Non-enzymatic colorimetric glucose detection based on Au/Ag nanoparticles using smartphone and machine learning

Area of Science:

• Nanotechnology
• Biomedical Engineering
• Machine Learning

Background:

• Conventional enzyme-based glucose quantification methods face limitations including high cost, specific temperature requirements, short shelf life, and poor stability.
• A portable, rapid, cost-efficient, and highly sensitive platform is crucial for diabetes management and healthcare.
• Non-enzymatic glucose detection offers a promising alternative to overcome the drawbacks of enzymatic methods.

Purpose of the Study:

• To develop a portable, non-enzymatic glucose quantification platform.
• To integrate gold (Au) and silver (Ag) nanoparticles (NPs) with a smartphone application utilizing machine learning.
• To enable rapid, cost-effective, and sensitive glucose detection for diabetes healthcare.

Main Methods:

• Utilized small and large gold (Au) and silver (Ag) nanoparticles (NPs) for glucose reaction.
• Captured color changes using a smartphone camera to generate a dataset for machine learning.
• Developed a user-friendly smartphone application "GlucoQuantifier" with cloud-based machine learning classification.
• Employed linear discriminant analysis (LDA) as the machine learning classifier.

Main Results:

• The reaction of Au/Ag NPs with glucose produced a measurable color change.
• A dataset was created from captured color changes for training machine learning classifiers.
• The "GlucoQuantifier" smartphone application facilitated remote analysis via a cloud system.
• Linear discriminant analysis achieved the highest classification performance of 93.63% with small Au/Ag NPs.

Conclusions:

• The developed portable platform effectively quantifies glucose non-enzymatically.
• Integration of Au/Ag NPs, machine learning, and a smartphone application is a viable approach for glucose monitoring.
• This method offers a cost-efficient, sensitive, and stable alternative for diabetes healthcare.