Classification of trimethylamine-N-oxide, a cardiometabolic disease biomarker, by Raman spectroscopy and support vector machines

Abstract In the present study, we investigated the ability of micro-Raman spectroscopy to identify low concentrations of trimethylamine-N-oxide (TMAO) mixed in synthetic urine composed of water, sodium chloride, urea, and creatinine using a support vector machine (SVM) as a discrimination tool to differentiate the Raman spectra of the different concentrations of TMAO. TMAO is a novel biomarker associated with cardiovascular diseases, kidney diseases, and complications of type 2 diabetes. We obtained the Raman spectra of four different concentrations of TMAO. The spectra were filtered before being classified using principal component analysis combined with the SVM method. We identify the spectral window that goes from 800 to 870 cm −1 where TMAO presents Raman activity in the synthetic urine mixture without the intervention of Raman activity of another molecule. We predicted the different concentrations of TMAO in the synthetic urine until 1 ppm (13.21 µ M) of TMAO, getting an accuracy of classification greater than 70% indicated by the confusion matrix, and the area under the receiver operating characteristic curve of 0.86 for 1 ppm (13.31 µ M) and 10 ppm (133.13 µ M) concentration. This study showed that Raman spectroscopy combined with SVM has the potential to detect low concentrations of TMAO in urine.