Noninvasive blood glucose sensing by secondary speckle pattern artificial intelligence analyses

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Artificial Intelligence in Healthcare

Background:

• Diabetes mellitus is a global health concern requiring accurate blood glucose monitoring.
• Traditional finger-prick methods for blood glucose measurement are invasive and can be painful.
• Noninvasive blood glucose detection offers a more convenient and patient-friendly alternative.

Purpose of the Study:

• To develop and evaluate a noncontact speckle-based blood glucose measurement system.
• To enhance glucose detection accuracy using artificial intelligence (AI) data processing.
• To investigate the effect of an alternating current (AC) magnetic field on detection sensitivity and selectivity.

Main Methods:

• A system comprising a digital camera, AC magnetic field source, laser, and computer was employed.
• Laser-induced speckle patterns from the finger were recorded under an applied magnetic field.
• Machine learning (ML) and deep neural networks (DNNs) were utilized for blood plasma glucose level classification, with finger-prick data as reference.

Main Results:

• The noncontact system with AI demonstrated high accuracy in detecting blood plasma glucose levels.
• The ML approach outperformed tested DNNs due to highly selective and efficient data preprocessing.
• The AC magnetic field influenced the sensitivity and selectivity of the glucose detection.

Conclusions:

• The proposed noncontact sensing mechanism offers a potentially more accurate, convenient, and less painful method for blood glucose monitoring.
• This AI-driven, magnetic field-assisted system enables inexpensive and rapid blood glucose screening.
• Further research with diverse demographic data could optimize noninvasive glucose detection for improved diabetes management.