Two-way detection of tuberculosis bacilli using fano resonance sensor and machine learning algorithms

A serious worldwide health concern, tuberculosis (TB) necessitates creative methods for quick and precise diagnosis. This paper presents the detection of Mycobacterium tuberculosis (mTB) using a two-way approach utilizing refractive index (RI) sensors and machine learning (ML) techniques. The approach is validated by adopting a Fano resonance (FR) sensor which consists of freshly prepared cerium oxide nanostructures (CNS) covered sensing slots. The CNS is prepared by using combustion technique and characterized through transmission electron microscopy (TEM) and x-ray diffraction (XRD). The RI values sensed through the sensor yield a high sensitivity and autocorrelation coefficient of 163 nm RIU ^−1 and ${R}^{2}=92.37\, \% $ , respectively. The microscopic sputum smear examination is used to detect mTB through five different ML classifiers. Among all, the random forest classifier provides the highest accuracy of 96%. According to preliminary findings, the suggested system provides a quick, affordable, and non-invasive diagnostic option with excellent accuracy, sensitivity, and specificity. The approach addresses important TB diagnostic problems and is more suitable for low-resource environments. This hybrid strategy can transform TB detection, enhance patient outcomes, and support TB control initiatives.