Comment on “Quantification of glycated hemoglobin and glucose in vivo using Raman spectroscopy and artificial neural networks”

In the paper by González-Viveros et al. [1], the authors demonstrated combination of conventional Raman (CR) spectroscopy and machine learning methods for the discriminating of healthy, prediabetes, and type 2 diabetes (T2D) patients based on the glycated hemoglobin (HbA1c) estimations. The authors utilize cost-effective Raman spectrometer (with signal-to-noise ratio about 3) for in vivo skin measurements. The authors demonstrated extremely high performance of the proposed technique; however, the presented results may be treated incorrectly due to the overestimation of the proposed classification models. …