Smart label-free SPR biosensing platform for hemoglobin and urine glucose detection via machine learning

Area of Science:

• Biomedical Engineering
• Nanotechnology
• Analytical Chemistry

Background:

• Accurate and rapid detection of biomarkers like hemoglobin and glucose is crucial for medical diagnostics.
• Existing biosensing methods can be time-consuming, costly, or require chemical labeling.
• Surface Plasmon Resonance (SPR) offers a label-free approach for real-time biomolecular detection.

Purpose of the Study:

• To design and numerically analyze a novel, label-free SPR sensor for quantifying hemoglobin in blood and glucose in urine.
• To optimize the sensor's design for enhanced sensitivity and performance using numerical simulations.
• To explore the application of machine learning for predicting sensor performance.

Main Methods:

• Design of a Surface Plasmon Resonance (SPR) sensor utilizing a silver (Ag) thin film on a prism.
• Numerical simulations employing the Finite Element Method (FEM) to optimize layer thickness and analyze sensor parameters.
• Application of machine learning models to predict sensor sensitivity based on structural and optical properties.

Main Results:

• The optimized SPR sensor design demonstrated a linear correlation between resonance wavelength shift and analyte refractive index.
• Achieved high performance metrics including sensitivity (288.29 °/RIU), QF (780.80), SNR (15.62), FoM (492.51), and CSF (539.20).
• Machine learning models effectively predicted sensor sensitivity, showcasing data-driven optimization capabilities.

Conclusions:

• The developed label-free SPR sensor is a rapid, economical, and sensitive tool for non-invasive biomedical applications.
• The sensor shows significant potential for use in diagnostics and point-of-care monitoring.
• Data-driven approaches, including machine learning, can accelerate the design and performance estimation of biosensors.