Enhance Signal Detection Research Articles

Integration of FAERS, DrugBank and SIDER Data for Machine Learning-based Detection of Adverse Drug Reactions

AbstractTraditionally, disproportionality analysis (DPA) methods are employed for signal detection in pharmacovigilance, but these methods utilize only a limited portion of the data available from spontaneous event reports (SERs). This research aims to enhance signal detection by applying machine learning (ML) methods that can leverage additional data. We create a dataset by integrating SER data from the FDA Adverse Event Reporting System (FAERS) with biological and chemical data from DrugBank, and information on known adverse drug reactions (ADRs) from Side Effect Resource (SIDER). The known ADRs from SIDER are used to label the dataset for ML training. Using the AutoML library TPOT, ML models are trained on this dataset. Our findings indicate that ML models, even when trained with the same features as DPA methods, achieve higher recall and precision. Moreover, incorporating additional features related to drugs and events significantly boosts the performance of ML models. Analysis using the explainable AI (XAI) technique SHAP reveals that the drug name, event name, and fifth-level ATC code are the most influential features for model predictions. These ML models offer a promising alternative or supplement to conventional DPA methods for signal detection in pharmacovigilance.

Characterization of a conductive hydrogel@Carbon fibers electrode as a novel intraneural interface

Peripheral neural interfaces facilitate bidirectional communication between the nervous system and external devices, enabling precise control for prosthetic limbs, sensory feedback systems, and therapeutic interventions in the field of Bioelectronic Medicine. Intraneural interfaces hold great promise since they ensure high selectivity in communicating only with the desired nerve fascicles. Despite significant advancements, challenges such as chronic immune response, signal degradation over time, and lack of long-term biocompatibility remain critical considerations in the development of such devices. Here we report on the development and benchtop characterization of a novel design of an intraneural interface based on carbon fiber bundles. Carbon fibers possess low impedance, enabling enhanced signal detection and stimulation efficacy compared to traditional metal electrodes. We provided a 3D-stabilizing structure for the carbon fiber bundles made of PEDOT:PSS hydrogel, to enhance the biocompatibility between the carbon fibers and the nervous tissue. We further coated the overall bundles with a thin layer of elastomeric material to provide electrical insulation. Taken together, our results demonstrated that our electrode possesses adequate structural and electrochemical properties to ensure proper stimulation and recording of peripheral nerve fibers and a biocompatible interface with the nervous tissue.

Characterization of Receptor Binding Affinity for Vascular Endothelial Growth Factor with Interferometric Imaging Sensor.

Wet Age-related macular degeneration (AMD) is the leading cause of vision loss in industrialized nations, often resulting in blindness. Biologics, therapeutic agents derived from biological sources, have been effective in AMD, albeit at a high cost. Due to the high cost of AMD treatment, it is critical to determine the binding affinity of biologics to ensure their efficacy and make quantitative comparisons between different drugs. This study evaluates the in vitro VEGF binding affinity of two drugs used for treating wet AMD, monoclonal antibody-based bevacizumab and fusion protein-based aflibercept, performing quantitative binding measurements on an Interferometric Reflectance Imaging Sensor (IRIS) system. Both biologics can inhibit Vascular Endothelial Growth Factor (VEGF). For comparison, the therapeutic molecules were immobilized on to the same support in a microarray format, and their real-time binding interactions with recombinant human VEGF (rhVEGF) were measured using an IRIS. The results indicated that aflibercept exhibited a higher binding affinity to VEGF than bevacizumab, consistent with previous studies using ELISA and SPR. The IRIS system's innovative and cost-effective features, such as silicon-based semiconductor chips for enhanced signal detection and multiplexed analysis capability, offer new prospects in sensor technologies. These attributes make IRISs a promising tool for future applications in the development of therapeutic agents, specifically biologics.

Wavelength-swept spontaneous Raman spectroscopy system improves fiber-based collection efficiency for whole brain tissue classification.

Raman spectroscopy is a valuable technique for tissue identification, but its conventional implementation is hindered by low efficiency due to scattering. Addressing this limitation, we are further developing the wavelength-swept Raman spectroscopy approach. We aim to enhance Raman signal detection by employing a laser capable of sweeping over a wide wavelength range to sequentially excite tissue with different wavelengths, paired with a photodetector featuring a fixed narrow-bandpass filter for collecting the Raman signal at a specific wavelength. We experimentally validate our technique using a fiber-based swept-source Raman spectroscopy setup. In addition, simulations are conducted to assess the efficacy of our approach in comparison with conventional spectrometer-based Raman spectroscopy. Our simulations reveal that the wavelength-swept configuration leads to a significantly stronger signal compared with conventional spectrometer-based Raman spectroscopy. Experimentally, our setup demonstrates an improvement of at least 200× in photon detection compared with the spectrometer-based setup. Furthermore, data acquired from different regions of a fixed monkey brain using our technique achieves 99% accuracy in classification via -nearest neighbor analysis. Our study showcases the potential of wavelength-swept Raman spectroscopy for tissue identification, particularly in highly scattering media, such as the brain. The developed technique offers enhanced signal detection capabilities, paving the way for future in vivo applications in tissue characterization.

Interface-constrained catalytic hairpin assembly permits highly sensitive SERS signaling of miRNA.

The rapid and precise monitoring of peripheral blood miRNA levels holds paramount importance for disease diagnosis and treatment monitoring. In this study, we propose an innovative research strategy that combines the catalytic hairpin assembly reaction with SERS signal congregation and enhancement. This combination can significantly enhance the stability of SERS detection, enabling stable and efficient detection of miRNA. Specifically, our paper-based SERS detection platform incorporates a streptavidin-modified substrate, biotin-labeled catalytic hairpin assembly reaction probes, 4-ATP, and primer-co-modified gold nanoparticles. In the presence of miRNA, the 4-ATP and primer-co-modified gold nanoparticles can specifically recognize the miRNA and interact with the biotin-labeled CHA probes to initiate an interfacial catalytic hairpin assembly reaction. This enzyme-free high-efficiency catalytic process can accumulate a large amount of biotin on the gold nanoparticles, which then bind to the streptavidin on the substrate with the assistance of the driving liquid, forming red gold nanoparticle stripes. These provide a multitude of hotspots for SERS, enabling enhanced signal detection. This innovative design achieves a low detection limit of 3.47 fM while maintaining excellent stability and repeatability. This conceptually innovative detection platform offers new technological possibilities and solutions for clinical miRNA detection.

Popcorn-like bimetallic palladium/platinum exhibiting enhanced peroxidase-like activity for signal enhancement in lateral flow immunoassay

BackgroundThe lateral flow immunoassay (LFIA) is widely employed as a point-of-care testing (POCT) technique. However, its limited sensitivity hinders its application in detecting biomarkers with low abundance. Recently, the utilization of nanozymes has been implemented to enhance the sensitivity of LFIA by catalyzing the oxidation of 3,3′,5,5′-tetramethylbenzidine (TMB). The catalytic performance of nanozymes plays a crucial role in influencing the sensitivity of LFIA. ResultsThe Cornus officinalis Sieb. et Zucc-Pd@Pt (CO–Pd@Pt) nanozyme with good peroxidase-like activity was synthesized herein through a facile one-pot method employing Cornus officinalis Sieb. et Zucc extract as a reducing agent. The morphology and composition of the CO–Pd@Pt nanozyme were characterized using TEM, SEM, XRD, and XPS. As a proof of concept, the as-synthesized CO–Pd@Pt nanozyme was utilized in LFIA (CO–Pd@Pt-LFIA) for the detection of human chorionic gonadotropin (hCG). Compared to conventional gold nanoparticles-based LFIA (AuNPs-LFIA), CO–Pd@Pt-LFIA demonstrated a significant enhancement in the limit of detection (LOD, 0.08 mIU/mL), which is approximately 160 times lower than that of AuNPs-LFIA. Furthermore, experiments evaluating accuracy, precision, selectivity, interference, and stability have confirmed the practical applicability of CO–Pd@Pt-LFIA for hCG content determination. SignificanceThe present study presents a novel approach for the synthesis of bimetallic nanozymes through environmentally friendly methods, utilizing plant extracts as both protective and reducing agents. Additionally, an easily implementable technique is proposed to enhance signal detection in lateral flow immunoassays.

Advancing mass spectrometry-based chemical imaging: A noncontact continuous flow surface probe in mass spectrometry for enhanced signal detection and spatial resolution

A new method has been developed for mass spectrometric imaging of small molecules and proteins on tissue or in thinly sliced materials. A laser desorption Venturi electrospray ionization-mass spectrometer was developed for molecular imaging. This method combines laser desorption (LD) and electrospray ionization (ESI) systems before a mass spectrometer (MS). To carry out laser desorption, samples are excited with a laser from the back side of a glass substrate. The desorbed molecules or particles are then captured by a solvent flow. In the ESI system, these desorbed particles and molecules are ionized. The spray part of the solvent system consists of two capillaries: one delivers solvent to the sample plate sides to capture desorbed molecules and particles, and the other carries the solution to the mass spectrometry side using the Venturi effect. A 2D stage facilitates sampling. The system is designed to minimize the sample size after desorption using a 355 nm diode laser, and it is optimized for molecules of various sizes, including organic molecules, amino acids, and proteins. Despite challenging atmospheric conditions for protein desorption, this specialized design enables the collection of protein spectra. The amino acids and other small molecules showed high sensitivity in the MSI measurements. This innovative MS imaging system can be directly applied to real tissue systems and other plant samples to visualize the molecular level distributions.

An evaluation research of constant false alarm rate algorithm for the application of underwater vehicles management system based on USBL sonar principles in shallow water

The paper presents an overview of underwater traffic management systems utilizing sonar principles. Based on the evaluation of the current situation both domestically and internationally, the research team has proposed the development of a underwater signal processing algorithm based on an adaptive threshold in the Ultra-short Baseline System (USBL). The theoretical results were implemented on FPGA hardware, not only enhancing signal detection capabilities but also meeting real-time signal processing requirements when deployed in complex underwater conditions in various shallow water regions of Vietnam. The research's future direction indicates substantial potential in the practical application of hydroacoustic signal processing algorithms when executing hardware implementations.

Advanced nanomaterials for imaging of eye diseases.

Vision impairment and blindness present significant global challenges, with common causes including age-related macular degeneration, diabetes, retinitis pigmentosa, and glaucoma. Advanced imaging tools, such as optical coherence tomography, fundus photography, photoacoustic microscopy, and fluorescence imaging, play a crucial role in improving therapeutic interventions and diagnostic methods. Contrast agents are often employed with these tools to enhance image clarity and signal detection. This review aims to explore the commonly used contrast agents in ocular disease imaging. The first section of the review delves into advanced ophthalmic imaging techniques, outlining their importance in addressing vision-related issues. The emphasis is on the efficacy of therapeutic interventions and diagnostic methods, establishing a foundation for the subsequent exploration of contrast agents. This review focuses on the role of contrast agents, with a specific emphasis on gold nanoparticles, particularly gold nanorods. The discussion highlights how these contrast agents optimize imaging in ocular disease diagnosis and monitoring, emphasizing their unique properties that enhance signal detection and imaging precision. The final section, we explores both organic and inorganic contrast agents and their applications in specific conditions such as choroidal neovascularization, retinal neovascularization, and stem cell tracking. The review concludes by addressing the limitations of current contrast agent usage and discussing potential future clinical applications. This comprehensive exploration contributes to advancing our understanding of contrast agents in ocular disease imaging and sets the stage for further research and development in the field.

Transforming Pharmacovigilance Using Gen AI: Innovations in Aggregate Reporting, Signal Detection, and Safety Surveillance

Pharmacovigilance, the science of monitoring and evaluating drug safety, plays a crucial role in ensuring patient well-being and public health. With the advent of artificial intelligence (AI), specifically Gen AI, pharmacovigilance has witnessed a transformative shift. Gen AI's advanced capabilities in real-time signal detection, automated reporting, and data integration have significantly enhanced the efficiency and accuracy of drug safety monitoring and surveillance. This article explores the role of Gen AI in pharmacovigilance, emphasizing its potential to revolutionize aggregate reporting, signal detection, risk assessment, and safety surveillance. It delves into the challenges and considerations that come with adopting AI in pharmacovigilance, such as ethical and regulatory implications, data privacy and security concerns, and the need for algorithm transparency and interpretability. The article also discusses the future directions and opportunities for Gen AI in pharmacovigilance which include enhanced signal detection algorithms, personalized safety assessments, and predictive risk modelling and incorporation of emerging technologies like blockchain and IoT that can complement Gen AI and improve data security and real-time monitoring. Collaborative efforts and data sharing among stakeholders are essential for maximizing Gen AI's potential in pharmacovigilance. Public-private partnerships and global pharmacovigilance networks can accelerate the adoption of AI technologies and drive innovation in drug safety monitoring. In conclusion, Gen AI presents a transformative opportunity for pharmacovigilance, promising safer medications and improved patient outcomes. Embracing responsible AI adoption, addressing ethical considerations, and encouraging further research are key to unlocking the full potential of Gen AI in advancing drug safety and public health.

Large-aperture wide-bandwidth densely populated coherent hydrophone array system for ocean acoustic monitoring

Underwater acoustic monitoring with a large-aperture coherent hydrophone array is advantageous because it enhances signal-to-noise ratio of received signals, provides estimates of signal bearing, and enhances signal detection ranges by one to two orders of magnitude over that of a single hydrophone. A large aperture coherent hydrophone array system comprising >160 elements has been developed inhouse at Northeastern University. The overall acoustic aperture length is 192 m with array elements nested into multiple uniformly spaced or log spaced subapertures. Hydrophones with integrated broadband pre-amplifiers designed with a linear frequency response from 10 Hz to 50 kHz send differential pair amplified and filtered analog signals to multiple 24-bit, 32-channel analog-to-digital converters with sampling rate that is programmable up to 100 kHz per channel. Array internals are designed using field replaceable pressure tolerant components verified by pressure chamber testing. Forward and aft modules are equipped with non-acoustic sensor elements to provide depth, heading, pitch, roll and temperature measurements. Acoustic aperture telemetry is user datagram protocol (UDP) converted to single-mode fiber for transmission along 600 m of faired tow cable to a shipboard data acquisition system. Examples of passive acoustic data from array deployment in the US Northeast coast are presented illustrating array capabilities.

Characteristic frequency detection of steady-state visual evoked potentials based on filter bank second-order underdamped tristable stochastic resonance

Brain computer interface (BCI) system based on steady-state visual evoked potentials (SSVEP) has great potential for communication and control applications. The actual intention of the user is determined by detecting the characteristic frequency of the SSVEP signals. However, the artifacts and spontaneous electroencephalogram (EEG) signals in SSVEP signals will degrade the performance of characteristic frequency detection. How to improve detection performance is a hot research topic. In this study, an innovative filter bank second-order underdamped tristable stochastic resonance (FBSUTSR) method is proposed and evaluated to enhance the characteristic frequency detection of SSVEP signals. FBSUTSR method uses noise to enhance signal detection and effectively utilizes the harmonic components contained in the signal. At the same time, the second-order underdamping characteristic makes the system form a unique filtering effect. In this study, the performance of SSVEP signals characteristic frequency detection with FBSUTSR, second-order underdamped tristable stochastic resonance (SUTSR), underdamped second-order stochastic resonance (USSR) and canonical coefficient analysis (CCA) is compared and analyzed using the public dataset of 40 types of stimulation targets of 35 subjects. When the stimulus data length is 5 s, FBSUTSR method achieves excellent performance with an average accuracy of 97.35±3.97% and an average information transmission rate (ITR) of 60.58±4.76bit·min-1, which is significantly higher than the performance of other three methods. This study verifies that the proposed FBSUTSR method has prominent performance advantages in implementing SSVEP signals characteristic frequency detection. The BCI system based on SSVEP using FBSUTSR method has great development potential in communication and control.

RoboFinch: A versatile audio‐visual synchronised robotic bird model for laboratory and field research on songbirds

Abstract Singing in birds is accompanied by beak, head and throat movements. The role of these visual cues has long been hypothesised to be an important facilitator in vocal communication, including social interactions and song acquisition, but has seen little experimental study. To address whether audio‐visual cues are relevant for birdsong we used high‐speed video recording, 3D scanning, 3D printing technology and colour‐realistic painting to create RoboFinch, an open source adult‐mimicking robot which matches temporal and chromatic properties of songbird vision. We exposed several groups of juvenile zebra finches during their song developmental phase to one of six singing robots that moved their beaks synchronised to their song and compared them with birds in a non‐synchronised and two control treatments. Juveniles in the synchronised treatment approached the robot setup from the start of the experiment and progressively increased the time they spent singing, contra to the other treatment groups. Interestingly, birds in the synchronised group seemed to actively listen during tutor song playback, as they sung less during the actual song playback compared to the birds in the asynchronous and audio‐only control treatments. Our open source RoboFinch setup thus provides an unprecedented tool for systematic study of the functionality and integration of audio‐visual cues associated with song behaviour. Realistic head and beak movements aligned to specific song elements may allow future studies to assess the importance of multisensory cues during song development, sexual signalling and social behaviour. All software and assembly instructions are open source, and the robot can be easily adapted to other species. Experimental manipulations of stimulus combinations and synchronisation can further elucidate how audio‐visual cues are integrated by receivers and how they may enhance signal detection, recognition, learning and memory.

Tuneable Cell-Laden Double-Emulsion Droplets for Enhanced Signal Detection.

Water-in-oil-in-water (w/o/w) or double-emulsion (DE) droplets have been widely used for cellular assays at a single-cell level because of their stability and biocompatibility. The oil shell of w/o/w droplets plays the role of a semipermeable membrane that allows substances with low molecular weight (e.g., water) to travel through but restricts those with high molecular weight (e.g., fluorescent biomarkers). Therefore, the core of DEs can be manipulated using osmosis, resulting in the shrinking or swelling of the core. Water leaves the inner aqueous phase to the outer phase via the oil shell when the osmotic pressure of the outer phase is higher than that in the inner phase, causing the shrinkage of DEs and vice versa. These processes can be achieved by transferring the DEs to hypertonic or hypotonic solutions. Manipulation of the core size of DEs can be beneficial to cellular assays. First, due to the selectivity of the oil shell of DEs, the concentration of biomarkers in the core increases when the inner aqueous phase is shrunk, resulting in the enhancement of biosignals. We demonstrate this by encapsulating the Bgl3 enzyme-secreting yeast with a substrate that displays fluorescence after hydrolyzation. In a second application, a single GFP-tagged yeast cell was encapsulated in DEs. After swelling the core of DEs, we observe that the larger core of DEs promotes cell growth compared to those with the smaller cores, leading to more intracellular proteins (green-fluorescent protein) for screening. These osmotic manipulations provide new tools for droplet-based biochemistry.

Enhanced Signal Detection and Constellation Design for Massive SIMO Communications With 1-Bit ADCs

In this paper, we investigate a transmitter and receiver design for a single-user massive SIMO (single-input multiple-output) system with 1-bit analog-to-digital converters (ADCs) at the base station (BS), where the user adopts higher-order modulation, e.g., 16-quadrature amplitude modulation (16-QAM), for the data transmission. For the channel estimation and the signal detection, linear least-squares (LS) estimation and maximum ratio combining (MRC) are respectively employed. In this context, we first introduce closed-form formulas for the mean of the estimated symbols and for the correlation matrix between their real and imaginary parts considering the effect of 1-bit quantization. The study of the distribution of the estimated symbols indicates that, in presence of 1-bit ADCs, the conventional 16-QAM detector and the typical square 16-QAM modulation are not adequate. In light of this, we propose three novel symbol detectors and re-design the 16-QAM modulation in order to improve the symbol error rate (SER). Furthermore, the upper bound on the SER is analyzed based on the pair-wise error probability and the boundary equation between two regions is also studied. Through numerical results, the proposed framework, i.e., the symbol detector and the transmit constellation design, shows a significant enhancement in the SER performance against the conventional detector and the typical square 16-QAM modulation.

Confirmation Bias as a Mechanism to Focus Attention Enhances Signal Detection

Confirmation Bias as a Mechanism to Focus Attention Enhances Signal Detection

Aromatase-Inhibitor-Induced Musculoskeletal Inflammation Is Observed Independent of Oophorectomy in a Novel Mouse Model.

Aromatase Inhibitors (AIs) block estrogen production and improve survival in patients with hormone-receptor-positive breast cancer. However, half of patients develop aromatase-inhibitor-induced arthralgia (AIIA), which is characterized by inflammation of the joints and the surrounding musculoskeletal tissue. To create a platform for future interventional strategies, our objective was to characterize a novel animal model of AIIA. Female BALB/C-Tg(NFκB-RE-luc)-Xen mice, which have a firefly luciferase NFκB reporter gene, were oophorectomized and treated with an AI (letrozole). Bioluminescent imaging showed significantly enhanced NFκB activation with AI treatment in the hind limbs. Moreover, an analysis of the knee joints and legs via MRI showed enhanced signal detection in the joint space and the surrounding tissue. Surprisingly, the responses observed with AI treatment were independent of oophorectomy, indicating that inflammation is not mediated by physiological estrogen levels. Histopathological and pro-inflammatory cytokine analyses further demonstrated the same trend, as tenosynovitis and musculoskeletal infiltrates were detected in all mice receiving AI, and serum cytokines were significantly upregulated. Human PBMCs treated with letrozole/estrogen combinations did not demonstrate an AI-specific gene expression pattern, suggesting AIIA-mediated pathogenesis through other cell types. Collectively, these data identify an AI-induced stimulation of disease pathology and suggest that AIIA pathogenesis may not be mediated by estrogen deficiency, as previously hypothesized.

Innovative Diagnostic Peptide-Based Technologies for Cancer Diagnosis: Focus on EGFR-Targeting Peptides.

Active targeting using biological ligands has emerged as a novel strategy for the targeted delivery of diagnostic agents to tumor cells. Conjugating functional targeting moieties with diagnostic probes can increase their accumulation in tumor cells and tissues, enhancing signal detection and, thus, the sensitivity of diagnosis. Due to their small size, ease of chemical synthesis and site-specific modification, high tissue penetration, low immunogenicity, rapid blood clearance, low cost, and biosafety, peptides offer several advantages over antibodies and proteins in diagnostic applications. Epidermal growth factor receptor (EGFR) is one of the most promising cancer biomarkers for actively targeting diagnostic and therapeutic agents to tumor cells due to its active involvement and overexpression in various cancers. Several peptides for EGFR-targeting have been identified in the last decades, which have been obtained by multiple means including derivation from natural proteins, phage display screening, positional scanning synthetic combinatorial library, and in silico screening. Many studies have used these peptides as a targeting moiety for diagnosing different cancers in vitro, in vivo, and in clinical trials. This review summarizes the progress of EGFR-targeting peptide-based assays in the molecular diagnosis of cancer.

Performance enhancement for differential energy signal detection of ambient backscatter communications

AbstractAmbient backscatter communications are a critical enabling technology of radio frequency identification in Internet of Things due to its full usage of ambient signals with low cost to transmit tag signals. However, the signal detection of tag signals is a challenging task as tag signals are usually superposed on the ambient radio frequency signals and the channel state information is unavailable due to its huge and unacceptable computation complexity at tags. To solve this, we in this article proposed an enhanced signal detection scheme for the differential energy detection in ambient backscatter communications. We discovered that the single detection threshold in classic detection will not provide desirable results in smart storage scenarios. We proposed a double‐threshold detection method which subtly integrates differential energy detection with maximum likelihood method to improve the detection accuracy while reducing the complexity of the reader. In addition, we derived the closed‐form expressions for bit error rate (BER), miss detection, and false alarm probabilities of the data bits sent by the tag. Simulation results are presented to show that our proposed detector can offer lower BER and computational complexity compared with single threshold detection schemes in low‐power and low‐storage scenarios, especially when the system is in a poor channel status environment.

Real-time validation of Surface-Enhanced Raman Scattering substrates via convolutional neural network algorithm

Development of next-generation sensor technologies could be advanced by exploiting surface enhanced Raman spectroscopy (SERS) substrates capable of considerably enhancing signal detection of analytes of interest down to single molecule levels. The widespread implementation of SERS, however, requires overcoming many of the existing and emerging challenges related to the enhancement inconsistency from the different parts of substrates, resulting in inconsistent readings. Here, a custom-designed convolutional neural networks (CNN) algorithm is developed to specifically analyse and validate substrates in real-time, providing a rapid output and identification of SERS active structures on each substrate enabling a consistently high signal enhancement. A computational algorithm has been developed to identify regions of high and consistent SERS activity from acquired optical microscopy images of SERS structures in real-time. The CNN model is tested on the inputted optical images and outputs the predicted consistent SERS-active structures or regions overlaid on top of the input image. The optimised CNN model used is a pre-trained neural network (VGG16) with the last layers fine-tuned. This model achieves an accuracy of 90% in classifying our uniquely developed electrohydrodynamically (EHD) fabricated, highly enhancing, SERS-active structures from optical images, enabling a real-time predictive tool for SERS substrates. The CNN output generates a heatmap, which provides feedback on the regions of high SERS activity, overlaid on the optical image to rapidly facilitate the SERS spectral acquisition measurements. Beyond its application for EHD, the versatility of the developed method renders it easily extendable for implementation with further, current and future, fabrication techniques of SERS substrates. The full script run yields an accuracy of 87.5 ± 12.0% for in-situ measurement acquisition, successfully demonstrating a proof-of-concept of rapidly identifying consistent SERS structures via the real-time CNN processing of optical images.