Vibrational spectroscopy of body fluids combined with machine learning for the early diagnosis of cystic echinococcosis.

Cystic echinococcosis (CE) is a globally prevalent zoonotic parasitic disease. Due to the covert symptoms and the inadequacies of screening technologies, accurate early diagnosis is crucial. This study explores the feasibility of employing body fluid vibrational spectroscopy techniques combined with machine learning algorithms for the early diagnosis of CE. Specifically, serum and urine samples from both early-stage CE and normal control mouse models were collected and analyzed using surface enhanced Raman spectroscopy (SERS) and Fourier transform infrared spectroscopy (FTIR). Four machine learning algorithms were employed to develop diagnostic models based on the spectroscopic data. The results indicate that the support vector machine (SVM) model achieved the optimal classification results, with diagnostic accuracies of 93.2 % for serum SERS and 95.5 % for serum FTIR datasets. The performance differences between the two spectroscopic techniques were not statistically significant (p > 0.05). However, both techniques yielded suboptimal classification results for urine samples, with diagnostic accuracies below 80 %. Additionally, analysis of serum vibrational spectra using a linear SVM-based importance map revealed potential early CE biomarkers, including purine metabolites (uric acid, hypoxanthine), protein-associated bands (amide I, CH3), and a lipid-related CH2. These findings suggest that the strategy of combining serum vibrational spectroscopy with machine learning holds broad prospects for application in the early diagnosis of CE.