Detection of Antibodies for COVID-19 from Reflectance Spectrum Using Supervised Machine Learning

Since the coronavirus disease 2019 occurred, the lateral flow immunoassay (LFIA) test strip has become a global testing tool for convenience and low cost. However, some studies have shown that LFIA strips perform poorly compared to other professional testing methods. This paper proposes a new method to improve the accuracy of LFIA strips using spectral signals. A spectrochip module is applied to disperse the reflected light from the LFIA strips. The obtained spectral signals will be used for supervised machine learning. After training, the trained model has 93.8% accuracy compared to the standard test. This result indicated that the evaluation method based on the spectrum of LFIA strips could enhance the detection performance.