Highly Sensitive Perovskite Photoplethysmography Sensor for Blood Glucose Sensing Using Machine Learning Techniques.

Area of Science:

• Materials Science
• Biomedical Engineering
• Medical Devices

Background:

• Accurate non-invasive blood glucose monitoring remains a significant challenge in diabetes management.
• Existing methods often require invasive procedures, impacting patient compliance and comfort.

Purpose of the Study:

• To develop a novel near-infrared (NIR) photoplethysmography (PPG) sensor for non-invasive blood glucose (BG) monitoring.
• To evaluate the sensor's performance using a hybrid perovskite photodetector and machine learning algorithms.

Main Methods:

• Fabrication of a vapor-deposited mixed tin-lead hybrid perovskite photodetector for NIR-PPG sensing.
• Extraction of eleven features from PPG signals for analysis.
• Application of machine learning algorithms for blood glucose level prediction.

Main Results:

• The perovskite photodetector exhibited high detectivity (5.32 × 10^12 Jones) and a large linear dynamic range (204 dB).
• Accurate blood glucose prediction was achieved with a mean absolute relative difference (MARD) of 2.48%.
• The self-powered sensor demonstrated functionality using sunlight for outdoor healthcare monitoring.

Conclusions:

• The developed perovskite-based NIR-PPG sensor offers a promising non-invasive solution for blood glucose monitoring.
• The sensor's high performance and self-powered nature support its use in real-time and outdoor healthcare applications.
• This technology holds potential for early diabetes diagnosis and improved diabetes management.