Combined Biospectroscopy with Multivariate Analysis for the Differential Diagnosis of Leptospirosis Disease: A Pilot Study

Currently, reported estimates indicate that there are 1.03 million annual cases of leptospirosis, with 58,900 deaths worldwide. The Pan American Health Organization (PAHO) in 2022 indicates that the febrile symptoms of leptospirosis are similar to other diseases such as influenza and dengue, among other diseases. Therefore, an early and accurate diagnosis of leptospirosis is essential for adequate and rapid treatment. In this study, attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy combined with multivariate analysis techniques was employed to classify between healthy controls and leptospirosis positive samples. Spectra from 113 dried blood plasma samples from patients (<i>n</i> = 43 leptospirosis, and <i>n</i> = 80 controls) were analyzed by linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), and support vector machine (SVM), combined with genetic algorithm (GA), successive projections algorithm (SPA), and principal component analysis (PCA) for feature selection/extraction. The GA-LDA model showed good sensitivity at 85.71% and specificity at 100% to discriminate both classes. Cross-validation was also performed, with the Venetian blinds method, showing the GA-LDA-CV model with better results in percentage sensitivity (76%), specificity (91%), precision (83%) and acurracy (86%), <i>F</i>-score (0.828), and AUC (0.835), when compared with the results of the preliminary GA-LDA model, and it is possible to rule out the possibility of overfitting of the preliminary GA-LDA model. Suggesting that combined ATR-FTIR spectroscopy with multivariate analysis has great potential to detect biochemical variations produced by the Leptospira pathogen in the blood of infected patients. These findings emerge as a new potential tool for improving leptospirosis diagnosis in the future using a rapid, low-cost, and minimally invasive methodology.