralizing antibody levels, the study leveraged three machine learning (ML) algorithms: SVM, random forest, and MLP. Notably, the MLP model outperformed the others in classifying antibody levels from LFIA spectra, achieving an accuracy of 94.3%. The study advanced quantitative analyses using both conventional image processing and spectral ML regressors, such as SVR, random forest, and MLPR. Spectral methods outperformed image processing, achieving R-2 values over 0.8. Moreover, a new hybrid model combining MLP classification with regression improved prediction accuracy (R-2 of 0.923). This research underscores the efficacy of combining spectroscopy and ML to detect neutralizing antibodies, paving the way for a potential alternative to traditional laboratory testing.