Non-Invasive In-Vivo Detection of Random Blood Glucose using Photoacoustic Spectroscopy

Area of Science:

• Biomedical Engineering
• Spectroscopy
• Medical Diagnostics

Background:

• Random blood glucose monitoring is crucial for detecting hyperglycemia and preventing complications.
• Current methods often require invasive procedures, posing challenges for frequent testing.
• Non-invasive techniques are highly sought after for improved patient compliance and accessibility.

Purpose of the Study:

• To develop and validate a non-invasive in vivo random blood glucose estimation system.
• To utilize Photoacoustic Spectroscopy (PAS) for accurate glucose level detection.
• To enhance the accuracy of glucose estimation by incorporating Body Mass Index (BMI).

Main Methods:

• A portable system was designed to acquire pre-processed Photoacoustic (PA) signals from a finger using a 905 nm laser.
• Partial Least Square Regression (PLSR) was implemented, using PA signal features and BMI for calibration.
• The system's performance was evaluated on 32 subjects using Clarke Error Grid and Bland-Altman Plots.

Main Results:

• The non-invasive method demonstrated promising accuracy in random blood glucose estimation.
• Key performance metrics included Mean Absolute Relative Difference (8.9%), Mean Absolute Difference (13.75 mg/dl), and Root Mean Square (16.13 mg/dl).
• The inclusion of BMI improved the overall accuracy of the glucose estimation model.

Conclusions:

• The proposed non-invasive random blood glucose monitoring technique using PAS is effective.
• This technology can significantly advance point-of-care applications for frequent glucose monitoring.
• The system holds particular promise for diabetic management in resource-limited settings.