Foodborne pathogens cause diseases that affect both human health and the economy. Food and water are an essential part of life, and their contamination by bacteria poses a serious threat to human health and lifestyles ^[1][2][1,2]. Food-industry operators require rapid testing devices to monitor the quality of food for the presence of pathogenic bacteria [3]. Every year, millions of people worldwide get infected by contaminated food and water by microorganisms that cause various diseases. It is estimated that around 600 million foodborne diseases occur annually, with a mortality rate of 420,000 [4]. The Centers for Disease Control and Prevention estimated that approximately 2.5 billion people lack access to healthy and safe water in developing nations. Every year, more than 2.2 million mortality rates are reported due to waterborne diseases [5]. In China, a summary of studies estimated that the prevalence of pathogens in the food was 8.5% [6]. Controlling food safety is a persistent and challenging task in China due to the diversity of foods and food-production industries. Microbes such as Escherichia coli, Staphylococcus aureus, Salmonella enterica, and Listeria monocytogenes are usually highly infectious, and the presence of few colony-forming units (CFUs) can cause disease [7]. Therefore, it is crucial to identify pathogens at an initial stage with highly sensitive techniques to avoid diseases and outbreaks ^[8][9][8,9].

Conventional bacteria detection methods include cultivation, Gram staining, and biochemical analysis. These methods are reliable and have made great contributions to pathogen detection, but they are time-consuming and often take 2 to 3 days or more, which is not convenient for the rapid identification of microbes. The current techniques that are being used as clinical methods include polymerase chain reaction (PCR), enzyme-linked immunosorbent assay (ELISA), and matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry ^{[10][11][12][13][14]}[10,11,12,13,14]. The latest techniques have revolutionized the field of diagnosis due to high sensitivity and specificity. However, some of the drawbacks associated with current clinical diagnostic techniques include high cost, controlled sampling conditions, being laborious and time-consuming, and the requirement for a skilled operator. Therefore, further research progress is required to develop a user-friendly, portable, and economic diagnostic development system. The development of simple techniques and systems for the rapid, economical, and on-site analytical approaches are essential in public health safety, medical diagnostics, and food safety ^{[15][16][17][18][19]}[15,16,17,18,19].

Surface-enhanced Raman spectroscopy is a real-time detection method that depends on the inelastic scattering of excitation light and molecular resonance. SERS inherits the significant chemical fingerprint information on Raman spectroscopy and enhances sensitivity using plasmon-enhanced excitation and scattering. In SERS, the inelastic scattering from the molecules is greatly enhanced by a factor of up to 10⁸ when the molecules are adsorbed onto corrugated metal surfaces. The method can rapidly and efficiently detect a range of chemical structures and material compositions with high accuracy and reproducibility. These advantages make SERS a very promising tool for developing microbial detection techniques. SERS has been applied to different applications for detecting and classifying various pathogens ^{[20][21][22][23][24]}[27,28,29,30,31].

Wang et al. developed a surface-enhanced Raman scattering (SERS)-based LFA strip for the detection of such pathogens as Yersinia pestis, Francisella tularensis, and Bacillus anthracis. Target-specific SERS nanotags (Raman reporter-labeled gold nanoparticles) were utilized instead of gold nanoparticles. The method detected the pathogens in a short duration of 15 min using a minimum sample volume of 40 µL. The obtained detection limits for Y. pestis, F. tularensis, and B. anthracis were 43.4 CFU/mL, 45.8 CFU/mL, and 357 CFU/mL, respectively ^[25][32]. A high-quality silver nanorod (AgNR)-based SERS substrate was prepared to acquire the chemical fingerprint information of 22 strains of common pathogens. The method was able to identify and discriminate 20 strains of pathogens (diluted to 10⁷ CFU/mL) with high sensitivity within 30 min ^[26][33]. Another SERS-based biosensor was fabricated using gold nanorods (GNRs) complexed with oligonucleotide aptamers. The SERS tags were combined with antibody-modified magnetic nanoparticles for the simultaneous detection of Escherichia coli and Salmonella typhimurium. The developed SERS biosensor showed a good linear response of 10¹ to 10⁶ CFU/mL, high detection sensitivity (<8 CFU/mL) and a recovery rate of 95.26–107.88%. That study on combining aptamers and Raman reporters in SERS tags makes it possible to simultaneously detect different pathogens using a single biosensor ^[27][34].

Artificial intelligence has been applied widely in different diagnostics applications. Machine learning and neural networks are emerging techniques for data analysis and classification ^[28][35]. Spectroscopy data acquired from the SERS biosensors and techniques have been applied in machine learning and neural network algorithms. Ding et al. developed a method by combining SERS with a multiscale convolutional neural network (CNN). The label-free Raman substrate was prepared using gold nanoparticles. Different 1854 SERS spectra of three Salmonella serovars were measured and a multiscale CNN model was applied to extract SERS spectral features. The prepared gold nanoparticles and the developed CNN model showed detection accuracy higher than 97%. The given outcomes showed that the combination of SERS spectroscopy with multiscale CNN is feasible for Salmonella serotyping (S. enteritidis, S. typhimurium, and S. Paratyphi) with bacterial concentration of 10⁸ CFU/mL ^[29][36]. A stacked autoencoder-based deep neural networks algorithm was applied using SERS for the detection of methicillin-resistant Staphylococcus aureus and methicillin-sensitive S. aureus. The developed algorithm can evaluate features from the acquired signals and classify the data with an accuracy of 97.99%. The developed deep learning model classifies the pathogens with an area under the curve of 0.99 ^[30][37]. Ciloglu et al. combined SERS with machine learning techniques to classify Staphylococcus aureus and Legionella pneumophila. The technique gives higher classification accuracy of 97.8% by applying the k-nearest neighbors classifier ^[31][38].

Raman spectroscopy has been utilized extensively for microbiological diagnostics. A point-of-care testing technique has been developed using adhesive tape as a single platform for fast sampling, photocontrolled release, and SERS detection of pathogens from infected wounds. Pathogenic infections of P. aeruginosa and S. aureus were detected using gold nanostars on the adhesive tape as SERS substrate. The detection limit of the technique is 1.8 nM ^[32][39]. Duan et al. developed a SERS aptasensor for simultaneous detection of various pathogens using gold-decorated PDMS substrate. The fabricated film bound with the SERS probe to detect Vibrio parahaemolyticus and Salmonella typhimurium with a selectively detection limit of 18 CFU/mL and 27 CFU/mL, respectively ^[33][40]. The advancement of materials in SERS technology has increased accuracy and sensitivity for the detection of pathogens. The binding of SERS probes with fabricated chips and PDMS materials has enabled continuously miniaturization of detection prototypes ^[34][35][41,42].

Nakar et al. obtained spectra from pathogens E. coli, Klebsiella pneumoniae, and Klebsiella oxytoca isolates using UV-resonance Raman spectroscopy and single-cell Raman microspectroscopy. The obtained spectra were analyzed by machine learning algorithms for the classification of bacteria at the genus and species levels. The technique provides higher classification with 92% accuracy ^[36][43]. The method was further applied for the detection of clinical strains of E. coli ^[37][44]. Shen et al. created a fiber-probe-based method of Raman spectroscopy for the identification of six pathogens (S. epidermidis, S. aureus, E. faecalis, E. faecium, P. aeruginosa, and the yeast C. albicans). The collected signals from the fiber probe were analyzed using principal component analysis and linear discrimination models. The classification model acquired results with an accuracy of 93.8% ^[38][45]. The given studies were further extended and applied on agar plates to classify pathogenic infections ^[39][46].

SERS scattering has been incorporated with a microfluidic chip for the identification and discrimination of pathogens using tagged gold nanostars. The testing sample flowed continuously through the microfluidic channel, and the SERS signal was acquired corresponding to the SERS-tagged nanostars coated with antibody-binding protein. The system is capable of discriminating between L. monocytogenes and Listeria innocua with a concentration of 10⁵ CFU/mL. Analyzing the data for the detection of pathogens requires less than 2 min. However, overall sample preparation and system operation time requires 30 min. A schematic illustration of the on-chip detection of L. monocytogenes using SERS can be found in ^[40][47]. Bai et al. developed a sandwich immunoassay platform using functionalized SERS probes and magnetic beads for the simultaneous detection of E. coli and S. aureus. The technique uses two SERS probes for acquiring the signal following the immunomagnetic separation of the sample. The method identifies the pathogen with a detection limit of 10 and 25 CFU/mL for the simultaneous detection of E. coli and S. aureus, respectively ^[41][48]. Overall, SERS has emerged as a powerful analytical tool for rapidly detecting pathogens. Recent progress in the field of micro- to nanofabrication methodologies has enabled SERS applicable to various applications, such as rapid detection, point-of-care detection, and in situ detection. Currently, commercially available pathogen-detection techniques using SERS do not yet exist, but the improvement in SERS techniques has made it possible to develop handheld and portable prototypes to detect pathogens rapidly.