SPR-based refractive index sensor design with grated Au-ZnS for dengue detection using machine learning

In this paper, we propose a Surface Plasmon Resonance (SPR) fiber optic refractive index (RI) sensor. It consists of a multi-mode fiber (MMF) sensor with a bi-metallic nanostructure with Gold (Au) and Zinc Sulphide (ZnS) as the plasmonic sensing layer. The effect of the design parameter variation is evaluated to achieve the maximum wavelength sensitivity (WS). The testing sample's RI is taken from 1.34 to 1.40. We found that RI sensitivity varies non-linearly from 9000 to 18,000 nm/RIU. This sensor is additionally capable of detecting the dengue virus with a highest WS of 14285.71 nm/RIU for the infected haemoglobin. Moreover, the incorporation of machine learning (ML) techniques signifies a substantial progression in sensor design. The neural network (NN) model exhibited best performance, with a minimal mean square error (MSE) of 0.2828 and an R² value of 0.9998. These algorithms offer flexible adaptation and the derivation of insights based on data, enhancing the sensor's effectiveness in forecasting resonance wavelength (RW) for various analytes, as well as errors, classification metrics like accuracy, precision, F1_score, and computation duration. Performance metrics such as Detection Accuracy (DA), Figure of Merit (FOM), signal-to-noise ratio (SNR) and quality factor (QF) are also determined having maximum values 14.285 μm^-1, 158.73 RIU^-1, 5.55 and 79365.055 nm/RIU respectively. The recommended biosensor, which uses the NN model, performed better than all the other models and may be used to biological applications.