Detecção e tipagem de arbovírus (dengue, zika e chikungunya) por infravermelho em conjunto com técnicas de análise multivariada

Abstract

The objective of the studies reported in this dissertation is to evaluate the use of infrared spectroscopy in conjunction with chemometric techniques of multivariate analysis, as a new tool for detection and typing of arboviruses present in clinical samples. In this dissertation there are 4 articles: 1 review article and 3 research articles. The review article is a survey of the main techniques spectroscopic and multivariate analysis techniques used in studies in the field of virology in the last 10 years, as well as the advantages of these techniques against standard techniques. In the first research article, multivariate models based on discriminant analysis were constructed with the objective of quantitatively discriminating the DENV-3 serotype present in four different concentrations in serum and blood samples. In the second study, variable selection techniques were applied with the objective of discriminating infected serum and blood samples in the laboratory, and also to predict which serotype is responsible for the infection. In the third study, the ability of the technique to discriminate between 4 groups of samples: dengue (blood samples from patients diagnosed with dengue), Chikungunya (blood samples from patients diagnosed with Chikungunya), Zika (blood samples from patients diagnosed with Zika) and healthy (blood samples from healthy volunteers). The multivariate analysis algorithms used were Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA), Successive Projections Algorithm-Linear Discriminant Analysis (SPA-LDA) and Genetic Algorithm-Linear Discriminant Analysis (GA-LDA). The performance of the technique was evaluated through calculations of sensitivity, specificity, positive and negative predictive values, Youden index and positive and negative likelihood ratio. The results were encouraging, and showed that the spectroscopy used in conjunction with multivariate analysis techniques has the potential to detect and identify the variations caused by the presence of dengue virus in biological samples, and to provide a fast result in comparison to the diagnostic techniques used in clinical routines.