Despite the reduced impact of COVID-19 due to widespread vaccination and improved treatments, a critical need remains for accessible, scalable, and rapid screening tools to address current and future infectious disease threats. Hyperspectral Imaging (HSI) may be such a tool, but the current limited availability of data from COVID-19 positive individuals hinders traditional supervised learning approaches. To overcome this, we designed a novel framework that integrates HSI with Artificial Intelligence (AI) analysis for detecting COVID-19 status from images of fingertips, thereby demonstrating infection detection through biometric data. By analyzing high-dimensional spectral signatures from the fingertip, the approach identifies distinctive patterns linked to physiological changes caused by the virus. In this pilot study, a Support Vector Machine (SVM) and a Logistic Regression classification algorithm demonstrated high accuracy in classifying HSI images, underscoring the potential of hyperspectral features for non-invasive, real-time health monitoring, even with the limitations of a small dataset. We introduced the first publicly available dataset of HSI images from COVID-19-positive individuals. This contribution sets a foundation for advancing biometric spectral imaging in biomedical research and AI-powered diagnostics. The datasets used during this study is available from the corresponding author upon reasonable request.
Marasco et al. (Thu,) studied this question.