Machine Learning Classifiers for Noninvasive Glucose Detection Using A Single Wavelength Mid-infrared Photoacoustic Spectroscopy

Despite recent outstanding development, machine learning (ML) has not been utilized in mid-infrared and photoacoustic spectroscopy for noninvasive glucose detection. ML models can assist in improving the detection sensitivity to meet FDA requirements. Furthermore, the employment of ML can help to solve the complexity of detecting glucose in the presence of different blood components or at various environmental conditions. In noninvasive optical spectroscopy, ML models can be developed to distinguish glucose signals despite the variations in human skin properties for in vivo measurements. Different ML classification algorithms have been developed and employed to detect glucose levels using MIR-infrared photoacoustic spectroscopy. The photoacoustic system has been developed using a single wavelength quantum cascade laser, lasing at a glucose fingerprint of 1080 cm-1 for noninvasive glucose monitoring. Artificial skin phantoms have been prepared as test models for the system with different glucose concentrations, covering the normal and hyperglycemia blood glucose ranges. Support vector machine, narrow neural network, and medium neural network algorithms have achieved high prediction accuracy in classifying glucose levels.