Smart optical biosensor for edible oil detection with machine learning integration.

Area of Science:

• Optoelectronics and Nanotechnology
• Biosensing Technologies
• Machine Learning Applications

Background:

• Photonic Crystal Fibers (PCFs) offer high sensitivity for material detection.
• Terahertz (THz) frequency range enables non-ionizing and label-free sensing.
• Accurate detection of edible oils is crucial for quality control and authenticity.

Purpose of the Study:

• To design and analyze a heptagonal core PCF biosensor for edible oil detection in the THz range.
• To evaluate key optical properties and optimize sensor performance.
• To develop a machine learning framework for efficient sensor simulation and prediction.

Main Methods:

• Numerical analysis using the finite element method (FEM) in COMSOL Multiphysics.
• Simulation of a heptagonal core PCF biosensor across 1.0-3.0 THz.
• Development and training of machine learning models (ANN, SVR, RF, XGBoost, Bagging SVR) using simulated data.

Main Results:

• Optimized sensor achieved a maximum relative sensitivity (RS) of 98.14% and minimum effective material loss (EML) of 0.0046758 cm⁻¹ at 2.2 THz.
• The Artificial Neural Network (ANN) model demonstrated superior predictive performance with R² of 0.9781 and MAE of 0.0636.
• The integrated FEM and ML approach significantly reduced simulation time for sensor optimization.

Conclusions:

• The proposed PCF biosensor shows high sensitivity for edible oil detection.
• Machine learning accelerates the design and optimization process for optical biosensors.
• This integrated framework offers an efficient pathway for developing advanced THz sensing devices.