Smart SPR Biosensors: Machine Learning‐Augmented Dual‐Channel PCF Technology

We present a modern dual‐channel surface plasmon resonance (SPR) biosensor that combines advanced photonic crystal fiber (PCF) design with machine learning (ML) techniques to achieve high sensitivity and accurate prediction. The proposed sensor performs very well over a wide refractive index range of 1.33–1.45, showing a wavelength sensitivity of 32 000 nm/RIU, amplitude sensitivity of 2338 RIU −1 , and a spectral resolution of 3.13 × 10 −6 RIU. It operates effectively within the wavelength range of 550–1310 nm using gold as the plasmonic material and achieves a high figure of merit of 1333 due to strong light–plasmon interaction. The model design and simulation are performed by (FEM) finite element method method in COMSOL Multiphysics v6.2. After that, to automate the optimization process, ML models are applied in combination of random forest, gradient boosting, and XGBoost model; the combined model has R 2 value of 0.9771 based on training data from existing studies. By combining this three different ML models, the prediction error minimized than the single individuals’ performance. Overall, integrating PCF design with ML offers a promising path toward automated optimization for PCF‐based SPR sensor, resulting in highly sensitive and reliable biosensing for various biological and chemical sensing.