Deep Learning-Enhanced Portable Chemiluminescence Biosensor: 3D-Printed, Smartphone-Integrated Platform for Glucose...

Area of Science:

• Analytical Chemistry
• Biomedical Engineering
• Artificial Intelligence

Background:

• Traditional glucose detection methods often require bulky, expensive equipment, limiting point-of-care applications.
• There is a need for portable, cost-effective, and accurate diagnostic tools for widespread healthcare access.

Purpose of the Study:

• To develop a novel, portable chemiluminescence (CL) sensing platform for selective glucose detection.
• To integrate deep learning and smartphone technology for enhanced accuracy and user-friendliness.
• To enable cost-effective point-of-care (PoC) diagnostics.

Main Methods:

• Development of a miniaturized CL sensor using low-cost, wax-printed micro-pads (WPµ-pads) on paper substrates.
• Utilization of a 3D-printed chamber and smartphone integration for data acquisition and analysis.
• Training and testing deep learning models (Random Forest, SVM, InceptionV3, VGG16, ResNet-50) on a dataset of 600 CL images.

Main Results:

• The platform achieved a linear glucose detection range of 10-1000 µM with a low detection limit of 8.68 µM.
• Deep learning models demonstrated superior accuracy and efficiency in CL image analysis.
• The sensor exhibited excellent stability, repeatability, and reliability in real-world evaluations.

Conclusions:

• The developed deep learning-powered CL sensing platform provides a cost-effective, portable, and accurate solution for glucose detection.
• This technology democratizes access to advanced diagnostics, with transformative potential for healthcare and biosensing.
• The platform's design facilitates seamless smartphone integration for user-friendly PoC applications.