Calibration of a non-invasive Raman spectroscopy device realized using pre-collected clinical data

This study investigates the use of Raman spectroscopy combined with multivariate analysis for non-invasive glucose monitoring, focusing on reducing calibration requirements. Building on recent advances by RSP Systems, the approach uses pre-collected clinical data to enable a shorter and more efficient calibration process.

A regression model is trained on paired Raman spectra and reference glucose values and then personalized using 10 calibration points collected within four hours. The method was tested in a clinical study with 50 participants with type 2 diabetes over two days, where glucose variations were induced through meal challenges. The results demonstrate that accurate measurements can be achieved with minimal calibration, supporting the potential of Raman spectroscopy in non-invasive diabetes management.