This multidisciplinary doctoral dissertation focused on development of untargeted ultrahigh-performance liquid chromatography–tandem mass spectrometry (UHPLC–MS/MS) -based metabolomics of serum and cerebrospinal fluid (CSF), integrated with supervised machine learning (ML), to identify metabolic alterations and candidate molecular features (MFs) for enhanced diagnostics of Lyme neuroborreliosis (LNB). Due to limitations of current biomarkers in Lyme borreliosis (LB) diagnostics, a paired, within-patient design was implemented to enhance sensitivity to disease-related changes. Serum profiling paired pretreatment LNB with 12 months after treatment samples. CSF profiling paired pretreatment LNB with samples three weeks after treatment initiation. LNB cohorts were companied with Borrelia antibody-negative, non-LNB controls in serum and CSF, and with other laboratory confirmed central nervous system infections in CSF, enabling assessment of disease specificity. Across matrices, pathway alterations were characterized by the tryptophankynurenine axis, broad lipid-signaling alterations (lysophospholipids, sphingomyelins, sphingoid bases, fatty acid amides, cyclic phosphatidic acids), and amino acid metabolism. Acetylcarnitine exhibited matrix-specific dynamics, elevated in pretreatment CSF, declining after treatment and higher in other CNS infections, but increased in post-treatment serum, indicating distinct compartmental responses. Several CSF MFs associated with chemokine CXCL13 concentration, thus supporting linkage to neuroinflammation. Supervised ML classifiers trained on serum profiles yielded strong discrimination with high sensitivity and specificity, and robust performance across comparisons. LNB induced detectable metabolic alterations, however, treated LNB retained a distinct serum signature versus non-LNB controls. Overlapping discriminatory features across models indicate an infection-linked signal rather than purely treatment-driven observations. UHPLC–MS/MS-based MF panels and ML classifiers emerge as potential complementary tools to enhance diagnostics and monitoring of disseminated LB.