Design and Analysis of 2D Photonic Biosensor with ML for Respiratory Virus Detection

Abstract

In this study, we have designed and integrated a novel photonic biosensor with a Machine Learning approach for the detection of five common respiratory viruses. The biosensor has been developed using a two-dimensional hexagonal photonic crystal defect structure, which has been designed through the use of Finite Difference Time Domain (FDTD) and Plane Wave Expansion (PWE) techniques to monitor wavelength shifts during virus detection. The analytes have been efficiently captured within the sensor's pores to optimize performance. The uniqueness of our sensor has been demonstrated through enhanced sensitivity (584nm/RIU) and a remarkable quality factor (9734). We have employed the naïve Bayes classifier Machine Learning algorithm to achieve accurate virus detection, leveraging parameters that have been extracted from the sensor design. Our integrated sensor and classifier have provided robust classification of virus types, outperforming existing methods, and yielding highly accurate results. Furthermore, to enhance user accessibility, we have developed a graphical user interface for intuitive result interpretation.