Aplicação da espectroscopia ATR-FTIR combinada a algoritmos de aprendizado de máquina para identificação da doença renal crônica em amostras de saliva

Abstract

Introduction: Chronic Kidney Disease (CKD) is a progressive condition that presents nonspecific symptoms in its early stages, making diagnosis difficult. This highlights the need for noninvasive, accessible, and effective diagnostic methods for CKD detection. Objectives: This study aimed to investigate the potential of attenuated total reflection- Fourier transform infrared (ATR-FTIR) spectroscopy of saliva, combined with machine learning algorithms, to discriminate patients with CKD from healthy individuals. Materials and Methods: Samples were collected at the Hemodialysis Unit of the Hospital das Clínicas at UFU. Participants followed prior restrictions on eating, drinking, smoking, and the use of oral products. Unstimulated whole saliva was collected in sterile tubes under standardized conditions, in accordance with international protocols for salivary biomarker studies. Spectra were acquired by ATR- FTIR and subjected to spectral preprocessing. Machine learning algorithms, including Principal Component Analysis (PCA), Support Vector Machine (SVM), Random Forest, AdaBoost, and Neural Networks, were used for statistical and predictive analysis. Results: Twenty-nine patients with CKD diagnosed according to the NKF criteria and 24 healthy controls, matched for age and sex, with no history of renal, systemic, or oral diseases, were included. The study identified ten salivary spectral biomarkers associated with CKD, mainly related to proteins, carbohydrates, lipids, and nucleic acids. One of the protein markers stood out in the amide III region, involving vibrations of the N–H and C–N bonds. Spectral classification achieved 98% accuracy, 100% sensitivity, and 96% specificity. Conclusion: This study demonstrated the potential of ATR-FTIR spectroscopy applied to saliva samples, combined with machine learning algorithms, as a promising approach for the non-invasive diagnosis of Chronic Kidney Disease (CKD). Keywords: Chronic kidney disease, Spectroscopy, Fourier Transform Infrared, Saliva