Detection Limits of Blood Metabolites at Physiological Concentrations Using Benchtop 1$$ {}^1 $$H NMR

Commercial low-field (LF) magnetic resonance spectroscopy (MRS) offers a route to rapid and repeated in vivo metabolite tracking; however, its sensitivity and interpretability at physiological concentrations remain underexplored. Here, we evaluate the performance of an 80-MHz benchtop nuclear magnetic resonance (NMR) spectrometer (Bruker Fourier 80) across several key blood metabolites at physiological concentrations, ranging between 0.05 and 10.0 mmol/L. We characterise the relationship between metabolite concentration, acquisition time and signal-to-noise ratio (SNR) for multiple pulse sequences and assess how the choice of SNR definition influences reported detection and quantification thresholds. Metabolites present at millimolar levels, such as glucose and lactate, were readily detectable within 20 s, with the water-suppressing wet pulse sequence yielding the highest SNR at fixed acquisition time. Concentration differences were also readily distinguishable. In contrast, submillimolar metabolites such as citrate require over 4 min to reach conventional detection thresholds, constraining their applicability to rapid metabolite tracking via MRS. To address interpretability in low-SNR data, we introduce a template-fitting approach based on simulated standards from CcpNMR AnalysisAssign, which stabilised relative metabolite quantification under low-SNR conditions. These results establish quantitative benchmarks for LF NMR metabolite detection and demonstrate how simulation-assisted analysis can extend its utility. These findings inform both the selection of target metabolites and optimisation strategies for emerging commercial in vivo MRS devices, including the DigiScan^TM finger-scanner, supporting their development as accessible tools for real-time metabolic tracking in personalised healthcare.