For noninvasive glucose sensing, we investigate the relationship between Raman spectrometer specification, especially collection efficiency, and glucose prediction performance in computer simulation. The computer simulation for synthetic Raman spectrum generation was designed to include glucose Raman signals at specific glucose concentrations, and Monte-Carlo simulation was used to simulate spectrum generation in a multi-layer model of skin tissue. We quantified the tendency between the decrease in collection efficiency and the deterioration of glucose prediction performance. This study can be important in the analysis of spectrometer specification for required performance for noninvasive glucose sensing, especially in a miniaturized portable or wearable device.
|