Abstract
Arboviruses such as dengue, Zika, chikungunya and yellow fever share similar clinical manifestations, making differential diagnosis challenging, particularly in endemic regions with viral co-circulation. Conventional laboratory methods present important limitations, including cross-reactivity and reliance on specialized infrastructure. In this scenario, spectroscopic techniques such as Fourier-transform attenuated total reflectance infrared spectroscopy (ATR-FTIR) and Raman, when combined with artificial intelligence (AI), have shown promise by enabling rapid, low-cost analyses. This systematic review (PROSPERO CRD420251006929) aimed to qualitatively and quantitatively synthesize studies that applied infrared and Raman spectroscopy to clinical samples, supported by chemometric models. All 23 included studies investigated dengue patients, with only one also assessing Zika and chikungunya. Most studies employed Raman spectroscopy and multivariate analysis methods, such as principal component analysis with linear discriminant analysis (PCA-LDA, 39.1%) and partial least squares with discriminant analysis (PLS-DA, 21.7%), with an overall sensitivity of 0.94 (95% CI: 0.91-0.96) and overall specificity of 0.97 (95% CI: 0.95-0.98) for Raman spectroscopy. The risk of bias across all studies was high, according to PROBAST-AI development and evaluation assessment. These findings highlight the potential of spectroscopic approaches combined with AI for diagnosing arboviral infections, although further robust studies are required to support broader clinical validation.