Abstract
The combination of mid-infrared and photoacoustic spectroscopy has shown promising developments as a substitute for invasive glucose detection technology. A dual single-wavelength quantum cascade laser system has been developed using photoacoustic spectroscopy for noninvasive glucose monitoring. Biomedical skin phantoms with similar properties to human skin have been prepared with blood components at different glucose concentrations as test models for the setup. The detection sensitivity of the system has been improved to ± 12.5 mg/dL in the hyperglycemia blood glucose ranges. An ensemble machine learning classifier has been developed to predict the glucose level in the presence of blood components. The model, which was trained with 72,360 unprocessed datasets, achieved a 96.7% prediction accuracy with 100% of the predicted data located in zones A and B of Clarke's error grid analysis. These findings fulfill both the US Food and Drug Administration and Health Canada requirements for glucose monitors.