Desenvolvimento de testes para detecção da COVID-19 empregando análise multivariada em associação com espectrofluorimetria ou NIR portátil

Abstract

The outbreak of Coronavirus Disease (COVID-19) triggered a global health emergency, resulting in millions of deaths. During this period, efforts to control the spread of the virus and resume activities were significantly impacted by the availability of supplies and skilled labor necessary for accurately diagnosing the population. The challenges faced by many countries in conducting mass diagnostics through techniques such as RT-PCR highlight the ongoing need to seek alternative testing strategies that are effective, less reliant on high-cost raw materials, and scalable for widespread application. In this study, we explore the use of spectroscopic analysis of blood serum samples combined with multivariate analysis techniques for COVID-19 diagnosis. Two approaches were tested:

the first using molecular fluorescence spectroscopy, and the second employing near- infrared spectroscopy with a portable device. In each approach, spectra from samples of

patients with positive and negative COVID-19 diagnoses were obtained and used to train and validate classification models based on SIMCA, DD-SIMCA, PCA-DA, and PLS-DA modeling techniques. Among the tested approaches, analyses with excitation at 280 nm allowed the development of a PCA-DA classification model with a sensitivity of 0.98, selectivity of 0.84, and efficiency of 95%. This model demonstrated high efficiency in detecting seropositive samples, with only minor difficulty in classifying seronegative samples. In the analyses using excitation-emission matrix spectra, the DD-SIMCA model generated through PARAFAC decomposition best met the desired objectives, achieving an overall efficiency of 96%, sensitivity of 0.98, and selectivity of 0.92. Finally, a new method is presented for classifying blood serum samples as positive or negative for COVID-19 infection, using fluorescence spectroscopy analysis combined with multivariate analysis techniques. The use of spectroscopy in the analysis of serum samples proved to be effective in generating discrimination models that can be used to analyze new samples and determine whether the recorded signal is characteristic of a patient infected with SARS-CoV-2. The findings of this study proved to be promising and may contribute to the development of auxiliary diagnostic methods that are cost-effective, easy to perform, and provide rapid and reliable results.