An accurate assay was developed by integrating a novel lateral flow immunoassay (LFIA) design, smartphone-based photoluminescence detection, and computer-aided analysis using machine learning algorithms for the quantitative measurement of cortisol levels in human saliva samples. The unique LFIA strip incorporates a photoluminescent film, which enables photoluminescence detection without an external light source, beneath a nitrocellulose membrane. A smartphone is used to capture images of the LFIA test strips, and specific regions in the captured images are analyzed. The digitized data are then processed using a computer. Machine learning algorithms were employed to interpret the data and quantify cortisol levels in saliva samples obtained from 14 volunteers. The developed assay was shown to be highly accurate, and a low average difference of 18.12% was observed between the predicted cortisol levels and those measured using an established enzyme-linked immunosorbent assay (ELISA) in real saliva samples. The assay has a calculated limit of detection of approximately 139 pg/mL. Furthermore, the strong correlation (r = 0.935) between the results of the developed assay and the ELISA results supports its validity.
Keywords: Competitive lateral flow immunoassay; Cortisol; Human saliva; Photoluminescence detection; Smartphone.
Copyright © 2024 Elsevier B.V. All rights reserved.