Converging Multidimensional Sensor and Machine Learning Toward High-Throughput and Biorecognition Element-Free Multidetermination of Extracellular Vesicle Biomarkers

Immunoassays for Extracellular Vesicle Detection via Transmembrane Proteins Using Surface Plasmon Resonance Biosensors

Extracellular vesicles (EVs) are preeminent carriers of biomarkers and have become the subject of intense biomedical research for medical diagnostics using biosensors. To create effective EV-based immunoassays, it is imperative to develop surface chemistry approaches with optimal EV detection targeting transmembrane protein biomarkers that are not affected by cell-to-cell variability. Here, we developed a series of immunoassays for the detection of EVs derived from mouse monocyte cells using surface plasmon resonance (SPR) biosensors. We chemically immobilized antibodies onto mixed self-assembled monolayers of oligo ethylene glycol (OEG) alkanethiolates with carboxylic and hydroxylic terminal groups. The effects of antibody clonality (monoclonal vs polyclonal) and antibody surface coverage in targeting EVs via CD81 tetraspanins were investigated. We determined binding kinetic parameters, establishing trends from steric hindrance effects and epitope recognition properties of antibodies. Our results indicate that a 40% surface coverage of polyclonal antibodies covalently linked onto a mixed SAM with 10% of terminated -COOH groups yields a promising approach for EV detection with a linear range of 1.9 × 108-1.9 × 109 EVs/mL and a limit of detection of 5.9 × 106 EVs/mL. This optimal immunoassay exhibits a 1.92 nM equilibrium dissociation constant for bound EVs, suggesting a high binding affinity when CD81 is targeted. Our study provides important insights into surface chemistry development for EV detection targeted via transmembrane protein biomarkers using antibodies, which has promising applications for disease diagnostics.

Artificial Intelligence in Point-of-Care Biosensing: Challenges and Opportunities

The integration of artificial intelligence (AI) into point-of-care (POC) biosensing has the potential to revolutionize diagnostic methodologies by offering rapid, accurate, and accessible health assessment directly at the patient level. This review paper explores the transformative impact of AI technologies on POC biosensing, emphasizing recent computational advancements, ongoing challenges, and future prospects in the field. We provide an overview of core biosensing technologies and their use at the POC, highlighting ongoing issues and challenges that may be solved with AI. We follow with an overview of AI methodologies that can be applied to biosensing, including machine learning algorithms, neural networks, and data processing frameworks that facilitate real-time analytical decision-making. We explore the applications of AI at each stage of the biosensor development process, highlighting the diverse opportunities beyond simple data analysis procedures. We include a thorough analysis of outstanding challenges in the field of AI-assisted biosensing, focusing on the technical and ethical challenges regarding the widespread adoption of these technologies, such as data security, algorithmic bias, and regulatory compliance. Through this review, we aim to emphasize the role of AI in advancing POC biosensing and inform researchers, clinicians, and policymakers about the potential of these technologies in reshaping global healthcare landscapes.

Algorithms for predicting COVID outcome using ready-to-use laboratorial and clinical data

The pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is an emerging crisis affecting the public health system. The clinical features of COVID-19 can range from an asymptomatic state to acute respiratory syndrome and multiple organ dysfunction. Although some hematological and biochemical parameters are altered during moderate and severe COVID-19, there is still a lack of tools to combine these parameters to predict the clinical outcome of a patient with COVID-19. Thus, this study aimed at employing hematological and biochemical parameters of patients diagnosed with COVID-19 in order to build machine learning algorithms for predicting COVID mortality or survival. Patients included in the study had a diagnosis of SARS-CoV-2 infection confirmed by RT-PCR and biochemical and hematological measurements were performed in three different time points upon hospital admission. Among the parameters evaluated, the ones that stand out the most are the important features of the T1 time point (urea, lymphocytes, glucose, basophils and age), which could be possible biomarkers for the severity of COVID-19 patients. This study shows that urea is the parameter that best classifies patient severity and rises over time, making it a crucial analyte to be used in machine learning algorithms to predict patient outcome. In this study optimal and medically interpretable machine learning algorithms for outcome prediction are presented for each time point. It was found that urea is the most paramount variable for outcome prediction over all three time points. However, the order of importance of other variables changes for each time point, demonstrating the importance of a dynamic approach for an effective patient's outcome prediction. All in all, the use of machine learning algorithms can be a defining tool for laboratory monitoring and clinical outcome prediction, which may bring benefits to public health in future pandemics with newly emerging and reemerging SARS-CoV-2 variants of concern.

Single-Response Duplexing of Electrochemical Label-Free Biosensor from the Same Tag

Multiplexing is a valuable strategy to boost throughput and improve clinical accuracy. Exploiting the vertical, meshed design of reproducible and low-cost ultra-dense electrochemical chips, the unprecedented single-response multiplexing of typical label-free biosensors is reported. Using a cheap, handheld one-channel workstation and a single redox probe, that is, ferro/ferricyanide, the recognition events taking place on two spatially resolved locations of the same working electrode can be tracked along a single voltammetry scan by collecting the electrochemical signatures of the probe in relation to different quasi-reference electrodes, Au (0 V) and Ag/AgCl ink (+0.2 V). This spatial isolation prevents crosstalk between the redox tags and interferences over functionalization and binding steps, representing an advantage over the existing non-spatially resolved single-response multiplex strategies. As proof of concept, peptide-tethered immunosensors are demonstrated to provide the duplex detection of COVID-19 antibodies, thereby doubling the throughput while achieving 100% accuracy in serum samples. The approach is envisioned to enable broad applications in high-throughput and multi-analyte platforms, as it can be tailored to other biosensing devices and formats.

High-throughput microfluidic systems accelerated by artificial intelligence for biomedical applications

High-throughput microfluidic systems are widely used in biomedical fields for tasks like disease detection, drug testing, and material discovery. Despite the great advances in automation and throughput, the large amounts of data generated by the high-throughput microfluidic systems generally outpace the abilities of manual analysis. Recently, the convergence of microfluidic systems and artificial intelligence (AI) has been promising in solving the issue by significantly accelerating the process of data analysis as well as improving the capability of intelligent decision. This review offers a comprehensive introduction on AI methods and outlines the current advances of high-throughput microfluidic systems accelerated by AI, covering biomedical detection, drug screening, and automated system control and design. Furthermore, the challenges and opportunities in this field are critically discussed as well.

Non-contact microfluidic analysis of the stiffness of single large extracellular vesicles from IDH1-mutated glioblastoma cells

In preparation for leveraging extracellular vesicles (EVs) for disease diagnostics and therapeutics, fundamental research is being done to understand EV biological, chemical, and physical properties. Most published studies have investigated nanoscale EVs and focused on EV biochemical content. There is much less understanding of large microscale EV characteristics and EV mechanical properties. We recently introduced a non-contact microfluidic technique that measures the stiffness of large EVs (>1 μm diameter). This pilot study probes the robustness of the microfluidic technique to distinguish between EV populations by comparing stiffness distributions of large EVs derived from glioblastoma cell lines. EVs derived from cells expressing the IDH1 mutation, a common glioblastoma mutation known to disrupt lipid metabolism, were stiffer than those expressed from wild-type cells in a statistical comparison of sample medians. A supporting lipidomics analysis showed that the IDH1 mutation increased the amount of saturated lipids in EVs. Taken together, these data encourage further investigation into the potential of high-throughput microfluidics to distinguish between large EV populations that differ in biomolecular composition. These findings contribute to the understanding of EV biomechanics, in particular for the less studied microscale EVs.

Single-Response Electronic Tongue and Machine Learning Enable the Multidetermination of Extracellular Vesicle Biomarkers for Cancer Diagnostics Without Recognition Elements

Platforms based on impedimetric electronic tongue (nonselective sensor) and machine learning are promising to bring disease screening biosensors into mainstream use toward straightforward, fast, and accurate analyses at the point-of-care, thus contributing to rationalize and decentralize laboratory tests with social and economic impacts being achieved. By combining a low-cost and scalable electronic tongue with machine learning, in this chapter, we describe the simultaneous determination of two extracellular vesicle (EV) biomarkers, i.e., the concentrations of EV and carried proteins, in mice blood with Ehrlich tumor from a single impedance spectrum without using biorecognizing elements. This tumor shows primary features of mammary tumor cells. Pencil HB core electrodes are integrated into polydimethylsiloxane (PDMS) microfluidic chip. The platform shows the highest throughput in comparison with the methods addressed in the literature to determine EV biomarkers.

Modular Label-Free Electrochemical Biosensor Loading Nature-Inspired Peptide toward the Widespread Use of COVID-19 Antibody Tests

Limitations of the recognition elements in terms of synthesis, cost, availability, and stability have impaired the translation of biosensors into practical use. Inspired by nature to mimic the molecular recognition of the anti-SARS-CoV-2 S protein antibody (AbS) by the S protein binding site, we synthesized the peptide sequence of Asn-Asn-Ala-Thr-Asn-COOH (abbreviated as PEP2003) to create COVID-19 screening label-free (LF) biosensors based on a carbon electrode, gold nanoparticles (AuNPs), and electrochemical impedance spectroscopy. The PEP2003 is easily obtained by chemical synthesis, and it can be adsorbed on electrodes while maintaining its ability for AbS recognition, further leading to a sensitivity 3.4-fold higher than the full-length S protein, which is in agreement with the increase in the target-to-receptor size ratio. Peptide-loaded LF devices based on noncovalent immobilization were developed by affording fast and simple analyses, along with a modular functionalization. From studies by molecular docking, the peptide-AbS binding was found to be driven by hydrogen bonds and hydrophobic interactions. Moreover, the peptide is not amenable to denaturation, thus addressing the trade-off between scalability, cost, and robustness. The biosensor preserves 95.1% of the initial signal for 20 days when stored dry at 4 °C. With the aid of two simple equations fitted by machine learning (ML), the method was able to make the COVID-19 screening of 39 biological samples into healthy and infected groups with 100.0% accuracy. By taking advantage of peptide-related merits combined with advances in surface chemistry and ML-aided accuracy, this platform is promising to bring COVID-19 biosensors into mainstream use toward straightforward, fast, and accurate analyses at the point of care, with social and economic impacts being achieved.

Microfluidic Approaches for Affinity-Based Exosome Separation

As a subspecies of extracellular vesicles (EVs), exosomes have provided promising results in diagnostic and theranostic applications in recent years. The nanometer-sized exosomes can be extracted by liquid biopsy from almost all body fluids, making them especially suitable for mainly non-invasive point-of-care (POC) applications. To achieve this, exosomes must first be separated from the respective biofluid. Impurities with similar properties, heterogeneity of exosome characteristics, and time-related biofouling complicate the separation. This practical review presents the state-of-the-art methods available for the separation of exosomes. Furthermore, it is shown how new separation methods can be developed. A particular focus lies on the fabrication and design of microfluidic devices using highly selective affinity separation. Due to their compactness, quick analysis time and portable form factor, these microfluidic devices are particularly suitable to deliver fast and reliable results for POC applications. For these devices, new manufacturing methods (e.g., laminating, replica molding and 3D printing) that use low-cost materials and do not require clean rooms are presented. Additionally, special flow routes and patterns that increase contact surfaces, as well as residence time, and thus improve affinity purification are displayed. Finally, various analyses are shown that can be used to evaluate the separation results of a newly developed device. Overall, this review paper provides a toolbox for developing new microfluidic affinity devices for exosome separation.

Real-Time and In Situ Monitoring of the Synthesis of Silica Nanoparticles

The real-time and in situ monitoring of the synthesis of nanomaterials (NMs) remains a challenging task, which is of pivotal importance by assisting fundamental studies (e.g., synthesis kinetics and colloidal phenomena) and providing optimized quality control. In fact, the lack of reproducibility in the synthesis of NMs is a bottleneck against the translation of nanotechnologies into the market toward daily practice. Here, we address an impedimetric millifluidic sensor with data processing by machine learning (ML) as a sensing platform to monitor silica nanoparticles (SiO₂NPs) over a 24 h synthesis from a single measurement. The SiO₂NPs were selected as a model NM because of their extensive applications. Impressively, simple ML-fitted descriptors were capable of overcoming interferences derived from SiO₂NP adsorption over the signals of polarizable Au interdigitate electrodes to assure the determination of the size and concentration of nanoparticles over synthesis while meeting the trade-off between accuracy and speed/simplicity of computation. The root-mean-square errors were calculated as ∼2.0 nm (size) and 2.6 × 10¹⁰ nanoparticles mL^-1 (concentration). Further, the robustness of the ML size descriptor was successfully challenged in data obtained along independent syntheses using different devices, with the global average accuracy being 103.7 ± 1.9%. Our work advances the developments required to transform a closed flow system basically encompassing the reactional flask and an impedimetric sensor into a scalable and user-friendly platform to assess the in situ synthesis of SiO₂NPs. Since the sensor presents a universal response principle, the method is expected to enable the monitoring of other NMs. Such a platform may help to pave the way for translating "sense-act" systems into practice use in nanotechnology.

Biocompatible Wearable Electrodes on Leaves toward the On-Site Monitoring of Water Loss from Plants

Impedimetric wearable sensors are a promising strategy for determining the loss of water content (LWC) from leaves because they can afford on-site and nondestructive quantification of cellular water from a single measurement. Because the water content is a key marker of leaf health, monitoring of the LWC can lend key insights into daily practice in precision agriculture, toxicity studies, and the development of agricultural inputs. Ongoing challenges with this monitoring are the on-leaf adhesion, compatibility, scalability, and reproducibility of the electrodes, especially when subjected to long-term measurements. This paper introduces a set of sensing material, technological, and data processing solutions that overwhelm such obstacles. Mass-production-suitable electrodes consisting of stand-alone Ni films obtained by well-established microfabrication methods or ecofriendly pyrolyzed paper enabled reproducible determination of the LWC from soy leaves with optimized sensibilities of 27.0 (Ni) and 17.5 kΩ %-1 (paper). The freestanding design of the Ni electrodes was further key to delivering high on-leaf adhesion and long-term compatibility. Their impedances remained unchanged under the action of wind at velocities of up to 2.00 m s-1, whereas X-ray nanoprobe fluorescence assays allowed us to confirm the Ni sensor compatibility by the monitoring of the soy leaf health in an electrode-exposed area. Both electrodes operated through direct transfer of the conductive materials on hairy soy leaves using an ordinary adhesive tape. We used a hand-held and low-power potentiostat with wireless connection to a smartphone to determine the LWC over 24 h. Impressively, a machine-learning model was able to convert the sensing responses into a simple mathematical equation that gauged the impairments on the water content at two temperatures (30 and 20 °C) with reduced root-mean-square errors (0.1% up to 0.3%). These data suggest broad applicability of the platform by enabling direct determination of the LWC from leaves even at variable temperatures. Overall, our findings may help to pave the way for translating "sense-act" technologies into practice toward the on-site and remote investigation of plant drought stress. These platforms can provide key information for aiding efficient data-driven management and guiding decision-making steps.

Ratiometric 3D DNA Machine Combined with Machine Learning Algorithm for Ultrasensitive and High-Precision Screening of Early Urinary Diseases

Urinary extracellular vesicles (uEVs) have received considerable attention as a potential biomarker source for the diagnosis of urinary diseases. Present studies mainly focus on the discovery of biomarkers based on high-throughput proteomics, while limited efforts have been paid to applying the uEVs' biomarkers for the diagnosis of early urinary disease. Herein, we demonstrate a diagnosis protocol to realize ultrasensitive detection of uEVs and accurate classification of early urinary diseases, by combing a ratiometric three-dimensional (3D) DNA machine with machine learning (ML). The ratiometric 3D DNA machine platform is constructed by conjugating a padlock probe (PLP) containing cytosine-rich (C-rich) sequences, anchor strands, and nucleic-acid-stabilized silver nanoclusters (DNA_AgNCs) onto the magnetic nanoparticles (MNPs). The competitive binding of uEVs with the aptamer releases the walker strand, thus the ratiometric 3D DNA machine was activated to undergo an accurate amplification reaction and produce a ratiometric fluorescence signal. The present biosensor offers a detection limit of 9.9 × 10³ particles mL^-1 with a linear range of 10⁴-10⁸ particles mL^-1 for uEVs. Two ML algorithms, K-nearest neighbor (KNN) and support vector machine (SVM), were subsequently applied for analyzing the correlation between the sensing signals of uEV multibiomarkers and the clinical state. The disease diagnostic accuracy of optimal biomarker combination reaches 95% and 100% by analyzing with KNN and SVM, and the disease type classification exhibits an accuracy of 94.7% and 89.5%, respectively. Moreover, the protocol results in 100% accurate visual identification of clinical uEV samples from individuals with disease or as healthy control by a t-distributed stochastic neighbor embedding (tSNE) algorithm.

Electrically driven microfluidic platforms for exosome manipulation and characterization

Exosomes are small extracellular vesicles that can be obtained from several body fluids such as blood and urine. Since these vesicles can carry biomarkers and other cargo, they have application in healthcare diagnostics and therapeutics, such as liquid biopsies and drug delivery. Yet, their identification and separation from a sample remain challenging due to their high degree of heterogeneity and their co-existence with other bioparticles. In this contribution, we review the state-of-the-art on electrical techniques and methods to displace, selectively trap/isolate, and detect/characterize exosomes in microfluidic devices. Although there are many reviews focused on exosome separation using benchtop equipment, such as ultracentrifugation, there are limited reviews focusing on the use of electrical phenomena in microfluidic devices for exosome manipulation and detection. Here, we highlight contributions published during the past decade and present perspectives for this research field for the near future, outlining challenges to address in years to come.

Advanced Nanotechnologies for Extracellular Vesicle-Based Liquid Biopsy

Extracellular vesicles (EVs) are emerging as a new source of biomarkers in liquid biopsy because of their wide presence in most body fluids and their ability to load cargoes from disease-related cells. Owing to the crucial role of EVs in disease diagnosis and treatment, significant efforts have been made to isolate, detect, and analyze EVs with high efficiency. A recent overview of advanced EV detection nanotechnologies is discussed here. First, several key challenges in EV-based liquid biopsies are introduced. Then, the related pivotal advances in nanotechnologies for EV isolation based on physical features, chemical affinity, and the combination of nanostructures and chemical affinity are summarized. Next, a summary of high-sensitivity sensors for EV detection and advanced approaches for single EV detection are provided. Later, EV analysis is introduced in practical clinical scenarios, and the application of machine learning in this field is highlighted. Finally, future opportunities for the development of next-generation nanotechnologies for EV detection are presented.

Bifunctional Metal Meshes Acting as a Semipermeable Membrane and Electrode for Sensitive Electrochemical Determination of Volatile Compounds

The monitoring of toxic inorganic gases and volatile organic compounds has brought the development of field-deployable, sensitive, and scalable sensors into focus. Here, we attempted to meet these requirements by using concurrently microhole-structured meshes as (i) a membrane for the gas diffusion extraction of an analyte from a donor sample and (ii) an electrode for the sensitive electrochemical determination of this target with the receptor electrolyte at rest. We used two types of meshes with complementary benefits, i.e., Ni mesh fabricated by robust, scalable, and well-established methods for manufacturing specific designs and stainless steel wire mesh (SSWM), which is commercially available at a low cost. The diffusion of gas (from a donor) was conducted in headspace mode, thus minimizing issues related to mesh fouling. When compared with the conventional polytetrafluoroethylene (PTFE) membrane, both the meshes (40 μm hole diameter) led to a higher amount of vapor collected into the electrolyte for subsequent detection. This inedited fashion produced a kind of reverse diffusion of the analyte dissolved into the electrolyte (receptor), i.e., from the electrode to bulk, which further enabled highly sensitive analyses. Using Ni mesh coated with Ni(OH)₂ nanoparticles, the limit of detection reached for ethanol was 24-fold lower than the data attained by a platform with a PTFE membrane and placement of the electrode into electrolyte bulk. This system was applied in the determination of ethanol in complex samples related to the production of ethanol biofuel. It is noteworthy that a simple equation fitted by machine learning was able to provide accurate assays (accuracies from 97 to 102%) by overcoming matrix effect-related interferences on detection performance. Furthermore, preliminary measurements demonstrated the successful coating of the meshes with gold films as an alternative raw electrode material and the monitoring of HCl utilizing Au-coated SSWMs. These strategies extend the applicability of the platform that may help to develop valuable volatile sensing solutions.

Development and Application of Mobile Apps for Molecular Sensing: A Review

Modern smartphone-based sensing devices are generally standalone detection platforms that can transduce signals (via the built-in USB port, audio jack, or camera), perform analysis through mobile applications (apps), and display results on the screen/user interface. The advancement toward this ultimate form of on-site chemical analysis and point-of-care diagnosis is tied closely with the evolution of mobile technology. Previous reviews in the field mainly focused on the physical platforms while overlooking the role of mobile apps in such devices. There exist three general stages throughout the development: (1) early generation telemedicine, (2) mobile phone-assisted clinical diagnosis (without apps), and (3) mobile app-based sensing devices for various analytes. This review presents the key breakthroughs during each stage, recent development, remaining challenges, and future perspectives of the field. Representative examples, spanning from the pioneering point-of-care testing to the latest devices with integrated mobile apps, are classified by their sensing mechanisms. The review also discusses the scarcity of open-source apps dedicated to molecular sensing. With the introduction of more open-source and commercial apps, the mobile app-based detection system is anticipated to dominate point-of-care diagnosis and on-site molecular sensing in our opinion.

Big data and machine learning for materials science

Herein, we review aspects of leading-edge research and innovation in materials science that exploit big data and machine learning (ML), two computer science concepts that combine to yield computational intelligence. ML can accelerate the solution of intricate chemical problems and even solve problems that otherwise would not be tractable. However, the potential benefits of ML come at the cost of big data production; that is, the algorithms demand large volumes of data of various natures and from different sources, from material properties to sensor data. In the survey, we propose a roadmap for future developments with emphasis on computer-aided discovery of new materials and analysis of chemical sensing compounds, both prominent research fields for ML in the context of materials science. In addition to providing an overview of recent advances, we elaborate upon the conceptual and practical limitations of big data and ML applied to materials science, outlining processes, discussing pitfalls, and reviewing cases of success and failure.